[ { "id": "L4_T1_Insight_Synthesis_S3_00009", "level": 4, "level_name": "Insight Synthesis", "category": "Insight Synthesis", "subtask": "Insight Synthesis", "question": "Analyze the behavior of channel 441 for the period 2020-11.\nPlease use ONLY the following phrases for trend description: rapid rise, gradual rise, rapid fall, gradual fall, steady stable, fluctuating stable.\nProvide a structured report covering:\n1. Trend Segmentation: Describe each stage with precise start/end timestamps (HH:MM:SS) using the phrases above.\n2. Outlier Audit: Identify only significant outliers that deviate sharply from the local trend. Ignore minor background noise.\n(Output format: structured natural language report.)", "answer": "1. Trend Segmentation: from 2020-11-01 00:00:00 to 2020-11-03 20:30:00, the trend showed a steady stable; from 2020-11-03 20:30:00 to 2020-11-12 10:15:00, the trend showed a rapid fall; from 2020-11-12 10:15:00 to 2020-11-18 03:30:00, the trend showed a fluctuating stable; from 2020-11-18 03:30:00 to 2020-11-21 00:00:00, the trend showed a rapid rise; from 2020-11-21 00:00:00 to 2020-11-26 17:15:00, the trend showed a steady stable; from 2020-11-26 17:15:00 to 2020-11-30 23:45:00, the trend showed a rapid fall.\n2. Outlier Audit: A significant spike was detected at 2020-11-26 21:30:00 (value: 90.23).", "ground_truth": { "trend_segments": [ "from 2020-11-01 00:00:00 to 2020-11-03 20:30:00, the trend showed a steady stable", "from 2020-11-03 20:30:00 to 2020-11-12 10:15:00, the trend showed a rapid fall", "from 2020-11-12 10:15:00 to 2020-11-18 03:30:00, the trend showed a fluctuating stable", "from 2020-11-18 03:30:00 to 2020-11-21 00:00:00, the trend showed a rapid rise", "from 2020-11-21 00:00:00 to 2020-11-26 17:15:00, the trend showed a steady stable", "from 2020-11-26 17:15:00 to 2020-11-30 23:45:00, the trend showed a rapid fall" ], "segments_meta": [ { "adj": "steady", "kind": "stable", "start_idx": 0, "end_idx": 274 }, { "adj": "rapid", "kind": "fall", "start_idx": 274, "end_idx": 1097 }, { "adj": "fluctuating", "kind": "stable", "start_idx": 1097, "end_idx": 1646 }, { "adj": "rapid", "kind": "rise", "start_idx": 1646, "end_idx": 1920 }, { "adj": "steady", "kind": "stable", "start_idx": 1920, "end_idx": 2469 }, { "adj": "rapid", "kind": "fall", "start_idx": 2469, "end_idx": 2880 } ], "significant_anomaly": { "timestamp": "2020-11-26 21:30:00", "kind": "significant spike", "value": 90.2302588163616 } }, "eval_metric": "report", "channel": "441", "ts_data_path": "ts_data/L4_T1_Insight_Synthesis_S3_00009.csv" }, { "id": "L1_T2_Temporal_Localization_S1_00009", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Temporal Localization", "question": "At what exact timestamp did channel 155 last fall below 0.031 in 2019-08? (Output format: a specific timestamp, e.g., YYYY-MM-DD HH:MM:SS)", "answer": "2019-08-31 02:15:00", "ground_truth": "2019-08-31 02:15:00", "eval_metric": "hit", "channel": "155", "ts_data_path": "ts_data/L1_T2_Temporal_Localization_S1_00009.csv" }, { "id": "L2_T1_Shape_Identification_S2_00009", "level": 2, "level_name": "Pattern Recognition", "category": "Shape Identification", "subtask": "Shape Identification", "question": "Identify the time range of the deepest deep valley in channel 173 within [2020-01-19 10:00:00 to 2020-01-25 00:30:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2020-01-23 08:15:00', '2020-01-24 04:00:00']", "ground_truth": [ "2020-01-23 08:15:00", "2020-01-24 04:00:00" ], "eval_metric": "iou", "channel": "173", "ts_data_path": "ts_data/L2_T1_Shape_Identification_S2_00009.csv" }, { "id": "L3_T2_Contextual_Anomaly_S2_00003", "level": 3, "level_name": "Semantic Reasoning", "category": "Contextual Anomaly", "subtask": "Contextual Anomaly", "question": "Identify the period in channel 501 during 2020 that experienced the most significant extreme surge in flow. (Output format: [YYYY-MM-DD, YYYY-MM-DD])", "answer": "[2020-06-24 00:15:00, 2020-07-25 00:00:00]", "ground_truth": [ "2020-06-24 00:15:00", "2020-07-25 00:00:00" ], "eval_metric": "iou", "channel": "501", "ts_data_path": "ts_data/L3_T2_Contextual_Anomaly_S2_00003.csv" }, { "id": "L3_T3_Causal_Anomaly_S0_00003", "level": 3, "level_name": "Semantic Reasoning", "category": "Causal Anomaly", "subtask": "Causal Anomaly", "question": "Given that channel 430_up_2021 is the upstream source of channel 430_down_2021, identify the time period in 2021 where 430_down_2021 shows a significant causal anomaly, such as an flat line during high activity. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2021-05-24 13:45:00, 2021-05-26 09:00:00]", "ground_truth": [ "2021-05-24 13:45:00", "2021-05-26 09:00:00" ], "eval_metric": "iou", "channel": "430_down_2021", "ts_data_path": "ts_data/L3_T3_Causal_Anomaly_S0_00003.csv" }, { "id": "L3_T1_Composite_Trend_S0_00005", "level": 3, "level_name": "Semantic Reasoning", "category": "Composite Trend", "subtask": "Composite Trend", "question": "Among days in channel 441 during 2020 that exhibit the trend pattern 'slow fall, then fluctuating stable, then steady stable, then slow fall', identify the top-3 days where the fluctuating stable segment is the most fluctuating. (Output format: a ranked list of dates, e.g., ['YYYY-MM-DD', ...])", "answer": "['2020-12-16', '2020-07-10', '2020-08-16']", "ground_truth": [ "2020-12-16", "2020-07-10", "2020-08-16" ], "eval_metric": "set_f1", "channel": "441", "ts_data_path": "ts_data/L3_T1_Composite_Trend_S0_00005.csv" }, { "id": "L3_T1_Composite_Trend_S3_00002", "level": 3, "level_name": "Semantic Reasoning", "category": "Composite Trend", "subtask": "Composite Trend", "question": "Among days in channel 71 during 2020 that exhibit the trend pattern 'rapid fall, then slow rise, then rise, then rapid rise', identify the top-6 days where the rapid rise segment is the fastest. (Output format: a ranked list of dates, e.g., ['YYYY-MM-DD', ...])", "answer": "['2020-03-22', '2020-07-08', '2020-01-19', '2020-06-26', '2020-10-08', '2020-07-30']", "ground_truth": [ "2020-03-22", "2020-07-08", "2020-01-19", "2020-06-26", "2020-10-08", "2020-07-30" ], "eval_metric": "set_f1", "channel": "71", "ts_data_path": "ts_data/L3_T1_Composite_Trend_S3_00002.csv" }, { "id": "L3_T3_Causal_Anomaly_S1_00008", "level": 3, "level_name": "Semantic Reasoning", "category": "Causal Anomaly", "subtask": "Causal Anomaly", "question": "Given that channel 154_up_2020 is the upstream source of channel 154_down_2020, identify the time period in 2020 where 154_down_2020 shows a significant causal anomaly, such as an flat line during high activity. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2020-07-26 08:00:00, 2020-07-29 15:00:00]", "ground_truth": [ "2020-07-26 08:00:00", "2020-07-29 15:00:00" ], "eval_metric": "iou", "channel": "154_down_2020", "ts_data_path": "ts_data/L3_T3_Causal_Anomaly_S1_00008.csv" }, { "id": "L2_T3_Subsequence_Matching_S2_00002", "level": 2, "level_name": "Pattern Recognition", "category": "Subsequence Matching", "subtask": "Subsequence Matching", "question": "Analyze the reference pattern in [2022-06-11 23:00:00 to 2022-06-12 22:45:00]. Find the time interval where channel 591 exhibits the most similar pattern within the search context [2022-06-15 17:30:00 to 2022-06-25 20:15:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2022-06-16 12:15:00', '2022-06-17 12:00:00']", "ground_truth": [ "2022-06-16 12:15:00", "2022-06-17 12:00:00" ], "eval_metric": "iou", "channel": "591", "ts_data_path": "ts_data/L2_T3_Subsequence_Matching_S2_00002.csv" }, { "id": "L1_T3_Interval_Discovery_S3_00006", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Interval Discovery", "question": "Find the longest period where channel 762 remained above 0.292 in 2020-01-14 to 2020-02-18. (Output format: a time interval, e.g., [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2020-02-10 22:15:00, 2020-02-13 14:00:00]", "ground_truth": [ "2020-02-10 22:15:00", "2020-02-13 14:00:00" ], "eval_metric": "iou", "channel": "762", "ts_data_path": "ts_data/L1_T3_Interval_Discovery_S3_00006.csv" }, { "id": "L2_T1_Shape_Identification_S4_00004", "level": 2, "level_name": "Pattern Recognition", "category": "Shape Identification", "subtask": "Shape Identification", "question": "Identify the time range of the largest step ascent in channel 727 within [2020-11-28 12:45:00 to 2020-12-10 01:00:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2020-12-06 13:15:00', '2020-12-08 06:15:00']", "ground_truth": [ "2020-12-06 13:15:00", "2020-12-08 06:15:00" ], "eval_metric": "iou", "channel": "727", "ts_data_path": "ts_data/L2_T1_Shape_Identification_S4_00004.csv" }, { "id": "L1_T3_Interval_Discovery_S0_00008", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Interval Discovery", "question": "Find the longest period where channel LULL remained above 1.036 in 2016-12. (Output format: a time interval, e.g., [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2016-12-29 23:15:00, 2016-12-30 12:00:00]", "ground_truth": [ "2016-12-29 23:15:00", "2016-12-30 12:00:00" ], "eval_metric": "iou", "channel": "LULL", "ts_data_path": "ts_data/L1_T3_Interval_Discovery_S0_00008.csv" }, { "id": "L3_T1_Composite_Trend_S3_00001", "level": 3, "level_name": "Semantic Reasoning", "category": "Composite Trend", "subtask": "Composite Trend", "question": "Among days in channel 67 during 2022 that exhibit the trend pattern 'rise, then steady stable, then rapid rise, then steady stable', identify the top-6 days where the rapid rise segment is the fastest. (Output format: a ranked list of dates, e.g., ['YYYY-MM-DD', ...])", "answer": "['2022-09-07', '2022-08-24', '2022-08-01', '2022-11-08', '2022-04-17', '2022-04-07']", "ground_truth": [ "2022-09-07", "2022-08-24", "2022-08-01", "2022-11-08", "2022-04-17", "2022-04-07" ], "eval_metric": "set_f1", "channel": "67", "ts_data_path": "ts_data/L3_T1_Composite_Trend_S3_00001.csv" }, { "id": "L2_T1_Shape_Identification_S2_00002", "level": 2, "level_name": "Pattern Recognition", "category": "Shape Identification", "subtask": "Shape Identification", "question": "Identify the time range of the longest low plateau (bottom out) in channel 169 within [2022-10-09 13:30:00 to 2022-10-24 02:45:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2022-10-12 03:15:00', '2022-10-14 07:15:00']", "ground_truth": [ "2022-10-12 03:15:00", "2022-10-14 07:15:00" ], "eval_metric": "iou", "channel": "169", "ts_data_path": "ts_data/L2_T1_Shape_Identification_S2_00002.csv" }, { "id": "L1_T3_Interval_Discovery_S2_00003", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Interval Discovery", "question": "Find the longest period where channel 595 remained above 1.34 in 2022-07-13 to 2022-12-10. (Output format: a time interval, e.g., [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2022-10-23 18:45:00, 2022-11-01 13:45:00]", "ground_truth": [ "2022-10-23 18:45:00", "2022-11-01 13:45:00" ], "eval_metric": "iou", "channel": "595", "ts_data_path": "ts_data/L1_T3_Interval_Discovery_S2_00003.csv" }, { "id": "L1_T4_Sliding_Window_S1_00006", "level": 1, "level_name": "Basic Operations", "category": "Sliding Window", "subtask": "Sliding Window", "question": "Which 76-day period in 2019-08-26 to 2020-05-30 had the largest range for channel 897? (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2019-08-26 10:45:00, 2019-08-27 15:30:00]", "ground_truth": [ "2019-08-26 10:45:00", "2019-08-27 15:30:00" ], "eval_metric": "iou", "channel": "897", "ts_data_path": "ts_data/L1_T4_Sliding_Window_S1_00006.csv" }, { "id": "L1_T3_Interval_Discovery_S1_00005", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Interval Discovery", "question": "Find the longest period where channel 895 remained above 2.73 in 2022-04-08 to 2022-05-30. (Output format: a time interval, e.g., [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2022-04-08 06:15:00, 2022-04-13 07:15:00]", "ground_truth": [ "2022-04-08 06:15:00", "2022-04-13 07:15:00" ], "eval_metric": "iou", "channel": "895", "ts_data_path": "ts_data/L1_T3_Interval_Discovery_S1_00005.csv" }, { "id": "L3_T1_Composite_Trend_S1_00006", "level": 3, "level_name": "Semantic Reasoning", "category": "Composite Trend", "subtask": "Composite Trend", "question": "Among days in channel 155 during 2019 that exhibit the trend pattern 'slow rise, then rapid fall', identify the top-2 days where the slow rise segment is the slowest. (Output format: a ranked list of dates, e.g., ['YYYY-MM-DD', ...])", "answer": "['2019-09-24', '2019-04-20']", "ground_truth": [ "2019-09-24", "2019-04-20" ], "eval_metric": "set_f1", "channel": "155", "ts_data_path": "ts_data/L3_T1_Composite_Trend_S1_00006.csv" }, { "id": "L2_T3_Subsequence_Matching_S2_00008", "level": 2, "level_name": "Pattern Recognition", "category": "Subsequence Matching", "subtask": "Subsequence Matching", "question": "Analyze the reference pattern in [2022-07-02 17:45:00 to 2022-07-03 11:30:00]. Find the time interval where channel 626 exhibits the most similar pattern within the search context [2022-07-06 01:30:00 to 2022-07-15 14:15:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2022-07-07 16:00:00', '2022-07-08 09:45:00']", "ground_truth": [ "2022-07-07 16:00:00", "2022-07-08 09:45:00" ], "eval_metric": "iou", "channel": "626", "ts_data_path": "ts_data/L2_T3_Subsequence_Matching_S2_00008.csv" }, { "id": "L3_T2_Contextual_Anomaly_S1_00001", "level": 3, "level_name": "Semantic Reasoning", "category": "Contextual Anomaly", "subtask": "Contextual Anomaly", "question": "Identify the period in channel 123 during 2019 that experienced the most significant historically low water level. (Output format: [YYYY-MM-DD, YYYY-MM-DD])", "answer": "[2019-11-06 07:15:00, 2019-12-11 07:00:00]", "ground_truth": [ "2019-11-06 07:15:00", "2019-12-11 07:00:00" ], "eval_metric": "iou", "channel": "123", "ts_data_path": "ts_data/L3_T2_Contextual_Anomaly_S1_00001.csv" }, { "id": "L3_T1_Composite_Trend_S2_00008", "level": 3, "level_name": "Semantic Reasoning", "category": "Composite Trend", "subtask": "Composite Trend", "question": "Among days in channel MULL during 2017 that exhibit the trend pattern 'rapid fall, then rapid rise, then fluctuating stable, then fall', identify the top-4 days where the rapid rise segment is the fastest. (Output format: a ranked list of dates, e.g., ['YYYY-MM-DD', ...])", "answer": "['2017-04-08', '2017-04-02', '2017-08-11', '2017-07-07']", "ground_truth": [ "2017-04-08", "2017-04-02", "2017-08-11", "2017-07-07" ], "eval_metric": "set_f1", "channel": "MULL", "ts_data_path": "ts_data/L3_T1_Composite_Trend_S2_00008.csv" }, { "id": "L3_T2_Contextual_Anomaly_S1_00009", "level": 3, "level_name": "Semantic Reasoning", "category": "Contextual Anomaly", "subtask": "Contextual Anomaly", "question": "Identify the period in channel 147 during 2019 that experienced the most significant severe drought. (Output format: [YYYY-MM-DD, YYYY-MM-DD])", "answer": "[2019-10-30 13:00:00, 2019-11-21 12:45:00]", "ground_truth": [ "2019-10-30 13:00:00", "2019-11-21 12:45:00" ], "eval_metric": "iou", "channel": "147", "ts_data_path": "ts_data/L3_T2_Contextual_Anomaly_S1_00009.csv" }, { "id": "L4_T1_Insight_Synthesis_S4_00006", "level": 4, "level_name": "Insight Synthesis", "category": "Insight Synthesis", "subtask": "Insight Synthesis", "question": "Analyze the behavior of channel LUFL for the period 2017-12.\nPlease use ONLY the following phrases for trend description: rapid rise, gradual rise, rapid fall, gradual fall, steady stable, fluctuating stable.\nProvide a structured report covering:\n1. Trend Segmentation: Describe each stage with precise start/end timestamps (HH:MM:SS) using the phrases above.\n2. Outlier Audit: Identify only significant outliers that deviate sharply from the local trend. Ignore minor background noise.\n(Output format: structured natural language report.)", "answer": "1. Trend Segmentation: from 2017-12-01 00:00:00 to 2017-12-16 12:00:00, the trend showed a rapid fall; from 2017-12-16 12:00:00 to 2017-12-22 16:45:00, the trend showed a gradual fall; from 2017-12-22 16:45:00 to 2017-12-31 23:45:00, the trend showed a rapid rise.\n2. Outlier Audit: A significant spike was detected at 2017-12-22 21:00:00 (value: 9.39).", "ground_truth": { "trend_segments": [ "from 2017-12-01 00:00:00 to 2017-12-16 12:00:00, the trend showed a rapid fall", "from 2017-12-16 12:00:00 to 2017-12-22 16:45:00, the trend showed a gradual fall", "from 2017-12-22 16:45:00 to 2017-12-31 23:45:00, the trend showed a rapid rise" ], "segments_meta": [ { "adj": "rapid", "kind": "fall", "start_idx": 0, "end_idx": 1488 }, { "adj": "gradual", "kind": "fall", "start_idx": 1488, "end_idx": 2083 }, { "adj": "rapid", "kind": "rise", "start_idx": 2083, "end_idx": 2976 } ], "significant_anomaly": { "timestamp": "2017-12-22 21:00:00", "kind": "significant spike", "value": 9.391955758213612 } }, "eval_metric": "report", "channel": "LUFL", "ts_data_path": "ts_data/L4_T1_Insight_Synthesis_S4_00006.csv" }, { "id": "L3_T1_Composite_Trend_S1_00002", "level": 3, "level_name": "Semantic Reasoning", "category": "Composite Trend", "subtask": "Composite Trend", "question": "Among days in channel LULL during 2017 that exhibit the trend pattern 'slow rise, then fluctuating stable', identify the top-2 days where the fluctuating stable segment is the most fluctuating. (Output format: a ranked list of dates, e.g., ['YYYY-MM-DD', ...])", "answer": "['2017-11-20', '2017-08-30']", "ground_truth": [ "2017-11-20", "2017-08-30" ], "eval_metric": "set_f1", "channel": "LULL", "ts_data_path": "ts_data/L3_T1_Composite_Trend_S1_00002.csv" }, { "id": "L3_T3_Causal_Anomaly_S3_00002", "level": 3, "level_name": "Semantic Reasoning", "category": "Causal Anomaly", "subtask": "Causal Anomaly", "question": "Given that channel 169_up_2021 is the upstream source of channel 169_down_2021, identify the time period in 2021 where 169_down_2021 shows a significant causal anomaly, such as an flat line during high activity. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2021-09-19 16:45:00, 2021-09-25 10:15:00]", "ground_truth": [ "2021-09-19 16:45:00", "2021-09-25 10:15:00" ], "eval_metric": "iou", "channel": "169_down_2021", "ts_data_path": "ts_data/L3_T3_Causal_Anomaly_S3_00002.csv" }, { "id": "L3_T2_Contextual_Anomaly_S0_00005", "level": 3, "level_name": "Semantic Reasoning", "category": "Contextual Anomaly", "subtask": "Contextual Anomaly", "question": "Identify the period in channel 680 during 2021 that experienced the most significant severe drought. (Output format: [YYYY-MM-DD, YYYY-MM-DD])", "answer": "[2021-10-28 03:30:00, 2021-12-12 03:15:00]", "ground_truth": [ "2021-10-28 03:30:00", "2021-12-12 03:15:00" ], "eval_metric": "iou", "channel": "680", "ts_data_path": "ts_data/L3_T2_Contextual_Anomaly_S0_00005.csv" }, { "id": "L1_T4_Sliding_Window_S4_00003", "level": 1, "level_name": "Basic Operations", "category": "Sliding Window", "subtask": "Sliding Window", "question": "Which 20-day period in 2022-08-17 to 2023-02-22 had the lowest average for channel 627? (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2022-08-17 19:30:00, 2022-08-18 00:00:00]", "ground_truth": [ "2022-08-17 19:30:00", "2022-08-18 00:00:00" ], "eval_metric": "iou", "channel": "627", "ts_data_path": "ts_data/L1_T4_Sliding_Window_S4_00003.csv" }, { "id": "L3_T3_Causal_Anomaly_S3_00007", "level": 3, "level_name": "Semantic Reasoning", "category": "Causal Anomaly", "subtask": "Causal Anomaly", "question": "Given that channel 728_up_2019 is the upstream source of channel 728_down_2019, identify the time period in 2019 where 728_down_2019 shows a significant causal anomaly, such as an inverse trend against the source. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2019-01-13 16:15:00, 2019-01-16 00:30:00]", "ground_truth": [ "2019-01-13 16:15:00", "2019-01-16 00:30:00" ], "eval_metric": "iou", "channel": "728_down_2019", "ts_data_path": "ts_data/L3_T3_Causal_Anomaly_S3_00007.csv" }, { "id": "L2_T1_Shape_Identification_S2_00003", "level": 2, "level_name": "Pattern Recognition", "category": "Shape Identification", "subtask": "Shape Identification", "question": "Identify the time range of the largest step descent in channel 173 within [2020-12-03 08:15:00 to 2020-12-11 10:45:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2020-12-04 07:00:00', '2020-12-05 11:45:00']", "ground_truth": [ "2020-12-04 07:00:00", "2020-12-05 11:45:00" ], "eval_metric": "iou", "channel": "173", "ts_data_path": "ts_data/L2_T1_Shape_Identification_S2_00003.csv" }, { "id": "L4_T1_Insight_Synthesis_S1_00001", "level": 4, "level_name": "Insight Synthesis", "category": "Insight Synthesis", "subtask": "Insight Synthesis", "question": "Analyze the behavior of channel 147 for the period 2020-06.\nPlease use ONLY the following phrases for trend description: rapid rise, gradual rise, rapid fall, gradual fall, steady stable, fluctuating stable.\nProvide a structured report covering:\n1. Trend Segmentation: Describe each stage with precise start/end timestamps (HH:MM:SS) using the phrases above.\n2. Outlier Audit: Identify only significant outliers that deviate sharply from the local trend. Ignore minor background noise.\n(Output format: structured natural language report.)", "answer": "1. Trend Segmentation: from 2020-06-01 00:00:00 to 2020-06-10 00:00:00, the trend showed a gradual rise; from 2020-06-10 00:00:00 to 2020-06-16 00:00:00, the trend showed a fluctuating stable; from 2020-06-16 00:00:00 to 2020-06-22 00:00:00, the trend showed a gradual fall; from 2020-06-22 00:00:00 to 2020-06-30 23:45:00, the trend showed a gradual rise.\n2. Outlier Audit: A significant drop was detected at 2020-06-14 12:30:00 (value: -0.67).", "ground_truth": { "trend_segments": [ "from 2020-06-01 00:00:00 to 2020-06-10 00:00:00, the trend showed a gradual rise", "from 2020-06-10 00:00:00 to 2020-06-16 00:00:00, the trend showed a fluctuating stable", "from 2020-06-16 00:00:00 to 2020-06-22 00:00:00, the trend showed a gradual fall", "from 2020-06-22 00:00:00 to 2020-06-30 23:45:00, the trend showed a gradual rise" ], "segments_meta": [ { "adj": "gradual", "kind": "rise", "start_idx": 0, "end_idx": 864 }, { "adj": "fluctuating", "kind": "stable", "start_idx": 864, "end_idx": 1440 }, { "adj": "gradual", "kind": "fall", "start_idx": 1440, "end_idx": 2016 }, { "adj": "gradual", "kind": "rise", "start_idx": 2016, "end_idx": 2880 } ], "significant_anomaly": { "timestamp": "2020-06-14 12:30:00", "kind": "significant drop", "value": -0.6703728467203915 } }, "eval_metric": "report", "channel": "147", "ts_data_path": "ts_data/L4_T1_Insight_Synthesis_S1_00001.csv" }, { "id": "L1_T4_Sliding_Window_S2_00008", "level": 1, "level_name": "Basic Operations", "category": "Sliding Window", "subtask": "Sliding Window", "question": "Which 41-day period in 2022-07-09 to 2023-01-22 had the highest average for channel 625? (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2022-12-12 01:00:00, 2023-01-22 00:45:00]", "ground_truth": [ "2022-12-12 01:00:00", "2023-01-22 00:45:00" ], "eval_metric": "iou", "channel": "625", "ts_data_path": "ts_data/L1_T4_Sliding_Window_S2_00008.csv" }, { "id": "L3_T1_Composite_Trend_S4_00003", "level": 3, "level_name": "Semantic Reasoning", "category": "Composite Trend", "subtask": "Composite Trend", "question": "Among days in channel 166 during 2021 that exhibit the trend pattern 'rise, then rapid rise, then slow fall, then fall', identify the top-8 days where the slow fall segment is the slowest. (Output format: a ranked list of dates, e.g., ['YYYY-MM-DD', ...])", "answer": "['2021-08-10', '2021-01-17', '2021-08-05', '2021-10-28', '2021-12-17', '2021-09-30', '2021-02-18', '2021-10-23']", "ground_truth": [ "2021-08-10", "2021-01-17", "2021-08-05", "2021-10-28", "2021-12-17", "2021-09-30", "2021-02-18", "2021-10-23" ], "eval_metric": "set_f1", "channel": "166", "ts_data_path": "ts_data/L3_T1_Composite_Trend_S4_00003.csv" }, { "id": "L1_T1_Global_Aggregation_S2_00005", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Global Aggregation", "question": "What is the average value of channel HULL in 2017-05? (Output format: a single numeric value, rounded to 3 decimal places, e.g., x.xxx)", "answer": "2.155", "ground_truth": 2.155, "eval_metric": "rel_acc", "channel": "HULL", "ts_data_path": "ts_data/L1_T1_Global_Aggregation_S2_00005.csv" }, { "id": "L1_T4_Sliding_Window_S4_00005", "level": 1, "level_name": "Basic Operations", "category": "Sliding Window", "subtask": "Sliding Window", "question": "Which 18-day period in 2022-08-13 to 2023-01-21 had the lowest average for channel 865? (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2022-08-13 16:00:00, 2022-08-13 16:00:00]", "ground_truth": [ "2022-08-13 16:00:00", "2022-08-13 16:00:00" ], "eval_metric": "iou", "channel": "865", "ts_data_path": "ts_data/L1_T4_Sliding_Window_S4_00005.csv" }, { "id": "L1_T1_Global_Aggregation_S4_00007", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Global Aggregation", "question": "What is the median value of channel 811 in 2019-01? (Output format: a single numeric value, rounded to 3 decimal places, e.g., x.xxx)", "answer": "14.0", "ground_truth": 14.0, "eval_metric": "rel_acc", "channel": "811", "ts_data_path": "ts_data/L1_T1_Global_Aggregation_S4_00007.csv" }, { "id": "L2_T1_Shape_Identification_S0_00005", "level": 2, "level_name": "Pattern Recognition", "category": "Shape Identification", "subtask": "Shape Identification", "question": "Identify the time range of the largest step descent in channel 728 within [2022-07-07 13:30:00 to 2022-07-11 00:45:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2022-07-08 05:00:00', '2022-07-08 17:15:00']", "ground_truth": [ "2022-07-08 05:00:00", "2022-07-08 17:15:00" ], "eval_metric": "iou", "channel": "728", "ts_data_path": "ts_data/L2_T1_Shape_Identification_S0_00005.csv" }, { "id": "L1_T3_Interval_Discovery_S2_00006", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Interval Discovery", "question": "Find the longest period where channel HUFL remained above 11.588 in 2017-06-09 to 2017-09-07. (Output format: a time interval, e.g., [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2017-07-30 19:30:00, 2017-08-01 06:00:00]", "ground_truth": [ "2017-07-30 19:30:00", "2017-08-01 06:00:00" ], "eval_metric": "iou", "channel": "HUFL", "ts_data_path": "ts_data/L1_T3_Interval_Discovery_S2_00006.csv" }, { "id": "L1_T1_Global_Aggregation_S1_00007", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Global Aggregation", "question": "What is the median value of channel 728 in 2022-12-04 to 2022-12-24? (Output format: a single numeric value, rounded to 3 decimal places, e.g., x.xxx)", "answer": "45.2", "ground_truth": 45.2, "eval_metric": "rel_acc", "channel": "728", "ts_data_path": "ts_data/L1_T1_Global_Aggregation_S1_00007.csv" }, { "id": "L3_T3_Causal_Anomaly_S2_00003", "level": 3, "level_name": "Semantic Reasoning", "category": "Causal Anomaly", "subtask": "Causal Anomaly", "question": "Given that channel 595_up_2021 is the upstream source of channel 595_down_2021, identify the time period in 2021 where 595_down_2021 shows a significant causal anomaly, such as an inverse trend against the source. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2021-06-10 11:00:00, 2021-06-12 07:45:00]", "ground_truth": [ "2021-06-10 11:00:00", "2021-06-12 07:45:00" ], "eval_metric": "iou", "channel": "595_down_2021", "ts_data_path": "ts_data/L3_T3_Causal_Anomaly_S2_00003.csv" }, { "id": "L2_T1_Shape_Identification_S2_00001", "level": 2, "level_name": "Pattern Recognition", "category": "Shape Identification", "subtask": "Shape Identification", "question": "Identify the time range of the largest step ascent in channel 154 within [2021-08-01 00:00:00 to 2021-08-13 19:30:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2021-08-08 17:30:00', '2021-08-10 15:15:00']", "ground_truth": [ "2021-08-08 17:30:00", "2021-08-10 15:15:00" ], "eval_metric": "iou", "channel": "154", "ts_data_path": "ts_data/L2_T1_Shape_Identification_S2_00001.csv" }, { "id": "L2_T3_Subsequence_Matching_S3_00009", "level": 2, "level_name": "Pattern Recognition", "category": "Subsequence Matching", "subtask": "Subsequence Matching", "question": "Analyze the reference pattern in [2017-10-30 08:00:00 to 2017-10-30 13:45:00]. Find the time interval where channel LULL exhibits the most similar pattern within the search context [2017-10-30 23:00:00 to 2017-11-01 00:30:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2017-10-31 11:00:00', '2017-10-31 16:45:00']", "ground_truth": [ "2017-10-31 11:00:00", "2017-10-31 16:45:00" ], "eval_metric": "iou", "channel": "LULL", "ts_data_path": "ts_data/L2_T3_Subsequence_Matching_S3_00009.csv" }, { "id": "L3_T3_Causal_Anomaly_S1_00004", "level": 3, "level_name": "Semantic Reasoning", "category": "Causal Anomaly", "subtask": "Causal Anomaly", "question": "Given that channel 895_up_2023 is the upstream source of channel 895_down_2023, identify the time period in 2023 where 895_down_2023 shows a significant causal anomaly, such as an inverse trend against the source. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2023-10-11 09:00:00, 2023-10-15 04:15:00]", "ground_truth": [ "2023-10-11 09:00:00", "2023-10-15 04:15:00" ], "eval_metric": "iou", "channel": "895_down_2023", "ts_data_path": "ts_data/L3_T3_Causal_Anomaly_S1_00004.csv" }, { "id": "L1_T1_Global_Aggregation_S4_00004", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Global Aggregation", "question": "What is the average value of channel 589 in 2023-02? (Output format: a single numeric value, rounded to 3 decimal places, e.g., x.xxx)", "answer": "524.881", "ground_truth": 524.881, "eval_metric": "rel_acc", "channel": "589", "ts_data_path": "ts_data/L1_T1_Global_Aggregation_S4_00004.csv" }, { "id": "L1_T3_Interval_Discovery_S0_00003", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Interval Discovery", "question": "Find the longest period where channel 496 remained above 0.525 in 2020-11-17 to 2021-04-25. (Output format: a time interval, e.g., [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2021-02-14 23:45:00, 2021-03-08 15:00:00]", "ground_truth": [ "2021-02-14 23:45:00", "2021-03-08 15:00:00" ], "eval_metric": "iou", "channel": "496", "ts_data_path": "ts_data/L1_T3_Interval_Discovery_S0_00003.csv" }, { "id": "L3_T2_Contextual_Anomaly_S0_00007", "level": 3, "level_name": "Semantic Reasoning", "category": "Contextual Anomaly", "subtask": "Contextual Anomaly", "question": "Identify the period in channel 897 during 2023 that experienced the most significant severe drought. (Output format: [YYYY-MM-DD, YYYY-MM-DD])", "answer": "[2023-07-26 07:15:00, 2023-08-23 07:00:00]", "ground_truth": [ "2023-07-26 07:15:00", "2023-08-23 07:00:00" ], "eval_metric": "iou", "channel": "897", "ts_data_path": "ts_data/L3_T2_Contextual_Anomaly_S0_00007.csv" }, { "id": "L4_T1_Insight_Synthesis_S4_00008", "level": 4, "level_name": "Insight Synthesis", "category": "Insight Synthesis", "subtask": "Insight Synthesis", "question": "Analyze the behavior of channel 312 for the period 2019-08.\nPlease use ONLY the following phrases for trend description: rapid rise, gradual rise, rapid fall, gradual fall, steady stable, fluctuating stable.\nProvide a structured report covering:\n1. Trend Segmentation: Describe each stage with precise start/end timestamps (HH:MM:SS) using the phrases above.\n2. Outlier Audit: Identify only significant outliers that deviate sharply from the local trend. Ignore minor background noise.\n(Output format: structured natural language report.)", "answer": "1. Trend Segmentation: from 2019-08-01 00:00:00 to 2019-08-10 02:45:00, the trend showed a steady stable; from 2019-08-10 02:45:00 to 2019-08-17 09:45:00, the trend showed a gradual fall; from 2019-08-17 09:45:00 to 2019-08-26 12:30:00, the trend showed a rapid rise; from 2019-08-26 12:30:00 to 2019-08-31 23:45:00, the trend showed a rapid fall.\n2. Outlier Audit: A significant drop was detected at 2019-08-16 20:45:00 (value: -0.94).", "ground_truth": { "trend_segments": [ "from 2019-08-01 00:00:00 to 2019-08-10 02:45:00, the trend showed a steady stable", "from 2019-08-10 02:45:00 to 2019-08-17 09:45:00, the trend showed a gradual fall", "from 2019-08-17 09:45:00 to 2019-08-26 12:30:00, the trend showed a rapid rise", "from 2019-08-26 12:30:00 to 2019-08-31 23:45:00, the trend showed a rapid fall" ], "segments_meta": [ { "adj": "steady", "kind": "stable", "start_idx": 0, "end_idx": 875 }, { "adj": "gradual", "kind": "fall", "start_idx": 875, "end_idx": 1575 }, { "adj": "rapid", "kind": "rise", "start_idx": 1575, "end_idx": 2450 }, { "adj": "rapid", "kind": "fall", "start_idx": 2450, "end_idx": 2976 } ], "significant_anomaly": { "timestamp": "2019-08-16 20:45:00", "kind": "significant drop", "value": -0.9404632138319988 } }, "eval_metric": "report", "channel": "312", "ts_data_path": "ts_data/L4_T1_Insight_Synthesis_S4_00008.csv" }, { "id": "L1_T3_Interval_Discovery_S0_00001", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Interval Discovery", "question": "Find the longest period where channel 124 remained above 0.336 in 2021-11. (Output format: a time interval, e.g., [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2021-11-04 13:45:00, 2021-11-10 15:45:00]", "ground_truth": [ "2021-11-04 13:45:00", "2021-11-10 15:45:00" ], "eval_metric": "iou", "channel": "124", "ts_data_path": "ts_data/L1_T3_Interval_Discovery_S0_00001.csv" }, { "id": "L3_T3_Causal_Anomaly_S2_00007", "level": 3, "level_name": "Semantic Reasoning", "category": "Causal Anomaly", "subtask": "Causal Anomaly", "question": "Given that channel 496_up_2021 is the upstream source of channel 496_down_2021, identify the time period in 2021 where 496_down_2021 shows a significant causal anomaly, such as an inverse trend against the source. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2021-03-30 14:15:00, 2021-03-31 16:30:00]", "ground_truth": [ "2021-03-30 14:15:00", "2021-03-31 16:30:00" ], "eval_metric": "iou", "channel": "496_down_2021", "ts_data_path": "ts_data/L3_T3_Causal_Anomaly_S2_00007.csv" }, { "id": "L2_T2_Periodicity_Detection_S3_00001", "level": 2, "level_name": "Pattern Recognition", "category": "Periodicity Detection", "subtask": "Periodicity Detection", "question": "What is the dominant cycle period (in data points) of channel 123 within [2022-06-01 03:45:00 to 2022-06-09 10:45:00]? (Output format: integer)", "answer": "77", "ground_truth": 77, "eval_metric": "rel_acc", "channel": "123", "ts_data_path": "ts_data/L2_T2_Periodicity_Detection_S3_00001.csv" }, { "id": "L3_T2_Contextual_Anomaly_S1_00008", "level": 3, "level_name": "Semantic Reasoning", "category": "Contextual Anomaly", "subtask": "Contextual Anomaly", "question": "Identify the period in channel 146 during 2021 that experienced the most significant severe drought. (Output format: [YYYY-MM-DD, YYYY-MM-DD])", "answer": "[2021-07-11 08:30:00, 2021-08-06 08:15:00]", "ground_truth": [ "2021-07-11 08:30:00", "2021-08-06 08:15:00" ], "eval_metric": "iou", "channel": "146", "ts_data_path": "ts_data/L3_T2_Contextual_Anomaly_S1_00008.csv" }, { "id": "L1_T2_Temporal_Localization_S3_00005", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Temporal Localization", "question": "At what exact timestamp did channel 627 reach its minimum value in 2023-11? (Output format: a specific timestamp, e.g., YYYY-MM-DD HH:MM:SS)", "answer": "2023-11-04 20:30:00", "ground_truth": "2023-11-04 20:30:00", "eval_metric": "hit", "channel": "627", "ts_data_path": "ts_data/L1_T2_Temporal_Localization_S3_00005.csv" }, { "id": "L3_T2_Contextual_Anomaly_S0_00001", "level": 3, "level_name": "Semantic Reasoning", "category": "Contextual Anomaly", "subtask": "Contextual Anomaly", "question": "Identify the period in channel 166 during 2021 that experienced the most significant severe drought. (Output format: [YYYY-MM-DD, YYYY-MM-DD])", "answer": "[2021-03-28 19:15:00, 2021-04-30 19:00:00]", "ground_truth": [ "2021-03-28 19:15:00", "2021-04-30 19:00:00" ], "eval_metric": "iou", "channel": "166", "ts_data_path": "ts_data/L3_T2_Contextual_Anomaly_S0_00001.csv" }, { "id": "L3_T3_Causal_Anomaly_S3_00009", "level": 3, "level_name": "Semantic Reasoning", "category": "Causal Anomaly", "subtask": "Causal Anomaly", "question": "Given that channel 146_up_2022 is the upstream source of channel 146_down_2022, identify the time period in 2022 where 146_down_2022 shows a significant causal anomaly, such as an flat line during high activity. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2022-06-12 09:15:00, 2022-06-15 15:15:00]", "ground_truth": [ "2022-06-12 09:15:00", "2022-06-15 15:15:00" ], "eval_metric": "iou", "channel": "146_down_2022", "ts_data_path": "ts_data/L3_T3_Causal_Anomaly_S3_00009.csv" }, { "id": "L2_T2_Periodicity_Detection_S1_00007", "level": 2, "level_name": "Pattern Recognition", "category": "Periodicity Detection", "subtask": "Periodicity Detection", "question": "What is the dominant cycle period (in data points) of channel 146 within [2019-09-10 23:45:00 to 2019-09-19 11:15:00]? (Output format: integer)", "answer": "47", "ground_truth": 47, "eval_metric": "rel_acc", "channel": "146", "ts_data_path": "ts_data/L2_T2_Periodicity_Detection_S1_00007.csv" }, { "id": "L1_T4_Sliding_Window_S4_00008", "level": 1, "level_name": "Basic Operations", "category": "Sliding Window", "subtask": "Sliding Window", "question": "Which 19-day period in 2017-01-19 to 2017-06-19 had the largest range for channel LULL? (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2017-05-24 00:30:00, 2017-06-12 00:15:00]", "ground_truth": [ "2017-05-24 00:30:00", "2017-06-12 00:15:00" ], "eval_metric": "iou", "channel": "LULL", "ts_data_path": "ts_data/L1_T4_Sliding_Window_S4_00008.csv" }, { "id": "L1_T1_Global_Aggregation_S3_00007", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Global Aggregation", "question": "What is the minimum value of channel LUFL in 2017-09-24 to 2017-10-20? (Output format: a single numeric value, rounded to 3 decimal places, e.g., x.xxx)", "answer": "0.853", "ground_truth": 0.853, "eval_metric": "rel_acc", "channel": "LUFL", "ts_data_path": "ts_data/L1_T1_Global_Aggregation_S3_00007.csv" }, { "id": "L2_T2_Periodicity_Detection_S1_00004", "level": 2, "level_name": "Pattern Recognition", "category": "Periodicity Detection", "subtask": "Periodicity Detection", "question": "What is the dominant cycle period (in data points) of channel 627 within [2022-01-06 21:00:00 to 2022-01-11 03:30:00]? (Output format: integer)", "answer": "136", "ground_truth": 136, "eval_metric": "rel_acc", "channel": "627", "ts_data_path": "ts_data/L2_T2_Periodicity_Detection_S1_00004.csv" }, { "id": "L2_T2_Periodicity_Detection_S4_00004", "level": 2, "level_name": "Pattern Recognition", "category": "Periodicity Detection", "subtask": "Periodicity Detection", "question": "What is the dominant cycle period (in data points) of channel 166 within [2019-11-10 13:30:00 to 2019-11-19 13:45:00]? (Output format: integer)", "answer": "204", "ground_truth": 204, "eval_metric": "rel_acc", "channel": "166", "ts_data_path": "ts_data/L2_T2_Periodicity_Detection_S4_00004.csv" }, { "id": "L2_T2_Periodicity_Detection_S0_00007", "level": 2, "level_name": "Pattern Recognition", "category": "Periodicity Detection", "subtask": "Periodicity Detection", "question": "What is the dominant cycle period (in data points) of channel MUFL within [2016-08-05 10:30:00 to 2016-08-12 00:00:00]? (Output format: integer)", "answer": "89", "ground_truth": 89, "eval_metric": "rel_acc", "channel": "MUFL", "ts_data_path": "ts_data/L2_T2_Periodicity_Detection_S0_00007.csv" }, { "id": "L1_T4_Sliding_Window_S1_00003", "level": 1, "level_name": "Basic Operations", "category": "Sliding Window", "subtask": "Sliding Window", "question": "Which 35-day period in 2020-09-17 to 2021-08-12 had the largest range for channel 811? (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2021-01-01 02:15:00, 2021-02-05 02:00:00]", "ground_truth": [ "2021-01-01 02:15:00", "2021-02-05 02:00:00" ], "eval_metric": "iou", "channel": "811", "ts_data_path": "ts_data/L1_T4_Sliding_Window_S1_00003.csv" }, { "id": "L2_T3_Subsequence_Matching_S1_00008", "level": 2, "level_name": "Pattern Recognition", "category": "Subsequence Matching", "subtask": "Subsequence Matching", "question": "Analyze the reference pattern in [2019-07-04 16:30:00 to 2019-07-04 22:15:00]. Find the time interval where channel 173 exhibits the most similar pattern within the search context [2019-07-05 07:30:00 to 2019-07-06 06:00:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2019-07-05 09:00:00', '2019-07-05 14:45:00']", "ground_truth": [ "2019-07-05 09:00:00", "2019-07-05 14:45:00" ], "eval_metric": "iou", "channel": "173", "ts_data_path": "ts_data/L2_T3_Subsequence_Matching_S1_00008.csv" }, { "id": "L1_T3_Interval_Discovery_S4_00003", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Interval Discovery", "question": "Find the longest period where channel 172 remained above 332.0 in 2021-04-24 to 2021-11-01. (Output format: a time interval, e.g., [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2021-07-09 02:15:00, 2021-08-02 08:30:00]", "ground_truth": [ "2021-07-09 02:15:00", "2021-08-02 08:30:00" ], "eval_metric": "iou", "channel": "172", "ts_data_path": "ts_data/L1_T3_Interval_Discovery_S4_00003.csv" }, { "id": "L1_T1_Global_Aggregation_S1_00004", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Global Aggregation", "question": "What is the maximum value of channel 166 in 2021-09-26 to 2021-10-16? (Output format: a single numeric value, rounded to 3 decimal places, e.g., x.xxx)", "answer": "368.0", "ground_truth": 368.0, "eval_metric": "rel_acc", "channel": "166", "ts_data_path": "ts_data/L1_T1_Global_Aggregation_S1_00004.csv" }, { "id": "L3_T3_Causal_Anomaly_S4_00002", "level": 3, "level_name": "Semantic Reasoning", "category": "Causal Anomaly", "subtask": "Causal Anomaly", "question": "Given that channel 245_up_2022 is the upstream source of channel 245_down_2022, identify the time period in 2022 where 245_down_2022 shows a significant causal anomaly, such as an inverse trend against the source. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2022-04-30 16:30:00, 2022-05-11 04:30:00]", "ground_truth": [ "2022-04-30 16:30:00", "2022-05-11 04:30:00" ], "eval_metric": "iou", "channel": "245_down_2022", "ts_data_path": "ts_data/L3_T3_Causal_Anomaly_S4_00002.csv" }, { "id": "L3_T1_Composite_Trend_S4_00007", "level": 3, "level_name": "Semantic Reasoning", "category": "Composite Trend", "subtask": "Composite Trend", "question": "Among days in channel LUFL during 2017 that exhibit the trend pattern 'rapid fall, then steady stable, then fluctuating stable, then steady stable', identify the top-8 days where the rapid fall segment is the fastest. (Output format: a ranked list of dates, e.g., ['YYYY-MM-DD', ...])", "answer": "['2017-06-03', '2017-08-29', '2017-04-30', '2017-12-25', '2017-11-02', '2017-03-15', '2017-03-19', '2017-08-24']", "ground_truth": [ "2017-06-03", "2017-08-29", "2017-04-30", "2017-12-25", "2017-11-02", "2017-03-15", "2017-03-19", "2017-08-24" ], "eval_metric": "set_f1", "channel": "LUFL", "ts_data_path": "ts_data/L3_T1_Composite_Trend_S4_00007.csv" }, { "id": "L2_T2_Periodicity_Detection_S0_00006", "level": 2, "level_name": "Pattern Recognition", "category": "Periodicity Detection", "subtask": "Periodicity Detection", "question": "What is the dominant cycle period (in data points) of channel HULL within [2016-12-26 15:30:00 to 2016-12-29 06:45:00]? (Output format: integer)", "answer": "84", "ground_truth": 84, "eval_metric": "rel_acc", "channel": "HULL", "ts_data_path": "ts_data/L2_T2_Periodicity_Detection_S0_00006.csv" }, { "id": "L3_T1_Composite_Trend_S3_00010", "level": 3, "level_name": "Semantic Reasoning", "category": "Composite Trend", "subtask": "Composite Trend", "question": "Among days in channel MUFL during 2017 that exhibit the trend pattern 'fall, then rapid rise, then fluctuating stable', identify the top-6 days where the fluctuating stable segment is the most fluctuating. (Output format: a ranked list of dates, e.g., ['YYYY-MM-DD', ...])", "answer": "['2017-11-03', '2017-08-13', '2017-04-03', '2017-09-05', '2017-05-28', '2017-03-08']", "ground_truth": [ "2017-11-03", "2017-08-13", "2017-04-03", "2017-09-05", "2017-05-28", "2017-03-08" ], "eval_metric": "set_f1", "channel": "MUFL", "ts_data_path": "ts_data/L3_T1_Composite_Trend_S3_00010.csv" }, { "id": "L2_T2_Periodicity_Detection_S3_00005", "level": 2, "level_name": "Pattern Recognition", "category": "Periodicity Detection", "subtask": "Periodicity Detection", "question": "What is the dominant cycle period (in data points) of channel 894 within [2021-09-14 04:30:00 to 2021-09-21 05:30:00]? (Output format: integer)", "answer": "64", "ground_truth": 64, "eval_metric": "rel_acc", "channel": "894", "ts_data_path": "ts_data/L2_T2_Periodicity_Detection_S3_00005.csv" }, { "id": "L4_T1_Insight_Synthesis_S1_00009", "level": 4, "level_name": "Insight Synthesis", "category": "Insight Synthesis", "subtask": "Insight Synthesis", "question": "Analyze the behavior of channel 166 for the period 2023-04.\nPlease use ONLY the following phrases for trend description: rapid rise, gradual rise, rapid fall, gradual fall, steady stable, fluctuating stable.\nProvide a structured report covering:\n1. Trend Segmentation: Describe each stage with precise start/end timestamps (HH:MM:SS) using the phrases above.\n2. Outlier Audit: Identify only significant outliers that deviate sharply from the local trend. Ignore minor background noise.\n(Output format: structured natural language report.)", "answer": "1. Trend Segmentation: from 2023-04-01 00:00:00 to 2023-04-05 14:45:00, the trend showed a rapid rise; from 2023-04-05 14:45:00 to 2023-04-09 01:45:00, the trend showed a gradual rise; from 2023-04-09 01:45:00 to 2023-04-14 20:15:00, the trend showed a gradual fall; from 2023-04-14 20:15:00 to 2023-04-20 14:45:00, the trend showed a steady stable; from 2023-04-20 14:45:00 to 2023-04-25 05:30:00, the trend showed a rapid rise; from 2023-04-25 05:30:00 to 2023-04-30 23:45:00, the trend showed a steady stable.\n2. Outlier Audit: A significant spike was detected at 2023-04-14 15:45:00 (value: 7112.70).", "ground_truth": { "trend_segments": [ "from 2023-04-01 00:00:00 to 2023-04-05 14:45:00, the trend showed a rapid rise", "from 2023-04-05 14:45:00 to 2023-04-09 01:45:00, the trend showed a gradual rise", "from 2023-04-09 01:45:00 to 2023-04-14 20:15:00, the trend showed a gradual fall", "from 2023-04-14 20:15:00 to 2023-04-20 14:45:00, the trend showed a steady stable", "from 2023-04-20 14:45:00 to 2023-04-25 05:30:00, the trend showed a rapid rise", "from 2023-04-25 05:30:00 to 2023-04-30 23:45:00, the trend showed a steady stable" ], "segments_meta": [ { "adj": "rapid", "kind": "rise", "start_idx": 0, "end_idx": 443 }, { "adj": "gradual", "kind": "rise", "start_idx": 443, "end_idx": 775 }, { "adj": "gradual", "kind": "fall", "start_idx": 775, "end_idx": 1329 }, { "adj": "steady", "kind": "stable", "start_idx": 1329, "end_idx": 1883 }, { "adj": "rapid", "kind": "rise", "start_idx": 1883, "end_idx": 2326 }, { "adj": "steady", "kind": "stable", "start_idx": 2326, "end_idx": 2880 } ], "significant_anomaly": { "timestamp": "2023-04-14 15:45:00", "kind": "significant spike", "value": 7112.702781581503 } }, "eval_metric": "report", "channel": "166", "ts_data_path": "ts_data/L4_T1_Insight_Synthesis_S1_00009.csv" }, { "id": "L3_T2_Contextual_Anomaly_S1_00003", "level": 3, "level_name": "Semantic Reasoning", "category": "Contextual Anomaly", "subtask": "Contextual Anomaly", "question": "Identify the period in channel 501 during 2022 that experienced the most significant extreme surge in flow. (Output format: [YYYY-MM-DD, YYYY-MM-DD])", "answer": "[2022-01-29 14:45:00, 2022-02-20 14:30:00]", "ground_truth": [ "2022-01-29 14:45:00", "2022-02-20 14:30:00" ], "eval_metric": "iou", "channel": "501", "ts_data_path": "ts_data/L3_T2_Contextual_Anomaly_S1_00003.csv" }, { "id": "L3_T3_Causal_Anomaly_S2_00004", "level": 3, "level_name": "Semantic Reasoning", "category": "Causal Anomaly", "subtask": "Causal Anomaly", "question": "Given that channel 762_up_2023 is the upstream source of channel 762_down_2023, identify the time period in 2023 where 762_down_2023 shows a significant causal anomaly, such as an inverse trend against the source. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2023-02-02 20:00:00, 2023-02-04 03:45:00]", "ground_truth": [ "2023-02-02 20:00:00", "2023-02-04 03:45:00" ], "eval_metric": "iou", "channel": "762_down_2023", "ts_data_path": "ts_data/L3_T3_Causal_Anomaly_S2_00004.csv" }, { "id": "L3_T3_Causal_Anomaly_S0_00009", "level": 3, "level_name": "Semantic Reasoning", "category": "Causal Anomaly", "subtask": "Causal Anomaly", "question": "Given that channel 591_up_2022 is the upstream source of channel 591_down_2022, identify the time period in 2022 where 591_down_2022 shows a significant causal anomaly, such as an flat line during high activity. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2022-02-20 08:30:00, 2022-02-22 04:00:00]", "ground_truth": [ "2022-02-20 08:30:00", "2022-02-22 04:00:00" ], "eval_metric": "iou", "channel": "591_down_2022", "ts_data_path": "ts_data/L3_T3_Causal_Anomaly_S0_00009.csv" }, { "id": "L1_T1_Global_Aggregation_S0_00004", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Global Aggregation", "question": "What is the maximum value of channel 683 in 2019-09-08 to 2019-10-17? (Output format: a single numeric value, rounded to 3 decimal places, e.g., x.xxx)", "answer": "1.41", "ground_truth": 1.41, "eval_metric": "rel_acc", "channel": "683", "ts_data_path": "ts_data/L1_T1_Global_Aggregation_S0_00004.csv" }, { "id": "L3_T2_Contextual_Anomaly_S4_00001", "level": 3, "level_name": "Semantic Reasoning", "category": "Contextual Anomaly", "subtask": "Contextual Anomaly", "question": "Identify the period in channel 67 during 2022 that experienced the most significant severe flood. (Output format: [YYYY-MM-DD, YYYY-MM-DD])", "answer": "[2022-02-02 14:45:00, 2022-03-01 14:30:00]", "ground_truth": [ "2022-02-02 14:45:00", "2022-03-01 14:30:00" ], "eval_metric": "iou", "channel": "67", "ts_data_path": "ts_data/L3_T2_Contextual_Anomaly_S4_00001.csv" }, { "id": "L2_T3_Subsequence_Matching_S3_00003", "level": 2, "level_name": "Pattern Recognition", "category": "Subsequence Matching", "subtask": "Subsequence Matching", "question": "Analyze the reference pattern in [2021-03-25 00:45:00 to 2021-03-25 10:30:00]. Find the time interval where channel 245 exhibits the most similar pattern within the search context [2021-03-27 05:30:00 to 2021-04-02 23:15:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2021-03-30 00:00:00', '2021-03-30 09:45:00']", "ground_truth": [ "2021-03-30 00:00:00", "2021-03-30 09:45:00" ], "eval_metric": "iou", "channel": "245", "ts_data_path": "ts_data/L2_T3_Subsequence_Matching_S3_00003.csv" }, { "id": "L2_T1_Shape_Identification_S4_00008", "level": 2, "level_name": "Pattern Recognition", "category": "Shape Identification", "subtask": "Shape Identification", "question": "Identify the time range of the largest step ascent in channel 441 within [2020-01-13 12:00:00 to 2020-01-19 18:30:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2020-01-15 22:15:00', '2020-01-16 20:30:00']", "ground_truth": [ "2020-01-15 22:15:00", "2020-01-16 20:30:00" ], "eval_metric": "iou", "channel": "441", "ts_data_path": "ts_data/L2_T1_Shape_Identification_S4_00008.csv" }, { "id": "L1_T3_Interval_Discovery_S3_00005", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Interval Discovery", "question": "Find the longest period where channel 627 remained above 0.545 in 2021-03. (Output format: a time interval, e.g., [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2021-03-14 03:30:00, 2021-03-17 13:00:00]", "ground_truth": [ "2021-03-14 03:30:00", "2021-03-17 13:00:00" ], "eval_metric": "iou", "channel": "627", "ts_data_path": "ts_data/L1_T3_Interval_Discovery_S3_00005.csv" }, { "id": "L1_T3_Interval_Discovery_S1_00004", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Interval Discovery", "question": "Find the longest period where channel 430 remained above 0.156 in 2021-06. (Output format: a time interval, e.g., [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2021-06-01 09:45:00, 2021-06-02 23:30:00]", "ground_truth": [ "2021-06-01 09:45:00", "2021-06-02 23:30:00" ], "eval_metric": "iou", "channel": "430", "ts_data_path": "ts_data/L1_T3_Interval_Discovery_S1_00004.csv" }, { "id": "L1_T2_Temporal_Localization_S3_00009", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Temporal Localization", "question": "At what exact timestamp did channel MULL last fall below 1.208 in 2017-06? (Output format: a specific timestamp, e.g., YYYY-MM-DD HH:MM:SS)", "answer": "2017-06-30 23:00:00", "ground_truth": "2017-06-30 23:00:00", "eval_metric": "hit", "channel": "MULL", "ts_data_path": "ts_data/L1_T2_Temporal_Localization_S3_00009.csv" }, { "id": "L1_T4_Sliding_Window_S0_00006", "level": 1, "level_name": "Basic Operations", "category": "Sliding Window", "subtask": "Sliding Window", "question": "Which 12-day period in 2020-12-07 to 2021-04-16 had the lowest average for channel 891? (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2020-12-07 21:30:00, 2020-12-07 21:30:00]", "ground_truth": [ "2020-12-07 21:30:00", "2020-12-07 21:30:00" ], "eval_metric": "iou", "channel": "891", "ts_data_path": "ts_data/L1_T4_Sliding_Window_S0_00006.csv" }, { "id": "L1_T1_Global_Aggregation_S2_00006", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Global Aggregation", "question": "What is the median value of channel 99 in 2023-10-23 to 2023-12-06? (Output format: a single numeric value, rounded to 3 decimal places, e.g., x.xxx)", "answer": "0.72", "ground_truth": 0.72, "eval_metric": "rel_acc", "channel": "99", "ts_data_path": "ts_data/L1_T1_Global_Aggregation_S2_00006.csv" }, { "id": "L1_T2_Temporal_Localization_S2_00007", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Temporal Localization", "question": "At what exact timestamp did channel LUFL reach its minimum value in 2017-11-03 to 2018-02-25? (Output format: a specific timestamp, e.g., YYYY-MM-DD HH:MM:SS)", "answer": "2018-01-12 18:30:00", "ground_truth": "2018-01-12 18:30:00", "eval_metric": "hit", "channel": "LUFL", "ts_data_path": "ts_data/L1_T2_Temporal_Localization_S2_00007.csv" }, { "id": "L2_T2_Periodicity_Detection_S4_00003", "level": 2, "level_name": "Pattern Recognition", "category": "Periodicity Detection", "subtask": "Periodicity Detection", "question": "What is the dominant cycle period (in data points) of channel 154 within [2022-03-20 08:45:00 to 2022-04-03 19:30:00]? (Output format: integer)", "answer": "45", "ground_truth": 45, "eval_metric": "rel_acc", "channel": "154", "ts_data_path": "ts_data/L2_T2_Periodicity_Detection_S4_00003.csv" }, { "id": "L1_T1_Global_Aggregation_S2_00001", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Global Aggregation", "question": "What is the maximum value of channel 99 in 2019-02? (Output format: a single numeric value, rounded to 3 decimal places, e.g., x.xxx)", "answer": "0.135", "ground_truth": 0.135, "eval_metric": "rel_acc", "channel": "99", "ts_data_path": "ts_data/L1_T1_Global_Aggregation_S2_00001.csv" }, { "id": "L2_T1_Shape_Identification_S4_00007", "level": 2, "level_name": "Pattern Recognition", "category": "Shape Identification", "subtask": "Shape Identification", "question": "Identify the time range of the largest step descent in channel MULL within [2018-01-17 06:00:00 to 2018-01-21 05:15:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2018-01-17 11:30:00', '2018-01-18 01:30:00']", "ground_truth": [ "2018-01-17 11:30:00", "2018-01-18 01:30:00" ], "eval_metric": "iou", "channel": "MULL", "ts_data_path": "ts_data/L2_T1_Shape_Identification_S4_00007.csv" }, { "id": "L3_T1_Composite_Trend_S0_00004", "level": 3, "level_name": "Semantic Reasoning", "category": "Composite Trend", "subtask": "Composite Trend", "question": "Among days in channel 430 during 2021 that exhibit the trend pattern 'rapid fall, then slow fall, then rise', identify the top-3 days where the slow fall segment is the slowest. (Output format: a ranked list of dates, e.g., ['YYYY-MM-DD', ...])", "answer": "['2021-10-16', '2021-05-30', '2021-11-07']", "ground_truth": [ "2021-10-16", "2021-05-30", "2021-11-07" ], "eval_metric": "set_f1", "channel": "430", "ts_data_path": "ts_data/L3_T1_Composite_Trend_S0_00004.csv" }, { "id": "L2_T3_Subsequence_Matching_S2_00003", "level": 2, "level_name": "Pattern Recognition", "category": "Subsequence Matching", "subtask": "Subsequence Matching", "question": "Analyze the reference pattern in [2022-06-07 08:00:00 to 2022-06-08 01:45:00]. Find the time interval where channel 811 exhibits the most similar pattern within the search context [2022-06-09 10:00:00 to 2022-06-14 00:30:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2022-06-10 10:30:00', '2022-06-11 04:15:00']", "ground_truth": [ "2022-06-10 10:30:00", "2022-06-11 04:15:00" ], "eval_metric": "iou", "channel": "811", "ts_data_path": "ts_data/L2_T3_Subsequence_Matching_S2_00003.csv" }, { "id": "L4_T1_Insight_Synthesis_S3_00004", "level": 4, "level_name": "Insight Synthesis", "category": "Insight Synthesis", "subtask": "Insight Synthesis", "question": "Analyze the behavior of channel 172 for the period 2022-11.\nPlease use ONLY the following phrases for trend description: rapid rise, gradual rise, rapid fall, gradual fall, steady stable, fluctuating stable.\nProvide a structured report covering:\n1. Trend Segmentation: Describe each stage with precise start/end timestamps (HH:MM:SS) using the phrases above.\n2. Outlier Audit: Identify only significant outliers that deviate sharply from the local trend. Ignore minor background noise.\n(Output format: structured natural language report.)", "answer": "1. Trend Segmentation: from 2022-11-01 00:00:00 to 2022-11-08 12:00:00, the trend showed a rapid rise; from 2022-11-08 12:00:00 to 2022-11-11 12:00:00, the trend showed a fluctuating stable; from 2022-11-11 12:00:00 to 2022-11-17 12:00:00, the trend showed a gradual rise; from 2022-11-17 12:00:00 to 2022-11-26 12:00:00, the trend showed a gradual fall; from 2022-11-26 12:00:00 to 2022-11-30 23:45:00, the trend showed a rapid fall.\n2. Outlier Audit: A significant spike was detected at 2022-11-09 10:15:00 (value: 2209.16).", "ground_truth": { "trend_segments": [ "from 2022-11-01 00:00:00 to 2022-11-08 12:00:00, the trend showed a rapid rise", "from 2022-11-08 12:00:00 to 2022-11-11 12:00:00, the trend showed a fluctuating stable", "from 2022-11-11 12:00:00 to 2022-11-17 12:00:00, the trend showed a gradual rise", "from 2022-11-17 12:00:00 to 2022-11-26 12:00:00, the trend showed a gradual fall", "from 2022-11-26 12:00:00 to 2022-11-30 23:45:00, the trend showed a rapid fall" ], "segments_meta": [ { "adj": "rapid", "kind": "rise", "start_idx": 0, "end_idx": 720 }, { "adj": "fluctuating", "kind": "stable", "start_idx": 720, "end_idx": 1008 }, { "adj": "gradual", "kind": "rise", "start_idx": 1008, "end_idx": 1584 }, { "adj": "gradual", "kind": "fall", "start_idx": 1584, "end_idx": 2448 }, { "adj": "rapid", "kind": "fall", "start_idx": 2448, "end_idx": 2880 } ], "significant_anomaly": { "timestamp": "2022-11-09 10:15:00", "kind": "significant spike", "value": 2209.155516942588 } }, "eval_metric": "report", "channel": "172", "ts_data_path": "ts_data/L4_T1_Insight_Synthesis_S3_00004.csv" }, { "id": "L4_T1_Insight_Synthesis_S4_00009", "level": 4, "level_name": "Insight Synthesis", "category": "Insight Synthesis", "subtask": "Insight Synthesis", "question": "Analyze the behavior of channel 495 for the period 2023-12.\nPlease use ONLY the following phrases for trend description: rapid rise, gradual rise, rapid fall, gradual fall, steady stable, fluctuating stable.\nProvide a structured report covering:\n1. Trend Segmentation: Describe each stage with precise start/end timestamps (HH:MM:SS) using the phrases above.\n2. Outlier Audit: Identify only significant outliers that deviate sharply from the local trend. Ignore minor background noise.\n(Output format: structured natural language report.)", "answer": "1. Trend Segmentation: from 2023-12-01 00:00:00 to 2023-12-11 22:30:00, the trend showed a fluctuating stable; from 2023-12-11 22:30:00 to 2023-12-21 01:15:00, the trend showed a gradual fall; from 2023-12-21 01:15:00 to 2023-12-24 16:45:00, the trend showed a fluctuating stable; from 2023-12-24 16:45:00 to 2023-12-31 23:45:00, the trend showed a gradual rise.\n2. Outlier Audit: A significant drop was detected at 2023-12-23 10:00:00 (value: -13.05).", "ground_truth": { "trend_segments": [ "from 2023-12-01 00:00:00 to 2023-12-11 22:30:00, the trend showed a fluctuating stable", "from 2023-12-11 22:30:00 to 2023-12-21 01:15:00, the trend showed a gradual fall", "from 2023-12-21 01:15:00 to 2023-12-24 16:45:00, the trend showed a fluctuating stable", "from 2023-12-24 16:45:00 to 2023-12-31 23:45:00, the trend showed a gradual rise" ], "segments_meta": [ { "adj": "fluctuating", "kind": "stable", "start_idx": 0, "end_idx": 1050 }, { "adj": "gradual", "kind": "fall", "start_idx": 1050, "end_idx": 1925 }, { "adj": "fluctuating", "kind": "stable", "start_idx": 1925, "end_idx": 2275 }, { "adj": "gradual", "kind": "rise", "start_idx": 2275, "end_idx": 2976 } ], "significant_anomaly": { "timestamp": "2023-12-23 10:00:00", "kind": "significant drop", "value": -13.054362714854749 } }, "eval_metric": "report", "channel": "495", "ts_data_path": "ts_data/L4_T1_Insight_Synthesis_S4_00009.csv" }, { "id": "L3_T3_Causal_Anomaly_S2_00006", "level": 3, "level_name": "Semantic Reasoning", "category": "Causal Anomaly", "subtask": "Causal Anomaly", "question": "Given that channel 495_up_2021 is the upstream source of channel 495_down_2021, identify the time period in 2021 where 495_down_2021 shows a significant causal anomaly, such as an flat line during high activity. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2021-04-01 03:15:00, 2021-04-02 22:15:00]", "ground_truth": [ "2021-04-01 03:15:00", "2021-04-02 22:15:00" ], "eval_metric": "iou", "channel": "495_down_2021", "ts_data_path": "ts_data/L3_T3_Causal_Anomaly_S2_00006.csv" }, { "id": "L1_T2_Temporal_Localization_S4_00007", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Temporal Localization", "question": "At what exact timestamp did channel 151 reach its minimum value in 2022-04-05 to 2022-06-20? (Output format: a specific timestamp, e.g., YYYY-MM-DD HH:MM:SS)", "answer": "2022-06-19 20:15:00", "ground_truth": "2022-06-19 20:15:00", "eval_metric": "hit", "channel": "151", "ts_data_path": "ts_data/L1_T2_Temporal_Localization_S4_00007.csv" }, { "id": "L4_T1_Insight_Synthesis_S1_00003", "level": 4, "level_name": "Insight Synthesis", "category": "Insight Synthesis", "subtask": "Insight Synthesis", "question": "Analyze the behavior of channel 501 for the period 2022-05.\nPlease use ONLY the following phrases for trend description: rapid rise, gradual rise, rapid fall, gradual fall, steady stable, fluctuating stable.\nProvide a structured report covering:\n1. Trend Segmentation: Describe each stage with precise start/end timestamps (HH:MM:SS) using the phrases above.\n2. Outlier Audit: Identify only significant outliers that deviate sharply from the local trend. Ignore minor background noise.\n(Output format: structured natural language report.)", "answer": "1. Trend Segmentation: from 2022-05-01 00:00:00 to 2022-05-08 07:00:00, the trend showed a steady stable; from 2022-05-08 07:00:00 to 2022-05-17 09:45:00, the trend showed a fluctuating stable; from 2022-05-17 09:45:00 to 2022-05-28 08:15:00, the trend showed a steady stable; from 2022-05-28 08:15:00 to 2022-05-31 23:45:00, the trend showed a rapid fall.\n2. Outlier Audit: A significant spike was detected at 2022-05-26 19:30:00 (value: 10.42).", "ground_truth": { "trend_segments": [ "from 2022-05-01 00:00:00 to 2022-05-08 07:00:00, the trend showed a steady stable", "from 2022-05-08 07:00:00 to 2022-05-17 09:45:00, the trend showed a fluctuating stable", "from 2022-05-17 09:45:00 to 2022-05-28 08:15:00, the trend showed a steady stable", "from 2022-05-28 08:15:00 to 2022-05-31 23:45:00, the trend showed a rapid fall" ], "segments_meta": [ { "adj": "steady", "kind": "stable", "start_idx": 0, "end_idx": 700 }, { "adj": "fluctuating", "kind": "stable", "start_idx": 700, "end_idx": 1575 }, { "adj": "steady", "kind": "stable", "start_idx": 1575, "end_idx": 2625 }, { "adj": "rapid", "kind": "fall", "start_idx": 2625, "end_idx": 2976 } ], "significant_anomaly": { "timestamp": "2022-05-26 19:30:00", "kind": "significant spike", "value": 10.418563015195192 } }, "eval_metric": "report", "channel": "501", "ts_data_path": "ts_data/L4_T1_Insight_Synthesis_S1_00003.csv" }, { "id": "L1_T2_Temporal_Localization_S2_00001", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Temporal Localization", "question": "At what exact timestamp did channel 146 first rise above 0.508 in 2021-07-16 to 2021-10-26? (Output format: a specific timestamp, e.g., YYYY-MM-DD HH:MM:SS)", "answer": "2021-07-26 17:45:00", "ground_truth": "2021-07-26 17:45:00", "eval_metric": "hit", "channel": "146", "ts_data_path": "ts_data/L1_T2_Temporal_Localization_S2_00001.csv" }, { "id": "L1_T3_Interval_Discovery_S1_00001", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Interval Discovery", "question": "Find the longest period where channel 123 remained above 0.412 in 2023-05-21 to 2023-08-11. (Output format: a time interval, e.g., [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2023-05-21 18:45:00, 2023-05-28 21:00:00]", "ground_truth": [ "2023-05-21 18:45:00", "2023-05-28 21:00:00" ], "eval_metric": "iou", "channel": "123", "ts_data_path": "ts_data/L1_T3_Interval_Discovery_S1_00001.csv" }, { "id": "L1_T3_Interval_Discovery_S4_00004", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Interval Discovery", "question": "Find the longest period where channel 680 remained above 0.86 in 2022-01-28 to 2022-09-21. (Output format: a time interval, e.g., [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2022-01-28 20:45:00, 2022-03-25 00:30:00]", "ground_truth": [ "2022-01-28 20:45:00", "2022-03-25 00:30:00" ], "eval_metric": "iou", "channel": "680", "ts_data_path": "ts_data/L1_T3_Interval_Discovery_S4_00004.csv" }, { "id": "L1_T3_Interval_Discovery_S4_00002", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Interval Discovery", "question": "Find the longest period where channel 99 remained above 0.368 in 2023-03-22 to 2023-09-13. (Output format: a time interval, e.g., [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2023-08-25 02:15:00, 2023-09-13 10:00:00]", "ground_truth": [ "2023-08-25 02:15:00", "2023-09-13 10:00:00" ], "eval_metric": "iou", "channel": "99", "ts_data_path": "ts_data/L1_T3_Interval_Discovery_S4_00002.csv" }, { "id": "L2_T1_Shape_Identification_S1_00004", "level": 2, "level_name": "Pattern Recognition", "category": "Shape Identification", "subtask": "Shape Identification", "question": "Identify the time range of the longest low plateau (bottom out) in channel 501 within [2021-05-09 15:00:00 to 2021-05-16 19:15:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2021-05-09 21:15:00', '2021-05-10 22:45:00']", "ground_truth": [ "2021-05-09 21:15:00", "2021-05-10 22:45:00" ], "eval_metric": "iou", "channel": "501", "ts_data_path": "ts_data/L2_T1_Shape_Identification_S1_00004.csv" }, { "id": "L1_T3_Interval_Discovery_S2_00008", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Interval Discovery", "question": "Find the longest period where channel MULL remained above 2.097 in 2017-11-13 to 2018-05-29. (Output format: a time interval, e.g., [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2018-03-30 22:15:00, 2018-04-01 07:00:00]", "ground_truth": [ "2018-03-30 22:15:00", "2018-04-01 07:00:00" ], "eval_metric": "iou", "channel": "MULL", "ts_data_path": "ts_data/L1_T3_Interval_Discovery_S2_00008.csv" }, { "id": "L2_T3_Subsequence_Matching_S1_00004", "level": 2, "level_name": "Pattern Recognition", "category": "Subsequence Matching", "subtask": "Subsequence Matching", "question": "Analyze the reference pattern in [2017-05-23 13:30:00 to 2017-05-24 13:15:00]. Find the time interval where channel HULL exhibits the most similar pattern within the search context [2017-05-26 13:00:00 to 2017-06-02 11:45:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2017-05-30 08:30:00', '2017-05-31 08:15:00']", "ground_truth": [ "2017-05-30 08:30:00", "2017-05-31 08:15:00" ], "eval_metric": "iou", "channel": "HULL", "ts_data_path": "ts_data/L2_T3_Subsequence_Matching_S1_00004.csv" }, { "id": "L4_T1_Insight_Synthesis_S2_00003", "level": 4, "level_name": "Insight Synthesis", "category": "Insight Synthesis", "subtask": "Insight Synthesis", "question": "Analyze the behavior of channel 441 for the period 2022-02.\nPlease use ONLY the following phrases for trend description: rapid rise, gradual rise, rapid fall, gradual fall, steady stable, fluctuating stable.\nProvide a structured report covering:\n1. Trend Segmentation: Describe each stage with precise start/end timestamps (HH:MM:SS) using the phrases above.\n2. Outlier Audit: Identify only significant outliers that deviate sharply from the local trend. Ignore minor background noise.\n(Output format: structured natural language report.)", "answer": "1. Trend Segmentation: from 2022-02-01 00:00:00 to 2022-02-16 06:30:00, the trend showed a steady stable; from 2022-02-16 06:30:00 to 2022-02-28 23:45:00, the trend showed a rapid fall.\n2. Outlier Audit: A significant spike was detected at 2022-02-18 21:15:00 (value: 121.20).", "ground_truth": { "trend_segments": [ "from 2022-02-01 00:00:00 to 2022-02-16 06:30:00, the trend showed a steady stable", "from 2022-02-16 06:30:00 to 2022-02-28 23:45:00, the trend showed a rapid fall" ], "segments_meta": [ { "adj": "steady", "kind": "stable", "start_idx": 0, "end_idx": 1466 }, { "adj": "rapid", "kind": "fall", "start_idx": 1466, "end_idx": 2688 } ], "significant_anomaly": { "timestamp": "2022-02-18 21:15:00", "kind": "significant spike", "value": 121.1971153816874 } }, "eval_metric": "report", "channel": "441", "ts_data_path": "ts_data/L4_T1_Insight_Synthesis_S2_00003.csv" }, { "id": "L2_T3_Subsequence_Matching_S0_00006", "level": 2, "level_name": "Pattern Recognition", "category": "Subsequence Matching", "subtask": "Subsequence Matching", "question": "Analyze the reference pattern in [2022-06-28 04:00:00 to 2022-06-28 21:45:00]. Find the time interval where channel 626 exhibits the most similar pattern within the search context [2022-06-30 01:30:00 to 2022-07-03 21:30:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2022-07-01 20:30:00', '2022-07-02 14:15:00']", "ground_truth": [ "2022-07-01 20:30:00", "2022-07-02 14:15:00" ], "eval_metric": "iou", "channel": "626", "ts_data_path": "ts_data/L2_T3_Subsequence_Matching_S0_00006.csv" }, { "id": "L3_T2_Contextual_Anomaly_S0_00002", "level": 3, "level_name": "Semantic Reasoning", "category": "Contextual Anomaly", "subtask": "Contextual Anomaly", "question": "Identify the period in channel 173 during 2022 that experienced the most significant historically high water level. (Output format: [YYYY-MM-DD, YYYY-MM-DD])", "answer": "[2022-05-24 16:00:00, 2022-07-01 15:45:00]", "ground_truth": [ "2022-05-24 16:00:00", "2022-07-01 15:45:00" ], "eval_metric": "iou", "channel": "173", "ts_data_path": "ts_data/L3_T2_Contextual_Anomaly_S0_00002.csv" }, { "id": "L1_T2_Temporal_Localization_S0_00004", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Temporal Localization", "question": "At what exact timestamp did channel 580 last fall below 0.268 in 2021-12? (Output format: a specific timestamp, e.g., YYYY-MM-DD HH:MM:SS)", "answer": "2021-12-23 18:00:00", "ground_truth": "2021-12-23 18:00:00", "eval_metric": "hit", "channel": "580", "ts_data_path": "ts_data/L1_T2_Temporal_Localization_S0_00004.csv" }, { "id": "L3_T1_Composite_Trend_S1_00008", "level": 3, "level_name": "Semantic Reasoning", "category": "Composite Trend", "subtask": "Composite Trend", "question": "Among days in channel 170 during 2019 that exhibit the trend pattern 'fluctuating stable, then slow rise', identify the top-2 days where the slow rise segment is the slowest. (Output format: a ranked list of dates, e.g., ['YYYY-MM-DD', ...])", "answer": "['2019-01-16', '2019-03-14']", "ground_truth": [ "2019-01-16", "2019-03-14" ], "eval_metric": "set_f1", "channel": "170", "ts_data_path": "ts_data/L3_T1_Composite_Trend_S1_00008.csv" }, { "id": "L3_T2_Contextual_Anomaly_S4_00004", "level": 3, "level_name": "Semantic Reasoning", "category": "Contextual Anomaly", "subtask": "Contextual Anomaly", "question": "Identify the period in channel 154 during 2020 that experienced the most significant historically low water level. (Output format: [YYYY-MM-DD, YYYY-MM-DD])", "answer": "[2020-10-10 02:30:00, 2020-10-24 02:15:00]", "ground_truth": [ "2020-10-10 02:30:00", "2020-10-24 02:15:00" ], "eval_metric": "iou", "channel": "154", "ts_data_path": "ts_data/L3_T2_Contextual_Anomaly_S4_00004.csv" }, { "id": "L1_T4_Sliding_Window_S0_00003", "level": 1, "level_name": "Basic Operations", "category": "Sliding Window", "subtask": "Sliding Window", "question": "Which 4-day period in 2020-12-22 to 2021-03-06 had the lowest average for channel 430? (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2021-02-12 21:00:00, 2021-02-16 20:45:00]", "ground_truth": [ "2021-02-12 21:00:00", "2021-02-16 20:45:00" ], "eval_metric": "iou", "channel": "430", "ts_data_path": "ts_data/L1_T4_Sliding_Window_S0_00003.csv" }, { "id": "L3_T1_Composite_Trend_S4_00008", "level": 3, "level_name": "Semantic Reasoning", "category": "Composite Trend", "subtask": "Composite Trend", "question": "Among days in channel 155 during 2022 that exhibit the trend pattern 'steady stable, then rise, then rapid fall', identify the top-8 days where the rapid fall segment is the fastest. (Output format: a ranked list of dates, e.g., ['YYYY-MM-DD', ...])", "answer": "['2022-05-20', '2022-10-24', '2022-07-06', '2022-09-16', '2022-01-07', '2022-02-26', '2022-04-22', '2022-03-06']", "ground_truth": [ "2022-05-20", "2022-10-24", "2022-07-06", "2022-09-16", "2022-01-07", "2022-02-26", "2022-04-22", "2022-03-06" ], "eval_metric": "set_f1", "channel": "155", "ts_data_path": "ts_data/L3_T1_Composite_Trend_S4_00008.csv" }, { "id": "L1_T3_Interval_Discovery_S0_00004", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Interval Discovery", "question": "Find the longest period where channel 626 remained above 0.396 in 2022-10. (Output format: a time interval, e.g., [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2022-10-20 05:00:00, 2022-10-28 12:45:00]", "ground_truth": [ "2022-10-20 05:00:00", "2022-10-28 12:45:00" ], "eval_metric": "iou", "channel": "626", "ts_data_path": "ts_data/L1_T3_Interval_Discovery_S0_00004.csv" }, { "id": "L4_T1_Insight_Synthesis_S3_00005", "level": 4, "level_name": "Insight Synthesis", "category": "Insight Synthesis", "subtask": "Insight Synthesis", "question": "Analyze the behavior of channel HULL for the period 2017-05.\nPlease use ONLY the following phrases for trend description: rapid rise, gradual rise, rapid fall, gradual fall, steady stable, fluctuating stable.\nProvide a structured report covering:\n1. Trend Segmentation: Describe each stage with precise start/end timestamps (HH:MM:SS) using the phrases above.\n2. Outlier Audit: Identify only significant outliers that deviate sharply from the local trend. Ignore minor background noise.\n(Output format: structured natural language report.)", "answer": "1. Trend Segmentation: from 2017-05-01 00:00:00 to 2017-05-03 22:45:00, the trend showed a rapid fall; from 2017-05-03 22:45:00 to 2017-05-06 21:30:00, the trend showed a steady stable; from 2017-05-06 21:30:00 to 2017-05-14 06:45:00, the trend showed a fluctuating stable; from 2017-05-14 06:45:00 to 2017-05-20 04:30:00, the trend showed a steady stable; from 2017-05-20 04:30:00 to 2017-05-24 14:45:00, the trend showed a rapid fall; from 2017-05-24 14:45:00 to 2017-05-31 23:45:00, the trend showed a gradual rise.\n2. Outlier Audit: A significant drop was detected at 2017-05-09 12:30:00 (value: -26.36).", "ground_truth": { "trend_segments": [ "from 2017-05-01 00:00:00 to 2017-05-03 22:45:00, the trend showed a rapid fall", "from 2017-05-03 22:45:00 to 2017-05-06 21:30:00, the trend showed a steady stable", "from 2017-05-06 21:30:00 to 2017-05-14 06:45:00, the trend showed a fluctuating stable", "from 2017-05-14 06:45:00 to 2017-05-20 04:30:00, the trend showed a steady stable", "from 2017-05-20 04:30:00 to 2017-05-24 14:45:00, the trend showed a rapid fall", "from 2017-05-24 14:45:00 to 2017-05-31 23:45:00, the trend showed a gradual rise" ], "segments_meta": [ { "adj": "rapid", "kind": "fall", "start_idx": 0, "end_idx": 283 }, { "adj": "steady", "kind": "stable", "start_idx": 283, "end_idx": 566 }, { "adj": "fluctuating", "kind": "stable", "start_idx": 566, "end_idx": 1275 }, { "adj": "steady", "kind": "stable", "start_idx": 1275, "end_idx": 1842 }, { "adj": "rapid", "kind": "fall", "start_idx": 1842, "end_idx": 2267 }, { "adj": "gradual", "kind": "rise", "start_idx": 2267, "end_idx": 2976 } ], "significant_anomaly": { "timestamp": "2017-05-09 12:30:00", "kind": "significant drop", "value": -26.36213293324823 } }, "eval_metric": "report", "channel": "HULL", "ts_data_path": "ts_data/L4_T1_Insight_Synthesis_S3_00005.csv" }, { "id": "L1_T4_Sliding_Window_S3_00004", "level": 1, "level_name": "Basic Operations", "category": "Sliding Window", "subtask": "Sliding Window", "question": "Which 25-day period in 2021-08-30 to 2022-07-26 had the highest average for channel 495? (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2022-02-03 18:00:00, 2022-02-28 17:45:00]", "ground_truth": [ "2022-02-03 18:00:00", "2022-02-28 17:45:00" ], "eval_metric": "iou", "channel": "495", "ts_data_path": "ts_data/L1_T4_Sliding_Window_S3_00004.csv" }, { "id": "L1_T1_Global_Aggregation_S1_00001", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Global Aggregation", "question": "What is the minimum value of channel 895 in 2020-04-18 to 2020-05-05? (Output format: a single numeric value, rounded to 3 decimal places, e.g., x.xxx)", "answer": "1.68", "ground_truth": 1.68, "eval_metric": "rel_acc", "channel": "895", "ts_data_path": "ts_data/L1_T1_Global_Aggregation_S1_00001.csv" }, { "id": "L2_T1_Shape_Identification_S0_00003", "level": 2, "level_name": "Pattern Recognition", "category": "Shape Identification", "subtask": "Shape Identification", "question": "Identify the time range of the longest plateau (stable period) in channel 154 within [2023-07-08 16:00:00 to 2023-07-11 14:45:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2023-07-09 08:15:00', '2023-07-09 18:30:00']", "ground_truth": [ "2023-07-09 08:15:00", "2023-07-09 18:30:00" ], "eval_metric": "iou", "channel": "154", "ts_data_path": "ts_data/L2_T1_Shape_Identification_S0_00003.csv" }, { "id": "L2_T3_Subsequence_Matching_S4_00004", "level": 2, "level_name": "Pattern Recognition", "category": "Subsequence Matching", "subtask": "Subsequence Matching", "question": "Analyze the reference pattern in [2022-07-29 00:00:00 to 2022-07-29 23:45:00]. Find the time interval where channel 894 exhibits the most similar pattern within the search context [2022-07-31 12:00:00 to 2022-08-05 01:15:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2022-08-01 08:15:00', '2022-08-02 08:00:00']", "ground_truth": [ "2022-08-01 08:15:00", "2022-08-02 08:00:00" ], "eval_metric": "iou", "channel": "894", "ts_data_path": "ts_data/L2_T3_Subsequence_Matching_S4_00004.csv" }, { "id": "L1_T2_Temporal_Localization_S1_00008", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Temporal Localization", "question": "At what exact timestamp did channel 99 first rise above 0.724 in 2023-07-18 to 2023-09-19? (Output format: a specific timestamp, e.g., YYYY-MM-DD HH:MM:SS)", "answer": "2023-08-29 04:15:00", "ground_truth": "2023-08-29 04:15:00", "eval_metric": "hit", "channel": "99", "ts_data_path": "ts_data/L1_T2_Temporal_Localization_S1_00008.csv" }, { "id": "L3_T1_Composite_Trend_S0_00003", "level": 3, "level_name": "Semantic Reasoning", "category": "Composite Trend", "subtask": "Composite Trend", "question": "Among days in channel 170 during 2020 that exhibit the trend pattern 'rapid rise, then steady stable, then fluctuating stable', identify the top-3 days where the rapid rise segment is the fastest. (Output format: a ranked list of dates, e.g., ['YYYY-MM-DD', ...])", "answer": "['2020-06-05', '2020-01-25', '2020-11-14']", "ground_truth": [ "2020-06-05", "2020-01-25", "2020-11-14" ], "eval_metric": "set_f1", "channel": "170", "ts_data_path": "ts_data/L3_T1_Composite_Trend_S0_00003.csv" }, { "id": "L2_T2_Periodicity_Detection_S4_00001", "level": 2, "level_name": "Pattern Recognition", "category": "Periodicity Detection", "subtask": "Periodicity Detection", "question": "What is the dominant cycle period (in data points) of channel 67 within [2022-09-12 00:00:00 to 2022-09-16 04:00:00]? (Output format: integer)", "answer": "99", "ground_truth": 99, "eval_metric": "rel_acc", "channel": "67", "ts_data_path": "ts_data/L2_T2_Periodicity_Detection_S4_00001.csv" }, { "id": "L4_T1_Insight_Synthesis_S2_00004", "level": 4, "level_name": "Insight Synthesis", "category": "Insight Synthesis", "subtask": "Insight Synthesis", "question": "Analyze the behavior of channel 727 for the period 2022-08.\nPlease use ONLY the following phrases for trend description: rapid rise, gradual rise, rapid fall, gradual fall, steady stable, fluctuating stable.\nProvide a structured report covering:\n1. Trend Segmentation: Describe each stage with precise start/end timestamps (HH:MM:SS) using the phrases above.\n2. Outlier Audit: Identify only significant outliers that deviate sharply from the local trend. Ignore minor background noise.\n(Output format: structured natural language report.)", "answer": "1. Trend Segmentation: from 2022-08-01 00:00:00 to 2022-08-09 20:30:00, the trend showed a rapid fall; from 2022-08-09 20:30:00 to 2022-08-12 19:15:00, the trend showed a fluctuating stable; from 2022-08-12 19:15:00 to 2022-08-15 18:00:00, the trend showed a gradual rise; from 2022-08-15 18:00:00 to 2022-08-23 03:15:00, the trend showed a gradual fall; from 2022-08-23 03:15:00 to 2022-08-31 23:45:00, the trend showed a rapid rise.\n2. Outlier Audit: A significant spike was detected at 2022-08-15 17:45:00 (value: 486.04).", "ground_truth": { "trend_segments": [ "from 2022-08-01 00:00:00 to 2022-08-09 20:30:00, the trend showed a rapid fall", "from 2022-08-09 20:30:00 to 2022-08-12 19:15:00, the trend showed a fluctuating stable", "from 2022-08-12 19:15:00 to 2022-08-15 18:00:00, the trend showed a gradual rise", "from 2022-08-15 18:00:00 to 2022-08-23 03:15:00, the trend showed a gradual fall", "from 2022-08-23 03:15:00 to 2022-08-31 23:45:00, the trend showed a rapid rise" ], "segments_meta": [ { "adj": "rapid", "kind": "fall", "start_idx": 0, "end_idx": 850 }, { "adj": "fluctuating", "kind": "stable", "start_idx": 850, "end_idx": 1133 }, { "adj": "gradual", "kind": "rise", "start_idx": 1133, "end_idx": 1416 }, { "adj": "gradual", "kind": "fall", "start_idx": 1416, "end_idx": 2125 }, { "adj": "rapid", "kind": "rise", "start_idx": 2125, "end_idx": 2976 } ], "significant_anomaly": { "timestamp": "2022-08-15 17:45:00", "kind": "significant spike", "value": 486.0412234079833 } }, "eval_metric": "report", "channel": "727", "ts_data_path": "ts_data/L4_T1_Insight_Synthesis_S2_00004.csv" }, { "id": "L3_T1_Composite_Trend_S2_00010", "level": 3, "level_name": "Semantic Reasoning", "category": "Composite Trend", "subtask": "Composite Trend", "question": "Among days in channel 124 during 2022 that exhibit the trend pattern 'fall, then rapid rise, then fluctuating stable', identify the top-4 days where the rapid rise segment is the fastest. (Output format: a ranked list of dates, e.g., ['YYYY-MM-DD', ...])", "answer": "['2022-02-16', '2022-04-10', '2022-04-01', '2022-07-21']", "ground_truth": [ "2022-02-16", "2022-04-10", "2022-04-01", "2022-07-21" ], "eval_metric": "set_f1", "channel": "124", "ts_data_path": "ts_data/L3_T1_Composite_Trend_S2_00010.csv" }, { "id": "L4_T1_Insight_Synthesis_S4_00004", "level": 4, "level_name": "Insight Synthesis", "category": "Insight Synthesis", "subtask": "Insight Synthesis", "question": "Analyze the behavior of channel 501 for the period 2021-06.\nPlease use ONLY the following phrases for trend description: rapid rise, gradual rise, rapid fall, gradual fall, steady stable, fluctuating stable.\nProvide a structured report covering:\n1. Trend Segmentation: Describe each stage with precise start/end timestamps (HH:MM:SS) using the phrases above.\n2. Outlier Audit: Identify only significant outliers that deviate sharply from the local trend. Ignore minor background noise.\n(Output format: structured natural language report.)", "answer": "1. Trend Segmentation: from 2021-06-01 00:00:00 to 2021-06-07 22:15:00, the trend showed a rapid rise; from 2021-06-07 22:15:00 to 2021-06-21 18:30:00, the trend showed a gradual rise; from 2021-06-21 18:30:00 to 2021-06-30 23:45:00, the trend showed a gradual fall.\n2. Outlier Audit: A significant spike was detected at 2021-06-05 16:30:00 (value: 11.71).", "ground_truth": { "trend_segments": [ "from 2021-06-01 00:00:00 to 2021-06-07 22:15:00, the trend showed a rapid rise", "from 2021-06-07 22:15:00 to 2021-06-21 18:30:00, the trend showed a gradual rise", "from 2021-06-21 18:30:00 to 2021-06-30 23:45:00, the trend showed a gradual fall" ], "segments_meta": [ { "adj": "rapid", "kind": "rise", "start_idx": 0, "end_idx": 665 }, { "adj": "gradual", "kind": "rise", "start_idx": 665, "end_idx": 1994 }, { "adj": "gradual", "kind": "fall", "start_idx": 1994, "end_idx": 2880 } ], "significant_anomaly": { "timestamp": "2021-06-05 16:30:00", "kind": "significant spike", "value": 11.712324557130762 } }, "eval_metric": "report", "channel": "501", "ts_data_path": "ts_data/L4_T1_Insight_Synthesis_S4_00004.csv" }, { "id": "L3_T3_Causal_Anomaly_S2_00001", "level": 3, "level_name": "Semantic Reasoning", "category": "Causal Anomaly", "subtask": "Causal Anomaly", "question": "Given that channel 173_up_2021 is the upstream source of channel 173_down_2021, identify the time period in 2021 where 173_down_2021 shows a significant causal anomaly, such as an inverse trend against the source. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2021-09-15 02:45:00, 2021-09-16 02:30:00]", "ground_truth": [ "2021-09-15 02:45:00", "2021-09-16 02:30:00" ], "eval_metric": "iou", "channel": "173_down_2021", "ts_data_path": "ts_data/L3_T3_Causal_Anomaly_S2_00001.csv" }, { "id": "L2_T3_Subsequence_Matching_S3_00004", "level": 2, "level_name": "Pattern Recognition", "category": "Subsequence Matching", "subtask": "Subsequence Matching", "question": "Analyze the reference pattern in [2022-12-03 16:45:00 to 2022-12-04 10:30:00]. Find the time interval where channel 430 exhibits the most similar pattern within the search context [2022-12-05 13:45:00 to 2022-12-07 20:00:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2022-12-06 00:15:00', '2022-12-06 18:00:00']", "ground_truth": [ "2022-12-06 00:15:00", "2022-12-06 18:00:00" ], "eval_metric": "iou", "channel": "430", "ts_data_path": "ts_data/L2_T3_Subsequence_Matching_S3_00004.csv" }, { "id": "L2_T1_Shape_Identification_S0_00004", "level": 2, "level_name": "Pattern Recognition", "category": "Shape Identification", "subtask": "Shape Identification", "question": "Identify the time range of the longest low plateau (bottom out) in channel 680 within [2020-04-13 13:30:00 to 2020-04-20 03:00:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2020-04-17 00:15:00', '2020-04-17 23:30:00']", "ground_truth": [ "2020-04-17 00:15:00", "2020-04-17 23:30:00" ], "eval_metric": "iou", "channel": "680", "ts_data_path": "ts_data/L2_T1_Shape_Identification_S0_00004.csv" }, { "id": "L3_T3_Causal_Anomaly_S4_00008", "level": 3, "level_name": "Semantic Reasoning", "category": "Causal Anomaly", "subtask": "Causal Anomaly", "question": "Given that channel 495_up_2023 is the upstream source of channel 495_down_2023, identify the time period in 2023 where 495_down_2023 shows a significant causal anomaly, such as an flat line during high activity. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2023-08-19 15:45:00, 2023-09-04 12:00:00]", "ground_truth": [ "2023-08-19 15:45:00", "2023-09-04 12:00:00" ], "eval_metric": "iou", "channel": "495_down_2023", "ts_data_path": "ts_data/L3_T3_Causal_Anomaly_S4_00008.csv" }, { "id": "L3_T2_Contextual_Anomaly_S1_00005", "level": 3, "level_name": "Semantic Reasoning", "category": "Contextual Anomaly", "subtask": "Contextual Anomaly", "question": "Identify the period in channel 762 during 2023 that experienced the most significant extreme surge in flow. (Output format: [YYYY-MM-DD, YYYY-MM-DD])", "answer": "[2023-05-23 02:00:00, 2023-06-24 01:45:00]", "ground_truth": [ "2023-05-23 02:00:00", "2023-06-24 01:45:00" ], "eval_metric": "iou", "channel": "762", "ts_data_path": "ts_data/L3_T2_Contextual_Anomaly_S1_00005.csv" }, { "id": "L1_T3_Interval_Discovery_S0_00007", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Interval Discovery", "question": "Find the longest period where channel 897 remained above 0.198 in 2023-01-14 to 2023-04-23. (Output format: a time interval, e.g., [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2023-03-13 05:15:00, 2023-03-29 19:45:00]", "ground_truth": [ "2023-03-13 05:15:00", "2023-03-29 19:45:00" ], "eval_metric": "iou", "channel": "897", "ts_data_path": "ts_data/L1_T3_Interval_Discovery_S0_00007.csv" }, { "id": "L1_T4_Sliding_Window_S1_00002", "level": 1, "level_name": "Basic Operations", "category": "Sliding Window", "subtask": "Sliding Window", "question": "Which 16-day period in 2019-03-01 to 2019-10-10 had the highest variance for channel 762? (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2019-03-03 02:00:00, 2019-03-19 01:45:00]", "ground_truth": [ "2019-03-03 02:00:00", "2019-03-19 01:45:00" ], "eval_metric": "iou", "channel": "762", "ts_data_path": "ts_data/L1_T4_Sliding_Window_S1_00002.csv" }, { "id": "L3_T3_Causal_Anomaly_S0_00004", "level": 3, "level_name": "Semantic Reasoning", "category": "Causal Anomaly", "subtask": "Causal Anomaly", "question": "Given that channel 865_up_2019 is the upstream source of channel 865_down_2019, identify the time period in 2019 where 865_down_2019 shows a significant causal anomaly, such as an inverse trend against the source. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2019-02-05 22:00:00, 2019-02-08 15:15:00]", "ground_truth": [ "2019-02-05 22:00:00", "2019-02-08 15:15:00" ], "eval_metric": "iou", "channel": "865_down_2019", "ts_data_path": "ts_data/L3_T3_Causal_Anomaly_S0_00004.csv" }, { "id": "L1_T1_Global_Aggregation_S3_00003", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Global Aggregation", "question": "What is the median value of channel 169 in 2022-06-23 to 2022-08-06? (Output format: a single numeric value, rounded to 3 decimal places, e.g., x.xxx)", "answer": "202.0", "ground_truth": 202.0, "eval_metric": "rel_acc", "channel": "169", "ts_data_path": "ts_data/L1_T1_Global_Aggregation_S3_00003.csv" }, { "id": "L1_T4_Sliding_Window_S2_00009", "level": 1, "level_name": "Basic Operations", "category": "Sliding Window", "subtask": "Sliding Window", "question": "Which 28-day period in 2022-04-13 to 2022-11-01 had the largest range for channel 865? (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2022-09-29 15:00:00, 2022-10-27 14:45:00]", "ground_truth": [ "2022-09-29 15:00:00", "2022-10-27 14:45:00" ], "eval_metric": "iou", "channel": "865", "ts_data_path": "ts_data/L1_T4_Sliding_Window_S2_00009.csv" }, { "id": "L1_T3_Interval_Discovery_S4_00007", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Interval Discovery", "question": "Find the longest period where channel MULL remained above 1.031 in 2017-04-14 to 2017-11-30. (Output format: a time interval, e.g., [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2017-08-20 07:00:00, 2017-08-29 00:00:00]", "ground_truth": [ "2017-08-20 07:00:00", "2017-08-29 00:00:00" ], "eval_metric": "iou", "channel": "MULL", "ts_data_path": "ts_data/L1_T3_Interval_Discovery_S4_00007.csv" }, { "id": "L1_T2_Temporal_Localization_S4_00003", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Temporal Localization", "question": "At what exact timestamp did channel 626 reach its maximum value in 2021-12-07 to 2022-01-31? (Output format: a specific timestamp, e.g., YYYY-MM-DD HH:MM:SS)", "answer": "2021-12-25 09:30:00", "ground_truth": "2021-12-25 09:30:00", "eval_metric": "hit", "channel": "626", "ts_data_path": "ts_data/L1_T2_Temporal_Localization_S4_00003.csv" }, { "id": "L3_T2_Contextual_Anomaly_S3_00004", "level": 3, "level_name": "Semantic Reasoning", "category": "Contextual Anomaly", "subtask": "Contextual Anomaly", "question": "Identify the period in channel 496 during 2021 that experienced the most significant dry-out period. (Output format: [YYYY-MM-DD, YYYY-MM-DD])", "answer": "[2021-08-11 05:45:00, 2021-09-17 05:30:00]", "ground_truth": [ "2021-08-11 05:45:00", "2021-09-17 05:30:00" ], "eval_metric": "iou", "channel": "496", "ts_data_path": "ts_data/L3_T2_Contextual_Anomaly_S3_00004.csv" }, { "id": "L3_T2_Contextual_Anomaly_S0_00003", "level": 3, "level_name": "Semantic Reasoning", "category": "Contextual Anomaly", "subtask": "Contextual Anomaly", "question": "Identify the period in channel 245 during 2020 that experienced the most significant severe drought. (Output format: [YYYY-MM-DD, YYYY-MM-DD])", "answer": "[2020-02-05 16:00:00, 2020-02-29 15:45:00]", "ground_truth": [ "2020-02-05 16:00:00", "2020-02-29 15:45:00" ], "eval_metric": "iou", "channel": "245", "ts_data_path": "ts_data/L3_T2_Contextual_Anomaly_S0_00003.csv" }, { "id": "L3_T2_Contextual_Anomaly_S1_00006", "level": 3, "level_name": "Semantic Reasoning", "category": "Contextual Anomaly", "subtask": "Contextual Anomaly", "question": "Identify the period in channel 933 during 2019 that experienced the most significant severe drought. (Output format: [YYYY-MM-DD, YYYY-MM-DD])", "answer": "[2019-07-30 01:30:00, 2019-08-30 01:15:00]", "ground_truth": [ "2019-07-30 01:30:00", "2019-08-30 01:15:00" ], "eval_metric": "iou", "channel": "933", "ts_data_path": "ts_data/L3_T2_Contextual_Anomaly_S1_00006.csv" }, { "id": "L2_T3_Subsequence_Matching_S3_00005", "level": 2, "level_name": "Pattern Recognition", "category": "Subsequence Matching", "subtask": "Subsequence Matching", "question": "Analyze the reference pattern in [2023-10-06 13:00:00 to 2023-10-07 06:45:00]. Find the time interval where channel 496 exhibits the most similar pattern within the search context [2023-10-09 00:30:00 to 2023-10-15 04:45:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2023-10-14 09:00:00', '2023-10-15 02:45:00']", "ground_truth": [ "2023-10-14 09:00:00", "2023-10-15 02:45:00" ], "eval_metric": "iou", "channel": "496", "ts_data_path": "ts_data/L2_T3_Subsequence_Matching_S3_00005.csv" }, { "id": "L1_T2_Temporal_Localization_S1_00007", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Temporal Localization", "question": "At what exact timestamp did channel 811 reach its maximum value in 2023-06? (Output format: a specific timestamp, e.g., YYYY-MM-DD HH:MM:SS)", "answer": "2023-06-24 12:00:00", "ground_truth": "2023-06-24 12:00:00", "eval_metric": "hit", "channel": "811", "ts_data_path": "ts_data/L1_T2_Temporal_Localization_S1_00007.csv" }, { "id": "L3_T2_Contextual_Anomaly_S1_00004", "level": 3, "level_name": "Semantic Reasoning", "category": "Contextual Anomaly", "subtask": "Contextual Anomaly", "question": "Identify the period in channel 627 during 2019 that experienced the most significant extreme surge in flow. (Output format: [YYYY-MM-DD, YYYY-MM-DD])", "answer": "[2019-07-15 13:45:00, 2019-08-13 13:30:00]", "ground_truth": [ "2019-07-15 13:45:00", "2019-08-13 13:30:00" ], "eval_metric": "iou", "channel": "627", "ts_data_path": "ts_data/L3_T2_Contextual_Anomaly_S1_00004.csv" }, { "id": "L3_T1_Composite_Trend_S2_00006", "level": 3, "level_name": "Semantic Reasoning", "category": "Composite Trend", "subtask": "Composite Trend", "question": "Among days in channel 647 during 2020 that exhibit the trend pattern 'slow fall, then steady stable, then fall', identify the top-4 days where the slow fall segment is the slowest. (Output format: a ranked list of dates, e.g., ['YYYY-MM-DD', ...])", "answer": "['2020-05-14', '2020-10-13', '2020-04-17', '2020-12-13']", "ground_truth": [ "2020-05-14", "2020-10-13", "2020-04-17", "2020-12-13" ], "eval_metric": "set_f1", "channel": "647", "ts_data_path": "ts_data/L3_T1_Composite_Trend_S2_00006.csv" }, { "id": "L1_T3_Interval_Discovery_S3_00002", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Interval Discovery", "question": "Find the longest period where channel 172 remained above 507.0 in 2020-11. (Output format: a time interval, e.g., [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2020-11-06 21:30:00, 2020-11-10 17:45:00]", "ground_truth": [ "2020-11-06 21:30:00", "2020-11-10 17:45:00" ], "eval_metric": "iou", "channel": "172", "ts_data_path": "ts_data/L1_T3_Interval_Discovery_S3_00002.csv" }, { "id": "L1_T2_Temporal_Localization_S2_00004", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Temporal Localization", "question": "At what exact timestamp did channel 430 last fall below 0.051 in 2022-06-22 to 2022-08-24? (Output format: a specific timestamp, e.g., YYYY-MM-DD HH:MM:SS)", "answer": "2022-08-17 10:30:00", "ground_truth": "2022-08-17 10:30:00", "eval_metric": "hit", "channel": "430", "ts_data_path": "ts_data/L1_T2_Temporal_Localization_S2_00004.csv" }, { "id": "L1_T3_Interval_Discovery_S1_00002", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Interval Discovery", "question": "Find the longest period where channel 172 remained above 306.0 in 2019-05-02 to 2019-06-20. (Output format: a time interval, e.g., [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2019-05-31 14:15:00, 2019-06-03 18:15:00]", "ground_truth": [ "2019-05-31 14:15:00", "2019-06-03 18:15:00" ], "eval_metric": "iou", "channel": "172", "ts_data_path": "ts_data/L1_T3_Interval_Discovery_S1_00002.csv" }, { "id": "L2_T1_Shape_Identification_S0_00002", "level": 2, "level_name": "Pattern Recognition", "category": "Shape Identification", "subtask": "Shape Identification", "question": "Identify the time range of the longest plateau (stable period) in channel 147 within [2020-09-29 00:00:00 to 2020-10-06 16:15:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2020-10-01 16:15:00', '2020-10-02 19:30:00']", "ground_truth": [ "2020-10-01 16:15:00", "2020-10-02 19:30:00" ], "eval_metric": "iou", "channel": "147", "ts_data_path": "ts_data/L2_T1_Shape_Identification_S0_00002.csv" }, { "id": "L2_T2_Periodicity_Detection_S2_00008", "level": 2, "level_name": "Pattern Recognition", "category": "Periodicity Detection", "subtask": "Periodicity Detection", "question": "What is the dominant cycle period (in data points) of channel 169 within [2020-12-17 09:45:00 to 2020-12-22 21:15:00]? (Output format: integer)", "answer": "127", "ground_truth": 127, "eval_metric": "rel_acc", "channel": "169", "ts_data_path": "ts_data/L2_T2_Periodicity_Detection_S2_00008.csv" }, { "id": "L2_T3_Subsequence_Matching_S4_00009", "level": 2, "level_name": "Pattern Recognition", "category": "Subsequence Matching", "subtask": "Subsequence Matching", "question": "Analyze the reference pattern in [2022-10-12 05:15:00 to 2022-10-12 23:00:00]. Find the time interval where channel 155 exhibits the most similar pattern within the search context [2022-10-14 02:15:00 to 2022-10-16 18:15:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2022-10-14 14:30:00', '2022-10-15 08:15:00']", "ground_truth": [ "2022-10-14 14:30:00", "2022-10-15 08:15:00" ], "eval_metric": "iou", "channel": "155", "ts_data_path": "ts_data/L2_T3_Subsequence_Matching_S4_00009.csv" }, { "id": "L1_T1_Global_Aggregation_S4_00001", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Global Aggregation", "question": "What is the median value of channel 170 in 2019-01-27 to 2019-02-09? (Output format: a single numeric value, rounded to 3 decimal places, e.g., x.xxx)", "answer": "498.0", "ground_truth": 498.0, "eval_metric": "rel_acc", "channel": "170", "ts_data_path": "ts_data/L1_T1_Global_Aggregation_S4_00001.csv" }, { "id": "L1_T2_Temporal_Localization_S1_00002", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Temporal Localization", "question": "At what exact timestamp did channel 147 reach its minimum value in 2023-02-25 to 2023-03-21? (Output format: a specific timestamp, e.g., YYYY-MM-DD HH:MM:SS)", "answer": "2023-03-04 20:45:00", "ground_truth": "2023-03-04 20:45:00", "eval_metric": "hit", "channel": "147", "ts_data_path": "ts_data/L1_T2_Temporal_Localization_S1_00002.csv" }, { "id": "L1_T1_Global_Aggregation_S2_00004", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Global Aggregation", "question": "What is the minimum value of channel 897 in 2022-12-10 to 2022-12-25? (Output format: a single numeric value, rounded to 3 decimal places, e.g., x.xxx)", "answer": "0.026", "ground_truth": 0.026, "eval_metric": "rel_acc", "channel": "897", "ts_data_path": "ts_data/L1_T1_Global_Aggregation_S2_00004.csv" }, { "id": "L1_T3_Interval_Discovery_S1_00008", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Interval Discovery", "question": "Find the longest period where channel OT remained above 12.17 in 2017-01. (Output format: a time interval, e.g., [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2017-01-01 12:15:00, 2017-01-02 03:15:00]", "ground_truth": [ "2017-01-01 12:15:00", "2017-01-02 03:15:00" ], "eval_metric": "iou", "channel": "OT", "ts_data_path": "ts_data/L1_T3_Interval_Discovery_S1_00008.csv" }, { "id": "L1_T4_Sliding_Window_S1_00005", "level": 1, "level_name": "Basic Operations", "category": "Sliding Window", "subtask": "Sliding Window", "question": "Which 20-day period in 2022-04-05 to 2022-10-27 had the highest variance for channel 865? (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2022-10-07 03:00:00, 2022-10-27 02:45:00]", "ground_truth": [ "2022-10-07 03:00:00", "2022-10-27 02:45:00" ], "eval_metric": "iou", "channel": "865", "ts_data_path": "ts_data/L1_T4_Sliding_Window_S1_00005.csv" }, { "id": "L4_T1_Insight_Synthesis_S2_00009", "level": 4, "level_name": "Insight Synthesis", "category": "Insight Synthesis", "subtask": "Insight Synthesis", "question": "Analyze the behavior of channel 166 for the period 2021-08.\nPlease use ONLY the following phrases for trend description: rapid rise, gradual rise, rapid fall, gradual fall, steady stable, fluctuating stable.\nProvide a structured report covering:\n1. Trend Segmentation: Describe each stage with precise start/end timestamps (HH:MM:SS) using the phrases above.\n2. Outlier Audit: Identify only significant outliers that deviate sharply from the local trend. Ignore minor background noise.\n(Output format: structured natural language report.)", "answer": "1. Trend Segmentation: from 2021-08-01 00:00:00 to 2021-08-18 05:15:00, the trend showed a rapid rise; from 2021-08-18 05:15:00 to 2021-08-31 23:45:00, the trend showed a steady stable.\n2. Outlier Audit: A significant spike was detected at 2021-08-25 23:15:00 (value: 3845.34).", "ground_truth": { "trend_segments": [ "from 2021-08-01 00:00:00 to 2021-08-18 05:15:00, the trend showed a rapid rise", "from 2021-08-18 05:15:00 to 2021-08-31 23:45:00, the trend showed a steady stable" ], "segments_meta": [ { "adj": "rapid", "kind": "rise", "start_idx": 0, "end_idx": 1653 }, { "adj": "steady", "kind": "stable", "start_idx": 1653, "end_idx": 2976 } ], "significant_anomaly": { "timestamp": "2021-08-25 23:15:00", "kind": "significant spike", "value": 3845.335832252088 } }, "eval_metric": "report", "channel": "166", "ts_data_path": "ts_data/L4_T1_Insight_Synthesis_S2_00009.csv" }, { "id": "L1_T2_Temporal_Localization_S3_00003", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Temporal Localization", "question": "At what exact timestamp did channel 170 first rise above 486.0 in 2019-04-03 to 2019-04-26? (Output format: a specific timestamp, e.g., YYYY-MM-DD HH:MM:SS)", "answer": "2019-04-03 21:15:00", "ground_truth": "2019-04-03 21:15:00", "eval_metric": "hit", "channel": "170", "ts_data_path": "ts_data/L1_T2_Temporal_Localization_S3_00003.csv" }, { "id": "L1_T3_Interval_Discovery_S1_00003", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Interval Discovery", "question": "Find the longest period where channel 177 remained above 168.0 in 2019-12. (Output format: a time interval, e.g., [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2019-12-25 18:00:00, 2019-12-26 16:30:00]", "ground_truth": [ "2019-12-25 18:00:00", "2019-12-26 16:30:00" ], "eval_metric": "iou", "channel": "177", "ts_data_path": "ts_data/L1_T3_Interval_Discovery_S1_00003.csv" }, { "id": "L1_T1_Global_Aggregation_S3_00009", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Global Aggregation", "question": "What is the minimum value of channel 146 in 2019-08-10 to 2019-09-14? (Output format: a single numeric value, rounded to 3 decimal places, e.g., x.xxx)", "answer": "0.094", "ground_truth": 0.094, "eval_metric": "rel_acc", "channel": "146", "ts_data_path": "ts_data/L1_T1_Global_Aggregation_S3_00009.csv" }, { "id": "L3_T1_Composite_Trend_S1_00009", "level": 3, "level_name": "Semantic Reasoning", "category": "Composite Trend", "subtask": "Composite Trend", "question": "Among days in channel 312 during 2021 that exhibit the trend pattern 'slow fall, then rise, then slow rise, then fall', identify the top-2 days where the slow rise segment is the slowest. (Output format: a ranked list of dates, e.g., ['YYYY-MM-DD', ...])", "answer": "['2021-11-16', '2021-03-19']", "ground_truth": [ "2021-11-16", "2021-03-19" ], "eval_metric": "set_f1", "channel": "312", "ts_data_path": "ts_data/L3_T1_Composite_Trend_S1_00009.csv" }, { "id": "L3_T3_Causal_Anomaly_S2_00002", "level": 3, "level_name": "Semantic Reasoning", "category": "Causal Anomaly", "subtask": "Causal Anomaly", "question": "Given that channel 501_up_2023 is the upstream source of channel 501_down_2023, identify the time period in 2023 where 501_down_2023 shows a significant causal anomaly, such as an inverse trend against the source. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2023-11-30 02:30:00, 2023-12-01 13:15:00]", "ground_truth": [ "2023-11-30 02:30:00", "2023-12-01 13:15:00" ], "eval_metric": "iou", "channel": "501_down_2023", "ts_data_path": "ts_data/L3_T3_Causal_Anomaly_S2_00002.csv" }, { "id": "L1_T2_Temporal_Localization_S3_00004", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Temporal Localization", "question": "At what exact timestamp did channel 237 reach its maximum value in 2019-08? (Output format: a specific timestamp, e.g., YYYY-MM-DD HH:MM:SS)", "answer": "2019-08-05 01:30:00", "ground_truth": "2019-08-05 01:30:00", "eval_metric": "hit", "channel": "237", "ts_data_path": "ts_data/L1_T2_Temporal_Localization_S3_00004.csv" }, { "id": "L3_T1_Composite_Trend_S0_00002", "level": 3, "level_name": "Semantic Reasoning", "category": "Composite Trend", "subtask": "Composite Trend", "question": "Among days in channel 169 during 2020 that exhibit the trend pattern 'fall, then fluctuating stable, then rise', identify the top-3 days where the fluctuating stable segment is the most fluctuating. (Output format: a ranked list of dates, e.g., ['YYYY-MM-DD', ...])", "answer": "['2020-02-12', '2020-07-31', '2020-12-04']", "ground_truth": [ "2020-02-12", "2020-07-31", "2020-12-04" ], "eval_metric": "set_f1", "channel": "169", "ts_data_path": "ts_data/L3_T1_Composite_Trend_S0_00002.csv" }, { "id": "L3_T1_Composite_Trend_S3_00004", "level": 3, "level_name": "Semantic Reasoning", "category": "Composite Trend", "subtask": "Composite Trend", "question": "Among days in channel 245 during 2021 that exhibit the trend pattern 'steady stable, then fluctuating stable', identify the top-6 days where the fluctuating stable segment is the most fluctuating. (Output format: a ranked list of dates, e.g., ['YYYY-MM-DD', ...])", "answer": "['2021-03-08', '2021-09-03', '2021-04-25', '2021-08-04', '2021-08-11', '2021-12-16']", "ground_truth": [ "2021-03-08", "2021-09-03", "2021-04-25", "2021-08-04", "2021-08-11", "2021-12-16" ], "eval_metric": "set_f1", "channel": "245", "ts_data_path": "ts_data/L3_T1_Composite_Trend_S3_00004.csv" }, { "id": "L1_T2_Temporal_Localization_S4_00008", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Temporal Localization", "question": "At what exact timestamp did channel 496 reach its maximum value in 2020-01-01 to 2020-04-21? (Output format: a specific timestamp, e.g., YYYY-MM-DD HH:MM:SS)", "answer": "2020-03-02 10:15:00", "ground_truth": "2020-03-02 10:15:00", "eval_metric": "hit", "channel": "496", "ts_data_path": "ts_data/L1_T2_Temporal_Localization_S4_00008.csv" }, { "id": "L1_T4_Sliding_Window_S1_00008", "level": 1, "level_name": "Basic Operations", "category": "Sliding Window", "subtask": "Sliding Window", "question": "Which 20-day period in 2017-02-07 to 2017-10-01 had the lowest average for channel LUFL? (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2017-09-11 11:15:00, 2017-10-01 11:00:00]", "ground_truth": [ "2017-09-11 11:15:00", "2017-10-01 11:00:00" ], "eval_metric": "iou", "channel": "LUFL", "ts_data_path": "ts_data/L1_T4_Sliding_Window_S1_00008.csv" }, { "id": "L3_T2_Contextual_Anomaly_S3_00002", "level": 3, "level_name": "Semantic Reasoning", "category": "Contextual Anomaly", "subtask": "Contextual Anomaly", "question": "Identify the period in channel 151 during 2022 that experienced the most significant severe drought. (Output format: [YYYY-MM-DD, YYYY-MM-DD])", "answer": "[2022-07-26 01:15:00, 2022-08-24 01:00:00]", "ground_truth": [ "2022-07-26 01:15:00", "2022-08-24 01:00:00" ], "eval_metric": "iou", "channel": "151", "ts_data_path": "ts_data/L3_T2_Contextual_Anomaly_S3_00002.csv" }, { "id": "L1_T2_Temporal_Localization_S4_00005", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Temporal Localization", "question": "At what exact timestamp did channel 897 last fall below 0.003 in 2022-08-03 to 2022-10-19? (Output format: a specific timestamp, e.g., YYYY-MM-DD HH:MM:SS)", "answer": "2022-08-19 11:30:00", "ground_truth": "2022-08-19 11:30:00", "eval_metric": "hit", "channel": "897", "ts_data_path": "ts_data/L1_T2_Temporal_Localization_S4_00005.csv" }, { "id": "L1_T3_Interval_Discovery_S2_00001", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Interval Discovery", "question": "Find the longest period where channel 99 remained above 0.128 in 2019-11-20 to 2020-05-23. (Output format: a time interval, e.g., [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2020-03-05 12:45:00, 2020-03-22 18:00:00]", "ground_truth": [ "2020-03-05 12:45:00", "2020-03-22 18:00:00" ], "eval_metric": "iou", "channel": "99", "ts_data_path": "ts_data/L1_T3_Interval_Discovery_S2_00001.csv" }, { "id": "L4_T1_Insight_Synthesis_S0_00005", "level": 4, "level_name": "Insight Synthesis", "category": "Insight Synthesis", "subtask": "Insight Synthesis", "question": "Analyze the behavior of channel 173 for the period 2019-09.\nPlease use ONLY the following phrases for trend description: rapid rise, gradual rise, rapid fall, gradual fall, steady stable, fluctuating stable.\nProvide a structured report covering:\n1. Trend Segmentation: Describe each stage with precise start/end timestamps (HH:MM:SS) using the phrases above.\n2. Outlier Audit: Identify only significant outliers that deviate sharply from the local trend. Ignore minor background noise.\n(Output format: structured natural language report.)", "answer": "1. Trend Segmentation: from 2019-09-01 00:00:00 to 2019-09-06 05:15:00, the trend showed a rapid rise; from 2019-09-06 05:15:00 to 2019-09-14 01:00:00, the trend showed a gradual rise; from 2019-09-14 01:00:00 to 2019-09-21 20:45:00, the trend showed a rapid rise; from 2019-09-21 20:45:00 to 2019-09-28 09:15:00, the trend showed a fluctuating stable; from 2019-09-28 09:15:00 to 2019-09-30 23:45:00, the trend showed a steady stable.\n2. Outlier Audit: A significant spike was detected at 2019-09-19 03:00:00 (value: 2719.78).", "ground_truth": { "trend_segments": [ "from 2019-09-01 00:00:00 to 2019-09-06 05:15:00, the trend showed a rapid rise", "from 2019-09-06 05:15:00 to 2019-09-14 01:00:00, the trend showed a gradual rise", "from 2019-09-14 01:00:00 to 2019-09-21 20:45:00, the trend showed a rapid rise", "from 2019-09-21 20:45:00 to 2019-09-28 09:15:00, the trend showed a fluctuating stable", "from 2019-09-28 09:15:00 to 2019-09-30 23:45:00, the trend showed a steady stable" ], "segments_meta": [ { "adj": "rapid", "kind": "rise", "start_idx": 0, "end_idx": 501 }, { "adj": "gradual", "kind": "rise", "start_idx": 501, "end_idx": 1252 }, { "adj": "rapid", "kind": "rise", "start_idx": 1252, "end_idx": 2003 }, { "adj": "fluctuating", "kind": "stable", "start_idx": 2003, "end_idx": 2629 }, { "adj": "steady", "kind": "stable", "start_idx": 2629, "end_idx": 2880 } ], "significant_anomaly": { "timestamp": "2019-09-19 03:00:00", "kind": "significant spike", "value": 2719.7761530018133 } }, "eval_metric": "report", "channel": "173", "ts_data_path": "ts_data/L4_T1_Insight_Synthesis_S0_00005.csv" }, { "id": "L3_T2_Contextual_Anomaly_S0_00008", "level": 3, "level_name": "Semantic Reasoning", "category": "Contextual Anomaly", "subtask": "Contextual Anomaly", "question": "Identify the period in channel 67 during 2023 that experienced the most significant dry-out period. (Output format: [YYYY-MM-DD, YYYY-MM-DD])", "answer": "[2023-09-03 02:00:00, 2023-10-14 01:45:00]", "ground_truth": [ "2023-09-03 02:00:00", "2023-10-14 01:45:00" ], "eval_metric": "iou", "channel": "67", "ts_data_path": "ts_data/L3_T2_Contextual_Anomaly_S0_00008.csv" }, { "id": "L1_T2_Temporal_Localization_S0_00008", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Temporal Localization", "question": "At what exact timestamp did channel LULL reach its minimum value in 2018-01? (Output format: a specific timestamp, e.g., YYYY-MM-DD HH:MM:SS)", "answer": "2018-01-12 17:00:00", "ground_truth": "2018-01-12 17:00:00", "eval_metric": "hit", "channel": "LULL", "ts_data_path": "ts_data/L1_T2_Temporal_Localization_S0_00008.csv" }, { "id": "L4_T1_Insight_Synthesis_S4_00005", "level": 4, "level_name": "Insight Synthesis", "category": "Insight Synthesis", "subtask": "Insight Synthesis", "question": "Analyze the behavior of channel 625 for the period 2021-09.\nPlease use ONLY the following phrases for trend description: rapid rise, gradual rise, rapid fall, gradual fall, steady stable, fluctuating stable.\nProvide a structured report covering:\n1. Trend Segmentation: Describe each stage with precise start/end timestamps (HH:MM:SS) using the phrases above.\n2. Outlier Audit: Identify only significant outliers that deviate sharply from the local trend. Ignore minor background noise.\n(Output format: structured natural language report.)", "answer": "1. Trend Segmentation: from 2021-09-01 00:00:00 to 2021-09-09 08:00:00, the trend showed a steady stable; from 2021-09-09 08:00:00 to 2021-09-12 16:00:00, the trend showed a rapid fall; from 2021-09-12 16:00:00 to 2021-09-22 16:00:00, the trend showed a fluctuating stable; from 2021-09-22 16:00:00 to 2021-09-30 23:45:00, the trend showed a gradual rise.\n2. Outlier Audit: A significant spike was detected at 2021-09-22 19:30:00 (value: 5.69).", "ground_truth": { "trend_segments": [ "from 2021-09-01 00:00:00 to 2021-09-09 08:00:00, the trend showed a steady stable", "from 2021-09-09 08:00:00 to 2021-09-12 16:00:00, the trend showed a rapid fall", "from 2021-09-12 16:00:00 to 2021-09-22 16:00:00, the trend showed a fluctuating stable", "from 2021-09-22 16:00:00 to 2021-09-30 23:45:00, the trend showed a gradual rise" ], "segments_meta": [ { "adj": "steady", "kind": "stable", "start_idx": 0, "end_idx": 800 }, { "adj": "rapid", "kind": "fall", "start_idx": 800, "end_idx": 1120 }, { "adj": "fluctuating", "kind": "stable", "start_idx": 1120, "end_idx": 2080 }, { "adj": "gradual", "kind": "rise", "start_idx": 2080, "end_idx": 2880 } ], "significant_anomaly": { "timestamp": "2021-09-22 19:30:00", "kind": "significant spike", "value": 5.69488443396203 } }, "eval_metric": "report", "channel": "625", "ts_data_path": "ts_data/L4_T1_Insight_Synthesis_S4_00005.csv" }, { "id": "L3_T3_Causal_Anomaly_S3_00003", "level": 3, "level_name": "Semantic Reasoning", "category": "Causal Anomaly", "subtask": "Causal Anomaly", "question": "Given that channel 170_up_2021 is the upstream source of channel 170_down_2021, identify the time period in 2021 where 170_down_2021 shows a significant causal anomaly, such as an flat line during high activity. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2021-03-28 07:45:00, 2021-04-01 16:15:00]", "ground_truth": [ "2021-03-28 07:45:00", "2021-04-01 16:15:00" ], "eval_metric": "iou", "channel": "170_down_2021", "ts_data_path": "ts_data/L3_T3_Causal_Anomaly_S3_00003.csv" }, { "id": "L1_T1_Global_Aggregation_S2_00002", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Global Aggregation", "question": "What is the minimum value of channel 578 in 2019-07-07 to 2019-07-24? (Output format: a single numeric value, rounded to 3 decimal places, e.g., x.xxx)", "answer": "0.063", "ground_truth": 0.063, "eval_metric": "rel_acc", "channel": "578", "ts_data_path": "ts_data/L1_T1_Global_Aggregation_S2_00002.csv" }, { "id": "L1_T4_Sliding_Window_S1_00004", "level": 1, "level_name": "Basic Operations", "category": "Sliding Window", "subtask": "Sliding Window", "question": "Which 7-day period in 2022-01-04 to 2022-07-26 had the lowest average for channel 813? (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2022-06-22 13:30:00, 2022-06-29 13:15:00]", "ground_truth": [ "2022-06-22 13:30:00", "2022-06-29 13:15:00" ], "eval_metric": "iou", "channel": "813", "ts_data_path": "ts_data/L1_T4_Sliding_Window_S1_00004.csv" }, { "id": "L2_T2_Periodicity_Detection_S2_00005", "level": 2, "level_name": "Pattern Recognition", "category": "Periodicity Detection", "subtask": "Periodicity Detection", "question": "What is the dominant cycle period (in data points) of channel 627 within [2022-11-13 00:00:00 to 2022-11-23 00:45:00]? (Output format: integer)", "answer": "27", "ground_truth": 27, "eval_metric": "rel_acc", "channel": "627", "ts_data_path": "ts_data/L2_T2_Periodicity_Detection_S2_00005.csv" }, { "id": "L3_T3_Causal_Anomaly_S3_00008", "level": 3, "level_name": "Semantic Reasoning", "category": "Causal Anomaly", "subtask": "Causal Anomaly", "question": "Given that channel 99_up_2023 is the upstream source of channel 99_down_2023, identify the time period in 2023 where 99_down_2023 shows a significant causal anomaly, such as an flat line during high activity. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2023-05-10 18:45:00, 2023-05-13 10:15:00]", "ground_truth": [ "2023-05-10 18:45:00", "2023-05-13 10:15:00" ], "eval_metric": "iou", "channel": "99_down_2023", "ts_data_path": "ts_data/L3_T3_Causal_Anomaly_S3_00008.csv" }, { "id": "L1_T2_Temporal_Localization_S3_00008", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Temporal Localization", "question": "At what exact timestamp did channel HUFL first rise above 10.114 in 2016-10? (Output format: a specific timestamp, e.g., YYYY-MM-DD HH:MM:SS)", "answer": "2016-10-01 06:15:00", "ground_truth": "2016-10-01 06:15:00", "eval_metric": "hit", "channel": "HUFL", "ts_data_path": "ts_data/L1_T2_Temporal_Localization_S3_00008.csv" }, { "id": "L1_T1_Global_Aggregation_S3_00006", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Global Aggregation", "question": "What is the range value of channel MULL in 2017-11-03 to 2018-01-23? (Output format: a single numeric value, rounded to 3 decimal places, e.g., x.xxx)", "answer": "8.351", "ground_truth": 8.351, "eval_metric": "rel_acc", "channel": "MULL", "ts_data_path": "ts_data/L1_T1_Global_Aggregation_S3_00006.csv" }, { "id": "L2_T3_Subsequence_Matching_S3_00002", "level": 2, "level_name": "Pattern Recognition", "category": "Subsequence Matching", "subtask": "Subsequence Matching", "question": "Analyze the reference pattern in [2020-03-20 22:30:00 to 2020-03-21 22:15:00]. Find the time interval where channel 155 exhibits the most similar pattern within the search context [2020-03-23 21:45:00 to 2020-03-30 19:15:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2020-03-27 11:00:00', '2020-03-28 10:45:00']", "ground_truth": [ "2020-03-27 11:00:00", "2020-03-28 10:45:00" ], "eval_metric": "iou", "channel": "155", "ts_data_path": "ts_data/L2_T3_Subsequence_Matching_S3_00002.csv" }, { "id": "L1_T1_Global_Aggregation_S4_00003", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Global Aggregation", "question": "What is the median value of channel 501 in 2019-07? (Output format: a single numeric value, rounded to 3 decimal places, e.g., x.xxx)", "answer": "1.09", "ground_truth": 1.09, "eval_metric": "rel_acc", "channel": "501", "ts_data_path": "ts_data/L1_T1_Global_Aggregation_S4_00003.csv" }, { "id": "L2_T1_Shape_Identification_S3_00009", "level": 2, "level_name": "Pattern Recognition", "category": "Shape Identification", "subtask": "Shape Identification", "question": "Identify the time range of the highest upward spike in channel 173 within [2020-01-13 22:15:00 to 2020-01-17 17:45:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2020-01-14 23:30:00', '2020-01-15 13:00:00']", "ground_truth": [ "2020-01-14 23:30:00", "2020-01-15 13:00:00" ], "eval_metric": "iou", "channel": "173", "ts_data_path": "ts_data/L2_T1_Shape_Identification_S3_00009.csv" }, { "id": "L3_T1_Composite_Trend_S4_00002", "level": 3, "level_name": "Semantic Reasoning", "category": "Composite Trend", "subtask": "Composite Trend", "question": "Among days in channel 123 during 2021 that exhibit the trend pattern 'rapid fall, then slow rise, then steady stable, then rapid fall', identify the top-8 days where the slow rise segment is the slowest. (Output format: a ranked list of dates, e.g., ['YYYY-MM-DD', ...])", "answer": "['2021-02-11', '2021-12-04', '2021-10-18', '2021-02-26', '2021-06-13', '2021-06-20', '2021-01-29', '2021-08-20']", "ground_truth": [ "2021-02-11", "2021-12-04", "2021-10-18", "2021-02-26", "2021-06-13", "2021-06-20", "2021-01-29", "2021-08-20" ], "eval_metric": "set_f1", "channel": "123", "ts_data_path": "ts_data/L3_T1_Composite_Trend_S4_00002.csv" }, { "id": "L3_T2_Contextual_Anomaly_S2_00009", "level": 3, "level_name": "Semantic Reasoning", "category": "Contextual Anomaly", "subtask": "Contextual Anomaly", "question": "Identify the period in channel 155 during 2020 that experienced the most significant dry-out period. (Output format: [YYYY-MM-DD, YYYY-MM-DD])", "answer": "[2020-08-20 12:15:00, 2020-09-25 12:00:00]", "ground_truth": [ "2020-08-20 12:15:00", "2020-09-25 12:00:00" ], "eval_metric": "iou", "channel": "155", "ts_data_path": "ts_data/L3_T2_Contextual_Anomaly_S2_00009.csv" }, { "id": "L2_T2_Periodicity_Detection_S0_00005", "level": 2, "level_name": "Pattern Recognition", "category": "Periodicity Detection", "subtask": "Periodicity Detection", "question": "What is the dominant cycle period (in data points) of channel 591 within [2022-07-09 15:00:00 to 2022-07-14 22:15:00]? (Output format: integer)", "answer": "107", "ground_truth": 107, "eval_metric": "rel_acc", "channel": "591", "ts_data_path": "ts_data/L2_T2_Periodicity_Detection_S0_00005.csv" }, { "id": "L1_T3_Interval_Discovery_S3_00004", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Interval Discovery", "question": "Find the longest period where channel 496 remained above 0.89 in 2020-01-10 to 2020-03-03. (Output format: a time interval, e.g., [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2020-02-24 19:45:00, 2020-03-03 01:15:00]", "ground_truth": [ "2020-02-24 19:45:00", "2020-03-03 01:15:00" ], "eval_metric": "iou", "channel": "496", "ts_data_path": "ts_data/L1_T3_Interval_Discovery_S3_00004.csv" }, { "id": "L1_T2_Temporal_Localization_S0_00005", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Temporal Localization", "question": "At what exact timestamp did channel 591 reach its minimum value in 2021-09? (Output format: a specific timestamp, e.g., YYYY-MM-DD HH:MM:SS)", "answer": "2021-09-26 09:45:00", "ground_truth": "2021-09-26 09:45:00", "eval_metric": "hit", "channel": "591", "ts_data_path": "ts_data/L1_T2_Temporal_Localization_S0_00005.csv" }, { "id": "L1_T1_Global_Aggregation_S0_00007", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Global Aggregation", "question": "What is the range value of channel MUFL in 2017-07-27 to 2017-09-05? (Output format: a single numeric value, rounded to 3 decimal places, e.g., x.xxx)", "answer": "35.251", "ground_truth": 35.251, "eval_metric": "rel_acc", "channel": "MUFL", "ts_data_path": "ts_data/L1_T1_Global_Aggregation_S0_00007.csv" }, { "id": "L1_T1_Global_Aggregation_S0_00009", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Global Aggregation", "question": "What is the minimum value of channel 67 in 2021-02-01 to 2021-02-17? (Output format: a single numeric value, rounded to 3 decimal places, e.g., x.xxx)", "answer": "0.163", "ground_truth": 0.163, "eval_metric": "rel_acc", "channel": "67", "ts_data_path": "ts_data/L1_T1_Global_Aggregation_S0_00009.csv" }, { "id": "L1_T3_Interval_Discovery_S3_00003", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Interval Discovery", "question": "Find the longest period where channel 177 remained above 203.0 in 2022-05-09 to 2022-08-06. (Output format: a time interval, e.g., [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2022-06-29 20:30:00, 2022-07-05 08:30:00]", "ground_truth": [ "2022-06-29 20:30:00", "2022-07-05 08:30:00" ], "eval_metric": "iou", "channel": "177", "ts_data_path": "ts_data/L1_T3_Interval_Discovery_S3_00003.csv" }, { "id": "L2_T3_Subsequence_Matching_S0_00009", "level": 2, "level_name": "Pattern Recognition", "category": "Subsequence Matching", "subtask": "Subsequence Matching", "question": "Analyze the reference pattern in [2019-07-30 20:00:00 to 2019-07-31 01:45:00]. Find the time interval where channel 891 exhibits the most similar pattern within the search context [2019-08-02 09:00:00 to 2019-08-11 07:45:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2019-08-03 09:15:00', '2019-08-03 15:00:00']", "ground_truth": [ "2019-08-03 09:15:00", "2019-08-03 15:00:00" ], "eval_metric": "iou", "channel": "891", "ts_data_path": "ts_data/L2_T3_Subsequence_Matching_S0_00009.csv" }, { "id": "L4_T1_Insight_Synthesis_S1_00007", "level": 4, "level_name": "Insight Synthesis", "category": "Insight Synthesis", "subtask": "Insight Synthesis", "question": "Analyze the behavior of channel 71 for the period 2021-08.\nPlease use ONLY the following phrases for trend description: rapid rise, gradual rise, rapid fall, gradual fall, steady stable, fluctuating stable.\nProvide a structured report covering:\n1. Trend Segmentation: Describe each stage with precise start/end timestamps (HH:MM:SS) using the phrases above.\n2. Outlier Audit: Identify only significant outliers that deviate sharply from the local trend. Ignore minor background noise.\n(Output format: structured natural language report.)", "answer": "1. Trend Segmentation: from 2021-08-01 00:00:00 to 2021-08-08 03:45:00, the trend showed a fluctuating stable; from 2021-08-08 03:45:00 to 2021-08-12 22:15:00, the trend showed a rapid rise; from 2021-08-12 22:15:00 to 2021-08-24 20:30:00, the trend showed a gradual fall; from 2021-08-24 20:30:00 to 2021-08-31 23:45:00, the trend showed a rapid rise.\n2. Outlier Audit: A significant spike was detected at 2021-08-24 12:00:00 (value: 7.63).", "ground_truth": { "trend_segments": [ "from 2021-08-01 00:00:00 to 2021-08-08 03:45:00, the trend showed a fluctuating stable", "from 2021-08-08 03:45:00 to 2021-08-12 22:15:00, the trend showed a rapid rise", "from 2021-08-12 22:15:00 to 2021-08-24 20:30:00, the trend showed a gradual fall", "from 2021-08-24 20:30:00 to 2021-08-31 23:45:00, the trend showed a rapid rise" ], "segments_meta": [ { "adj": "fluctuating", "kind": "stable", "start_idx": 0, "end_idx": 687 }, { "adj": "rapid", "kind": "rise", "start_idx": 687, "end_idx": 1145 }, { "adj": "gradual", "kind": "fall", "start_idx": 1145, "end_idx": 2290 }, { "adj": "rapid", "kind": "rise", "start_idx": 2290, "end_idx": 2976 } ], "significant_anomaly": { "timestamp": "2021-08-24 12:00:00", "kind": "significant spike", "value": 7.634634868096812 } }, "eval_metric": "report", "channel": "71", "ts_data_path": "ts_data/L4_T1_Insight_Synthesis_S1_00007.csv" }, { "id": "L2_T1_Shape_Identification_S1_00003", "level": 2, "level_name": "Pattern Recognition", "category": "Shape Identification", "subtask": "Shape Identification", "question": "Identify the time range of the longest low plateau (bottom out) in channel 441 within [2020-01-04 10:30:00 to 2020-01-09 03:00:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2020-01-05 02:45:00', '2020-01-05 19:15:00']", "ground_truth": [ "2020-01-05 02:45:00", "2020-01-05 19:15:00" ], "eval_metric": "iou", "channel": "441", "ts_data_path": "ts_data/L2_T1_Shape_Identification_S1_00003.csv" }, { "id": "L1_T3_Interval_Discovery_S2_00004", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Interval Discovery", "question": "Find the longest period where channel 626 remained above 1.14 in 2022-11-04 to 2023-06-11. (Output format: a time interval, e.g., [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2023-02-19 01:15:00, 2023-04-03 13:15:00]", "ground_truth": [ "2023-02-19 01:15:00", "2023-04-03 13:15:00" ], "eval_metric": "iou", "channel": "626", "ts_data_path": "ts_data/L1_T3_Interval_Discovery_S2_00004.csv" }, { "id": "L4_T1_Insight_Synthesis_S4_00007", "level": 4, "level_name": "Insight Synthesis", "category": "Insight Synthesis", "subtask": "Insight Synthesis", "question": "Analyze the behavior of channel 146 for the period 2022-07.\nPlease use ONLY the following phrases for trend description: rapid rise, gradual rise, rapid fall, gradual fall, steady stable, fluctuating stable.\nProvide a structured report covering:\n1. Trend Segmentation: Describe each stage with precise start/end timestamps (HH:MM:SS) using the phrases above.\n2. Outlier Audit: Identify only significant outliers that deviate sharply from the local trend. Ignore minor background noise.\n(Output format: structured natural language report.)", "answer": "1. Trend Segmentation: from 2022-07-01 00:00:00 to 2022-07-10 13:00:00, the trend showed a steady stable; from 2022-07-10 13:00:00 to 2022-07-17 16:45:00, the trend showed a gradual fall; from 2022-07-17 16:45:00 to 2022-07-22 11:15:00, the trend showed a gradual rise; from 2022-07-22 11:15:00 to 2022-07-31 23:45:00, the trend showed a gradual fall.\n2. Outlier Audit: A significant spike was detected at 2022-07-21 08:15:00 (value: 6.38).", "ground_truth": { "trend_segments": [ "from 2022-07-01 00:00:00 to 2022-07-10 13:00:00, the trend showed a steady stable", "from 2022-07-10 13:00:00 to 2022-07-17 16:45:00, the trend showed a gradual fall", "from 2022-07-17 16:45:00 to 2022-07-22 11:15:00, the trend showed a gradual rise", "from 2022-07-22 11:15:00 to 2022-07-31 23:45:00, the trend showed a gradual fall" ], "segments_meta": [ { "adj": "steady", "kind": "stable", "start_idx": 0, "end_idx": 916 }, { "adj": "gradual", "kind": "fall", "start_idx": 916, "end_idx": 1603 }, { "adj": "gradual", "kind": "rise", "start_idx": 1603, "end_idx": 2061 }, { "adj": "gradual", "kind": "fall", "start_idx": 2061, "end_idx": 2976 } ], "significant_anomaly": { "timestamp": "2022-07-21 08:15:00", "kind": "significant spike", "value": 6.379802705483224 } }, "eval_metric": "report", "channel": "146", "ts_data_path": "ts_data/L4_T1_Insight_Synthesis_S4_00007.csv" }, { "id": "L3_T2_Contextual_Anomaly_S2_00006", "level": 3, "level_name": "Semantic Reasoning", "category": "Contextual Anomaly", "subtask": "Contextual Anomaly", "question": "Identify the period in channel 891 during 2022 that experienced the most significant historically low water level. (Output format: [YYYY-MM-DD, YYYY-MM-DD])", "answer": "[2022-10-27 01:00:00, 2022-11-26 00:45:00]", "ground_truth": [ "2022-10-27 01:00:00", "2022-11-26 00:45:00" ], "eval_metric": "iou", "channel": "891", "ts_data_path": "ts_data/L3_T2_Contextual_Anomaly_S2_00006.csv" }, { "id": "L2_T3_Subsequence_Matching_S1_00009", "level": 2, "level_name": "Pattern Recognition", "category": "Subsequence Matching", "subtask": "Subsequence Matching", "question": "Analyze the reference pattern in [2021-04-26 22:15:00 to 2021-04-27 22:00:00]. Find the time interval where channel 578 exhibits the most similar pattern within the search context [2021-04-29 10:15:00 to 2021-05-02 05:00:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2021-04-29 13:15:00', '2021-04-30 13:00:00']", "ground_truth": [ "2021-04-29 13:15:00", "2021-04-30 13:00:00" ], "eval_metric": "iou", "channel": "578", "ts_data_path": "ts_data/L2_T3_Subsequence_Matching_S1_00009.csv" }, { "id": "L2_T2_Periodicity_Detection_S0_00002", "level": 2, "level_name": "Pattern Recognition", "category": "Periodicity Detection", "subtask": "Periodicity Detection", "question": "What is the dominant cycle period (in data points) of channel 172 within [2020-12-17 14:30:00 to 2020-12-20 16:00:00]? (Output format: integer)", "answer": "98", "ground_truth": 98, "eval_metric": "rel_acc", "channel": "172", "ts_data_path": "ts_data/L2_T2_Periodicity_Detection_S0_00002.csv" }, { "id": "L3_T2_Contextual_Anomaly_S3_00001", "level": 3, "level_name": "Semantic Reasoning", "category": "Contextual Anomaly", "subtask": "Contextual Anomaly", "question": "Identify the period in channel 99 during 2023 that experienced the most significant historically low water level. (Output format: [YYYY-MM-DD, YYYY-MM-DD])", "answer": "[2023-07-02 17:30:00, 2023-08-17 17:15:00]", "ground_truth": [ "2023-07-02 17:30:00", "2023-08-17 17:15:00" ], "eval_metric": "iou", "channel": "99", "ts_data_path": "ts_data/L3_T2_Contextual_Anomaly_S3_00001.csv" }, { "id": "L1_T1_Global_Aggregation_S2_00009", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Global Aggregation", "question": "What is the minimum value of channel 894 in 2022-04-05 to 2022-04-24? (Output format: a single numeric value, rounded to 3 decimal places, e.g., x.xxx)", "answer": "6.15", "ground_truth": 6.15, "eval_metric": "rel_acc", "channel": "894", "ts_data_path": "ts_data/L1_T1_Global_Aggregation_S2_00009.csv" }, { "id": "L2_T1_Shape_Identification_S0_00009", "level": 2, "level_name": "Pattern Recognition", "category": "Shape Identification", "subtask": "Shape Identification", "question": "Identify the time range of the longest low plateau (bottom out) in channel OT within [2017-10-20 09:00:00 to 2017-10-31 02:15:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2017-10-24 17:00:00', '2017-10-26 07:15:00']", "ground_truth": [ "2017-10-24 17:00:00", "2017-10-26 07:15:00" ], "eval_metric": "iou", "channel": "OT", "ts_data_path": "ts_data/L2_T1_Shape_Identification_S0_00009.csv" }, { "id": "L2_T1_Shape_Identification_S3_00008", "level": 2, "level_name": "Pattern Recognition", "category": "Shape Identification", "subtask": "Shape Identification", "question": "Identify the time range of the largest step descent in channel LUFL within [2017-11-11 05:15:00 to 2017-11-16 19:30:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2017-11-11 07:45:00', '2017-11-12 03:30:00']", "ground_truth": [ "2017-11-11 07:45:00", "2017-11-12 03:30:00" ], "eval_metric": "iou", "channel": "LUFL", "ts_data_path": "ts_data/L2_T1_Shape_Identification_S3_00008.csv" }, { "id": "L1_T3_Interval_Discovery_S3_00001", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Interval Discovery", "question": "Find the longest period where channel 147 remained above 0.198 in 2019-11-07 to 2020-01-22. (Output format: a time interval, e.g., [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2019-12-12 08:30:00, 2019-12-17 08:30:00]", "ground_truth": [ "2019-12-12 08:30:00", "2019-12-17 08:30:00" ], "eval_metric": "iou", "channel": "147", "ts_data_path": "ts_data/L1_T3_Interval_Discovery_S3_00001.csv" }, { "id": "L1_T2_Temporal_Localization_S0_00007", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Temporal Localization", "question": "At what exact timestamp did channel MUFL reach its minimum value in 2017-09-18 to 2017-11-08? (Output format: a specific timestamp, e.g., YYYY-MM-DD HH:MM:SS)", "answer": "2017-09-28 12:15:00", "ground_truth": "2017-09-28 12:15:00", "eval_metric": "hit", "channel": "MUFL", "ts_data_path": "ts_data/L1_T2_Temporal_Localization_S0_00007.csv" }, { "id": "L3_T1_Composite_Trend_S1_00001", "level": 3, "level_name": "Semantic Reasoning", "category": "Composite Trend", "subtask": "Composite Trend", "question": "Among days in channel 169 during 2021 that exhibit the trend pattern 'fluctuating stable, then rapid rise, then fall, then steady stable', identify the top-2 days where the rapid rise segment is the fastest. (Output format: a ranked list of dates, e.g., ['YYYY-MM-DD', ...])", "answer": "['2021-09-07', '2021-12-15']", "ground_truth": [ "2021-09-07", "2021-12-15" ], "eval_metric": "set_f1", "channel": "169", "ts_data_path": "ts_data/L3_T1_Composite_Trend_S1_00001.csv" }, { "id": "L2_T2_Periodicity_Detection_S2_00001", "level": 2, "level_name": "Pattern Recognition", "category": "Periodicity Detection", "subtask": "Periodicity Detection", "question": "What is the dominant cycle period (in data points) of channel 147 within [2020-09-11 17:00:00 to 2020-09-22 21:15:00]? (Output format: integer)", "answer": "32", "ground_truth": 32, "eval_metric": "rel_acc", "channel": "147", "ts_data_path": "ts_data/L2_T2_Periodicity_Detection_S2_00001.csv" }, { "id": "L2_T2_Periodicity_Detection_S4_00006", "level": 2, "level_name": "Pattern Recognition", "category": "Periodicity Detection", "subtask": "Periodicity Detection", "question": "What is the dominant cycle period (in data points) of channel 591 within [2022-06-23 02:00:00 to 2022-07-03 05:15:00]? (Output format: integer)", "answer": "46", "ground_truth": 46, "eval_metric": "rel_acc", "channel": "591", "ts_data_path": "ts_data/L2_T2_Periodicity_Detection_S4_00006.csv" }, { "id": "L1_T2_Temporal_Localization_S2_00003", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Temporal Localization", "question": "At what exact timestamp did channel 155 reach its minimum value in 2022-02-04 to 2022-03-11? (Output format: a specific timestamp, e.g., YYYY-MM-DD HH:MM:SS)", "answer": "2022-03-11 06:15:00", "ground_truth": "2022-03-11 06:15:00", "eval_metric": "hit", "channel": "155", "ts_data_path": "ts_data/L1_T2_Temporal_Localization_S2_00003.csv" }, { "id": "L2_T3_Subsequence_Matching_S4_00006", "level": 2, "level_name": "Pattern Recognition", "category": "Subsequence Matching", "subtask": "Subsequence Matching", "question": "Analyze the reference pattern in [2017-11-28 03:00:00 to 2017-11-28 08:45:00]. Find the time interval where channel LUFL exhibits the most similar pattern within the search context [2017-11-28 22:00:00 to 2017-11-30 20:45:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2017-11-29 00:00:00', '2017-11-29 05:45:00']", "ground_truth": [ "2017-11-29 00:00:00", "2017-11-29 05:45:00" ], "eval_metric": "iou", "channel": "LUFL", "ts_data_path": "ts_data/L2_T3_Subsequence_Matching_S4_00006.csv" }, { "id": "L2_T2_Periodicity_Detection_S0_00009", "level": 2, "level_name": "Pattern Recognition", "category": "Periodicity Detection", "subtask": "Periodicity Detection", "question": "What is the dominant cycle period (in data points) of channel LULL within [2017-03-26 11:30:00 to 2017-04-09 04:00:00]? (Output format: integer)", "answer": "160", "ground_truth": 160, "eval_metric": "rel_acc", "channel": "LULL", "ts_data_path": "ts_data/L2_T2_Periodicity_Detection_S0_00009.csv" }, { "id": "L2_T1_Shape_Identification_S3_00002", "level": 2, "level_name": "Pattern Recognition", "category": "Shape Identification", "subtask": "Shape Identification", "question": "Identify the time range of the highest upward spike in channel 441 within [2019-08-24 15:00:00 to 2019-08-27 08:15:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2019-08-26 05:15:00', '2019-08-26 14:45:00']", "ground_truth": [ "2019-08-26 05:15:00", "2019-08-26 14:45:00" ], "eval_metric": "iou", "channel": "441", "ts_data_path": "ts_data/L2_T1_Shape_Identification_S3_00002.csv" }, { "id": "L1_T4_Sliding_Window_S3_00003", "level": 1, "level_name": "Basic Operations", "category": "Sliding Window", "subtask": "Sliding Window", "question": "Which 50-day period in 2022-01-15 to 2022-10-20 had the highest average for channel 312? (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2022-02-02 11:30:00, 2022-03-24 11:15:00]", "ground_truth": [ "2022-02-02 11:30:00", "2022-03-24 11:15:00" ], "eval_metric": "iou", "channel": "312", "ts_data_path": "ts_data/L1_T4_Sliding_Window_S3_00003.csv" }, { "id": "L4_T1_Insight_Synthesis_S4_00001", "level": 4, "level_name": "Insight Synthesis", "category": "Insight Synthesis", "subtask": "Insight Synthesis", "question": "Analyze the behavior of channel 151 for the period 2021-06.\nPlease use ONLY the following phrases for trend description: rapid rise, gradual rise, rapid fall, gradual fall, steady stable, fluctuating stable.\nProvide a structured report covering:\n1. Trend Segmentation: Describe each stage with precise start/end timestamps (HH:MM:SS) using the phrases above.\n2. Outlier Audit: Identify only significant outliers that deviate sharply from the local trend. Ignore minor background noise.\n(Output format: structured natural language report.)", "answer": "1. Trend Segmentation: from 2021-06-01 00:00:00 to 2021-06-08 12:00:00, the trend showed a rapid rise; from 2021-06-08 12:00:00 to 2021-06-17 21:00:00, the trend showed a steady stable; from 2021-06-17 21:00:00 to 2021-06-23 12:00:00, the trend showed a fluctuating stable; from 2021-06-23 12:00:00 to 2021-06-30 23:45:00, the trend showed a rapid fall.\n2. Outlier Audit: A significant spike was detected at 2021-06-13 23:15:00 (value: 4.92).", "ground_truth": { "trend_segments": [ "from 2021-06-01 00:00:00 to 2021-06-08 12:00:00, the trend showed a rapid rise", "from 2021-06-08 12:00:00 to 2021-06-17 21:00:00, the trend showed a steady stable", "from 2021-06-17 21:00:00 to 2021-06-23 12:00:00, the trend showed a fluctuating stable", "from 2021-06-23 12:00:00 to 2021-06-30 23:45:00, the trend showed a rapid fall" ], "segments_meta": [ { "adj": "rapid", "kind": "rise", "start_idx": 0, "end_idx": 720 }, { "adj": "steady", "kind": "stable", "start_idx": 720, "end_idx": 1620 }, { "adj": "fluctuating", "kind": "stable", "start_idx": 1620, "end_idx": 2160 }, { "adj": "rapid", "kind": "fall", "start_idx": 2160, "end_idx": 2880 } ], "significant_anomaly": { "timestamp": "2021-06-13 23:15:00", "kind": "significant spike", "value": 4.916388575508013 } }, "eval_metric": "report", "channel": "151", "ts_data_path": "ts_data/L4_T1_Insight_Synthesis_S4_00001.csv" }, { "id": "L2_T3_Subsequence_Matching_S1_00007", "level": 2, "level_name": "Pattern Recognition", "category": "Subsequence Matching", "subtask": "Subsequence Matching", "question": "Analyze the reference pattern in [2019-11-21 21:00:00 to 2019-11-22 20:45:00]. Find the time interval where channel 166 exhibits the most similar pattern within the search context [2019-11-25 05:30:00 to 2019-12-03 15:30:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2019-11-27 11:00:00', '2019-11-28 10:45:00']", "ground_truth": [ "2019-11-27 11:00:00", "2019-11-28 10:45:00" ], "eval_metric": "iou", "channel": "166", "ts_data_path": "ts_data/L2_T3_Subsequence_Matching_S1_00007.csv" }, { "id": "L1_T2_Temporal_Localization_S4_00004", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Temporal Localization", "question": "At what exact timestamp did channel 683 reach its minimum value in 2019-11-04 to 2020-03-14? (Output format: a specific timestamp, e.g., YYYY-MM-DD HH:MM:SS)", "answer": "2019-11-13 11:00:00", "ground_truth": "2019-11-13 11:00:00", "eval_metric": "hit", "channel": "683", "ts_data_path": "ts_data/L1_T2_Temporal_Localization_S4_00004.csv" }, { "id": "L2_T3_Subsequence_Matching_S1_00002", "level": 2, "level_name": "Pattern Recognition", "category": "Subsequence Matching", "subtask": "Subsequence Matching", "question": "Analyze the reference pattern in [2020-01-15 16:30:00 to 2020-01-16 02:15:00]. Find the time interval where channel 627 exhibits the most similar pattern within the search context [2020-01-16 21:15:00 to 2020-01-19 14:15:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2020-01-17 15:30:00', '2020-01-18 01:15:00']", "ground_truth": [ "2020-01-17 15:30:00", "2020-01-18 01:15:00" ], "eval_metric": "iou", "channel": "627", "ts_data_path": "ts_data/L2_T3_Subsequence_Matching_S1_00002.csv" }, { "id": "L1_T2_Temporal_Localization_S3_00002", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Temporal Localization", "question": "At what exact timestamp did channel 123 reach its maximum value in 2019-03-24 to 2019-05-08? (Output format: a specific timestamp, e.g., YYYY-MM-DD HH:MM:SS)", "answer": "2019-04-03 00:15:00", "ground_truth": "2019-04-03 00:15:00", "eval_metric": "hit", "channel": "123", "ts_data_path": "ts_data/L1_T2_Temporal_Localization_S3_00002.csv" }, { "id": "L3_T3_Causal_Anomaly_S1_00006", "level": 3, "level_name": "Semantic Reasoning", "category": "Causal Anomaly", "subtask": "Causal Anomaly", "question": "Given that channel 147_up_2021 is the upstream source of channel 147_down_2021, identify the time period in 2021 where 147_down_2021 shows a significant causal anomaly, such as an inverse trend against the source. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2021-04-22 10:30:00, 2021-04-24 08:45:00]", "ground_truth": [ "2021-04-22 10:30:00", "2021-04-24 08:45:00" ], "eval_metric": "iou", "channel": "147_down_2021", "ts_data_path": "ts_data/L3_T3_Causal_Anomaly_S1_00006.csv" }, { "id": "L2_T3_Subsequence_Matching_S1_00003", "level": 2, "level_name": "Pattern Recognition", "category": "Subsequence Matching", "subtask": "Subsequence Matching", "question": "Analyze the reference pattern in [2020-11-06 11:00:00 to 2020-11-07 10:45:00]. Find the time interval where channel 895 exhibits the most similar pattern within the search context [2020-11-09 06:00:00 to 2020-11-15 10:15:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2020-11-13 22:00:00', '2020-11-14 21:45:00']", "ground_truth": [ "2020-11-13 22:00:00", "2020-11-14 21:45:00" ], "eval_metric": "iou", "channel": "895", "ts_data_path": "ts_data/L2_T3_Subsequence_Matching_S1_00003.csv" }, { "id": "L1_T2_Temporal_Localization_S4_00002", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Temporal Localization", "question": "At what exact timestamp did channel 430 first rise above 0.11 in 2022-05-21 to 2022-08-10? (Output format: a specific timestamp, e.g., YYYY-MM-DD HH:MM:SS)", "answer": "2022-05-21 09:30:00", "ground_truth": "2022-05-21 09:30:00", "eval_metric": "hit", "channel": "430", "ts_data_path": "ts_data/L1_T2_Temporal_Localization_S4_00002.csv" }, { "id": "L2_T2_Periodicity_Detection_S2_00003", "level": 2, "level_name": "Pattern Recognition", "category": "Periodicity Detection", "subtask": "Periodicity Detection", "question": "What is the dominant cycle period (in data points) of channel 245 within [2021-05-03 03:00:00 to 2021-05-10 22:30:00]? (Output format: integer)", "answer": "148", "ground_truth": 148, "eval_metric": "rel_acc", "channel": "245", "ts_data_path": "ts_data/L2_T2_Periodicity_Detection_S2_00003.csv" }, { "id": "L2_T1_Shape_Identification_S4_00009", "level": 2, "level_name": "Pattern Recognition", "category": "Shape Identification", "subtask": "Shape Identification", "question": "Identify the time range of the highest upward spike in channel 591 within [2022-07-15 13:00:00 to 2022-07-27 04:30:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2022-07-21 11:15:00', '2022-07-23 04:45:00']", "ground_truth": [ "2022-07-21 11:15:00", "2022-07-23 04:45:00" ], "eval_metric": "iou", "channel": "591", "ts_data_path": "ts_data/L2_T1_Shape_Identification_S4_00009.csv" }, { "id": "L2_T2_Periodicity_Detection_S3_00002", "level": 2, "level_name": "Pattern Recognition", "category": "Periodicity Detection", "subtask": "Periodicity Detection", "question": "What is the dominant cycle period (in data points) of channel 237 within [2022-07-09 00:00:00 to 2022-07-17 07:45:00]? (Output format: integer)", "answer": "50", "ground_truth": 50, "eval_metric": "rel_acc", "channel": "237", "ts_data_path": "ts_data/L2_T2_Periodicity_Detection_S3_00002.csv" }, { "id": "L1_T2_Temporal_Localization_S1_00003", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Temporal Localization", "question": "At what exact timestamp did channel 151 reach its maximum value in 2019-12? (Output format: a specific timestamp, e.g., YYYY-MM-DD HH:MM:SS)", "answer": "2019-12-10 03:30:00", "ground_truth": "2019-12-10 03:30:00", "eval_metric": "hit", "channel": "151", "ts_data_path": "ts_data/L1_T2_Temporal_Localization_S1_00003.csv" }, { "id": "L1_T4_Sliding_Window_S2_00005", "level": 1, "level_name": "Basic Operations", "category": "Sliding Window", "subtask": "Sliding Window", "question": "Which 24-day period in 2016-08-01 to 2017-02-28 had the largest range for channel HULL? (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2017-01-30 23:45:00, 2017-02-23 23:30:00]", "ground_truth": [ "2017-01-30 23:45:00", "2017-02-23 23:30:00" ], "eval_metric": "iou", "channel": "HULL", "ts_data_path": "ts_data/L1_T4_Sliding_Window_S2_00005.csv" }, { "id": "L2_T3_Subsequence_Matching_S0_00003", "level": 2, "level_name": "Pattern Recognition", "category": "Subsequence Matching", "subtask": "Subsequence Matching", "question": "Analyze the reference pattern in [2019-11-17 08:15:00 to 2019-11-18 02:00:00]. Find the time interval where channel 169 exhibits the most similar pattern within the search context [2019-11-20 04:30:00 to 2019-11-27 20:15:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2019-11-23 13:15:00', '2019-11-24 07:00:00']", "ground_truth": [ "2019-11-23 13:15:00", "2019-11-24 07:00:00" ], "eval_metric": "iou", "channel": "169", "ts_data_path": "ts_data/L2_T3_Subsequence_Matching_S0_00003.csv" }, { "id": "L2_T1_Shape_Identification_S0_00001", "level": 2, "level_name": "Pattern Recognition", "category": "Shape Identification", "subtask": "Shape Identification", "question": "Identify the time range of the largest step ascent in channel 71 within [2019-07-01 06:00:00 to 2019-07-13 08:30:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2019-07-05 09:30:00', '2019-07-07 04:45:00']", "ground_truth": [ "2019-07-05 09:30:00", "2019-07-07 04:45:00" ], "eval_metric": "iou", "channel": "71", "ts_data_path": "ts_data/L2_T1_Shape_Identification_S0_00001.csv" }, { "id": "L1_T1_Global_Aggregation_S4_00005", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Global Aggregation", "question": "What is the maximum value of channel 627 in 2023-10-09 to 2023-10-14? (Output format: a single numeric value, rounded to 3 decimal places, e.g., x.xxx)", "answer": "0.451", "ground_truth": 0.451, "eval_metric": "rel_acc", "channel": "627", "ts_data_path": "ts_data/L1_T1_Global_Aggregation_S4_00005.csv" }, { "id": "L3_T1_Composite_Trend_S3_00008", "level": 3, "level_name": "Semantic Reasoning", "category": "Composite Trend", "subtask": "Composite Trend", "question": "Among days in channel HUFL during 2017 that exhibit the trend pattern 'slow rise, then fall', identify the top-6 days where the slow rise segment is the slowest. (Output format: a ranked list of dates, e.g., ['YYYY-MM-DD', ...])", "answer": "['2017-02-12', '2017-12-07', '2017-02-07', '2017-09-23', '2017-05-27', '2017-10-14']", "ground_truth": [ "2017-02-12", "2017-12-07", "2017-02-07", "2017-09-23", "2017-05-27", "2017-10-14" ], "eval_metric": "set_f1", "channel": "HUFL", "ts_data_path": "ts_data/L3_T1_Composite_Trend_S3_00008.csv" }, { "id": "L2_T1_Shape_Identification_S0_00007", "level": 2, "level_name": "Pattern Recognition", "category": "Shape Identification", "subtask": "Shape Identification", "question": "Identify the time range of the longest low plateau (bottom out) in channel HULL within [2017-07-08 04:00:00 to 2017-07-10 10:45:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2017-07-10 02:00:00', '2017-07-10 10:00:00']", "ground_truth": [ "2017-07-10 02:00:00", "2017-07-10 10:00:00" ], "eval_metric": "iou", "channel": "HULL", "ts_data_path": "ts_data/L2_T1_Shape_Identification_S0_00007.csv" }, { "id": "L2_T3_Subsequence_Matching_S4_00002", "level": 2, "level_name": "Pattern Recognition", "category": "Subsequence Matching", "subtask": "Subsequence Matching", "question": "Analyze the reference pattern in [2021-08-06 18:00:00 to 2021-08-07 03:45:00]. Find the time interval where channel 154 exhibits the most similar pattern within the search context [2021-08-09 11:00:00 to 2021-08-18 05:00:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2021-08-15 09:30:00', '2021-08-15 19:15:00']", "ground_truth": [ "2021-08-15 09:30:00", "2021-08-15 19:15:00" ], "eval_metric": "iou", "channel": "154", "ts_data_path": "ts_data/L2_T3_Subsequence_Matching_S4_00002.csv" }, { "id": "L2_T3_Subsequence_Matching_S3_00008", "level": 2, "level_name": "Pattern Recognition", "category": "Subsequence Matching", "subtask": "Subsequence Matching", "question": "Analyze the reference pattern in [2018-02-27 19:30:00 to 2018-02-28 05:15:00]. Find the time interval where channel HULL exhibits the most similar pattern within the search context [2018-03-01 14:45:00 to 2018-03-06 18:00:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2018-03-04 12:15:00', '2018-03-04 22:00:00']", "ground_truth": [ "2018-03-04 12:15:00", "2018-03-04 22:00:00" ], "eval_metric": "iou", "channel": "HULL", "ts_data_path": "ts_data/L2_T3_Subsequence_Matching_S3_00008.csv" }, { "id": "L1_T4_Sliding_Window_S0_00004", "level": 1, "level_name": "Basic Operations", "category": "Sliding Window", "subtask": "Sliding Window", "question": "Which 10-day period in 2021-12-11 to 2022-05-24 had the highest variance for channel 647? (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2022-02-14 04:15:00, 2022-02-24 04:00:00]", "ground_truth": [ "2022-02-14 04:15:00", "2022-02-24 04:00:00" ], "eval_metric": "iou", "channel": "647", "ts_data_path": "ts_data/L1_T4_Sliding_Window_S0_00004.csv" }, { "id": "L2_T1_Shape_Identification_S1_00006", "level": 2, "level_name": "Pattern Recognition", "category": "Shape Identification", "subtask": "Shape Identification", "question": "Identify the time range of the deepest deep valley in channel 683 within [2019-12-09 00:00:00 to 2019-12-16 22:15:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2019-12-10 21:00:00', '2019-12-12 01:15:00']", "ground_truth": [ "2019-12-10 21:00:00", "2019-12-12 01:15:00" ], "eval_metric": "iou", "channel": "683", "ts_data_path": "ts_data/L2_T1_Shape_Identification_S1_00006.csv" }, { "id": "L1_T3_Interval_Discovery_S3_00009", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Interval Discovery", "question": "Find the longest period where channel MULL remained above 3.127 in 2016-10. (Output format: a time interval, e.g., [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2016-10-13 22:15:00, 2016-10-14 05:00:00]", "ground_truth": [ "2016-10-13 22:15:00", "2016-10-14 05:00:00" ], "eval_metric": "iou", "channel": "MULL", "ts_data_path": "ts_data/L1_T3_Interval_Discovery_S3_00009.csv" }, { "id": "L1_T3_Interval_Discovery_S4_00001", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Interval Discovery", "question": "Find the longest period where channel 67 remained above 0.244 in 2022-08-09 to 2023-03-25. (Output format: a time interval, e.g., [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2022-11-11 17:00:00, 2022-12-15 00:15:00]", "ground_truth": [ "2022-11-11 17:00:00", "2022-12-15 00:15:00" ], "eval_metric": "iou", "channel": "67", "ts_data_path": "ts_data/L1_T3_Interval_Discovery_S4_00001.csv" }, { "id": "L1_T3_Interval_Discovery_S2_00005", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Interval Discovery", "question": "Find the longest period where channel 891 remained above 3.96 in 2022-09-07 to 2023-03-30. (Output format: a time interval, e.g., [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2023-03-07 21:30:00, 2023-03-30 15:15:00]", "ground_truth": [ "2023-03-07 21:30:00", "2023-03-30 15:15:00" ], "eval_metric": "iou", "channel": "891", "ts_data_path": "ts_data/L1_T3_Interval_Discovery_S2_00005.csv" }, { "id": "L2_T1_Shape_Identification_S1_00008", "level": 2, "level_name": "Pattern Recognition", "category": "Shape Identification", "subtask": "Shape Identification", "question": "Identify the time range of the longest low plateau (bottom out) in channel 71 within [2022-08-18 15:00:00 to 2022-08-23 07:15:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2022-08-18 19:45:00', '2022-08-19 12:15:00']", "ground_truth": [ "2022-08-18 19:45:00", "2022-08-19 12:15:00" ], "eval_metric": "iou", "channel": "71", "ts_data_path": "ts_data/L2_T1_Shape_Identification_S1_00008.csv" }, { "id": "L2_T1_Shape_Identification_S0_00006", "level": 2, "level_name": "Pattern Recognition", "category": "Shape Identification", "subtask": "Shape Identification", "question": "Identify the time range of the highest upward spike in channel 894 within [2020-10-15 18:00:00 to 2020-10-17 07:30:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2020-10-16 00:15:00', '2020-10-16 05:30:00']", "ground_truth": [ "2020-10-16 00:15:00", "2020-10-16 05:30:00" ], "eval_metric": "iou", "channel": "894", "ts_data_path": "ts_data/L2_T1_Shape_Identification_S0_00006.csv" }, { "id": "L3_T1_Composite_Trend_S0_00001", "level": 3, "level_name": "Semantic Reasoning", "category": "Composite Trend", "subtask": "Composite Trend", "question": "Among days in channel 71 during 2023 that exhibit the trend pattern 'rapid rise, then fluctuating stable, then steady stable', identify the top-3 days where the fluctuating stable segment is the most fluctuating. (Output format: a ranked list of dates, e.g., ['YYYY-MM-DD', ...])", "answer": "['2023-07-09', '2023-06-22', '2023-09-19']", "ground_truth": [ "2023-07-09", "2023-06-22", "2023-09-19" ], "eval_metric": "set_f1", "channel": "71", "ts_data_path": "ts_data/L3_T1_Composite_Trend_S0_00001.csv" }, { "id": "L2_T1_Shape_Identification_S4_00002", "level": 2, "level_name": "Pattern Recognition", "category": "Shape Identification", "subtask": "Shape Identification", "question": "Identify the time range of the highest upward spike in channel 173 within [2019-09-24 07:00:00 to 2019-10-02 23:15:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2019-09-25 16:15:00', '2019-09-26 23:15:00']", "ground_truth": [ "2019-09-25 16:15:00", "2019-09-26 23:15:00" ], "eval_metric": "iou", "channel": "173", "ts_data_path": "ts_data/L2_T1_Shape_Identification_S4_00002.csv" }, { "id": "L2_T2_Periodicity_Detection_S1_00008", "level": 2, "level_name": "Pattern Recognition", "category": "Periodicity Detection", "subtask": "Periodicity Detection", "question": "What is the dominant cycle period (in data points) of channel 170 within [2019-08-07 08:30:00 to 2019-08-09 19:00:00]? (Output format: integer)", "answer": "78", "ground_truth": 78, "eval_metric": "rel_acc", "channel": "170", "ts_data_path": "ts_data/L2_T2_Periodicity_Detection_S1_00008.csv" }, { "id": "L2_T1_Shape_Identification_S4_00003", "level": 2, "level_name": "Pattern Recognition", "category": "Shape Identification", "subtask": "Shape Identification", "question": "Identify the time range of the largest step ascent in channel 245 within [2022-12-25 16:30:00 to 2022-12-31 04:30:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2022-12-26 22:15:00', '2022-12-27 17:45:00']", "ground_truth": [ "2022-12-26 22:15:00", "2022-12-27 17:45:00" ], "eval_metric": "iou", "channel": "245", "ts_data_path": "ts_data/L2_T1_Shape_Identification_S4_00003.csv" }, { "id": "L4_T1_Insight_Synthesis_S3_00008", "level": 4, "level_name": "Insight Synthesis", "category": "Insight Synthesis", "subtask": "Insight Synthesis", "question": "Analyze the behavior of channel 151 for the period 2019-10.\nPlease use ONLY the following phrases for trend description: rapid rise, gradual rise, rapid fall, gradual fall, steady stable, fluctuating stable.\nProvide a structured report covering:\n1. Trend Segmentation: Describe each stage with precise start/end timestamps (HH:MM:SS) using the phrases above.\n2. Outlier Audit: Identify only significant outliers that deviate sharply from the local trend. Ignore minor background noise.\n(Output format: structured natural language report.)", "answer": "1. Trend Segmentation: from 2019-10-01 00:00:00 to 2019-10-08 01:00:00, the trend showed a gradual rise; from 2019-10-08 01:00:00 to 2019-10-15 02:00:00, the trend showed a gradual fall; from 2019-10-15 02:00:00 to 2019-10-20 17:15:00, the trend showed a rapid fall; from 2019-10-20 17:15:00 to 2019-10-23 13:00:00, the trend showed a fluctuating stable; from 2019-10-23 13:00:00 to 2019-10-31 23:45:00, the trend showed a steady stable.\n2. Outlier Audit: A significant spike was detected at 2019-10-12 18:00:00 (value: 4.61).", "ground_truth": { "trend_segments": [ "from 2019-10-01 00:00:00 to 2019-10-08 01:00:00, the trend showed a gradual rise", "from 2019-10-08 01:00:00 to 2019-10-15 02:00:00, the trend showed a gradual fall", "from 2019-10-15 02:00:00 to 2019-10-20 17:15:00, the trend showed a rapid fall", "from 2019-10-20 17:15:00 to 2019-10-23 13:00:00, the trend showed a fluctuating stable", "from 2019-10-23 13:00:00 to 2019-10-31 23:45:00, the trend showed a steady stable" ], "segments_meta": [ { "adj": "gradual", "kind": "rise", "start_idx": 0, "end_idx": 676 }, { "adj": "gradual", "kind": "fall", "start_idx": 676, "end_idx": 1352 }, { "adj": "rapid", "kind": "fall", "start_idx": 1352, "end_idx": 1893 }, { "adj": "fluctuating", "kind": "stable", "start_idx": 1893, "end_idx": 2164 }, { "adj": "steady", "kind": "stable", "start_idx": 2164, "end_idx": 2976 } ], "significant_anomaly": { "timestamp": "2019-10-12 18:00:00", "kind": "significant spike", "value": 4.61148630678022 } }, "eval_metric": "report", "channel": "151", "ts_data_path": "ts_data/L4_T1_Insight_Synthesis_S3_00008.csv" }, { "id": "L4_T1_Insight_Synthesis_S3_00002", "level": 4, "level_name": "Insight Synthesis", "category": "Insight Synthesis", "subtask": "Insight Synthesis", "question": "Analyze the behavior of channel 123 for the period 2021-09.\nPlease use ONLY the following phrases for trend description: rapid rise, gradual rise, rapid fall, gradual fall, steady stable, fluctuating stable.\nProvide a structured report covering:\n1. Trend Segmentation: Describe each stage with precise start/end timestamps (HH:MM:SS) using the phrases above.\n2. Outlier Audit: Identify only significant outliers that deviate sharply from the local trend. Ignore minor background noise.\n(Output format: structured natural language report.)", "answer": "1. Trend Segmentation: from 2021-09-01 00:00:00 to 2021-09-07 19:45:00, the trend showed a rapid fall; from 2021-09-07 19:45:00 to 2021-09-16 00:00:00, the trend showed a gradual rise; from 2021-09-16 00:00:00 to 2021-09-21 11:00:00, the trend showed a steady stable; from 2021-09-21 11:00:00 to 2021-09-25 13:15:00, the trend showed a gradual fall; from 2021-09-25 13:15:00 to 2021-09-30 23:45:00, the trend showed a rapid rise.\n2. Outlier Audit: A significant spike was detected at 2021-09-06 03:45:00 (value: 4.68).", "ground_truth": { "trend_segments": [ "from 2021-09-01 00:00:00 to 2021-09-07 19:45:00, the trend showed a rapid fall", "from 2021-09-07 19:45:00 to 2021-09-16 00:00:00, the trend showed a gradual rise", "from 2021-09-16 00:00:00 to 2021-09-21 11:00:00, the trend showed a steady stable", "from 2021-09-21 11:00:00 to 2021-09-25 13:15:00, the trend showed a gradual fall", "from 2021-09-25 13:15:00 to 2021-09-30 23:45:00, the trend showed a rapid rise" ], "segments_meta": [ { "adj": "rapid", "kind": "fall", "start_idx": 0, "end_idx": 655 }, { "adj": "gradual", "kind": "rise", "start_idx": 655, "end_idx": 1440 }, { "adj": "steady", "kind": "stable", "start_idx": 1440, "end_idx": 1964 }, { "adj": "gradual", "kind": "fall", "start_idx": 1964, "end_idx": 2357 }, { "adj": "rapid", "kind": "rise", "start_idx": 2357, "end_idx": 2880 } ], "significant_anomaly": { "timestamp": "2021-09-06 03:45:00", "kind": "significant spike", "value": 4.676899944528263 } }, "eval_metric": "report", "channel": "123", "ts_data_path": "ts_data/L4_T1_Insight_Synthesis_S3_00002.csv" }, { "id": "L1_T1_Global_Aggregation_S1_00008", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Global Aggregation", "question": "What is the minimum value of channel 729 in 2020-06-15 to 2020-06-30? (Output format: a single numeric value, rounded to 3 decimal places, e.g., x.xxx)", "answer": "30.2", "ground_truth": 30.2, "eval_metric": "rel_acc", "channel": "729", "ts_data_path": "ts_data/L1_T1_Global_Aggregation_S1_00008.csv" }, { "id": "L1_T2_Temporal_Localization_S2_00006", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Temporal Localization", "question": "At what exact timestamp did channel 813 reach its minimum value in 2021-01-10 to 2021-04-04? (Output format: a specific timestamp, e.g., YYYY-MM-DD HH:MM:SS)", "answer": "2021-02-04 07:00:00", "ground_truth": "2021-02-04 07:00:00", "eval_metric": "hit", "channel": "813", "ts_data_path": "ts_data/L1_T2_Temporal_Localization_S2_00006.csv" }, { "id": "L1_T3_Interval_Discovery_S1_00009", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Interval Discovery", "question": "Find the longest period where channel 245 remained above 0.579 in 2020-06. (Output format: a time interval, e.g., [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2020-06-13 15:15:00, 2020-06-16 02:15:00]", "ground_truth": [ "2020-06-13 15:15:00", "2020-06-16 02:15:00" ], "eval_metric": "iou", "channel": "245", "ts_data_path": "ts_data/L1_T3_Interval_Discovery_S1_00009.csv" }, { "id": "L2_T1_Shape_Identification_S1_00005", "level": 2, "level_name": "Pattern Recognition", "category": "Shape Identification", "subtask": "Shape Identification", "question": "Identify the time range of the highest upward spike in channel 578 within [2019-04-28 06:00:00 to 2019-05-02 15:15:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2019-04-30 09:00:00', '2019-05-01 00:30:00']", "ground_truth": [ "2019-04-30 09:00:00", "2019-05-01 00:30:00" ], "eval_metric": "iou", "channel": "578", "ts_data_path": "ts_data/L2_T1_Shape_Identification_S1_00005.csv" }, { "id": "L2_T3_Subsequence_Matching_S1_00001", "level": 2, "level_name": "Pattern Recognition", "category": "Subsequence Matching", "subtask": "Subsequence Matching", "question": "Analyze the reference pattern in [2023-08-30 04:30:00 to 2023-08-31 04:15:00]. Find the time interval where channel 71 exhibits the most similar pattern within the search context [2023-09-01 16:30:00 to 2023-09-05 23:15:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2023-09-03 00:45:00', '2023-09-04 00:30:00']", "ground_truth": [ "2023-09-03 00:45:00", "2023-09-04 00:30:00" ], "eval_metric": "iou", "channel": "71", "ts_data_path": "ts_data/L2_T3_Subsequence_Matching_S1_00001.csv" }, { "id": "L2_T1_Shape_Identification_S2_00007", "level": 2, "level_name": "Pattern Recognition", "category": "Shape Identification", "subtask": "Shape Identification", "question": "Identify the time range of the largest step descent in channel 147 within [2023-06-10 21:00:00 to 2023-06-15 14:15:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2023-06-12 00:15:00', '2023-06-12 17:00:00']", "ground_truth": [ "2023-06-12 00:15:00", "2023-06-12 17:00:00" ], "eval_metric": "iou", "channel": "147", "ts_data_path": "ts_data/L2_T1_Shape_Identification_S2_00007.csv" }, { "id": "L1_T1_Global_Aggregation_S1_00003", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Global Aggregation", "question": "What is the average value of channel 151 in 2022-01-29 to 2022-02-12? (Output format: a single numeric value, rounded to 3 decimal places, e.g., x.xxx)", "answer": "1.242", "ground_truth": 1.242, "eval_metric": "rel_acc", "channel": "151", "ts_data_path": "ts_data/L1_T1_Global_Aggregation_S1_00003.csv" }, { "id": "L3_T2_Contextual_Anomaly_S1_00007", "level": 3, "level_name": "Semantic Reasoning", "category": "Contextual Anomaly", "subtask": "Contextual Anomaly", "question": "Identify the period in channel 67 during 2021 that experienced the most significant historically low water level. (Output format: [YYYY-MM-DD, YYYY-MM-DD])", "answer": "[2021-10-06 17:30:00, 2021-10-29 17:15:00]", "ground_truth": [ "2021-10-06 17:30:00", "2021-10-29 17:15:00" ], "eval_metric": "iou", "channel": "67", "ts_data_path": "ts_data/L3_T2_Contextual_Anomaly_S1_00007.csv" }, { "id": "L2_T2_Periodicity_Detection_S2_00004", "level": 2, "level_name": "Pattern Recognition", "category": "Periodicity Detection", "subtask": "Periodicity Detection", "question": "What is the dominant cycle period (in data points) of channel 496 within [2020-12-03 01:15:00 to 2020-12-09 08:15:00]? (Output format: integer)", "answer": "43", "ground_truth": 43, "eval_metric": "rel_acc", "channel": "496", "ts_data_path": "ts_data/L2_T2_Periodicity_Detection_S2_00004.csv" }, { "id": "L2_T1_Shape_Identification_S3_00003", "level": 2, "level_name": "Pattern Recognition", "category": "Shape Identification", "subtask": "Shape Identification", "question": "Identify the time range of the longest plateau (stable period) in channel 501 within [2022-07-21 22:00:00 to 2022-07-31 04:30:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2022-07-29 00:30:00', '2022-07-30 09:30:00']", "ground_truth": [ "2022-07-29 00:30:00", "2022-07-30 09:30:00" ], "eval_metric": "iou", "channel": "501", "ts_data_path": "ts_data/L2_T1_Shape_Identification_S3_00003.csv" }, { "id": "L1_T1_Global_Aggregation_S4_00006", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Global Aggregation", "question": "What is the range value of channel 680 in 2023-04-30 to 2023-05-07? (Output format: a single numeric value, rounded to 3 decimal places, e.g., x.xxx)", "answer": "2.6", "ground_truth": 2.6, "eval_metric": "rel_acc", "channel": "680", "ts_data_path": "ts_data/L1_T1_Global_Aggregation_S4_00006.csv" }, { "id": "L2_T3_Subsequence_Matching_S2_00005", "level": 2, "level_name": "Pattern Recognition", "category": "Subsequence Matching", "subtask": "Subsequence Matching", "question": "Analyze the reference pattern in [2016-12-23 08:45:00 to 2016-12-23 18:30:00]. Find the time interval where channel MULL exhibits the most similar pattern within the search context [2016-12-24 11:00:00 to 2016-12-26 18:00:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2016-12-25 06:00:00', '2016-12-25 15:45:00']", "ground_truth": [ "2016-12-25 06:00:00", "2016-12-25 15:45:00" ], "eval_metric": "iou", "channel": "MULL", "ts_data_path": "ts_data/L2_T3_Subsequence_Matching_S2_00005.csv" }, { "id": "L3_T1_Composite_Trend_S0_00007", "level": 3, "level_name": "Semantic Reasoning", "category": "Composite Trend", "subtask": "Composite Trend", "question": "Among days in channel 591 during 2020 that exhibit the trend pattern 'rapid rise, then slow rise', identify the top-3 days where the slow rise segment is the slowest. (Output format: a ranked list of dates, e.g., ['YYYY-MM-DD', ...])", "answer": "['2020-12-01', '2020-02-26', '2020-03-04']", "ground_truth": [ "2020-12-01", "2020-02-26", "2020-03-04" ], "eval_metric": "set_f1", "channel": "591", "ts_data_path": "ts_data/L3_T1_Composite_Trend_S0_00007.csv" }, { "id": "L3_T3_Causal_Anomaly_S1_00009", "level": 3, "level_name": "Semantic Reasoning", "category": "Causal Anomaly", "subtask": "Causal Anomaly", "question": "Given that channel 172_up_2022 is the upstream source of channel 172_down_2022, identify the time period in 2022 where 172_down_2022 shows a significant causal anomaly, such as an flat line during high activity. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2022-09-09 11:30:00, 2022-09-11 08:30:00]", "ground_truth": [ "2022-09-09 11:30:00", "2022-09-11 08:30:00" ], "eval_metric": "iou", "channel": "172_down_2022", "ts_data_path": "ts_data/L3_T3_Causal_Anomaly_S1_00009.csv" }, { "id": "L3_T2_Contextual_Anomaly_S3_00009", "level": 3, "level_name": "Semantic Reasoning", "category": "Contextual Anomaly", "subtask": "Contextual Anomaly", "question": "Identify the period in channel 123 during 2021 that experienced the most significant dry-out period. (Output format: [YYYY-MM-DD, YYYY-MM-DD])", "answer": "[2021-03-01 22:00:00, 2021-03-29 21:45:00]", "ground_truth": [ "2021-03-01 22:00:00", "2021-03-29 21:45:00" ], "eval_metric": "iou", "channel": "123", "ts_data_path": "ts_data/L3_T2_Contextual_Anomaly_S3_00009.csv" }, { "id": "L4_T1_Insight_Synthesis_S0_00001", "level": 4, "level_name": "Insight Synthesis", "category": "Insight Synthesis", "subtask": "Insight Synthesis", "question": "Analyze the behavior of channel 124 for the period 2020-08.\nPlease use ONLY the following phrases for trend description: rapid rise, gradual rise, rapid fall, gradual fall, steady stable, fluctuating stable.\nProvide a structured report covering:\n1. Trend Segmentation: Describe each stage with precise start/end timestamps (HH:MM:SS) using the phrases above.\n2. Outlier Audit: Identify only significant outliers that deviate sharply from the local trend. Ignore minor background noise.\n(Output format: structured natural language report.)", "answer": "1. Trend Segmentation: from 2020-08-01 00:00:00 to 2020-08-12 06:30:00, the trend showed a rapid fall; from 2020-08-12 06:30:00 to 2020-08-17 21:45:00, the trend showed a fluctuating stable; from 2020-08-17 21:45:00 to 2020-08-31 23:45:00, the trend showed a gradual rise.\n2. Outlier Audit: A significant drop was detected at 2020-08-17 07:00:00 (value: -3.14).", "ground_truth": { "trend_segments": [ "from 2020-08-01 00:00:00 to 2020-08-12 06:30:00, the trend showed a rapid fall", "from 2020-08-12 06:30:00 to 2020-08-17 21:45:00, the trend showed a fluctuating stable", "from 2020-08-17 21:45:00 to 2020-08-31 23:45:00, the trend showed a gradual rise" ], "segments_meta": [ { "adj": "rapid", "kind": "fall", "start_idx": 0, "end_idx": 1082 }, { "adj": "fluctuating", "kind": "stable", "start_idx": 1082, "end_idx": 1623 }, { "adj": "gradual", "kind": "rise", "start_idx": 1623, "end_idx": 2976 } ], "significant_anomaly": { "timestamp": "2020-08-17 07:00:00", "kind": "significant drop", "value": -3.143684731970337 } }, "eval_metric": "report", "channel": "124", "ts_data_path": "ts_data/L4_T1_Insight_Synthesis_S0_00001.csv" }, { "id": "L1_T1_Global_Aggregation_S4_00009", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Global Aggregation", "question": "What is the minimum value of channel MUFL in 2017-07-18 to 2017-07-22? (Output format: a single numeric value, rounded to 3 decimal places, e.g., x.xxx)", "answer": "-13.823", "ground_truth": -13.823, "eval_metric": "rel_acc", "channel": "MUFL", "ts_data_path": "ts_data/L1_T1_Global_Aggregation_S4_00009.csv" }, { "id": "L2_T1_Shape_Identification_S3_00005", "level": 2, "level_name": "Pattern Recognition", "category": "Shape Identification", "subtask": "Shape Identification", "question": "Identify the time range of the longest low plateau (bottom out) in channel 591 within [2022-07-24 02:30:00 to 2022-07-25 22:45:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2022-07-24 21:00:00', '2022-07-25 03:15:00']", "ground_truth": [ "2022-07-24 21:00:00", "2022-07-25 03:15:00" ], "eval_metric": "iou", "channel": "591", "ts_data_path": "ts_data/L2_T1_Shape_Identification_S3_00005.csv" }, { "id": "L4_T1_Insight_Synthesis_S3_00001", "level": 4, "level_name": "Insight Synthesis", "category": "Insight Synthesis", "subtask": "Insight Synthesis", "question": "Analyze the behavior of channel 99 for the period 2022-02.\nPlease use ONLY the following phrases for trend description: rapid rise, gradual rise, rapid fall, gradual fall, steady stable, fluctuating stable.\nProvide a structured report covering:\n1. Trend Segmentation: Describe each stage with precise start/end timestamps (HH:MM:SS) using the phrases above.\n2. Outlier Audit: Identify only significant outliers that deviate sharply from the local trend. Ignore minor background noise.\n(Output format: structured natural language report.)", "answer": "1. Trend Segmentation: from 2022-02-01 00:00:00 to 2022-02-08 00:00:00, the trend showed a steady stable; from 2022-02-08 00:00:00 to 2022-02-10 19:15:00, the trend showed a gradual rise; from 2022-02-10 19:15:00 to 2022-02-19 04:45:00, the trend showed a rapid fall; from 2022-02-19 04:45:00 to 2022-02-24 19:15:00, the trend showed a fluctuating stable; from 2022-02-24 19:15:00 to 2022-02-28 23:45:00, the trend showed a gradual fall.\n2. Outlier Audit: A significant drop was detected at 2022-02-21 12:15:00 (value: -2.03).", "ground_truth": { "trend_segments": [ "from 2022-02-01 00:00:00 to 2022-02-08 00:00:00, the trend showed a steady stable", "from 2022-02-08 00:00:00 to 2022-02-10 19:15:00, the trend showed a gradual rise", "from 2022-02-10 19:15:00 to 2022-02-19 04:45:00, the trend showed a rapid fall", "from 2022-02-19 04:45:00 to 2022-02-24 19:15:00, the trend showed a fluctuating stable", "from 2022-02-24 19:15:00 to 2022-02-28 23:45:00, the trend showed a gradual fall" ], "segments_meta": [ { "adj": "steady", "kind": "stable", "start_idx": 0, "end_idx": 672 }, { "adj": "gradual", "kind": "rise", "start_idx": 672, "end_idx": 941 }, { "adj": "rapid", "kind": "fall", "start_idx": 941, "end_idx": 1747 }, { "adj": "fluctuating", "kind": "stable", "start_idx": 1747, "end_idx": 2285 }, { "adj": "gradual", "kind": "fall", "start_idx": 2285, "end_idx": 2688 } ], "significant_anomaly": { "timestamp": "2022-02-21 12:15:00", "kind": "significant drop", "value": -2.0313182691767926 } }, "eval_metric": "report", "channel": "99", "ts_data_path": "ts_data/L4_T1_Insight_Synthesis_S3_00001.csv" }, { "id": "L2_T2_Periodicity_Detection_S1_00005", "level": 2, "level_name": "Pattern Recognition", "category": "Periodicity Detection", "subtask": "Periodicity Detection", "question": "What is the dominant cycle period (in data points) of channel 895 within [2021-03-05 22:30:00 to 2021-03-09 13:00:00]? (Output format: integer)", "answer": "115", "ground_truth": 115, "eval_metric": "rel_acc", "channel": "895", "ts_data_path": "ts_data/L2_T2_Periodicity_Detection_S1_00005.csv" }, { "id": "L3_T1_Composite_Trend_S0_00010", "level": 3, "level_name": "Semantic Reasoning", "category": "Composite Trend", "subtask": "Composite Trend", "question": "Among days in channel 441 during 2019 that exhibit the trend pattern 'rapid rise, then fall, then slow fall', identify the top-3 days where the rapid rise segment is the fastest. (Output format: a ranked list of dates, e.g., ['YYYY-MM-DD', ...])", "answer": "['2019-08-18', '2019-10-31', '2019-02-11']", "ground_truth": [ "2019-08-18", "2019-10-31", "2019-02-11" ], "eval_metric": "set_f1", "channel": "441", "ts_data_path": "ts_data/L3_T1_Composite_Trend_S0_00010.csv" }, { "id": "L1_T4_Sliding_Window_S2_00003", "level": 1, "level_name": "Basic Operations", "category": "Sliding Window", "subtask": "Sliding Window", "question": "Which 38-day period in 2020-05-11 to 2020-09-27 had the lowest average for channel 897? (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2020-07-11 19:45:00, 2020-08-18 19:30:00]", "ground_truth": [ "2020-07-11 19:45:00", "2020-08-18 19:30:00" ], "eval_metric": "iou", "channel": "897", "ts_data_path": "ts_data/L1_T4_Sliding_Window_S2_00003.csv" }, { "id": "L3_T2_Contextual_Anomaly_S0_00009", "level": 3, "level_name": "Semantic Reasoning", "category": "Contextual Anomaly", "subtask": "Contextual Anomaly", "question": "Identify the period in channel 170 during 2022 that experienced the most significant historically high water level. (Output format: [YYYY-MM-DD, YYYY-MM-DD])", "answer": "[2022-06-13 05:00:00, 2022-07-23 04:45:00]", "ground_truth": [ "2022-06-13 05:00:00", "2022-07-23 04:45:00" ], "eval_metric": "iou", "channel": "170", "ts_data_path": "ts_data/L3_T2_Contextual_Anomaly_S0_00009.csv" }, { "id": "L2_T3_Subsequence_Matching_S2_00009", "level": 2, "level_name": "Pattern Recognition", "category": "Subsequence Matching", "subtask": "Subsequence Matching", "question": "Analyze the reference pattern in [2020-05-14 09:00:00 to 2020-05-14 14:45:00]. Find the time interval where channel 680 exhibits the most similar pattern within the search context [2020-05-15 16:30:00 to 2020-05-19 17:15:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2020-05-15 22:15:00', '2020-05-16 04:00:00']", "ground_truth": [ "2020-05-15 22:15:00", "2020-05-16 04:00:00" ], "eval_metric": "iou", "channel": "680", "ts_data_path": "ts_data/L2_T3_Subsequence_Matching_S2_00009.csv" }, { "id": "L2_T2_Periodicity_Detection_S4_00007", "level": 2, "level_name": "Pattern Recognition", "category": "Periodicity Detection", "subtask": "Periodicity Detection", "question": "What is the dominant cycle period (in data points) of channel 891 within [2022-06-05 06:00:00 to 2022-06-11 12:00:00]? (Output format: integer)", "answer": "81", "ground_truth": 81, "eval_metric": "rel_acc", "channel": "891", "ts_data_path": "ts_data/L2_T2_Periodicity_Detection_S4_00007.csv" }, { "id": "L3_T2_Contextual_Anomaly_S3_00003", "level": 3, "level_name": "Semantic Reasoning", "category": "Contextual Anomaly", "subtask": "Contextual Anomaly", "question": "Identify the period in channel 173 during 2020 that experienced the most significant extreme surge in flow. (Output format: [YYYY-MM-DD, YYYY-MM-DD])", "answer": "[2020-09-20 06:15:00, 2020-10-25 06:00:00]", "ground_truth": [ "2020-09-20 06:15:00", "2020-10-25 06:00:00" ], "eval_metric": "iou", "channel": "173", "ts_data_path": "ts_data/L3_T2_Contextual_Anomaly_S3_00003.csv" }, { "id": "L2_T1_Shape_Identification_S1_00009", "level": 2, "level_name": "Pattern Recognition", "category": "Shape Identification", "subtask": "Shape Identification", "question": "Identify the time range of the deepest deep valley in channel 172 within [2019-11-29 12:15:00 to 2019-12-11 15:15:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2019-12-05 20:15:00', '2019-12-07 15:30:00']", "ground_truth": [ "2019-12-05 20:15:00", "2019-12-07 15:30:00" ], "eval_metric": "iou", "channel": "172", "ts_data_path": "ts_data/L2_T1_Shape_Identification_S1_00009.csv" }, { "id": "L1_T1_Global_Aggregation_S0_00003", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Global Aggregation", "question": "What is the range value of channel 680 in 2020-11-20 to 2021-01-12? (Output format: a single numeric value, rounded to 3 decimal places, e.g., x.xxx)", "answer": "3.042", "ground_truth": 3.042, "eval_metric": "rel_acc", "channel": "680", "ts_data_path": "ts_data/L1_T1_Global_Aggregation_S0_00003.csv" }, { "id": "L2_T3_Subsequence_Matching_S0_00001", "level": 2, "level_name": "Pattern Recognition", "category": "Subsequence Matching", "subtask": "Subsequence Matching", "question": "Analyze the reference pattern in [2020-07-13 03:00:00 to 2020-07-13 12:45:00]. Find the time interval where channel 123 exhibits the most similar pattern within the search context [2020-07-14 20:30:00 to 2020-07-19 16:30:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2020-07-15 08:15:00', '2020-07-15 18:00:00']", "ground_truth": [ "2020-07-15 08:15:00", "2020-07-15 18:00:00" ], "eval_metric": "iou", "channel": "123", "ts_data_path": "ts_data/L2_T3_Subsequence_Matching_S0_00001.csv" }, { "id": "L1_T1_Global_Aggregation_S0_00008", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Global Aggregation", "question": "What is the range value of channel MULL in 2017-07? (Output format: a single numeric value, rounded to 3 decimal places, e.g., x.xxx)", "answer": "9.31", "ground_truth": 9.31, "eval_metric": "rel_acc", "channel": "MULL", "ts_data_path": "ts_data/L1_T1_Global_Aggregation_S0_00008.csv" }, { "id": "L1_T4_Sliding_Window_S0_00008", "level": 1, "level_name": "Basic Operations", "category": "Sliding Window", "subtask": "Sliding Window", "question": "Which 8-day period in 2021-09-17 to 2021-12-26 had the highest average for channel 169? (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2021-11-05 17:45:00, 2021-11-13 17:30:00]", "ground_truth": [ "2021-11-05 17:45:00", "2021-11-13 17:30:00" ], "eval_metric": "iou", "channel": "169", "ts_data_path": "ts_data/L1_T4_Sliding_Window_S0_00008.csv" }, { "id": "L1_T3_Interval_Discovery_S4_00008", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Interval Discovery", "question": "Find the longest period where channel LULL remained above 0.853 in 2016-09-06 to 2017-01-24. (Output format: a time interval, e.g., [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2016-10-25 23:15:00, 2016-10-27 22:00:00]", "ground_truth": [ "2016-10-25 23:15:00", "2016-10-27 22:00:00" ], "eval_metric": "iou", "channel": "LULL", "ts_data_path": "ts_data/L1_T3_Interval_Discovery_S4_00008.csv" }, { "id": "L3_T3_Causal_Anomaly_S4_00007", "level": 3, "level_name": "Semantic Reasoning", "category": "Causal Anomaly", "subtask": "Causal Anomaly", "question": "Given that channel 237_up_2019 is the upstream source of channel 237_down_2019, identify the time period in 2019 where 237_down_2019 shows a significant causal anomaly, such as an inverse trend against the source. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2019-11-27 01:45:00, 2019-12-09 01:45:00]", "ground_truth": [ "2019-11-27 01:45:00", "2019-12-09 01:45:00" ], "eval_metric": "iou", "channel": "237_down_2019", "ts_data_path": "ts_data/L3_T3_Causal_Anomaly_S4_00007.csv" }, { "id": "L2_T3_Subsequence_Matching_S0_00007", "level": 2, "level_name": "Pattern Recognition", "category": "Subsequence Matching", "subtask": "Subsequence Matching", "question": "Analyze the reference pattern in [2023-10-06 09:00:00 to 2023-10-07 08:45:00]. Find the time interval where channel 727 exhibits the most similar pattern within the search context [2023-10-09 02:15:00 to 2023-10-14 23:00:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2023-10-09 17:30:00', '2023-10-10 17:15:00']", "ground_truth": [ "2023-10-09 17:30:00", "2023-10-10 17:15:00" ], "eval_metric": "iou", "channel": "727", "ts_data_path": "ts_data/L2_T3_Subsequence_Matching_S0_00007.csv" }, { "id": "L3_T3_Causal_Anomaly_S4_00009", "level": 3, "level_name": "Semantic Reasoning", "category": "Causal Anomaly", "subtask": "Causal Anomaly", "question": "Given that channel 580_up_2021 is the upstream source of channel 580_down_2021, identify the time period in 2021 where 580_down_2021 shows a significant causal anomaly, such as an flat line during high activity. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2021-07-27 00:30:00, 2021-08-12 19:00:00]", "ground_truth": [ "2021-07-27 00:30:00", "2021-08-12 19:00:00" ], "eval_metric": "iou", "channel": "580_down_2021", "ts_data_path": "ts_data/L3_T3_Causal_Anomaly_S4_00009.csv" }, { "id": "L2_T3_Subsequence_Matching_S4_00005", "level": 2, "level_name": "Pattern Recognition", "category": "Subsequence Matching", "subtask": "Subsequence Matching", "question": "Analyze the reference pattern in [2020-07-16 23:00:00 to 2020-07-17 04:45:00]. Find the time interval where channel 933 exhibits the most similar pattern within the search context [2020-07-19 16:00:00 to 2020-07-29 06:30:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2020-07-24 19:00:00', '2020-07-25 00:45:00']", "ground_truth": [ "2020-07-24 19:00:00", "2020-07-25 00:45:00" ], "eval_metric": "iou", "channel": "933", "ts_data_path": "ts_data/L2_T3_Subsequence_Matching_S4_00005.csv" }, { "id": "L2_T1_Shape_Identification_S2_00006", "level": 2, "level_name": "Pattern Recognition", "category": "Shape Identification", "subtask": "Shape Identification", "question": "Identify the time range of the largest step descent in channel 71 within [2022-08-19 13:30:00 to 2022-08-24 05:45:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2022-08-20 06:15:00', '2022-08-20 22:45:00']", "ground_truth": [ "2022-08-20 06:15:00", "2022-08-20 22:45:00" ], "eval_metric": "iou", "channel": "71", "ts_data_path": "ts_data/L2_T1_Shape_Identification_S2_00006.csv" }, { "id": "L3_T1_Composite_Trend_S2_00007", "level": 3, "level_name": "Semantic Reasoning", "category": "Composite Trend", "subtask": "Composite Trend", "question": "Among days in channel MUFL during 2017 that exhibit the trend pattern 'fluctuating stable, then slow fall, then fluctuating stable', identify the top-4 days where the slow fall segment is the slowest. (Output format: a ranked list of dates, e.g., ['YYYY-MM-DD', ...])", "answer": "['2017-09-10', '2017-07-08', '2017-08-24', '2017-04-17']", "ground_truth": [ "2017-09-10", "2017-07-08", "2017-08-24", "2017-04-17" ], "eval_metric": "set_f1", "channel": "MUFL", "ts_data_path": "ts_data/L3_T1_Composite_Trend_S2_00007.csv" }, { "id": "L3_T1_Composite_Trend_S2_00002", "level": 3, "level_name": "Semantic Reasoning", "category": "Composite Trend", "subtask": "Composite Trend", "question": "Among days in channel 441 during 2023 that exhibit the trend pattern 'rise, then rapid rise, then fall, then slow rise', identify the top-4 days where the rapid rise segment is the fastest. (Output format: a ranked list of dates, e.g., ['YYYY-MM-DD', ...])", "answer": "['2023-09-20', '2023-08-16', '2023-02-07', '2023-05-01']", "ground_truth": [ "2023-09-20", "2023-08-16", "2023-02-07", "2023-05-01" ], "eval_metric": "set_f1", "channel": "441", "ts_data_path": "ts_data/L3_T1_Composite_Trend_S2_00002.csv" }, { "id": "L3_T3_Causal_Anomaly_S2_00008", "level": 3, "level_name": "Semantic Reasoning", "category": "Causal Anomaly", "subtask": "Causal Anomaly", "question": "Given that channel 683_up_2020 is the upstream source of channel 683_down_2020, identify the time period in 2020 where 683_down_2020 shows a significant causal anomaly, such as an flat line during high activity. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2020-12-11 11:45:00, 2020-12-12 16:45:00]", "ground_truth": [ "2020-12-11 11:45:00", "2020-12-12 16:45:00" ], "eval_metric": "iou", "channel": "683_down_2020", "ts_data_path": "ts_data/L3_T3_Causal_Anomaly_S2_00008.csv" }, { "id": "L1_T2_Temporal_Localization_S1_00005", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Temporal Localization", "question": "At what exact timestamp did channel 496 reach its maximum value in 2020-08-02 to 2020-10-23? (Output format: a specific timestamp, e.g., YYYY-MM-DD HH:MM:SS)", "answer": "2020-10-09 17:00:00", "ground_truth": "2020-10-09 17:00:00", "eval_metric": "hit", "channel": "496", "ts_data_path": "ts_data/L1_T2_Temporal_Localization_S1_00005.csv" }, { "id": "L1_T3_Interval_Discovery_S1_00007", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Interval Discovery", "question": "Find the longest period where channel LUFL remained above 4.021 in 2016-12. (Output format: a time interval, e.g., [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2016-12-29 23:15:00, 2016-12-30 07:45:00]", "ground_truth": [ "2016-12-29 23:15:00", "2016-12-30 07:45:00" ], "eval_metric": "iou", "channel": "LUFL", "ts_data_path": "ts_data/L1_T3_Interval_Discovery_S1_00007.csv" }, { "id": "L3_T1_Composite_Trend_S1_00003", "level": 3, "level_name": "Semantic Reasoning", "category": "Composite Trend", "subtask": "Composite Trend", "question": "Among days in channel 71 during 2019 that exhibit the trend pattern 'steady stable, then rapid fall, then rise, then fall', identify the top-2 days where the rapid fall segment is the fastest. (Output format: a ranked list of dates, e.g., ['YYYY-MM-DD', ...])", "answer": "['2019-03-04', '2019-08-09']", "ground_truth": [ "2019-03-04", "2019-08-09" ], "eval_metric": "set_f1", "channel": "71", "ts_data_path": "ts_data/L3_T1_Composite_Trend_S1_00003.csv" }, { "id": "L1_T4_Sliding_Window_S2_00007", "level": 1, "level_name": "Basic Operations", "category": "Sliding Window", "subtask": "Sliding Window", "question": "Which 19-day period in 2021-07-04 to 2021-11-21 had the highest average for channel 589? (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2021-09-02 07:00:00, 2021-09-21 06:45:00]", "ground_truth": [ "2021-09-02 07:00:00", "2021-09-21 06:45:00" ], "eval_metric": "iou", "channel": "589", "ts_data_path": "ts_data/L1_T4_Sliding_Window_S2_00007.csv" }, { "id": "L1_T3_Interval_Discovery_S2_00007", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Interval Discovery", "question": "Find the longest period where channel MUFL remained above 8.884 in 2017-09-06 to 2018-03-31. (Output format: a time interval, e.g., [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2018-01-30 17:15:00, 2018-02-01 08:15:00]", "ground_truth": [ "2018-01-30 17:15:00", "2018-02-01 08:15:00" ], "eval_metric": "iou", "channel": "MUFL", "ts_data_path": "ts_data/L1_T3_Interval_Discovery_S2_00007.csv" }, { "id": "L4_T1_Insight_Synthesis_S3_00003", "level": 4, "level_name": "Insight Synthesis", "category": "Insight Synthesis", "subtask": "Insight Synthesis", "question": "Analyze the behavior of channel 170 for the period 2022-12.\nPlease use ONLY the following phrases for trend description: rapid rise, gradual rise, rapid fall, gradual fall, steady stable, fluctuating stable.\nProvide a structured report covering:\n1. Trend Segmentation: Describe each stage with precise start/end timestamps (HH:MM:SS) using the phrases above.\n2. Outlier Audit: Identify only significant outliers that deviate sharply from the local trend. Ignore minor background noise.\n(Output format: structured natural language report.)", "answer": "1. Trend Segmentation: from 2022-12-01 00:00:00 to 2022-12-08 03:45:00, the trend showed a gradual rise; from 2022-12-08 03:45:00 to 2022-12-14 02:45:00, the trend showed a rapid rise; from 2022-12-14 02:45:00 to 2022-12-21 06:30:00, the trend showed a steady stable; from 2022-12-21 06:30:00 to 2022-12-28 10:15:00, the trend showed a fluctuating stable; from 2022-12-28 10:15:00 to 2022-12-31 23:45:00, the trend showed a rapid rise.\n2. Outlier Audit: A significant drop was detected at 2022-12-08 03:45:00 (value: -1480.13).", "ground_truth": { "trend_segments": [ "from 2022-12-01 00:00:00 to 2022-12-08 03:45:00, the trend showed a gradual rise", "from 2022-12-08 03:45:00 to 2022-12-14 02:45:00, the trend showed a rapid rise", "from 2022-12-14 02:45:00 to 2022-12-21 06:30:00, the trend showed a steady stable", "from 2022-12-21 06:30:00 to 2022-12-28 10:15:00, the trend showed a fluctuating stable", "from 2022-12-28 10:15:00 to 2022-12-31 23:45:00, the trend showed a rapid rise" ], "segments_meta": [ { "adj": "gradual", "kind": "rise", "start_idx": 0, "end_idx": 687 }, { "adj": "rapid", "kind": "rise", "start_idx": 687, "end_idx": 1259 }, { "adj": "steady", "kind": "stable", "start_idx": 1259, "end_idx": 1946 }, { "adj": "fluctuating", "kind": "stable", "start_idx": 1946, "end_idx": 2633 }, { "adj": "rapid", "kind": "rise", "start_idx": 2633, "end_idx": 2976 } ], "significant_anomaly": { "timestamp": "2022-12-08 03:45:00", "kind": "significant drop", "value": -1480.1348438158388 } }, "eval_metric": "report", "channel": "170", "ts_data_path": "ts_data/L4_T1_Insight_Synthesis_S3_00003.csv" }, { "id": "L2_T1_Shape_Identification_S3_00006", "level": 2, "level_name": "Pattern Recognition", "category": "Shape Identification", "subtask": "Shape Identification", "question": "Identify the time range of the longest plateau (stable period) in channel 680 within [2022-11-07 08:15:00 to 2022-11-08 11:15:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2022-11-07 10:30:00', '2022-11-07 14:15:00']", "ground_truth": [ "2022-11-07 10:30:00", "2022-11-07 14:15:00" ], "eval_metric": "iou", "channel": "680", "ts_data_path": "ts_data/L2_T1_Shape_Identification_S3_00006.csv" }, { "id": "L3_T1_Composite_Trend_S1_00010", "level": 3, "level_name": "Semantic Reasoning", "category": "Composite Trend", "subtask": "Composite Trend", "question": "Among days in channel 595 during 2019 that exhibit the trend pattern 'slow rise, then rise, then fall', identify the top-2 days where the slow rise segment is the slowest. (Output format: a ranked list of dates, e.g., ['YYYY-MM-DD', ...])", "answer": "['2019-10-20', '2019-01-23']", "ground_truth": [ "2019-10-20", "2019-01-23" ], "eval_metric": "set_f1", "channel": "595", "ts_data_path": "ts_data/L3_T1_Composite_Trend_S1_00010.csv" }, { "id": "L3_T2_Contextual_Anomaly_S4_00002", "level": 3, "level_name": "Semantic Reasoning", "category": "Contextual Anomaly", "subtask": "Contextual Anomaly", "question": "Identify the period in channel 71 during 2022 that experienced the most significant severe flood. (Output format: [YYYY-MM-DD, YYYY-MM-DD])", "answer": "[2022-10-13 21:15:00, 2022-11-02 21:00:00]", "ground_truth": [ "2022-10-13 21:15:00", "2022-11-02 21:00:00" ], "eval_metric": "iou", "channel": "71", "ts_data_path": "ts_data/L3_T2_Contextual_Anomaly_S4_00002.csv" }, { "id": "L2_T2_Periodicity_Detection_S3_00004", "level": 2, "level_name": "Pattern Recognition", "category": "Periodicity Detection", "subtask": "Periodicity Detection", "question": "What is the dominant cycle period (in data points) of channel 762 within [2019-12-05 08:15:00 to 2019-12-06 18:30:00]? (Output format: integer)", "answer": "46", "ground_truth": 46, "eval_metric": "rel_acc", "channel": "762", "ts_data_path": "ts_data/L2_T2_Periodicity_Detection_S3_00004.csv" }, { "id": "L3_T1_Composite_Trend_S3_00006", "level": 3, "level_name": "Semantic Reasoning", "category": "Composite Trend", "subtask": "Composite Trend", "question": "Among days in channel 683 during 2022 that exhibit the trend pattern 'rise, then fluctuating stable', identify the top-6 days where the fluctuating stable segment is the most fluctuating. (Output format: a ranked list of dates, e.g., ['YYYY-MM-DD', ...])", "answer": "['2022-12-22', '2022-02-14', '2022-05-30', '2022-07-24', '2022-02-03', '2022-08-01']", "ground_truth": [ "2022-12-22", "2022-02-14", "2022-05-30", "2022-07-24", "2022-02-03", "2022-08-01" ], "eval_metric": "set_f1", "channel": "683", "ts_data_path": "ts_data/L3_T1_Composite_Trend_S3_00006.csv" }, { "id": "L3_T1_Composite_Trend_S4_00006", "level": 3, "level_name": "Semantic Reasoning", "category": "Composite Trend", "subtask": "Composite Trend", "question": "Among days in channel HULL during 2017 that exhibit the trend pattern 'rapid fall, then steady stable, then rise, then slow fall', identify the top-8 days where the rapid fall segment is the fastest. (Output format: a ranked list of dates, e.g., ['YYYY-MM-DD', ...])", "answer": "['2017-06-28', '2017-04-11', '2017-02-24', '2017-10-16', '2017-12-27', '2017-11-01', '2017-09-24', '2017-09-02']", "ground_truth": [ "2017-06-28", "2017-04-11", "2017-02-24", "2017-10-16", "2017-12-27", "2017-11-01", "2017-09-24", "2017-09-02" ], "eval_metric": "set_f1", "channel": "HULL", "ts_data_path": "ts_data/L3_T1_Composite_Trend_S4_00006.csv" }, { "id": "L3_T2_Contextual_Anomaly_S0_00004", "level": 3, "level_name": "Semantic Reasoning", "category": "Contextual Anomaly", "subtask": "Contextual Anomaly", "question": "Identify the period in channel 578 during 2022 that experienced the most significant historically low water level. (Output format: [YYYY-MM-DD, YYYY-MM-DD])", "answer": "[2022-03-17 20:00:00, 2022-04-06 19:45:00]", "ground_truth": [ "2022-03-17 20:00:00", "2022-04-06 19:45:00" ], "eval_metric": "iou", "channel": "578", "ts_data_path": "ts_data/L3_T2_Contextual_Anomaly_S0_00004.csv" }, { "id": "L4_T1_Insight_Synthesis_S2_00002", "level": 4, "level_name": "Insight Synthesis", "category": "Insight Synthesis", "subtask": "Insight Synthesis", "question": "Analyze the behavior of channel 71 for the period 2021-03.\nPlease use ONLY the following phrases for trend description: rapid rise, gradual rise, rapid fall, gradual fall, steady stable, fluctuating stable.\nProvide a structured report covering:\n1. Trend Segmentation: Describe each stage with precise start/end timestamps (HH:MM:SS) using the phrases above.\n2. Outlier Audit: Identify only significant outliers that deviate sharply from the local trend. Ignore minor background noise.\n(Output format: structured natural language report.)", "answer": "1. Trend Segmentation: from 2021-03-01 00:00:00 to 2021-03-13 09:30:00, the trend showed a rapid fall; from 2021-03-13 09:30:00 to 2021-03-31 23:45:00, the trend showed a steady stable.\n2. Outlier Audit: A significant spike was detected at 2021-03-12 01:00:00 (value: 7.19).", "ground_truth": { "trend_segments": [ "from 2021-03-01 00:00:00 to 2021-03-13 09:30:00, the trend showed a rapid fall", "from 2021-03-13 09:30:00 to 2021-03-31 23:45:00, the trend showed a steady stable" ], "segments_meta": [ { "adj": "rapid", "kind": "fall", "start_idx": 0, "end_idx": 1190 }, { "adj": "steady", "kind": "stable", "start_idx": 1190, "end_idx": 2976 } ], "significant_anomaly": { "timestamp": "2021-03-12 01:00:00", "kind": "significant spike", "value": 7.186938362634227 } }, "eval_metric": "report", "channel": "71", "ts_data_path": "ts_data/L4_T1_Insight_Synthesis_S2_00002.csv" }, { "id": "L2_T2_Periodicity_Detection_S0_00008", "level": 2, "level_name": "Pattern Recognition", "category": "Periodicity Detection", "subtask": "Periodicity Detection", "question": "What is the dominant cycle period (in data points) of channel MULL within [2018-02-04 12:45:00 to 2018-02-18 05:45:00]? (Output format: integer)", "answer": "156", "ground_truth": 156, "eval_metric": "rel_acc", "channel": "MULL", "ts_data_path": "ts_data/L2_T2_Periodicity_Detection_S0_00008.csv" }, { "id": "L3_T2_Contextual_Anomaly_S2_00002", "level": 3, "level_name": "Semantic Reasoning", "category": "Contextual Anomaly", "subtask": "Contextual Anomaly", "question": "Identify the period in channel 237 during 2021 that experienced the most significant severe drought. (Output format: [YYYY-MM-DD, YYYY-MM-DD])", "answer": "[2021-09-28 18:30:00, 2021-10-29 18:15:00]", "ground_truth": [ "2021-09-28 18:30:00", "2021-10-29 18:15:00" ], "eval_metric": "iou", "channel": "237", "ts_data_path": "ts_data/L3_T2_Contextual_Anomaly_S2_00002.csv" }, { "id": "L1_T2_Temporal_Localization_S2_00002", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Temporal Localization", "question": "At what exact timestamp did channel 147 first rise above 0.159 in 2019-02-12 to 2019-05-22? (Output format: a specific timestamp, e.g., YYYY-MM-DD HH:MM:SS)", "answer": "2019-02-13 13:00:00", "ground_truth": "2019-02-13 13:00:00", "eval_metric": "hit", "channel": "147", "ts_data_path": "ts_data/L1_T2_Temporal_Localization_S2_00002.csv" }, { "id": "L3_T3_Causal_Anomaly_S4_00004", "level": 3, "level_name": "Semantic Reasoning", "category": "Causal Anomaly", "subtask": "Causal Anomaly", "question": "Given that channel 627_up_2023 is the upstream source of channel 627_down_2023, identify the time period in 2023 where 627_down_2023 shows a significant causal anomaly, such as an inverse trend against the source. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2023-04-03 23:15:00, 2023-04-14 17:30:00]", "ground_truth": [ "2023-04-03 23:15:00", "2023-04-14 17:30:00" ], "eval_metric": "iou", "channel": "627_down_2023", "ts_data_path": "ts_data/L3_T3_Causal_Anomaly_S4_00004.csv" }, { "id": "L3_T1_Composite_Trend_S1_00004", "level": 3, "level_name": "Semantic Reasoning", "category": "Composite Trend", "subtask": "Composite Trend", "question": "Among days in channel 124 during 2019 that exhibit the trend pattern 'rapid fall, then fluctuating stable, then steady stable, then fluctuating stable', identify the top-2 days where the rapid fall segment is the fastest. (Output format: a ranked list of dates, e.g., ['YYYY-MM-DD', ...])", "answer": "['2019-12-16', '2019-07-06']", "ground_truth": [ "2019-12-16", "2019-07-06" ], "eval_metric": "set_f1", "channel": "124", "ts_data_path": "ts_data/L3_T1_Composite_Trend_S1_00004.csv" }, { "id": "L1_T4_Sliding_Window_S4_00004", "level": 1, "level_name": "Basic Operations", "category": "Sliding Window", "subtask": "Sliding Window", "question": "Which 21-day period in 2020-08-16 to 2020-11-28 had the highest variance for channel 727? (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2020-10-04 08:15:00, 2020-10-25 08:00:00]", "ground_truth": [ "2020-10-04 08:15:00", "2020-10-25 08:00:00" ], "eval_metric": "iou", "channel": "727", "ts_data_path": "ts_data/L1_T4_Sliding_Window_S4_00004.csv" }, { "id": "L2_T3_Subsequence_Matching_S2_00001", "level": 2, "level_name": "Pattern Recognition", "category": "Subsequence Matching", "subtask": "Subsequence Matching", "question": "Analyze the reference pattern in [2019-10-08 12:00:00 to 2019-10-08 21:45:00]. Find the time interval where channel 169 exhibits the most similar pattern within the search context [2019-10-11 01:00:00 to 2019-10-19 03:15:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2019-10-14 13:30:00', '2019-10-14 23:15:00']", "ground_truth": [ "2019-10-14 13:30:00", "2019-10-14 23:15:00" ], "eval_metric": "iou", "channel": "169", "ts_data_path": "ts_data/L2_T3_Subsequence_Matching_S2_00001.csv" }, { "id": "L1_T1_Global_Aggregation_S4_00008", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Global Aggregation", "question": "What is the range value of channel HULL in 2017-04? (Output format: a single numeric value, rounded to 3 decimal places, e.g., x.xxx)", "answer": "11.453", "ground_truth": 11.453, "eval_metric": "rel_acc", "channel": "HULL", "ts_data_path": "ts_data/L1_T1_Global_Aggregation_S4_00008.csv" }, { "id": "L4_T1_Insight_Synthesis_S3_00006", "level": 4, "level_name": "Insight Synthesis", "category": "Insight Synthesis", "subtask": "Insight Synthesis", "question": "Analyze the behavior of channel LULL for the period 2016-11.\nPlease use ONLY the following phrases for trend description: rapid rise, gradual rise, rapid fall, gradual fall, steady stable, fluctuating stable.\nProvide a structured report covering:\n1. Trend Segmentation: Describe each stage with precise start/end timestamps (HH:MM:SS) using the phrases above.\n2. Outlier Audit: Identify only significant outliers that deviate sharply from the local trend. Ignore minor background noise.\n(Output format: structured natural language report.)", "answer": "1. Trend Segmentation: from 2016-11-01 00:00:00 to 2016-11-05 06:45:00, the trend showed a rapid fall; from 2016-11-05 06:45:00 to 2016-11-09 13:30:00, the trend showed a gradual fall; from 2016-11-09 13:30:00 to 2016-11-20 06:45:00, the trend showed a rapid fall; from 2016-11-20 06:45:00 to 2016-11-26 17:00:00, the trend showed a rapid rise; from 2016-11-26 17:00:00 to 2016-11-30 23:45:00, the trend showed a rapid fall.\n2. Outlier Audit: A significant drop was detected at 2016-11-26 15:30:00 (value: -9.25).", "ground_truth": { "trend_segments": [ "from 2016-11-01 00:00:00 to 2016-11-05 06:45:00, the trend showed a rapid fall", "from 2016-11-05 06:45:00 to 2016-11-09 13:30:00, the trend showed a gradual fall", "from 2016-11-09 13:30:00 to 2016-11-20 06:45:00, the trend showed a rapid fall", "from 2016-11-20 06:45:00 to 2016-11-26 17:00:00, the trend showed a rapid rise", "from 2016-11-26 17:00:00 to 2016-11-30 23:45:00, the trend showed a rapid fall" ], "segments_meta": [ { "adj": "rapid", "kind": "fall", "start_idx": 0, "end_idx": 411 }, { "adj": "gradual", "kind": "fall", "start_idx": 411, "end_idx": 822 }, { "adj": "rapid", "kind": "fall", "start_idx": 822, "end_idx": 1851 }, { "adj": "rapid", "kind": "rise", "start_idx": 1851, "end_idx": 2468 }, { "adj": "rapid", "kind": "fall", "start_idx": 2468, "end_idx": 2880 } ], "significant_anomaly": { "timestamp": "2016-11-26 15:30:00", "kind": "significant drop", "value": -9.252551989697247 } }, "eval_metric": "report", "channel": "LULL", "ts_data_path": "ts_data/L4_T1_Insight_Synthesis_S3_00006.csv" }, { "id": "L2_T3_Subsequence_Matching_S1_00006", "level": 2, "level_name": "Pattern Recognition", "category": "Subsequence Matching", "subtask": "Subsequence Matching", "question": "Analyze the reference pattern in [2019-02-21 04:30:00 to 2019-02-21 10:15:00]. Find the time interval where channel 146 exhibits the most similar pattern within the search context [2019-02-22 18:00:00 to 2019-02-27 18:00:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2019-02-24 18:45:00', '2019-02-25 00:30:00']", "ground_truth": [ "2019-02-24 18:45:00", "2019-02-25 00:30:00" ], "eval_metric": "iou", "channel": "146", "ts_data_path": "ts_data/L2_T3_Subsequence_Matching_S1_00006.csv" }, { "id": "L1_T2_Temporal_Localization_S2_00008", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Temporal Localization", "question": "At what exact timestamp did channel OT reach its minimum value in 2018-03-27 to 2018-04-28? (Output format: a specific timestamp, e.g., YYYY-MM-DD HH:MM:SS)", "answer": "2018-04-05 15:15:00", "ground_truth": "2018-04-05 15:15:00", "eval_metric": "hit", "channel": "OT", "ts_data_path": "ts_data/L1_T2_Temporal_Localization_S2_00008.csv" }, { "id": "L3_T1_Composite_Trend_S4_00005", "level": 3, "level_name": "Semantic Reasoning", "category": "Composite Trend", "subtask": "Composite Trend", "question": "Among days in channel HUFL during 2017 that exhibit the trend pattern 'rapid fall, then steady stable, then slow fall, then fluctuating stable', identify the top-8 days where the slow fall segment is the slowest. (Output format: a ranked list of dates, e.g., ['YYYY-MM-DD', ...])", "answer": "['2017-05-24', '2017-12-03', '2017-03-19', '2017-11-17', '2017-06-16', '2017-02-12', '2017-03-11', '2017-09-25']", "ground_truth": [ "2017-05-24", "2017-12-03", "2017-03-19", "2017-11-17", "2017-06-16", "2017-02-12", "2017-03-11", "2017-09-25" ], "eval_metric": "set_f1", "channel": "HUFL", "ts_data_path": "ts_data/L3_T1_Composite_Trend_S4_00005.csv" }, { "id": "L2_T2_Periodicity_Detection_S1_00003", "level": 2, "level_name": "Pattern Recognition", "category": "Periodicity Detection", "subtask": "Periodicity Detection", "question": "What is the dominant cycle period (in data points) of channel 589 within [2020-06-15 17:15:00 to 2020-06-17 21:00:00]? (Output format: integer)", "answer": "69", "ground_truth": 69, "eval_metric": "rel_acc", "channel": "589", "ts_data_path": "ts_data/L2_T2_Periodicity_Detection_S1_00003.csv" }, { "id": "L3_T3_Causal_Anomaly_S4_00006", "level": 3, "level_name": "Semantic Reasoning", "category": "Causal Anomaly", "subtask": "Causal Anomaly", "question": "Given that channel 177_up_2022 is the upstream source of channel 177_down_2022, identify the time period in 2022 where 177_down_2022 shows a significant causal anomaly, such as an inverse trend against the source. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2022-10-25 06:30:00, 2022-11-04 18:00:00]", "ground_truth": [ "2022-10-25 06:30:00", "2022-11-04 18:00:00" ], "eval_metric": "iou", "channel": "177_down_2022", "ts_data_path": "ts_data/L3_T3_Causal_Anomaly_S4_00006.csv" }, { "id": "L1_T3_Interval_Discovery_S3_00008", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Interval Discovery", "question": "Find the longest period where channel MUFL remained above 8.315 in 2017-03-15 to 2017-05-10. (Output format: a time interval, e.g., [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2017-03-17 22:15:00, 2017-03-18 07:45:00]", "ground_truth": [ "2017-03-17 22:15:00", "2017-03-18 07:45:00" ], "eval_metric": "iou", "channel": "MUFL", "ts_data_path": "ts_data/L1_T3_Interval_Discovery_S3_00008.csv" }, { "id": "L2_T1_Shape_Identification_S3_00001", "level": 2, "level_name": "Pattern Recognition", "category": "Shape Identification", "subtask": "Shape Identification", "question": "Identify the time range of the deepest deep valley in channel 147 within [2019-05-05 21:45:00 to 2019-05-07 00:30:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2019-05-06 18:15:00', '2019-05-06 22:00:00']", "ground_truth": [ "2019-05-06 18:15:00", "2019-05-06 22:00:00" ], "eval_metric": "iou", "channel": "147", "ts_data_path": "ts_data/L2_T1_Shape_Identification_S3_00001.csv" }, { "id": "L1_T4_Sliding_Window_S4_00009", "level": 1, "level_name": "Basic Operations", "category": "Sliding Window", "subtask": "Sliding Window", "question": "Which 4-day period in 2020-08-05 to 2021-02-12 had the highest average for channel 591? (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2020-10-25 03:00:00, 2020-10-29 02:45:00]", "ground_truth": [ "2020-10-25 03:00:00", "2020-10-29 02:45:00" ], "eval_metric": "iou", "channel": "591", "ts_data_path": "ts_data/L1_T4_Sliding_Window_S4_00009.csv" }, { "id": "L4_T1_Insight_Synthesis_S1_00002", "level": 4, "level_name": "Insight Synthesis", "category": "Insight Synthesis", "subtask": "Insight Synthesis", "question": "Analyze the behavior of channel 495 for the period 2021-08.\nPlease use ONLY the following phrases for trend description: rapid rise, gradual rise, rapid fall, gradual fall, steady stable, fluctuating stable.\nProvide a structured report covering:\n1. Trend Segmentation: Describe each stage with precise start/end timestamps (HH:MM:SS) using the phrases above.\n2. Outlier Audit: Identify only significant outliers that deviate sharply from the local trend. Ignore minor background noise.\n(Output format: structured natural language report.)", "answer": "1. Trend Segmentation: from 2021-08-01 00:00:00 to 2021-08-05 03:15:00, the trend showed a fluctuating stable; from 2021-08-05 03:15:00 to 2021-08-09 06:30:00, the trend showed a gradual rise; from 2021-08-09 06:30:00 to 2021-08-13 09:45:00, the trend showed a gradual fall; from 2021-08-13 09:45:00 to 2021-08-25 19:15:00, the trend showed a rapid rise; from 2021-08-25 19:15:00 to 2021-08-31 23:45:00, the trend showed a steady stable.\n2. Outlier Audit: A significant drop was detected at 2021-08-24 03:15:00 (value: -3.33).", "ground_truth": { "trend_segments": [ "from 2021-08-01 00:00:00 to 2021-08-05 03:15:00, the trend showed a fluctuating stable", "from 2021-08-05 03:15:00 to 2021-08-09 06:30:00, the trend showed a gradual rise", "from 2021-08-09 06:30:00 to 2021-08-13 09:45:00, the trend showed a gradual fall", "from 2021-08-13 09:45:00 to 2021-08-25 19:15:00, the trend showed a rapid rise", "from 2021-08-25 19:15:00 to 2021-08-31 23:45:00, the trend showed a steady stable" ], "segments_meta": [ { "adj": "fluctuating", "kind": "stable", "start_idx": 0, "end_idx": 397 }, { "adj": "gradual", "kind": "rise", "start_idx": 397, "end_idx": 794 }, { "adj": "gradual", "kind": "fall", "start_idx": 794, "end_idx": 1191 }, { "adj": "rapid", "kind": "rise", "start_idx": 1191, "end_idx": 2381 }, { "adj": "steady", "kind": "stable", "start_idx": 2381, "end_idx": 2976 } ], "significant_anomaly": { "timestamp": "2021-08-24 03:15:00", "kind": "significant drop", "value": -3.3339344188083824 } }, "eval_metric": "report", "channel": "495", "ts_data_path": "ts_data/L4_T1_Insight_Synthesis_S1_00002.csv" }, { "id": "L1_T3_Interval_Discovery_S0_00005", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Interval Discovery", "question": "Find the longest period where channel 754 remained above 0.357 in 2021-07-20 to 2021-11-15. (Output format: a time interval, e.g., [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2021-08-30 17:45:00, 2021-09-17 09:30:00]", "ground_truth": [ "2021-08-30 17:45:00", "2021-09-17 09:30:00" ], "eval_metric": "iou", "channel": "754", "ts_data_path": "ts_data/L1_T3_Interval_Discovery_S0_00005.csv" }, { "id": "L1_T4_Sliding_Window_S1_00001", "level": 1, "level_name": "Basic Operations", "category": "Sliding Window", "subtask": "Sliding Window", "question": "Which 77-day period in 2022-02-13 to 2022-12-16 had the lowest average for channel 245? (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2022-06-10 09:30:00, 2022-08-26 09:15:00]", "ground_truth": [ "2022-06-10 09:30:00", "2022-08-26 09:15:00" ], "eval_metric": "iou", "channel": "245", "ts_data_path": "ts_data/L1_T4_Sliding_Window_S1_00001.csv" }, { "id": "L1_T1_Global_Aggregation_S0_00001", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Global Aggregation", "question": "What is the median value of channel 496 in 2023-10-28 to 2023-11-12? (Output format: a single numeric value, rounded to 3 decimal places, e.g., x.xxx)", "answer": "0.265", "ground_truth": 0.265, "eval_metric": "rel_acc", "channel": "496", "ts_data_path": "ts_data/L1_T1_Global_Aggregation_S0_00001.csv" }, { "id": "L1_T4_Sliding_Window_S3_00009", "level": 1, "level_name": "Basic Operations", "category": "Sliding Window", "subtask": "Sliding Window", "question": "Which 99-day period in 2020-12-20 to 2021-11-21 had the largest range for channel 625? (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2021-02-01 14:15:00, 2021-05-11 14:00:00]", "ground_truth": [ "2021-02-01 14:15:00", "2021-05-11 14:00:00" ], "eval_metric": "iou", "channel": "625", "ts_data_path": "ts_data/L1_T4_Sliding_Window_S3_00009.csv" }, { "id": "L1_T4_Sliding_Window_S2_00004", "level": 1, "level_name": "Basic Operations", "category": "Sliding Window", "subtask": "Sliding Window", "question": "Which 27-day period in 2016-10-19 to 2017-06-14 had the highest average for channel HUFL? (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2016-12-23 22:30:00, 2017-01-19 22:15:00]", "ground_truth": [ "2016-12-23 22:30:00", "2017-01-19 22:15:00" ], "eval_metric": "iou", "channel": "HUFL", "ts_data_path": "ts_data/L1_T4_Sliding_Window_S2_00004.csv" }, { "id": "L1_T1_Global_Aggregation_S3_00008", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Global Aggregation", "question": "What is the minimum value of channel 71 in 2022-09-08 to 2022-09-28? (Output format: a single numeric value, rounded to 3 decimal places, e.g., x.xxx)", "answer": "0.037", "ground_truth": 0.037, "eval_metric": "rel_acc", "channel": "71", "ts_data_path": "ts_data/L1_T1_Global_Aggregation_S3_00008.csv" }, { "id": "L2_T3_Subsequence_Matching_S1_00005", "level": 2, "level_name": "Pattern Recognition", "category": "Subsequence Matching", "subtask": "Subsequence Matching", "question": "Analyze the reference pattern in [2017-07-01 23:15:00 to 2017-07-02 05:00:00]. Find the time interval where channel LUFL exhibits the most similar pattern within the search context [2017-07-04 17:00:00 to 2017-07-14 10:45:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2017-07-10 03:00:00', '2017-07-10 08:45:00']", "ground_truth": [ "2017-07-10 03:00:00", "2017-07-10 08:45:00" ], "eval_metric": "iou", "channel": "LUFL", "ts_data_path": "ts_data/L2_T3_Subsequence_Matching_S1_00005.csv" }, { "id": "L2_T1_Shape_Identification_S1_00007", "level": 2, "level_name": "Pattern Recognition", "category": "Shape Identification", "subtask": "Shape Identification", "question": "Identify the time range of the longest plateau (stable period) in channel MULL within [2017-12-02 21:45:00 to 2017-12-09 18:00:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2017-12-03 11:30:00', '2017-12-04 11:45:00']", "ground_truth": [ "2017-12-03 11:30:00", "2017-12-04 11:45:00" ], "eval_metric": "iou", "channel": "MULL", "ts_data_path": "ts_data/L2_T1_Shape_Identification_S1_00007.csv" }, { "id": "L3_T2_Contextual_Anomaly_S4_00008", "level": 3, "level_name": "Semantic Reasoning", "category": "Contextual Anomaly", "subtask": "Contextual Anomaly", "question": "Identify the period in channel 728 during 2020 that experienced the most significant dry-out period. (Output format: [YYYY-MM-DD, YYYY-MM-DD])", "answer": "[2020-04-01 16:30:00, 2020-05-01 16:15:00]", "ground_truth": [ "2020-04-01 16:30:00", "2020-05-01 16:15:00" ], "eval_metric": "iou", "channel": "728", "ts_data_path": "ts_data/L3_T2_Contextual_Anomaly_S4_00008.csv" }, { "id": "L1_T4_Sliding_Window_S3_00005", "level": 1, "level_name": "Basic Operations", "category": "Sliding Window", "subtask": "Sliding Window", "question": "Which 22-day period in 2017-03-13 to 2018-02-04 had the largest range for channel LULL? (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2017-06-26 10:00:00, 2017-07-18 09:45:00]", "ground_truth": [ "2017-06-26 10:00:00", "2017-07-18 09:45:00" ], "eval_metric": "iou", "channel": "LULL", "ts_data_path": "ts_data/L1_T4_Sliding_Window_S3_00005.csv" }, { "id": "L2_T2_Periodicity_Detection_S2_00007", "level": 2, "level_name": "Pattern Recognition", "category": "Periodicity Detection", "subtask": "Periodicity Detection", "question": "What is the dominant cycle period (in data points) of channel MULL within [2016-08-28 16:00:00 to 2016-09-11 00:15:00]? (Output format: integer)", "answer": "70", "ground_truth": 70, "eval_metric": "rel_acc", "channel": "MULL", "ts_data_path": "ts_data/L2_T2_Periodicity_Detection_S2_00007.csv" }, { "id": "L2_T2_Periodicity_Detection_S1_00002", "level": 2, "level_name": "Pattern Recognition", "category": "Periodicity Detection", "subtask": "Periodicity Detection", "question": "What is the dominant cycle period (in data points) of channel 146 within [2019-02-17 02:30:00 to 2019-02-24 00:45:00]? (Output format: integer)", "answer": "26", "ground_truth": 26, "eval_metric": "rel_acc", "channel": "146", "ts_data_path": "ts_data/L2_T2_Periodicity_Detection_S1_00002.csv" }, { "id": "L1_T1_Global_Aggregation_S1_00009", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Global Aggregation", "question": "What is the range value of channel 813 in 2022-03-24 to 2022-03-29? (Output format: a single numeric value, rounded to 3 decimal places, e.g., x.xxx)", "answer": "0.335", "ground_truth": 0.335, "eval_metric": "rel_acc", "channel": "813", "ts_data_path": "ts_data/L1_T1_Global_Aggregation_S1_00009.csv" }, { "id": "L1_T2_Temporal_Localization_S1_00004", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Temporal Localization", "question": "At what exact timestamp did channel 312 reach its maximum value in 2021-01? (Output format: a specific timestamp, e.g., YYYY-MM-DD HH:MM:SS)", "answer": "2021-01-31 10:00:00", "ground_truth": "2021-01-31 10:00:00", "eval_metric": "hit", "channel": "312", "ts_data_path": "ts_data/L1_T2_Temporal_Localization_S1_00004.csv" }, { "id": "L3_T1_Composite_Trend_S2_00004", "level": 3, "level_name": "Semantic Reasoning", "category": "Composite Trend", "subtask": "Composite Trend", "question": "Among days in channel 501 during 2023 that exhibit the trend pattern 'steady stable, then rapid fall, then rapid rise', identify the top-4 days where the rapid rise segment is the fastest. (Output format: a ranked list of dates, e.g., ['YYYY-MM-DD', ...])", "answer": "['2023-09-21', '2023-05-22', '2023-04-06', '2023-10-26']", "ground_truth": [ "2023-09-21", "2023-05-22", "2023-04-06", "2023-10-26" ], "eval_metric": "set_f1", "channel": "501", "ts_data_path": "ts_data/L3_T1_Composite_Trend_S2_00004.csv" }, { "id": "L3_T1_Composite_Trend_S4_00001", "level": 3, "level_name": "Semantic Reasoning", "category": "Composite Trend", "subtask": "Composite Trend", "question": "Among days in channel 99 during 2021 that exhibit the trend pattern 'fluctuating stable, then slow fall, then rapid fall', identify the top-8 days where the rapid fall segment is the fastest. (Output format: a ranked list of dates, e.g., ['YYYY-MM-DD', ...])", "answer": "['2021-06-29', '2021-11-03', '2021-05-01', '2021-09-11', '2021-12-29', '2021-01-21', '2021-03-07', '2021-03-01']", "ground_truth": [ "2021-06-29", "2021-11-03", "2021-05-01", "2021-09-11", "2021-12-29", "2021-01-21", "2021-03-07", "2021-03-01" ], "eval_metric": "set_f1", "channel": "99", "ts_data_path": "ts_data/L3_T1_Composite_Trend_S4_00001.csv" }, { "id": "L1_T4_Sliding_Window_S4_00007", "level": 1, "level_name": "Basic Operations", "category": "Sliding Window", "subtask": "Sliding Window", "question": "Which 11-day period in 2016-08-31 to 2017-01-12 had the largest range for channel MUFL? (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2016-12-05 23:45:00, 2016-12-16 23:30:00]", "ground_truth": [ "2016-12-05 23:45:00", "2016-12-16 23:30:00" ], "eval_metric": "iou", "channel": "MUFL", "ts_data_path": "ts_data/L1_T4_Sliding_Window_S4_00007.csv" }, { "id": "L1_T2_Temporal_Localization_S0_00009", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Temporal Localization", "question": "At what exact timestamp did channel OT reach its maximum value in 2017-10? (Output format: a specific timestamp, e.g., YYYY-MM-DD HH:MM:SS)", "answer": "2017-10-09 13:30:00", "ground_truth": "2017-10-09 13:30:00", "eval_metric": "hit", "channel": "OT", "ts_data_path": "ts_data/L1_T2_Temporal_Localization_S0_00009.csv" }, { "id": "L2_T3_Subsequence_Matching_S0_00005", "level": 2, "level_name": "Pattern Recognition", "category": "Subsequence Matching", "subtask": "Subsequence Matching", "question": "Analyze the reference pattern in [2020-07-24 15:00:00 to 2020-07-25 14:45:00]. Find the time interval where channel 430 exhibits the most similar pattern within the search context [2020-07-27 22:00:00 to 2020-08-05 02:45:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2020-08-02 15:30:00', '2020-08-03 15:15:00']", "ground_truth": [ "2020-08-02 15:30:00", "2020-08-03 15:15:00" ], "eval_metric": "iou", "channel": "430", "ts_data_path": "ts_data/L2_T3_Subsequence_Matching_S0_00005.csv" }, { "id": "L2_T2_Periodicity_Detection_S3_00007", "level": 2, "level_name": "Pattern Recognition", "category": "Periodicity Detection", "subtask": "Periodicity Detection", "question": "What is the dominant cycle period (in data points) of channel MUFL within [2016-11-12 01:30:00 to 2016-11-19 12:15:00]? (Output format: integer)", "answer": "74", "ground_truth": 74, "eval_metric": "rel_acc", "channel": "MUFL", "ts_data_path": "ts_data/L2_T2_Periodicity_Detection_S3_00007.csv" }, { "id": "L2_T2_Periodicity_Detection_S4_00008", "level": 2, "level_name": "Pattern Recognition", "category": "Periodicity Detection", "subtask": "Periodicity Detection", "question": "What is the dominant cycle period (in data points) of channel HULL within [2017-04-21 06:00:00 to 2017-04-25 20:15:00]? (Output format: integer)", "answer": "147", "ground_truth": 147, "eval_metric": "rel_acc", "channel": "HULL", "ts_data_path": "ts_data/L2_T2_Periodicity_Detection_S4_00008.csv" }, { "id": "L2_T1_Shape_Identification_S0_00008", "level": 2, "level_name": "Pattern Recognition", "category": "Shape Identification", "subtask": "Shape Identification", "question": "Identify the time range of the largest step ascent in channel MUFL within [2016-07-01 00:00:00 to 2016-07-13 12:15:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2016-07-05 04:00:00', '2016-07-07 00:45:00']", "ground_truth": [ "2016-07-05 04:00:00", "2016-07-07 00:45:00" ], "eval_metric": "iou", "channel": "MUFL", "ts_data_path": "ts_data/L2_T1_Shape_Identification_S0_00008.csv" }, { "id": "L3_T3_Causal_Anomaly_S1_00003", "level": 3, "level_name": "Semantic Reasoning", "category": "Causal Anomaly", "subtask": "Causal Anomaly", "question": "Given that channel 625_up_2019 is the upstream source of channel 625_down_2019, identify the time period in 2019 where 625_down_2019 shows a significant causal anomaly, such as an inverse trend against the source. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2019-10-15 14:15:00, 2019-10-17 05:15:00]", "ground_truth": [ "2019-10-15 14:15:00", "2019-10-17 05:15:00" ], "eval_metric": "iou", "channel": "625_down_2019", "ts_data_path": "ts_data/L3_T3_Causal_Anomaly_S1_00003.csv" }, { "id": "L3_T3_Causal_Anomaly_S4_00005", "level": 3, "level_name": "Semantic Reasoning", "category": "Causal Anomaly", "subtask": "Causal Anomaly", "question": "Given that channel 99_up_2022 is the upstream source of channel 99_down_2022, identify the time period in 2022 where 99_down_2022 shows a significant causal anomaly, such as an inverse trend against the source. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2022-04-05 14:00:00, 2022-04-17 13:00:00]", "ground_truth": [ "2022-04-05 14:00:00", "2022-04-17 13:00:00" ], "eval_metric": "iou", "channel": "99_down_2022", "ts_data_path": "ts_data/L3_T3_Causal_Anomaly_S4_00005.csv" }, { "id": "L2_T2_Periodicity_Detection_S2_00009", "level": 2, "level_name": "Pattern Recognition", "category": "Periodicity Detection", "subtask": "Periodicity Detection", "question": "What is the dominant cycle period (in data points) of channel 441 within [2020-01-12 12:00:00 to 2020-01-25 14:30:00]? (Output format: integer)", "answer": "119", "ground_truth": 119, "eval_metric": "rel_acc", "channel": "441", "ts_data_path": "ts_data/L2_T2_Periodicity_Detection_S2_00009.csv" }, { "id": "L1_T1_Global_Aggregation_S3_00002", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Global Aggregation", "question": "What is the median value of channel 146 in 2020-05-07 to 2020-05-30? (Output format: a single numeric value, rounded to 3 decimal places, e.g., x.xxx)", "answer": "0.42", "ground_truth": 0.42, "eval_metric": "rel_acc", "channel": "146", "ts_data_path": "ts_data/L1_T1_Global_Aggregation_S3_00002.csv" }, { "id": "L2_T3_Subsequence_Matching_S3_00007", "level": 2, "level_name": "Pattern Recognition", "category": "Subsequence Matching", "subtask": "Subsequence Matching", "question": "Analyze the reference pattern in [2020-05-06 16:00:00 to 2020-05-07 09:45:00]. Find the time interval where channel 897 exhibits the most similar pattern within the search context [2020-05-08 13:00:00 to 2020-05-10 20:00:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2020-05-09 01:15:00', '2020-05-09 19:00:00']", "ground_truth": [ "2020-05-09 01:15:00", "2020-05-09 19:00:00" ], "eval_metric": "iou", "channel": "897", "ts_data_path": "ts_data/L2_T3_Subsequence_Matching_S3_00007.csv" }, { "id": "L2_T3_Subsequence_Matching_S2_00006", "level": 2, "level_name": "Pattern Recognition", "category": "Subsequence Matching", "subtask": "Subsequence Matching", "question": "Analyze the reference pattern in [2019-07-16 05:00:00 to 2019-07-16 10:45:00]. Find the time interval where channel 170 exhibits the most similar pattern within the search context [2019-07-19 08:15:00 to 2019-07-30 15:00:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2019-07-19 22:30:00', '2019-07-20 04:15:00']", "ground_truth": [ "2019-07-19 22:30:00", "2019-07-20 04:15:00" ], "eval_metric": "iou", "channel": "170", "ts_data_path": "ts_data/L2_T3_Subsequence_Matching_S2_00006.csv" }, { "id": "L2_T2_Periodicity_Detection_S4_00002", "level": 2, "level_name": "Pattern Recognition", "category": "Periodicity Detection", "subtask": "Periodicity Detection", "question": "What is the dominant cycle period (in data points) of channel 147 within [2022-07-27 21:00:00 to 2022-07-30 21:00:00]? (Output format: integer)", "answer": "96", "ground_truth": 96, "eval_metric": "rel_acc", "channel": "147", "ts_data_path": "ts_data/L2_T2_Periodicity_Detection_S4_00002.csv" }, { "id": "L1_T2_Temporal_Localization_S2_00009", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Temporal Localization", "question": "At what exact timestamp did channel 237 last fall below 0.143 in 2022-08-29 to 2022-11-09? (Output format: a specific timestamp, e.g., YYYY-MM-DD HH:MM:SS)", "answer": "2022-09-13 09:45:00", "ground_truth": "2022-09-13 09:45:00", "eval_metric": "hit", "channel": "237", "ts_data_path": "ts_data/L1_T2_Temporal_Localization_S2_00009.csv" }, { "id": "L1_T4_Sliding_Window_S0_00001", "level": 1, "level_name": "Basic Operations", "category": "Sliding Window", "subtask": "Sliding Window", "question": "Which 12-day period in 2019-10-05 to 2020-01-04 had the lowest average for channel 67? (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2019-10-06 18:45:00, 2019-10-18 18:30:00]", "ground_truth": [ "2019-10-06 18:45:00", "2019-10-18 18:30:00" ], "eval_metric": "iou", "channel": "67", "ts_data_path": "ts_data/L1_T4_Sliding_Window_S0_00001.csv" }, { "id": "L1_T4_Sliding_Window_S3_00006", "level": 1, "level_name": "Basic Operations", "category": "Sliding Window", "subtask": "Sliding Window", "question": "Which 117-day period in 2016-09-21 to 2017-06-16 had the lowest average for channel OT? (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2016-12-05 11:15:00, 2017-04-01 11:00:00]", "ground_truth": [ "2016-12-05 11:15:00", "2017-04-01 11:00:00" ], "eval_metric": "iou", "channel": "OT", "ts_data_path": "ts_data/L1_T4_Sliding_Window_S3_00006.csv" }, { "id": "L1_T4_Sliding_Window_S1_00007", "level": 1, "level_name": "Basic Operations", "category": "Sliding Window", "subtask": "Sliding Window", "question": "Which 38-day period in 2017-06-21 to 2018-03-10 had the lowest average for channel MULL? (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2017-08-29 00:15:00, 2017-10-06 00:00:00]", "ground_truth": [ "2017-08-29 00:15:00", "2017-10-06 00:00:00" ], "eval_metric": "iou", "channel": "MULL", "ts_data_path": "ts_data/L1_T4_Sliding_Window_S1_00007.csv" }, { "id": "L1_T4_Sliding_Window_S0_00007", "level": 1, "level_name": "Basic Operations", "category": "Sliding Window", "subtask": "Sliding Window", "question": "Which 12-day period in 2022-12-17 to 2023-05-07 had the highest variance for channel 147? (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2023-03-05 06:30:00, 2023-03-17 06:15:00]", "ground_truth": [ "2023-03-05 06:30:00", "2023-03-17 06:15:00" ], "eval_metric": "iou", "channel": "147", "ts_data_path": "ts_data/L1_T4_Sliding_Window_S0_00007.csv" }, { "id": "L1_T3_Interval_Discovery_S4_00009", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Interval Discovery", "question": "Find the longest period where channel 155 remained above 0.245 in 2022-11-28 to 2023-07-27. (Output format: a time interval, e.g., [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2023-03-06 22:45:00, 2023-04-19 15:45:00]", "ground_truth": [ "2023-03-06 22:45:00", "2023-04-19 15:45:00" ], "eval_metric": "iou", "channel": "155", "ts_data_path": "ts_data/L1_T3_Interval_Discovery_S4_00009.csv" }, { "id": "L1_T2_Temporal_Localization_S4_00009", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Temporal Localization", "question": "At what exact timestamp did channel 626 first rise above 1.53 in 2022-01-16 to 2022-03-11? (Output format: a specific timestamp, e.g., YYYY-MM-DD HH:MM:SS)", "answer": "2022-02-18 03:45:00", "ground_truth": "2022-02-18 03:45:00", "eval_metric": "hit", "channel": "626", "ts_data_path": "ts_data/L1_T2_Temporal_Localization_S4_00009.csv" }, { "id": "L2_T1_Shape_Identification_S1_00001", "level": 2, "level_name": "Pattern Recognition", "category": "Shape Identification", "subtask": "Shape Identification", "question": "Identify the time range of the largest step descent in channel 71 within [2023-10-19 23:45:00 to 2023-10-30 13:15:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2023-10-20 02:45:00', '2023-10-21 16:30:00']", "ground_truth": [ "2023-10-20 02:45:00", "2023-10-21 16:30:00" ], "eval_metric": "iou", "channel": "71", "ts_data_path": "ts_data/L2_T1_Shape_Identification_S1_00001.csv" }, { "id": "L3_T2_Contextual_Anomaly_S2_00001", "level": 3, "level_name": "Semantic Reasoning", "category": "Contextual Anomaly", "subtask": "Contextual Anomaly", "question": "Identify the period in channel 173 during 2023 that experienced the most significant historically low water level. (Output format: [YYYY-MM-DD, YYYY-MM-DD])", "answer": "[2023-07-11 21:15:00, 2023-08-20 21:00:00]", "ground_truth": [ "2023-07-11 21:15:00", "2023-08-20 21:00:00" ], "eval_metric": "iou", "channel": "173", "ts_data_path": "ts_data/L3_T2_Contextual_Anomaly_S2_00001.csv" }, { "id": "L2_T2_Periodicity_Detection_S1_00009", "level": 2, "level_name": "Pattern Recognition", "category": "Periodicity Detection", "subtask": "Periodicity Detection", "question": "What is the dominant cycle period (in data points) of channel 245 within [2022-09-11 00:00:00 to 2022-09-22 02:15:00]? (Output format: integer)", "answer": "167", "ground_truth": 167, "eval_metric": "rel_acc", "channel": "245", "ts_data_path": "ts_data/L2_T2_Periodicity_Detection_S1_00009.csv" }, { "id": "L3_T2_Contextual_Anomaly_S2_00004", "level": 3, "level_name": "Semantic Reasoning", "category": "Contextual Anomaly", "subtask": "Contextual Anomaly", "question": "Identify the period in channel 595 during 2023 that experienced the most significant dry-out period. (Output format: [YYYY-MM-DD, YYYY-MM-DD])", "answer": "[2023-09-26 08:45:00, 2023-10-26 08:30:00]", "ground_truth": [ "2023-09-26 08:45:00", "2023-10-26 08:30:00" ], "eval_metric": "iou", "channel": "595", "ts_data_path": "ts_data/L3_T2_Contextual_Anomaly_S2_00004.csv" }, { "id": "L2_T2_Periodicity_Detection_S3_00006", "level": 2, "level_name": "Pattern Recognition", "category": "Periodicity Detection", "subtask": "Periodicity Detection", "question": "What is the dominant cycle period (in data points) of channel HULL within [2016-12-26 03:00:00 to 2017-01-02 22:15:00]? (Output format: integer)", "answer": "19", "ground_truth": 19, "eval_metric": "rel_acc", "channel": "HULL", "ts_data_path": "ts_data/L2_T2_Periodicity_Detection_S3_00006.csv" }, { "id": "L3_T3_Causal_Anomaly_S0_00005", "level": 3, "level_name": "Semantic Reasoning", "category": "Causal Anomaly", "subtask": "Causal Anomaly", "question": "Given that channel 895_up_2021 is the upstream source of channel 895_down_2021, identify the time period in 2021 where 895_down_2021 shows a significant causal anomaly, such as an flat line during high activity. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2021-09-24 20:00:00, 2021-09-27 13:45:00]", "ground_truth": [ "2021-09-24 20:00:00", "2021-09-27 13:45:00" ], "eval_metric": "iou", "channel": "895_down_2021", "ts_data_path": "ts_data/L3_T3_Causal_Anomaly_S0_00005.csv" }, { "id": "L2_T1_Shape_Identification_S2_00004", "level": 2, "level_name": "Pattern Recognition", "category": "Shape Identification", "subtask": "Shape Identification", "question": "Identify the time range of the longest plateau (stable period) in channel 813 within [2022-09-27 00:00:00 to 2022-10-03 18:30:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2022-09-27 08:30:00', '2022-09-28 08:30:00']", "ground_truth": [ "2022-09-27 08:30:00", "2022-09-28 08:30:00" ], "eval_metric": "iou", "channel": "813", "ts_data_path": "ts_data/L2_T1_Shape_Identification_S2_00004.csv" }, { "id": "L1_T1_Global_Aggregation_S2_00007", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Global Aggregation", "question": "What is the median value of channel 589 in 2022-01? (Output format: a single numeric value, rounded to 3 decimal places, e.g., x.xxx)", "answer": "422.0", "ground_truth": 422.0, "eval_metric": "rel_acc", "channel": "589", "ts_data_path": "ts_data/L1_T1_Global_Aggregation_S2_00007.csv" }, { "id": "L4_T1_Insight_Synthesis_S2_00006", "level": 4, "level_name": "Insight Synthesis", "category": "Insight Synthesis", "subtask": "Insight Synthesis", "question": "Analyze the behavior of channel 762 for the period 2021-06.\nPlease use ONLY the following phrases for trend description: rapid rise, gradual rise, rapid fall, gradual fall, steady stable, fluctuating stable.\nProvide a structured report covering:\n1. Trend Segmentation: Describe each stage with precise start/end timestamps (HH:MM:SS) using the phrases above.\n2. Outlier Audit: Identify only significant outliers that deviate sharply from the local trend. Ignore minor background noise.\n(Output format: structured natural language report.)", "answer": "1. Trend Segmentation: from 2021-06-01 00:00:00 to 2021-06-09 19:45:00, the trend showed a fluctuating stable; from 2021-06-09 19:45:00 to 2021-06-20 09:45:00, the trend showed a gradual rise; from 2021-06-20 09:45:00 to 2021-06-27 11:15:00, the trend showed a fluctuating stable; from 2021-06-27 11:15:00 to 2021-06-30 23:45:00, the trend showed a steady stable.\n2. Outlier Audit: A significant drop was detected at 2021-06-27 17:00:00 (value: -0.91).", "ground_truth": { "trend_segments": [ "from 2021-06-01 00:00:00 to 2021-06-09 19:45:00, the trend showed a fluctuating stable", "from 2021-06-09 19:45:00 to 2021-06-20 09:45:00, the trend showed a gradual rise", "from 2021-06-20 09:45:00 to 2021-06-27 11:15:00, the trend showed a fluctuating stable", "from 2021-06-27 11:15:00 to 2021-06-30 23:45:00, the trend showed a steady stable" ], "segments_meta": [ { "adj": "fluctuating", "kind": "stable", "start_idx": 0, "end_idx": 847 }, { "adj": "gradual", "kind": "rise", "start_idx": 847, "end_idx": 1863 }, { "adj": "fluctuating", "kind": "stable", "start_idx": 1863, "end_idx": 2541 }, { "adj": "steady", "kind": "stable", "start_idx": 2541, "end_idx": 2880 } ], "significant_anomaly": { "timestamp": "2021-06-27 17:00:00", "kind": "significant drop", "value": -0.9146434328282702 } }, "eval_metric": "report", "channel": "762", "ts_data_path": "ts_data/L4_T1_Insight_Synthesis_S2_00006.csv" }, { "id": "L2_T1_Shape_Identification_S3_00007", "level": 2, "level_name": "Pattern Recognition", "category": "Shape Identification", "subtask": "Shape Identification", "question": "Identify the time range of the deepest deep valley in channel HULL within [2016-08-15 05:00:00 to 2016-08-18 20:15:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2016-08-17 00:30:00', '2016-08-17 13:15:00']", "ground_truth": [ "2016-08-17 00:30:00", "2016-08-17 13:15:00" ], "eval_metric": "iou", "channel": "HULL", "ts_data_path": "ts_data/L2_T1_Shape_Identification_S3_00007.csv" }, { "id": "L1_T4_Sliding_Window_S2_00002", "level": 1, "level_name": "Basic Operations", "category": "Sliding Window", "subtask": "Sliding Window", "question": "Which 18-day period in 2021-04-15 to 2021-12-02 had the lowest average for channel 891? (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2021-07-29 14:15:00, 2021-08-16 14:00:00]", "ground_truth": [ "2021-07-29 14:15:00", "2021-08-16 14:00:00" ], "eval_metric": "iou", "channel": "891", "ts_data_path": "ts_data/L1_T4_Sliding_Window_S2_00002.csv" }, { "id": "L2_T3_Subsequence_Matching_S0_00002", "level": 2, "level_name": "Pattern Recognition", "category": "Subsequence Matching", "subtask": "Subsequence Matching", "question": "Analyze the reference pattern in [2020-07-15 03:00:00 to 2020-07-15 12:45:00]. Find the time interval where channel 124 exhibits the most similar pattern within the search context [2020-07-17 15:15:00 to 2020-07-25 14:00:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2020-07-21 22:00:00', '2020-07-22 07:45:00']", "ground_truth": [ "2020-07-21 22:00:00", "2020-07-22 07:45:00" ], "eval_metric": "iou", "channel": "124", "ts_data_path": "ts_data/L2_T3_Subsequence_Matching_S0_00002.csv" }, { "id": "L2_T1_Shape_Identification_S3_00004", "level": 2, "level_name": "Pattern Recognition", "category": "Shape Identification", "subtask": "Shape Identification", "question": "Identify the time range of the highest upward spike in channel 589 within [2022-08-19 17:30:00 to 2022-08-24 05:30:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2022-08-21 04:45:00', '2022-08-21 20:30:00']", "ground_truth": [ "2022-08-21 04:45:00", "2022-08-21 20:30:00" ], "eval_metric": "iou", "channel": "589", "ts_data_path": "ts_data/L2_T1_Shape_Identification_S3_00004.csv" }, { "id": "L1_T1_Global_Aggregation_S1_00006", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Global Aggregation", "question": "What is the minimum value of channel 591 in 2023-05-30 to 2023-06-08? (Output format: a single numeric value, rounded to 3 decimal places, e.g., x.xxx)", "answer": "150.0", "ground_truth": 150.0, "eval_metric": "rel_acc", "channel": "591", "ts_data_path": "ts_data/L1_T1_Global_Aggregation_S1_00006.csv" }, { "id": "L2_T2_Periodicity_Detection_S0_00003", "level": 2, "level_name": "Pattern Recognition", "category": "Periodicity Detection", "subtask": "Periodicity Detection", "question": "What is the dominant cycle period (in data points) of channel 237 within [2020-07-26 16:30:00 to 2020-08-03 08:30:00]? (Output format: integer)", "answer": "168", "ground_truth": 168, "eval_metric": "rel_acc", "channel": "237", "ts_data_path": "ts_data/L2_T2_Periodicity_Detection_S0_00003.csv" }, { "id": "L1_T2_Temporal_Localization_S3_00007", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Temporal Localization", "question": "At what exact timestamp did channel 813 reach its maximum value in 2022-11-04 to 2022-11-16? (Output format: a specific timestamp, e.g., YYYY-MM-DD HH:MM:SS)", "answer": "2022-11-05 00:15:00", "ground_truth": "2022-11-05 00:15:00", "eval_metric": "hit", "channel": "813", "ts_data_path": "ts_data/L1_T2_Temporal_Localization_S3_00007.csv" }, { "id": "L3_T2_Contextual_Anomaly_S3_00006", "level": 3, "level_name": "Semantic Reasoning", "category": "Contextual Anomaly", "subtask": "Contextual Anomaly", "question": "Identify the period in channel 728 during 2023 that experienced the most significant severe flood. (Output format: [YYYY-MM-DD, YYYY-MM-DD])", "answer": "[2023-03-15 07:00:00, 2023-04-22 06:45:00]", "ground_truth": [ "2023-03-15 07:00:00", "2023-04-22 06:45:00" ], "eval_metric": "iou", "channel": "728", "ts_data_path": "ts_data/L3_T2_Contextual_Anomaly_S3_00006.csv" }, { "id": "L3_T1_Composite_Trend_S2_00005", "level": 3, "level_name": "Semantic Reasoning", "category": "Composite Trend", "subtask": "Composite Trend", "question": "Among days in channel 591 during 2020 that exhibit the trend pattern 'rise, then steady stable, then slow fall', identify the top-4 days where the slow fall segment is the slowest. (Output format: a ranked list of dates, e.g., ['YYYY-MM-DD', ...])", "answer": "['2020-02-05', '2020-01-28', '2020-11-03', '2020-07-30']", "ground_truth": [ "2020-02-05", "2020-01-28", "2020-11-03", "2020-07-30" ], "eval_metric": "set_f1", "channel": "591", "ts_data_path": "ts_data/L3_T1_Composite_Trend_S2_00005.csv" }, { "id": "L3_T2_Contextual_Anomaly_S2_00007", "level": 3, "level_name": "Semantic Reasoning", "category": "Contextual Anomaly", "subtask": "Contextual Anomaly", "question": "Identify the period in channel 895 during 2023 that experienced the most significant extreme surge in flow. (Output format: [YYYY-MM-DD, YYYY-MM-DD])", "answer": "[2023-06-10 15:00:00, 2023-07-21 14:45:00]", "ground_truth": [ "2023-06-10 15:00:00", "2023-07-21 14:45:00" ], "eval_metric": "iou", "channel": "895", "ts_data_path": "ts_data/L3_T2_Contextual_Anomaly_S2_00007.csv" }, { "id": "L1_T3_Interval_Discovery_S4_00006", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Interval Discovery", "question": "Find the longest period where channel MUFL remained above 4.904 in 2017-02-17 to 2017-07-05. (Output format: a time interval, e.g., [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2017-03-30 22:15:00, 2017-04-01 08:00:00]", "ground_truth": [ "2017-03-30 22:15:00", "2017-04-01 08:00:00" ], "eval_metric": "iou", "channel": "MUFL", "ts_data_path": "ts_data/L1_T3_Interval_Discovery_S4_00006.csv" }, { "id": "L1_T1_Global_Aggregation_S4_00002", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Global Aggregation", "question": "What is the minimum value of channel 441 in 2020-01-15 to 2020-01-29? (Output format: a single numeric value, rounded to 3 decimal places, e.g., x.xxx)", "answer": "4.7", "ground_truth": 4.7, "eval_metric": "rel_acc", "channel": "441", "ts_data_path": "ts_data/L1_T1_Global_Aggregation_S4_00002.csv" }, { "id": "L2_T2_Periodicity_Detection_S1_00006", "level": 2, "level_name": "Pattern Recognition", "category": "Periodicity Detection", "subtask": "Periodicity Detection", "question": "What is the dominant cycle period (in data points) of channel LULL within [2017-04-15 11:00:00 to 2017-04-24 10:15:00]? (Output format: integer)", "answer": "31", "ground_truth": 31, "eval_metric": "rel_acc", "channel": "LULL", "ts_data_path": "ts_data/L2_T2_Periodicity_Detection_S1_00006.csv" }, { "id": "L3_T3_Causal_Anomaly_S1_00002", "level": 3, "level_name": "Semantic Reasoning", "category": "Causal Anomaly", "subtask": "Causal Anomaly", "question": "Given that channel 578_up_2020 is the upstream source of channel 578_down_2020, identify the time period in 2020 where 578_down_2020 shows a significant causal anomaly, such as an inverse trend against the source. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2020-05-10 01:15:00, 2020-05-13 01:45:00]", "ground_truth": [ "2020-05-10 01:15:00", "2020-05-13 01:45:00" ], "eval_metric": "iou", "channel": "578_down_2020", "ts_data_path": "ts_data/L3_T3_Causal_Anomaly_S1_00002.csv" }, { "id": "L3_T1_Composite_Trend_S2_00001", "level": 3, "level_name": "Semantic Reasoning", "category": "Composite Trend", "subtask": "Composite Trend", "question": "Among days in channel 123 during 2020 that exhibit the trend pattern 'steady stable, then fluctuating stable, then slow fall', identify the top-4 days where the fluctuating stable segment is the most fluctuating. (Output format: a ranked list of dates, e.g., ['YYYY-MM-DD', ...])", "answer": "['2020-04-10', '2020-06-17', '2020-07-13', '2020-09-15']", "ground_truth": [ "2020-04-10", "2020-06-17", "2020-07-13", "2020-09-15" ], "eval_metric": "set_f1", "channel": "123", "ts_data_path": "ts_data/L3_T1_Composite_Trend_S2_00001.csv" }, { "id": "L3_T2_Contextual_Anomaly_S4_00005", "level": 3, "level_name": "Semantic Reasoning", "category": "Contextual Anomaly", "subtask": "Contextual Anomaly", "question": "Identify the period in channel 430 during 2019 that experienced the most significant severe flood. (Output format: [YYYY-MM-DD, YYYY-MM-DD])", "answer": "[2019-01-18 04:15:00, 2019-02-12 04:00:00]", "ground_truth": [ "2019-01-18 04:15:00", "2019-02-12 04:00:00" ], "eval_metric": "iou", "channel": "430", "ts_data_path": "ts_data/L3_T2_Contextual_Anomaly_S4_00005.csv" }, { "id": "L2_T3_Subsequence_Matching_S0_00004", "level": 2, "level_name": "Pattern Recognition", "category": "Subsequence Matching", "subtask": "Subsequence Matching", "question": "Analyze the reference pattern in [2019-12-13 02:45:00 to 2019-12-13 20:30:00]. Find the time interval where channel 177 exhibits the most similar pattern within the search context [2019-12-16 04:30:00 to 2019-12-24 18:15:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2019-12-19 06:30:00', '2019-12-20 00:15:00']", "ground_truth": [ "2019-12-19 06:30:00", "2019-12-20 00:15:00" ], "eval_metric": "iou", "channel": "177", "ts_data_path": "ts_data/L2_T3_Subsequence_Matching_S0_00004.csv" }, { "id": "L3_T2_Contextual_Anomaly_S4_00007", "level": 3, "level_name": "Semantic Reasoning", "category": "Contextual Anomaly", "subtask": "Contextual Anomaly", "question": "Identify the period in channel 595 during 2020 that experienced the most significant extreme surge in flow. (Output format: [YYYY-MM-DD, YYYY-MM-DD])", "answer": "[2020-06-12 23:45:00, 2020-07-03 23:30:00]", "ground_truth": [ "2020-06-12 23:45:00", "2020-07-03 23:30:00" ], "eval_metric": "iou", "channel": "595", "ts_data_path": "ts_data/L3_T2_Contextual_Anomaly_S4_00007.csv" }, { "id": "L3_T1_Composite_Trend_S0_00008", "level": 3, "level_name": "Semantic Reasoning", "category": "Composite Trend", "subtask": "Composite Trend", "question": "Among days in channel HULL during 2017 that exhibit the trend pattern 'fluctuating stable, then rise', identify the top-3 days where the fluctuating stable segment is the most fluctuating. (Output format: a ranked list of dates, e.g., ['YYYY-MM-DD', ...])", "answer": "['2017-02-26', '2017-07-28', '2017-06-29']", "ground_truth": [ "2017-02-26", "2017-07-28", "2017-06-29" ], "eval_metric": "set_f1", "channel": "HULL", "ts_data_path": "ts_data/L3_T1_Composite_Trend_S0_00008.csv" }, { "id": "L1_T4_Sliding_Window_S3_00001", "level": 1, "level_name": "Basic Operations", "category": "Sliding Window", "subtask": "Sliding Window", "question": "Which 62-day period in 2021-11-10 to 2022-07-27 had the highest variance for channel 173? (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2022-02-10 09:15:00, 2022-04-13 09:00:00]", "ground_truth": [ "2022-02-10 09:15:00", "2022-04-13 09:00:00" ], "eval_metric": "iou", "channel": "173", "ts_data_path": "ts_data/L1_T4_Sliding_Window_S3_00001.csv" }, { "id": "L4_T1_Insight_Synthesis_S2_00007", "level": 4, "level_name": "Insight Synthesis", "category": "Insight Synthesis", "subtask": "Insight Synthesis", "question": "Analyze the behavior of channel 813 for the period 2021-03.\nPlease use ONLY the following phrases for trend description: rapid rise, gradual rise, rapid fall, gradual fall, steady stable, fluctuating stable.\nProvide a structured report covering:\n1. Trend Segmentation: Describe each stage with precise start/end timestamps (HH:MM:SS) using the phrases above.\n2. Outlier Audit: Identify only significant outliers that deviate sharply from the local trend. Ignore minor background noise.\n(Output format: structured natural language report.)", "answer": "1. Trend Segmentation: from 2021-03-01 00:00:00 to 2021-03-09 20:30:00, the trend showed a fluctuating stable; from 2021-03-09 20:30:00 to 2021-03-18 17:00:00, the trend showed a rapid fall; from 2021-03-18 17:00:00 to 2021-03-23 03:15:00, the trend showed a rapid rise; from 2021-03-23 03:15:00 to 2021-03-31 23:45:00, the trend showed a gradual fall.\n2. Outlier Audit: A significant drop was detected at 2021-03-18 07:45:00 (value: -3.39).", "ground_truth": { "trend_segments": [ "from 2021-03-01 00:00:00 to 2021-03-09 20:30:00, the trend showed a fluctuating stable", "from 2021-03-09 20:30:00 to 2021-03-18 17:00:00, the trend showed a rapid fall", "from 2021-03-18 17:00:00 to 2021-03-23 03:15:00, the trend showed a rapid rise", "from 2021-03-23 03:15:00 to 2021-03-31 23:45:00, the trend showed a gradual fall" ], "segments_meta": [ { "adj": "fluctuating", "kind": "stable", "start_idx": 0, "end_idx": 850 }, { "adj": "rapid", "kind": "fall", "start_idx": 850, "end_idx": 1700 }, { "adj": "rapid", "kind": "rise", "start_idx": 1700, "end_idx": 2125 }, { "adj": "gradual", "kind": "fall", "start_idx": 2125, "end_idx": 2976 } ], "significant_anomaly": { "timestamp": "2021-03-18 07:45:00", "kind": "significant drop", "value": -3.3913546126925276 } }, "eval_metric": "report", "channel": "813", "ts_data_path": "ts_data/L4_T1_Insight_Synthesis_S2_00007.csv" }, { "id": "L2_T2_Periodicity_Detection_S3_00003", "level": 2, "level_name": "Pattern Recognition", "category": "Periodicity Detection", "subtask": "Periodicity Detection", "question": "What is the dominant cycle period (in data points) of channel 430 within [2020-09-08 16:00:00 to 2020-09-18 07:45:00]? (Output format: integer)", "answer": "22", "ground_truth": 22, "eval_metric": "rel_acc", "channel": "430", "ts_data_path": "ts_data/L2_T2_Periodicity_Detection_S3_00003.csv" }, { "id": "L3_T1_Composite_Trend_S3_00007", "level": 3, "level_name": "Semantic Reasoning", "category": "Composite Trend", "subtask": "Composite Trend", "question": "Among days in channel 728 during 2021 that exhibit the trend pattern 'rapid rise, then fall', identify the top-6 days where the rapid rise segment is the fastest. (Output format: a ranked list of dates, e.g., ['YYYY-MM-DD', ...])", "answer": "['2021-01-15', '2021-12-21', '2021-07-27', '2021-05-06', '2021-09-21', '2021-11-05']", "ground_truth": [ "2021-01-15", "2021-12-21", "2021-07-27", "2021-05-06", "2021-09-21", "2021-11-05" ], "eval_metric": "set_f1", "channel": "728", "ts_data_path": "ts_data/L3_T1_Composite_Trend_S3_00007.csv" }, { "id": "L1_T1_Global_Aggregation_S0_00006", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Global Aggregation", "question": "What is the median value of channel HULL in 2018-06? (Output format: a single numeric value, rounded to 3 decimal places, e.g., x.xxx)", "answer": "4.421", "ground_truth": 4.421, "eval_metric": "rel_acc", "channel": "HULL", "ts_data_path": "ts_data/L1_T1_Global_Aggregation_S0_00006.csv" }, { "id": "L1_T3_Interval_Discovery_S1_00006", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Interval Discovery", "question": "Find the longest period where channel 897 remained above 0.012 in 2020-10. (Output format: a time interval, e.g., [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2020-10-30 06:15:00, 2020-10-31 23:45:00]", "ground_truth": [ "2020-10-30 06:15:00", "2020-10-31 23:45:00" ], "eval_metric": "iou", "channel": "897", "ts_data_path": "ts_data/L1_T3_Interval_Discovery_S1_00006.csv" }, { "id": "L2_T3_Subsequence_Matching_S4_00001", "level": 2, "level_name": "Pattern Recognition", "category": "Subsequence Matching", "subtask": "Subsequence Matching", "question": "Analyze the reference pattern in [2020-01-20 18:15:00 to 2020-01-21 00:00:00]. Find the time interval where channel 67 exhibits the most similar pattern within the search context [2020-01-21 11:00:00 to 2020-01-23 00:45:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2020-01-21 23:15:00', '2020-01-22 05:00:00']", "ground_truth": [ "2020-01-21 23:15:00", "2020-01-22 05:00:00" ], "eval_metric": "iou", "channel": "67", "ts_data_path": "ts_data/L2_T3_Subsequence_Matching_S4_00001.csv" }, { "id": "L1_T3_Interval_Discovery_S0_00002", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Interval Discovery", "question": "Find the longest period where channel 245 remained above 0.53 in 2022-10-05 to 2023-02-25. (Output format: a time interval, e.g., [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2023-01-07 16:45:00, 2023-01-25 21:00:00]", "ground_truth": [ "2023-01-07 16:45:00", "2023-01-25 21:00:00" ], "eval_metric": "iou", "channel": "245", "ts_data_path": "ts_data/L1_T3_Interval_Discovery_S0_00002.csv" }, { "id": "L3_T1_Composite_Trend_S0_00006", "level": 3, "level_name": "Semantic Reasoning", "category": "Composite Trend", "subtask": "Composite Trend", "question": "Among days in channel 578 during 2022 that exhibit the trend pattern 'steady stable, then fluctuating stable, then steady stable', identify the top-3 days where the fluctuating stable segment is the most fluctuating. (Output format: a ranked list of dates, e.g., ['YYYY-MM-DD', ...])", "answer": "['2022-06-03', '2022-10-28', '2022-10-10']", "ground_truth": [ "2022-06-03", "2022-10-28", "2022-10-10" ], "eval_metric": "set_f1", "channel": "578", "ts_data_path": "ts_data/L3_T1_Composite_Trend_S0_00006.csv" }, { "id": "L2_T1_Shape_Identification_S4_00006", "level": 2, "level_name": "Pattern Recognition", "category": "Shape Identification", "subtask": "Shape Identification", "question": "Identify the time range of the largest step ascent in channel 813 within [2022-06-23 07:30:00 to 2022-06-30 12:15:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2022-06-27 06:15:00', '2022-06-28 07:45:00']", "ground_truth": [ "2022-06-27 06:15:00", "2022-06-28 07:45:00" ], "eval_metric": "iou", "channel": "813", "ts_data_path": "ts_data/L2_T1_Shape_Identification_S4_00006.csv" }, { "id": "L3_T2_Contextual_Anomaly_S3_00008", "level": 3, "level_name": "Semantic Reasoning", "category": "Contextual Anomaly", "subtask": "Contextual Anomaly", "question": "Identify the period in channel 99 during 2021 that experienced the most significant historically high water level. (Output format: [YYYY-MM-DD, YYYY-MM-DD])", "answer": "[2021-01-07 21:30:00, 2021-02-04 21:15:00]", "ground_truth": [ "2021-01-07 21:30:00", "2021-02-04 21:15:00" ], "eval_metric": "iou", "channel": "99", "ts_data_path": "ts_data/L3_T2_Contextual_Anomaly_S3_00008.csv" }, { "id": "L1_T4_Sliding_Window_S1_00009", "level": 1, "level_name": "Basic Operations", "category": "Sliding Window", "subtask": "Sliding Window", "question": "Which 50-day period in 2016-10-06 to 2017-05-12 had the highest average for channel LULL? (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2017-03-23 07:00:00, 2017-05-12 06:45:00]", "ground_truth": [ "2017-03-23 07:00:00", "2017-05-12 06:45:00" ], "eval_metric": "iou", "channel": "LULL", "ts_data_path": "ts_data/L1_T4_Sliding_Window_S1_00009.csv" }, { "id": "L1_T1_Global_Aggregation_S3_00004", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Global Aggregation", "question": "What is the maximum value of channel 495 in 2022-12-06 to 2023-02-17? (Output format: a single numeric value, rounded to 3 decimal places, e.g., x.xxx)", "answer": "1.17", "ground_truth": 1.17, "eval_metric": "rel_acc", "channel": "495", "ts_data_path": "ts_data/L1_T1_Global_Aggregation_S3_00004.csv" }, { "id": "L4_T1_Insight_Synthesis_S1_00006", "level": 4, "level_name": "Insight Synthesis", "category": "Insight Synthesis", "subtask": "Insight Synthesis", "question": "Analyze the behavior of channel LUFL for the period 2018-04.\nPlease use ONLY the following phrases for trend description: rapid rise, gradual rise, rapid fall, gradual fall, steady stable, fluctuating stable.\nProvide a structured report covering:\n1. Trend Segmentation: Describe each stage with precise start/end timestamps (HH:MM:SS) using the phrases above.\n2. Outlier Audit: Identify only significant outliers that deviate sharply from the local trend. Ignore minor background noise.\n(Output format: structured natural language report.)", "answer": "1. Trend Segmentation: from 2018-04-01 00:00:00 to 2018-04-06 15:00:00, the trend showed a rapid rise; from 2018-04-06 15:00:00 to 2018-04-14 03:00:00, the trend showed a fluctuating stable; from 2018-04-14 03:00:00 to 2018-04-21 15:00:00, the trend showed a rapid rise; from 2018-04-21 15:00:00 to 2018-04-30 23:45:00, the trend showed a steady stable.\n2. Outlier Audit: A significant spike was detected at 2018-04-21 11:00:00 (value: 28.39).", "ground_truth": { "trend_segments": [ "from 2018-04-01 00:00:00 to 2018-04-06 15:00:00, the trend showed a rapid rise", "from 2018-04-06 15:00:00 to 2018-04-14 03:00:00, the trend showed a fluctuating stable", "from 2018-04-14 03:00:00 to 2018-04-21 15:00:00, the trend showed a rapid rise", "from 2018-04-21 15:00:00 to 2018-04-30 23:45:00, the trend showed a steady stable" ], "segments_meta": [ { "adj": "rapid", "kind": "rise", "start_idx": 0, "end_idx": 540 }, { "adj": "fluctuating", "kind": "stable", "start_idx": 540, "end_idx": 1260 }, { "adj": "rapid", "kind": "rise", "start_idx": 1260, "end_idx": 1980 }, { "adj": "steady", "kind": "stable", "start_idx": 1980, "end_idx": 2880 } ], "significant_anomaly": { "timestamp": "2018-04-21 11:00:00", "kind": "significant spike", "value": 28.390044565358565 } }, "eval_metric": "report", "channel": "LUFL", "ts_data_path": "ts_data/L4_T1_Insight_Synthesis_S1_00006.csv" }, { "id": "L3_T3_Causal_Anomaly_S2_00005", "level": 3, "level_name": "Semantic Reasoning", "category": "Causal Anomaly", "subtask": "Causal Anomaly", "question": "Given that channel 177_up_2021 is the upstream source of channel 177_down_2021, identify the time period in 2021 where 177_down_2021 shows a significant causal anomaly, such as an flat line during high activity. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2021-09-18 23:45:00, 2021-09-20 06:45:00]", "ground_truth": [ "2021-09-18 23:45:00", "2021-09-20 06:45:00" ], "eval_metric": "iou", "channel": "177_down_2021", "ts_data_path": "ts_data/L3_T3_Causal_Anomaly_S2_00005.csv" }, { "id": "L4_T1_Insight_Synthesis_S3_00007", "level": 4, "level_name": "Insight Synthesis", "category": "Insight Synthesis", "subtask": "Insight Synthesis", "question": "Analyze the behavior of channel 123 for the period 2019-06.\nPlease use ONLY the following phrases for trend description: rapid rise, gradual rise, rapid fall, gradual fall, steady stable, fluctuating stable.\nProvide a structured report covering:\n1. Trend Segmentation: Describe each stage with precise start/end timestamps (HH:MM:SS) using the phrases above.\n2. Outlier Audit: Identify only significant outliers that deviate sharply from the local trend. Ignore minor background noise.\n(Output format: structured natural language report.)", "answer": "1. Trend Segmentation: from 2019-06-01 00:00:00 to 2019-06-05 14:45:00, the trend showed a steady stable; from 2019-06-05 14:45:00 to 2019-06-17 03:45:00, the trend showed a gradual rise; from 2019-06-17 03:45:00 to 2019-06-26 09:15:00, the trend showed a gradual fall; from 2019-06-26 09:15:00 to 2019-06-30 23:45:00, the trend showed a gradual rise.\n2. Outlier Audit: A significant spike was detected at 2019-06-12 06:00:00 (value: 9.05).", "ground_truth": { "trend_segments": [ "from 2019-06-01 00:00:00 to 2019-06-05 14:45:00, the trend showed a steady stable", "from 2019-06-05 14:45:00 to 2019-06-17 03:45:00, the trend showed a gradual rise", "from 2019-06-17 03:45:00 to 2019-06-26 09:15:00, the trend showed a gradual fall", "from 2019-06-26 09:15:00 to 2019-06-30 23:45:00, the trend showed a gradual rise" ], "segments_meta": [ { "adj": "steady", "kind": "stable", "start_idx": 0, "end_idx": 443 }, { "adj": "gradual", "kind": "rise", "start_idx": 443, "end_idx": 1551 }, { "adj": "gradual", "kind": "fall", "start_idx": 1551, "end_idx": 2437 }, { "adj": "gradual", "kind": "rise", "start_idx": 2437, "end_idx": 2880 } ], "significant_anomaly": { "timestamp": "2019-06-12 06:00:00", "kind": "significant spike", "value": 9.054966018810093 } }, "eval_metric": "report", "channel": "123", "ts_data_path": "ts_data/L4_T1_Insight_Synthesis_S3_00007.csv" }, { "id": "L3_T1_Composite_Trend_S1_00005", "level": 3, "level_name": "Semantic Reasoning", "category": "Composite Trend", "subtask": "Composite Trend", "question": "Among days in channel 147 during 2020 that exhibit the trend pattern 'fall, then steady stable, then slow rise', identify the top-2 days where the slow rise segment is the slowest. (Output format: a ranked list of dates, e.g., ['YYYY-MM-DD', ...])", "answer": "['2020-03-25', '2020-01-30']", "ground_truth": [ "2020-03-25", "2020-01-30" ], "eval_metric": "set_f1", "channel": "147", "ts_data_path": "ts_data/L3_T1_Composite_Trend_S1_00005.csv" }, { "id": "L2_T3_Subsequence_Matching_S0_00008", "level": 2, "level_name": "Pattern Recognition", "category": "Subsequence Matching", "subtask": "Subsequence Matching", "question": "Analyze the reference pattern in [2022-08-12 16:30:00 to 2022-08-12 22:15:00]. Find the time interval where channel 865 exhibits the most similar pattern within the search context [2022-08-14 05:15:00 to 2022-08-19 02:45:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2022-08-15 08:45:00', '2022-08-15 14:30:00']", "ground_truth": [ "2022-08-15 08:45:00", "2022-08-15 14:30:00" ], "eval_metric": "iou", "channel": "865", "ts_data_path": "ts_data/L2_T3_Subsequence_Matching_S0_00008.csv" }, { "id": "L4_T1_Insight_Synthesis_S0_00002", "level": 4, "level_name": "Insight Synthesis", "category": "Insight Synthesis", "subtask": "Insight Synthesis", "question": "Analyze the behavior of channel 151 for the period 2022-12.\nPlease use ONLY the following phrases for trend description: rapid rise, gradual rise, rapid fall, gradual fall, steady stable, fluctuating stable.\nProvide a structured report covering:\n1. Trend Segmentation: Describe each stage with precise start/end timestamps (HH:MM:SS) using the phrases above.\n2. Outlier Audit: Identify only significant outliers that deviate sharply from the local trend. Ignore minor background noise.\n(Output format: structured natural language report.)", "answer": "1. Trend Segmentation: from 2022-12-01 00:00:00 to 2022-12-07 17:45:00, the trend showed a gradual fall; from 2022-12-07 17:45:00 to 2022-12-14 11:30:00, the trend showed a rapid rise; from 2022-12-14 11:30:00 to 2022-12-17 04:15:00, the trend showed a gradual rise; from 2022-12-17 04:15:00 to 2022-12-25 06:15:00, the trend showed a gradual fall; from 2022-12-25 06:15:00 to 2022-12-31 23:45:00, the trend showed a gradual rise.\n2. Outlier Audit: A significant drop was detected at 2022-12-14 23:45:00 (value: -5.07).", "ground_truth": { "trend_segments": [ "from 2022-12-01 00:00:00 to 2022-12-07 17:45:00, the trend showed a gradual fall", "from 2022-12-07 17:45:00 to 2022-12-14 11:30:00, the trend showed a rapid rise", "from 2022-12-14 11:30:00 to 2022-12-17 04:15:00, the trend showed a gradual rise", "from 2022-12-17 04:15:00 to 2022-12-25 06:15:00, the trend showed a gradual fall", "from 2022-12-25 06:15:00 to 2022-12-31 23:45:00, the trend showed a gradual rise" ], "segments_meta": [ { "adj": "gradual", "kind": "fall", "start_idx": 0, "end_idx": 647 }, { "adj": "rapid", "kind": "rise", "start_idx": 647, "end_idx": 1294 }, { "adj": "gradual", "kind": "rise", "start_idx": 1294, "end_idx": 1553 }, { "adj": "gradual", "kind": "fall", "start_idx": 1553, "end_idx": 2329 }, { "adj": "gradual", "kind": "rise", "start_idx": 2329, "end_idx": 2976 } ], "significant_anomaly": { "timestamp": "2022-12-14 23:45:00", "kind": "significant drop", "value": -5.070681823677937 } }, "eval_metric": "report", "channel": "151", "ts_data_path": "ts_data/L4_T1_Insight_Synthesis_S0_00002.csv" }, { "id": "L3_T3_Causal_Anomaly_S0_00007", "level": 3, "level_name": "Semantic Reasoning", "category": "Causal Anomaly", "subtask": "Causal Anomaly", "question": "Given that channel 245_up_2023 is the upstream source of channel 245_down_2023, identify the time period in 2023 where 245_down_2023 shows a significant causal anomaly, such as an flat line during high activity. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2023-02-05 23:30:00, 2023-02-08 06:15:00]", "ground_truth": [ "2023-02-05 23:30:00", "2023-02-08 06:15:00" ], "eval_metric": "iou", "channel": "245_down_2023", "ts_data_path": "ts_data/L3_T3_Causal_Anomaly_S0_00007.csv" }, { "id": "L4_T1_Insight_Synthesis_S1_00008", "level": 4, "level_name": "Insight Synthesis", "category": "Insight Synthesis", "subtask": "Insight Synthesis", "question": "Analyze the behavior of channel 146 for the period 2022-09.\nPlease use ONLY the following phrases for trend description: rapid rise, gradual rise, rapid fall, gradual fall, steady stable, fluctuating stable.\nProvide a structured report covering:\n1. Trend Segmentation: Describe each stage with precise start/end timestamps (HH:MM:SS) using the phrases above.\n2. Outlier Audit: Identify only significant outliers that deviate sharply from the local trend. Ignore minor background noise.\n(Output format: structured natural language report.)", "answer": "1. Trend Segmentation: from 2022-09-01 00:00:00 to 2022-09-07 07:30:00, the trend showed a steady stable; from 2022-09-07 07:30:00 to 2022-09-16 18:45:00, the trend showed a rapid rise; from 2022-09-16 18:45:00 to 2022-09-24 16:15:00, the trend showed a steady stable; from 2022-09-24 16:15:00 to 2022-09-27 20:00:00, the trend showed a rapid fall; from 2022-09-27 20:00:00 to 2022-09-30 23:45:00, the trend showed a gradual fall.\n2. Outlier Audit: A significant drop was detected at 2022-09-27 06:30:00 (value: -4.86).", "ground_truth": { "trend_segments": [ "from 2022-09-01 00:00:00 to 2022-09-07 07:30:00, the trend showed a steady stable", "from 2022-09-07 07:30:00 to 2022-09-16 18:45:00, the trend showed a rapid rise", "from 2022-09-16 18:45:00 to 2022-09-24 16:15:00, the trend showed a steady stable", "from 2022-09-24 16:15:00 to 2022-09-27 20:00:00, the trend showed a rapid fall", "from 2022-09-27 20:00:00 to 2022-09-30 23:45:00, the trend showed a gradual fall" ], "segments_meta": [ { "adj": "steady", "kind": "stable", "start_idx": 0, "end_idx": 606 }, { "adj": "rapid", "kind": "rise", "start_idx": 606, "end_idx": 1515 }, { "adj": "steady", "kind": "stable", "start_idx": 1515, "end_idx": 2273 }, { "adj": "rapid", "kind": "fall", "start_idx": 2273, "end_idx": 2576 }, { "adj": "gradual", "kind": "fall", "start_idx": 2576, "end_idx": 2880 } ], "significant_anomaly": { "timestamp": "2022-09-27 06:30:00", "kind": "significant drop", "value": -4.8587445051339975 } }, "eval_metric": "report", "channel": "146", "ts_data_path": "ts_data/L4_T1_Insight_Synthesis_S1_00008.csv" }, { "id": "L3_T2_Contextual_Anomaly_S4_00009", "level": 3, "level_name": "Semantic Reasoning", "category": "Contextual Anomaly", "subtask": "Contextual Anomaly", "question": "Identify the period in channel 813 during 2023 that experienced the most significant severe flood. (Output format: [YYYY-MM-DD, YYYY-MM-DD])", "answer": "[2023-04-06 04:15:00, 2023-04-17 04:00:00]", "ground_truth": [ "2023-04-06 04:15:00", "2023-04-17 04:00:00" ], "eval_metric": "iou", "channel": "813", "ts_data_path": "ts_data/L3_T2_Contextual_Anomaly_S4_00009.csv" }, { "id": "L1_T2_Temporal_Localization_S1_00001", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Temporal Localization", "question": "At what exact timestamp did channel 146 reach its minimum value in 2022-07? (Output format: a specific timestamp, e.g., YYYY-MM-DD HH:MM:SS)", "answer": "2022-07-25 14:45:00", "ground_truth": "2022-07-25 14:45:00", "eval_metric": "hit", "channel": "146", "ts_data_path": "ts_data/L1_T2_Temporal_Localization_S1_00001.csv" }, { "id": "L2_T3_Subsequence_Matching_S4_00008", "level": 2, "level_name": "Pattern Recognition", "category": "Subsequence Matching", "subtask": "Subsequence Matching", "question": "Analyze the reference pattern in [2019-12-21 11:00:00 to 2019-12-22 04:45:00]. Find the time interval where channel 146 exhibits the most similar pattern within the search context [2019-12-24 04:00:00 to 2019-12-31 06:30:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2019-12-26 22:45:00', '2019-12-27 16:30:00']", "ground_truth": [ "2019-12-26 22:45:00", "2019-12-27 16:30:00" ], "eval_metric": "iou", "channel": "146", "ts_data_path": "ts_data/L2_T3_Subsequence_Matching_S4_00008.csv" }, { "id": "L3_T2_Contextual_Anomaly_S3_00007", "level": 3, "level_name": "Semantic Reasoning", "category": "Contextual Anomaly", "subtask": "Contextual Anomaly", "question": "Identify the period in channel 762 during 2022 that experienced the most significant dry-out period. (Output format: [YYYY-MM-DD, YYYY-MM-DD])", "answer": "[2022-09-27 01:30:00, 2022-11-13 01:15:00]", "ground_truth": [ "2022-09-27 01:30:00", "2022-11-13 01:15:00" ], "eval_metric": "iou", "channel": "762", "ts_data_path": "ts_data/L3_T2_Contextual_Anomaly_S3_00007.csv" }, { "id": "L1_T4_Sliding_Window_S2_00006", "level": 1, "level_name": "Basic Operations", "category": "Sliding Window", "subtask": "Sliding Window", "question": "Which 8-day period in 2017-02-16 to 2017-07-23 had the highest variance for channel OT? (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2017-04-08 21:30:00, 2017-04-16 21:15:00]", "ground_truth": [ "2017-04-08 21:30:00", "2017-04-16 21:15:00" ], "eval_metric": "iou", "channel": "OT", "ts_data_path": "ts_data/L1_T4_Sliding_Window_S2_00006.csv" }, { "id": "L2_T3_Subsequence_Matching_S4_00003", "level": 2, "level_name": "Pattern Recognition", "category": "Subsequence Matching", "subtask": "Subsequence Matching", "question": "Analyze the reference pattern in [2020-11-18 04:00:00 to 2020-11-18 13:45:00]. Find the time interval where channel 580 exhibits the most similar pattern within the search context [2020-11-19 18:00:00 to 2020-11-24 00:30:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2020-11-22 19:45:00', '2020-11-23 05:30:00']", "ground_truth": [ "2020-11-22 19:45:00", "2020-11-23 05:30:00" ], "eval_metric": "iou", "channel": "580", "ts_data_path": "ts_data/L2_T3_Subsequence_Matching_S4_00003.csv" }, { "id": "L1_T1_Global_Aggregation_S0_00002", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Global Aggregation", "question": "What is the range value of channel 626 in 2021-12? (Output format: a single numeric value, rounded to 3 decimal places, e.g., x.xxx)", "answer": "0.78", "ground_truth": 0.78, "eval_metric": "rel_acc", "channel": "626", "ts_data_path": "ts_data/L1_T1_Global_Aggregation_S0_00002.csv" }, { "id": "L1_T1_Global_Aggregation_S0_00005", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Global Aggregation", "question": "What is the average value of channel 813 in 2020-09-07 to 2020-09-26? (Output format: a single numeric value, rounded to 3 decimal places, e.g., x.xxx)", "answer": "0.495", "ground_truth": 0.495, "eval_metric": "rel_acc", "channel": "813", "ts_data_path": "ts_data/L1_T1_Global_Aggregation_S0_00005.csv" }, { "id": "L1_T4_Sliding_Window_S4_00002", "level": 1, "level_name": "Basic Operations", "category": "Sliding Window", "subtask": "Sliding Window", "question": "Which 28-day period in 2023-05-12 to 2023-08-14 had the highest variance for channel 147? (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2023-07-11 08:15:00, 2023-08-08 08:00:00]", "ground_truth": [ "2023-07-11 08:15:00", "2023-08-08 08:00:00" ], "eval_metric": "iou", "channel": "147", "ts_data_path": "ts_data/L1_T4_Sliding_Window_S4_00002.csv" }, { "id": "L3_T3_Causal_Anomaly_S1_00001", "level": 3, "level_name": "Semantic Reasoning", "category": "Causal Anomaly", "subtask": "Causal Anomaly", "question": "Given that channel 172_up_2021 is the upstream source of channel 172_down_2021, identify the time period in 2021 where 172_down_2021 shows a significant causal anomaly, such as an inverse trend against the source. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2021-10-23 06:15:00, 2021-10-24 23:30:00]", "ground_truth": [ "2021-10-23 06:15:00", "2021-10-24 23:30:00" ], "eval_metric": "iou", "channel": "172_down_2021", "ts_data_path": "ts_data/L3_T3_Causal_Anomaly_S1_00001.csv" }, { "id": "L4_T1_Insight_Synthesis_S0_00004", "level": 4, "level_name": "Insight Synthesis", "category": "Insight Synthesis", "subtask": "Insight Synthesis", "question": "Analyze the behavior of channel 172 for the period 2023-07.\nPlease use ONLY the following phrases for trend description: rapid rise, gradual rise, rapid fall, gradual fall, steady stable, fluctuating stable.\nProvide a structured report covering:\n1. Trend Segmentation: Describe each stage with precise start/end timestamps (HH:MM:SS) using the phrases above.\n2. Outlier Audit: Identify only significant outliers that deviate sharply from the local trend. Ignore minor background noise.\n(Output format: structured natural language report.)", "answer": "1. Trend Segmentation: from 2023-07-01 00:00:00 to 2023-07-06 04:00:00, the trend showed a gradual fall; from 2023-07-06 04:00:00 to 2023-07-14 18:45:00, the trend showed a steady stable; from 2023-07-14 18:45:00 to 2023-07-21 16:00:00, the trend showed a gradual rise; from 2023-07-21 16:00:00 to 2023-07-31 23:45:00, the trend showed a rapid fall.\n2. Outlier Audit: A significant drop was detected at 2023-07-15 06:15:00 (value: -3255.58).", "ground_truth": { "trend_segments": [ "from 2023-07-01 00:00:00 to 2023-07-06 04:00:00, the trend showed a gradual fall", "from 2023-07-06 04:00:00 to 2023-07-14 18:45:00, the trend showed a steady stable", "from 2023-07-14 18:45:00 to 2023-07-21 16:00:00, the trend showed a gradual rise", "from 2023-07-21 16:00:00 to 2023-07-31 23:45:00, the trend showed a rapid fall" ], "segments_meta": [ { "adj": "gradual", "kind": "fall", "start_idx": 0, "end_idx": 496 }, { "adj": "steady", "kind": "stable", "start_idx": 496, "end_idx": 1323 }, { "adj": "gradual", "kind": "rise", "start_idx": 1323, "end_idx": 1984 }, { "adj": "rapid", "kind": "fall", "start_idx": 1984, "end_idx": 2976 } ], "significant_anomaly": { "timestamp": "2023-07-15 06:15:00", "kind": "significant drop", "value": -3255.57709420622 } }, "eval_metric": "report", "channel": "172", "ts_data_path": "ts_data/L4_T1_Insight_Synthesis_S0_00004.csv" }, { "id": "L1_T2_Temporal_Localization_S0_00002", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Temporal Localization", "question": "At what exact timestamp did channel 177 reach its maximum value in 2022-07-18 to 2022-08-15? (Output format: a specific timestamp, e.g., YYYY-MM-DD HH:MM:SS)", "answer": "2022-08-04 21:45:00", "ground_truth": "2022-08-04 21:45:00", "eval_metric": "hit", "channel": "177", "ts_data_path": "ts_data/L1_T2_Temporal_Localization_S0_00002.csv" }, { "id": "L3_T3_Causal_Anomaly_S0_00006", "level": 3, "level_name": "Semantic Reasoning", "category": "Causal Anomaly", "subtask": "Causal Anomaly", "question": "Given that channel 166_up_2020 is the upstream source of channel 166_down_2020, identify the time period in 2020 where 166_down_2020 shows a significant causal anomaly, such as an inverse trend against the source. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2020-08-24 09:45:00, 2020-08-26 08:30:00]", "ground_truth": [ "2020-08-24 09:45:00", "2020-08-26 08:30:00" ], "eval_metric": "iou", "channel": "166_down_2020", "ts_data_path": "ts_data/L3_T3_Causal_Anomaly_S0_00006.csv" }, { "id": "L1_T3_Interval_Discovery_S2_00002", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Interval Discovery", "question": "Find the longest period where channel 580 remained above 0.333 in 2019-06-16 to 2020-01-28. (Output format: a time interval, e.g., [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2019-12-07 20:30:00, 2019-12-19 14:00:00]", "ground_truth": [ "2019-12-07 20:30:00", "2019-12-19 14:00:00" ], "eval_metric": "iou", "channel": "580", "ts_data_path": "ts_data/L1_T3_Interval_Discovery_S2_00002.csv" }, { "id": "L2_T1_Shape_Identification_S4_00005", "level": 2, "level_name": "Pattern Recognition", "category": "Shape Identification", "subtask": "Shape Identification", "question": "Identify the time range of the largest step ascent in channel 728 within [2021-09-12 12:00:00 to 2021-09-20 22:15:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2021-09-12 15:30:00', '2021-09-13 21:30:00']", "ground_truth": [ "2021-09-12 15:30:00", "2021-09-13 21:30:00" ], "eval_metric": "iou", "channel": "728", "ts_data_path": "ts_data/L2_T1_Shape_Identification_S4_00005.csv" }, { "id": "L2_T2_Periodicity_Detection_S3_00008", "level": 2, "level_name": "Pattern Recognition", "category": "Periodicity Detection", "subtask": "Periodicity Detection", "question": "What is the dominant cycle period (in data points) of channel OT within [2018-01-08 06:00:00 to 2018-01-17 12:15:00]? (Output format: integer)", "answer": "13", "ground_truth": 13, "eval_metric": "rel_acc", "channel": "OT", "ts_data_path": "ts_data/L2_T2_Periodicity_Detection_S3_00008.csv" }, { "id": "L3_T1_Composite_Trend_S4_00010", "level": 3, "level_name": "Semantic Reasoning", "category": "Composite Trend", "subtask": "Composite Trend", "question": "Among days in channel 237 during 2022 that exhibit the trend pattern 'fall, then slow fall, then fall, then fluctuating stable', identify the top-8 days where the slow fall segment is the slowest. (Output format: a ranked list of dates, e.g., ['YYYY-MM-DD', ...])", "answer": "['2022-01-07', '2022-02-05', '2022-10-22', '2022-05-18', '2022-09-30', '2022-08-13', '2022-10-10', '2022-06-02']", "ground_truth": [ "2022-01-07", "2022-02-05", "2022-10-22", "2022-05-18", "2022-09-30", "2022-08-13", "2022-10-10", "2022-06-02" ], "eval_metric": "set_f1", "channel": "237", "ts_data_path": "ts_data/L3_T1_Composite_Trend_S4_00010.csv" }, { "id": "L2_T2_Periodicity_Detection_S2_00006", "level": 2, "level_name": "Pattern Recognition", "category": "Periodicity Detection", "subtask": "Periodicity Detection", "question": "What is the dominant cycle period (in data points) of channel 813 within [2019-06-29 10:30:00 to 2019-07-04 14:45:00]? (Output format: integer)", "answer": "61", "ground_truth": 61, "eval_metric": "rel_acc", "channel": "813", "ts_data_path": "ts_data/L2_T2_Periodicity_Detection_S2_00006.csv" }, { "id": "L1_T3_Interval_Discovery_S3_00007", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Interval Discovery", "question": "Find the longest period where channel HUFL remained above 12.458 in 2016-10-19 to 2016-12-31. (Output format: a time interval, e.g., [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2016-12-23 23:00:00, 2016-12-24 08:45:00]", "ground_truth": [ "2016-12-23 23:00:00", "2016-12-24 08:45:00" ], "eval_metric": "iou", "channel": "HUFL", "ts_data_path": "ts_data/L1_T3_Interval_Discovery_S3_00007.csv" }, { "id": "L1_T4_Sliding_Window_S4_00001", "level": 1, "level_name": "Basic Operations", "category": "Sliding Window", "subtask": "Sliding Window", "question": "Which 8-day period in 2023-05-13 to 2023-11-23 had the highest average for channel 67? (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2023-11-13 17:45:00, 2023-11-21 17:30:00]", "ground_truth": [ "2023-11-13 17:45:00", "2023-11-21 17:30:00" ], "eval_metric": "iou", "channel": "67", "ts_data_path": "ts_data/L1_T4_Sliding_Window_S4_00001.csv" }, { "id": "L1_T2_Temporal_Localization_S3_00006", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Temporal Localization", "question": "At what exact timestamp did channel 683 first rise above 1.36 in 2020-05-29 to 2020-06-08? (Output format: a specific timestamp, e.g., YYYY-MM-DD HH:MM:SS)", "answer": "2020-05-29 13:00:00", "ground_truth": "2020-05-29 13:00:00", "eval_metric": "hit", "channel": "683", "ts_data_path": "ts_data/L1_T2_Temporal_Localization_S3_00006.csv" }, { "id": "L2_T1_Shape_Identification_S2_00005", "level": 2, "level_name": "Pattern Recognition", "category": "Shape Identification", "subtask": "Shape Identification", "question": "Identify the time range of the longest plateau (stable period) in channel HULL within [2017-05-04 19:30:00 to 2017-05-12 15:45:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2017-05-10 16:00:00', '2017-05-11 20:00:00']", "ground_truth": [ "2017-05-10 16:00:00", "2017-05-11 20:00:00" ], "eval_metric": "iou", "channel": "HULL", "ts_data_path": "ts_data/L2_T1_Shape_Identification_S2_00005.csv" }, { "id": "L1_T3_Interval_Discovery_S4_00005", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Interval Discovery", "question": "Find the longest period where channel HUFL remained above 10.114 in 2017-06-25 to 2018-01-21. (Output format: a time interval, e.g., [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2017-07-30 19:15:00, 2017-08-01 06:15:00]", "ground_truth": [ "2017-07-30 19:15:00", "2017-08-01 06:15:00" ], "eval_metric": "iou", "channel": "HUFL", "ts_data_path": "ts_data/L1_T3_Interval_Discovery_S4_00005.csv" }, { "id": "L3_T3_Causal_Anomaly_S0_00001", "level": 3, "level_name": "Semantic Reasoning", "category": "Causal Anomaly", "subtask": "Causal Anomaly", "question": "Given that channel 151_up_2022 is the upstream source of channel 151_down_2022, identify the time period in 2022 where 151_down_2022 shows a significant causal anomaly, such as an flat line during high activity. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2022-09-01 06:30:00, 2022-09-02 12:45:00]", "ground_truth": [ "2022-09-01 06:30:00", "2022-09-02 12:45:00" ], "eval_metric": "iou", "channel": "151_down_2022", "ts_data_path": "ts_data/L3_T3_Causal_Anomaly_S0_00001.csv" }, { "id": "L3_T1_Composite_Trend_S3_00003", "level": 3, "level_name": "Semantic Reasoning", "category": "Composite Trend", "subtask": "Composite Trend", "question": "Among days in channel 155 during 2022 that exhibit the trend pattern 'steady stable, then rapid rise', identify the top-6 days where the rapid rise segment is the fastest. (Output format: a ranked list of dates, e.g., ['YYYY-MM-DD', ...])", "answer": "['2022-07-28', '2022-11-13', '2022-06-05', '2022-04-15', '2022-01-06', '2022-03-04']", "ground_truth": [ "2022-07-28", "2022-11-13", "2022-06-05", "2022-04-15", "2022-01-06", "2022-03-04" ], "eval_metric": "set_f1", "channel": "155", "ts_data_path": "ts_data/L3_T1_Composite_Trend_S3_00003.csv" }, { "id": "L1_T4_Sliding_Window_S2_00001", "level": 1, "level_name": "Basic Operations", "category": "Sliding Window", "subtask": "Sliding Window", "question": "Which 41-day period in 2019-01-25 to 2019-06-26 had the highest variance for channel 626? (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2019-03-14 23:15:00, 2019-04-24 23:00:00]", "ground_truth": [ "2019-03-14 23:15:00", "2019-04-24 23:00:00" ], "eval_metric": "iou", "channel": "626", "ts_data_path": "ts_data/L1_T4_Sliding_Window_S2_00001.csv" }, { "id": "L1_T2_Temporal_Localization_S0_00003", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Temporal Localization", "question": "At what exact timestamp did channel 501 reach its minimum value in 2021-03-28 to 2021-04-19? (Output format: a specific timestamp, e.g., YYYY-MM-DD HH:MM:SS)", "answer": "2021-04-17 22:00:00", "ground_truth": "2021-04-17 22:00:00", "eval_metric": "hit", "channel": "501", "ts_data_path": "ts_data/L1_T2_Temporal_Localization_S0_00003.csv" }, { "id": "L2_T3_Subsequence_Matching_S4_00007", "level": 2, "level_name": "Pattern Recognition", "category": "Subsequence Matching", "subtask": "Subsequence Matching", "question": "Analyze the reference pattern in [2017-05-01 09:45:00 to 2017-05-01 15:30:00]. Find the time interval where channel LULL exhibits the most similar pattern within the search context [2017-05-03 06:00:00 to 2017-05-09 09:45:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2017-05-08 22:15:00', '2017-05-09 04:00:00']", "ground_truth": [ "2017-05-08 22:15:00", "2017-05-09 04:00:00" ], "eval_metric": "iou", "channel": "LULL", "ts_data_path": "ts_data/L2_T3_Subsequence_Matching_S4_00007.csv" }, { "id": "L4_T1_Insight_Synthesis_S0_00009", "level": 4, "level_name": "Insight Synthesis", "category": "Insight Synthesis", "subtask": "Insight Synthesis", "question": "Analyze the behavior of channel LULL for the period 2017-12.\nPlease use ONLY the following phrases for trend description: rapid rise, gradual rise, rapid fall, gradual fall, steady stable, fluctuating stable.\nProvide a structured report covering:\n1. Trend Segmentation: Describe each stage with precise start/end timestamps (HH:MM:SS) using the phrases above.\n2. Outlier Audit: Identify only significant outliers that deviate sharply from the local trend. Ignore minor background noise.\n(Output format: structured natural language report.)", "answer": "1. Trend Segmentation: from 2017-12-01 00:00:00 to 2017-12-07 15:30:00, the trend showed a fluctuating stable; from 2017-12-07 15:30:00 to 2017-12-18 17:15:00, the trend showed a rapid fall; from 2017-12-18 17:15:00 to 2017-12-31 23:45:00, the trend showed a gradual rise.\n2. Outlier Audit: A significant drop was detected at 2017-12-22 16:30:00 (value: -5.25).", "ground_truth": { "trend_segments": [ "from 2017-12-01 00:00:00 to 2017-12-07 15:30:00, the trend showed a fluctuating stable", "from 2017-12-07 15:30:00 to 2017-12-18 17:15:00, the trend showed a rapid fall", "from 2017-12-18 17:15:00 to 2017-12-31 23:45:00, the trend showed a gradual rise" ], "segments_meta": [ { "adj": "fluctuating", "kind": "stable", "start_idx": 0, "end_idx": 638 }, { "adj": "rapid", "kind": "fall", "start_idx": 638, "end_idx": 1701 }, { "adj": "gradual", "kind": "rise", "start_idx": 1701, "end_idx": 2976 } ], "significant_anomaly": { "timestamp": "2017-12-22 16:30:00", "kind": "significant drop", "value": -5.253295223252302 } }, "eval_metric": "report", "channel": "LULL", "ts_data_path": "ts_data/L4_T1_Insight_Synthesis_S0_00009.csv" }, { "id": "L4_T1_Insight_Synthesis_S0_00008", "level": 4, "level_name": "Insight Synthesis", "category": "Insight Synthesis", "subtask": "Insight Synthesis", "question": "Analyze the behavior of channel MULL for the period 2018-02.\nPlease use ONLY the following phrases for trend description: rapid rise, gradual rise, rapid fall, gradual fall, steady stable, fluctuating stable.\nProvide a structured report covering:\n1. Trend Segmentation: Describe each stage with precise start/end timestamps (HH:MM:SS) using the phrases above.\n2. Outlier Audit: Identify only significant outliers that deviate sharply from the local trend. Ignore minor background noise.\n(Output format: structured natural language report.)", "answer": "1. Trend Segmentation: from 2018-02-01 00:00:00 to 2018-02-04 17:30:00, the trend showed a rapid rise; from 2018-02-04 17:30:00 to 2018-02-12 04:45:00, the trend showed a rapid fall; from 2018-02-12 04:45:00 to 2018-02-23 09:30:00, the trend showed a fluctuating stable; from 2018-02-23 09:30:00 to 2018-02-28 23:45:00, the trend showed a rapid rise.\n2. Outlier Audit: A significant spike was detected at 2018-02-09 14:15:00 (value: 20.92).", "ground_truth": { "trend_segments": [ "from 2018-02-01 00:00:00 to 2018-02-04 17:30:00, the trend showed a rapid rise", "from 2018-02-04 17:30:00 to 2018-02-12 04:45:00, the trend showed a rapid fall", "from 2018-02-12 04:45:00 to 2018-02-23 09:30:00, the trend showed a fluctuating stable", "from 2018-02-23 09:30:00 to 2018-02-28 23:45:00, the trend showed a rapid rise" ], "segments_meta": [ { "adj": "rapid", "kind": "rise", "start_idx": 0, "end_idx": 358 }, { "adj": "rapid", "kind": "fall", "start_idx": 358, "end_idx": 1075 }, { "adj": "fluctuating", "kind": "stable", "start_idx": 1075, "end_idx": 2150 }, { "adj": "rapid", "kind": "rise", "start_idx": 2150, "end_idx": 2688 } ], "significant_anomaly": { "timestamp": "2018-02-09 14:15:00", "kind": "significant spike", "value": 20.918145602455713 } }, "eval_metric": "report", "channel": "MULL", "ts_data_path": "ts_data/L4_T1_Insight_Synthesis_S0_00008.csv" }, { "id": "L4_T1_Insight_Synthesis_S4_00003", "level": 4, "level_name": "Insight Synthesis", "category": "Insight Synthesis", "subtask": "Insight Synthesis", "question": "Analyze the behavior of channel 312 for the period 2023-08.\nPlease use ONLY the following phrases for trend description: rapid rise, gradual rise, rapid fall, gradual fall, steady stable, fluctuating stable.\nProvide a structured report covering:\n1. Trend Segmentation: Describe each stage with precise start/end timestamps (HH:MM:SS) using the phrases above.\n2. Outlier Audit: Identify only significant outliers that deviate sharply from the local trend. Ignore minor background noise.\n(Output format: structured natural language report.)", "answer": "1. Trend Segmentation: from 2023-08-01 00:00:00 to 2023-08-14 06:45:00, the trend showed a gradual fall; from 2023-08-14 06:45:00 to 2023-08-23 03:15:00, the trend showed a gradual rise; from 2023-08-23 03:15:00 to 2023-08-31 23:45:00, the trend showed a fluctuating stable.\n2. Outlier Audit: A significant spike was detected at 2023-08-10 16:00:00 (value: 4.05).", "ground_truth": { "trend_segments": [ "from 2023-08-01 00:00:00 to 2023-08-14 06:45:00, the trend showed a gradual fall", "from 2023-08-14 06:45:00 to 2023-08-23 03:15:00, the trend showed a gradual rise", "from 2023-08-23 03:15:00 to 2023-08-31 23:45:00, the trend showed a fluctuating stable" ], "segments_meta": [ { "adj": "gradual", "kind": "fall", "start_idx": 0, "end_idx": 1275 }, { "adj": "gradual", "kind": "rise", "start_idx": 1275, "end_idx": 2125 }, { "adj": "fluctuating", "kind": "stable", "start_idx": 2125, "end_idx": 2976 } ], "significant_anomaly": { "timestamp": "2023-08-10 16:00:00", "kind": "significant spike", "value": 4.046739232781836 } }, "eval_metric": "report", "channel": "312", "ts_data_path": "ts_data/L4_T1_Insight_Synthesis_S4_00003.csv" }, { "id": "L4_T1_Insight_Synthesis_S1_00005", "level": 4, "level_name": "Insight Synthesis", "category": "Insight Synthesis", "subtask": "Insight Synthesis", "question": "Analyze the behavior of channel 811 for the period 2023-06.\nPlease use ONLY the following phrases for trend description: rapid rise, gradual rise, rapid fall, gradual fall, steady stable, fluctuating stable.\nProvide a structured report covering:\n1. Trend Segmentation: Describe each stage with precise start/end timestamps (HH:MM:SS) using the phrases above.\n2. Outlier Audit: Identify only significant outliers that deviate sharply from the local trend. Ignore minor background noise.\n(Output format: structured natural language report.)", "answer": "1. Trend Segmentation: from 2023-06-01 00:00:00 to 2023-06-08 21:30:00, the trend showed a steady stable; from 2023-06-08 21:30:00 to 2023-06-16 19:00:00, the trend showed a rapid rise; from 2023-06-16 19:00:00 to 2023-06-21 12:45:00, the trend showed a steady stable; from 2023-06-21 12:45:00 to 2023-06-30 23:45:00, the trend showed a gradual fall.\n2. Outlier Audit: A significant drop was detected at 2023-06-27 01:45:00 (value: -78.49).", "ground_truth": { "trend_segments": [ "from 2023-06-01 00:00:00 to 2023-06-08 21:30:00, the trend showed a steady stable", "from 2023-06-08 21:30:00 to 2023-06-16 19:00:00, the trend showed a rapid rise", "from 2023-06-16 19:00:00 to 2023-06-21 12:45:00, the trend showed a steady stable", "from 2023-06-21 12:45:00 to 2023-06-30 23:45:00, the trend showed a gradual fall" ], "segments_meta": [ { "adj": "steady", "kind": "stable", "start_idx": 0, "end_idx": 758 }, { "adj": "rapid", "kind": "rise", "start_idx": 758, "end_idx": 1516 }, { "adj": "steady", "kind": "stable", "start_idx": 1516, "end_idx": 1971 }, { "adj": "gradual", "kind": "fall", "start_idx": 1971, "end_idx": 2880 } ], "significant_anomaly": { "timestamp": "2023-06-27 01:45:00", "kind": "significant drop", "value": -78.49421749703829 } }, "eval_metric": "report", "channel": "811", "ts_data_path": "ts_data/L4_T1_Insight_Synthesis_S1_00005.csv" }, { "id": "L2_T3_Subsequence_Matching_S3_00001", "level": 2, "level_name": "Pattern Recognition", "category": "Subsequence Matching", "subtask": "Subsequence Matching", "question": "Analyze the reference pattern in [2019-07-22 12:00:00 to 2019-07-23 05:45:00]. Find the time interval where channel 147 exhibits the most similar pattern within the search context [2019-07-25 03:00:00 to 2019-07-31 21:45:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2019-07-28 12:00:00', '2019-07-29 05:45:00']", "ground_truth": [ "2019-07-28 12:00:00", "2019-07-29 05:45:00" ], "eval_metric": "iou", "channel": "147", "ts_data_path": "ts_data/L2_T3_Subsequence_Matching_S3_00001.csv" }, { "id": "L1_T1_Global_Aggregation_S2_00003", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Global Aggregation", "question": "What is the minimum value of channel 754 in 2023-04-08 to 2023-05-07? (Output format: a single numeric value, rounded to 3 decimal places, e.g., x.xxx)", "answer": "0.207", "ground_truth": 0.207, "eval_metric": "rel_acc", "channel": "754", "ts_data_path": "ts_data/L1_T1_Global_Aggregation_S2_00003.csv" }, { "id": "L3_T2_Contextual_Anomaly_S4_00003", "level": 3, "level_name": "Semantic Reasoning", "category": "Contextual Anomaly", "subtask": "Contextual Anomaly", "question": "Identify the period in channel 124 during 2019 that experienced the most significant historically high water level. (Output format: [YYYY-MM-DD, YYYY-MM-DD])", "answer": "[2019-04-06 14:00:00, 2019-04-16 13:45:00]", "ground_truth": [ "2019-04-06 14:00:00", "2019-04-16 13:45:00" ], "eval_metric": "iou", "channel": "124", "ts_data_path": "ts_data/L3_T2_Contextual_Anomaly_S4_00003.csv" }, { "id": "L3_T3_Causal_Anomaly_S2_00009", "level": 3, "level_name": "Semantic Reasoning", "category": "Causal Anomaly", "subtask": "Causal Anomaly", "question": "Given that channel 729_up_2020 is the upstream source of channel 729_down_2020, identify the time period in 2020 where 729_down_2020 shows a significant causal anomaly, such as an inverse trend against the source. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2020-03-18 19:15:00, 2020-03-20 11:15:00]", "ground_truth": [ "2020-03-18 19:15:00", "2020-03-20 11:15:00" ], "eval_metric": "iou", "channel": "729_down_2020", "ts_data_path": "ts_data/L3_T3_Causal_Anomaly_S2_00009.csv" }, { "id": "L3_T3_Causal_Anomaly_S0_00008", "level": 3, "level_name": "Semantic Reasoning", "category": "Causal Anomaly", "subtask": "Causal Anomaly", "question": "Given that channel 495_up_2023 is the upstream source of channel 495_down_2023, identify the time period in 2023 where 495_down_2023 shows a significant causal anomaly, such as an inverse trend against the source. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2023-02-26 06:00:00, 2023-02-28 17:00:00]", "ground_truth": [ "2023-02-26 06:00:00", "2023-02-28 17:00:00" ], "eval_metric": "iou", "channel": "495_down_2023", "ts_data_path": "ts_data/L3_T3_Causal_Anomaly_S0_00008.csv" }, { "id": "L2_T2_Periodicity_Detection_S3_00009", "level": 2, "level_name": "Pattern Recognition", "category": "Periodicity Detection", "subtask": "Periodicity Detection", "question": "What is the dominant cycle period (in data points) of channel 154 within [2021-07-16 06:45:00 to 2021-07-26 03:15:00]? (Output format: integer)", "answer": "50", "ground_truth": 50, "eval_metric": "rel_acc", "channel": "154", "ts_data_path": "ts_data/L2_T2_Periodicity_Detection_S3_00009.csv" }, { "id": "L1_T4_Sliding_Window_S0_00005", "level": 1, "level_name": "Basic Operations", "category": "Sliding Window", "subtask": "Sliding Window", "question": "Which 6-day period in 2021-08-06 to 2021-12-30 had the lowest average for channel 811? (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2021-10-07 08:00:00, 2021-10-13 07:45:00]", "ground_truth": [ "2021-10-07 08:00:00", "2021-10-13 07:45:00" ], "eval_metric": "iou", "channel": "811", "ts_data_path": "ts_data/L1_T4_Sliding_Window_S0_00005.csv" }, { "id": "L3_T3_Causal_Anomaly_S3_00001", "level": 3, "level_name": "Semantic Reasoning", "category": "Causal Anomaly", "subtask": "Causal Anomaly", "question": "Given that channel 151_up_2022 is the upstream source of channel 151_down_2022, identify the time period in 2022 where 151_down_2022 shows a significant causal anomaly, such as an inverse trend against the source. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2022-07-20 12:00:00, 2022-07-24 02:45:00]", "ground_truth": [ "2022-07-20 12:00:00", "2022-07-24 02:45:00" ], "eval_metric": "iou", "channel": "151_down_2022", "ts_data_path": "ts_data/L3_T3_Causal_Anomaly_S3_00001.csv" }, { "id": "L3_T2_Contextual_Anomaly_S0_00006", "level": 3, "level_name": "Semantic Reasoning", "category": "Contextual Anomaly", "subtask": "Contextual Anomaly", "question": "Identify the period in channel 895 during 2023 that experienced the most significant severe drought. (Output format: [YYYY-MM-DD, YYYY-MM-DD])", "answer": "[2023-04-26 15:45:00, 2023-05-29 15:30:00]", "ground_truth": [ "2023-04-26 15:45:00", "2023-05-29 15:30:00" ], "eval_metric": "iou", "channel": "895", "ts_data_path": "ts_data/L3_T2_Contextual_Anomaly_S0_00006.csv" }, { "id": "L3_T1_Composite_Trend_S3_00005", "level": 3, "level_name": "Semantic Reasoning", "category": "Composite Trend", "subtask": "Composite Trend", "question": "Among days in channel 647 during 2019 that exhibit the trend pattern 'rise, then fluctuating stable, then rise', identify the top-6 days where the fluctuating stable segment is the most fluctuating. (Output format: a ranked list of dates, e.g., ['YYYY-MM-DD', ...])", "answer": "['2019-04-11', '2019-05-09', '2019-10-02', '2019-12-13', '2019-07-20', '2019-09-17']", "ground_truth": [ "2019-04-11", "2019-05-09", "2019-10-02", "2019-12-13", "2019-07-20", "2019-09-17" ], "eval_metric": "set_f1", "channel": "647", "ts_data_path": "ts_data/L3_T1_Composite_Trend_S3_00005.csv" }, { "id": "L1_T2_Temporal_Localization_S3_00001", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Temporal Localization", "question": "At what exact timestamp did channel 67 last fall below 0.199 in 2022-03? (Output format: a specific timestamp, e.g., YYYY-MM-DD HH:MM:SS)", "answer": "2022-03-31 10:00:00", "ground_truth": "2022-03-31 10:00:00", "eval_metric": "hit", "channel": "67", "ts_data_path": "ts_data/L1_T2_Temporal_Localization_S3_00001.csv" }, { "id": "L4_T1_Insight_Synthesis_S0_00003", "level": 4, "level_name": "Insight Synthesis", "category": "Insight Synthesis", "subtask": "Insight Synthesis", "question": "Analyze the behavior of channel 169 for the period 2019-10.\nPlease use ONLY the following phrases for trend description: rapid rise, gradual rise, rapid fall, gradual fall, steady stable, fluctuating stable.\nProvide a structured report covering:\n1. Trend Segmentation: Describe each stage with precise start/end timestamps (HH:MM:SS) using the phrases above.\n2. Outlier Audit: Identify only significant outliers that deviate sharply from the local trend. Ignore minor background noise.\n(Output format: structured natural language report.)", "answer": "1. Trend Segmentation: from 2019-10-01 00:00:00 to 2019-10-06 19:30:00, the trend showed a gradual rise; from 2019-10-06 19:30:00 to 2019-10-14 13:30:00, the trend showed a gradual fall; from 2019-10-14 13:30:00 to 2019-10-20 09:00:00, the trend showed a rapid rise; from 2019-10-20 09:00:00 to 2019-10-31 23:45:00, the trend showed a gradual fall.\n2. Outlier Audit: A significant spike was detected at 2019-10-04 04:45:00 (value: 2777.69).", "ground_truth": { "trend_segments": [ "from 2019-10-01 00:00:00 to 2019-10-06 19:30:00, the trend showed a gradual rise", "from 2019-10-06 19:30:00 to 2019-10-14 13:30:00, the trend showed a gradual fall", "from 2019-10-14 13:30:00 to 2019-10-20 09:00:00, the trend showed a rapid rise", "from 2019-10-20 09:00:00 to 2019-10-31 23:45:00, the trend showed a gradual fall" ], "segments_meta": [ { "adj": "gradual", "kind": "rise", "start_idx": 0, "end_idx": 558 }, { "adj": "gradual", "kind": "fall", "start_idx": 558, "end_idx": 1302 }, { "adj": "rapid", "kind": "rise", "start_idx": 1302, "end_idx": 1860 }, { "adj": "gradual", "kind": "fall", "start_idx": 1860, "end_idx": 2976 } ], "significant_anomaly": { "timestamp": "2019-10-04 04:45:00", "kind": "significant spike", "value": 2777.6910470956473 } }, "eval_metric": "report", "channel": "169", "ts_data_path": "ts_data/L4_T1_Insight_Synthesis_S0_00003.csv" }, { "id": "L1_T1_Global_Aggregation_S1_00005", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Global Aggregation", "question": "What is the range value of channel 441 in 2022-08-18 to 2022-08-29? (Output format: a single numeric value, rounded to 3 decimal places, e.g., x.xxx)", "answer": "20.08", "ground_truth": 20.08, "eval_metric": "rel_acc", "channel": "441", "ts_data_path": "ts_data/L1_T1_Global_Aggregation_S1_00005.csv" }, { "id": "L2_T1_Shape_Identification_S2_00008", "level": 2, "level_name": "Pattern Recognition", "category": "Shape Identification", "subtask": "Shape Identification", "question": "Identify the time range of the largest step descent in channel 151 within [2023-05-12 14:00:00 to 2023-05-24 16:30:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2023-05-12 18:30:00', '2023-05-14 13:45:00']", "ground_truth": [ "2023-05-12 18:30:00", "2023-05-14 13:45:00" ], "eval_metric": "iou", "channel": "151", "ts_data_path": "ts_data/L2_T1_Shape_Identification_S2_00008.csv" }, { "id": "L1_T1_Global_Aggregation_S3_00001", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Global Aggregation", "question": "What is the range value of channel 124 in 2022-02-11 to 2022-03-29? (Output format: a single numeric value, rounded to 3 decimal places, e.g., x.xxx)", "answer": "0.652", "ground_truth": 0.652, "eval_metric": "rel_acc", "channel": "124", "ts_data_path": "ts_data/L1_T1_Global_Aggregation_S3_00001.csv" }, { "id": "L4_T1_Insight_Synthesis_S2_00001", "level": 4, "level_name": "Insight Synthesis", "category": "Insight Synthesis", "subtask": "Insight Synthesis", "question": "Analyze the behavior of channel 67 for the period 2022-05.\nPlease use ONLY the following phrases for trend description: rapid rise, gradual rise, rapid fall, gradual fall, steady stable, fluctuating stable.\nProvide a structured report covering:\n1. Trend Segmentation: Describe each stage with precise start/end timestamps (HH:MM:SS) using the phrases above.\n2. Outlier Audit: Identify only significant outliers that deviate sharply from the local trend. Ignore minor background noise.\n(Output format: structured natural language report.)", "answer": "1. Trend Segmentation: from 2022-05-01 00:00:00 to 2022-05-08 18:00:00, the trend showed a gradual rise; from 2022-05-08 18:00:00 to 2022-05-14 22:45:00, the trend showed a fluctuating stable; from 2022-05-14 22:45:00 to 2022-05-18 01:15:00, the trend showed a rapid rise; from 2022-05-18 01:15:00 to 2022-05-22 16:45:00, the trend showed a fluctuating stable; from 2022-05-22 16:45:00 to 2022-05-31 23:45:00, the trend showed a steady stable.\n2. Outlier Audit: A significant drop was detected at 2022-05-23 03:30:00 (value: -0.24).", "ground_truth": { "trend_segments": [ "from 2022-05-01 00:00:00 to 2022-05-08 18:00:00, the trend showed a gradual rise", "from 2022-05-08 18:00:00 to 2022-05-14 22:45:00, the trend showed a fluctuating stable", "from 2022-05-14 22:45:00 to 2022-05-18 01:15:00, the trend showed a rapid rise", "from 2022-05-18 01:15:00 to 2022-05-22 16:45:00, the trend showed a fluctuating stable", "from 2022-05-22 16:45:00 to 2022-05-31 23:45:00, the trend showed a steady stable" ], "segments_meta": [ { "adj": "gradual", "kind": "rise", "start_idx": 0, "end_idx": 744 }, { "adj": "fluctuating", "kind": "stable", "start_idx": 744, "end_idx": 1339 }, { "adj": "rapid", "kind": "rise", "start_idx": 1339, "end_idx": 1637 }, { "adj": "fluctuating", "kind": "stable", "start_idx": 1637, "end_idx": 2083 }, { "adj": "steady", "kind": "stable", "start_idx": 2083, "end_idx": 2976 } ], "significant_anomaly": { "timestamp": "2022-05-23 03:30:00", "kind": "significant drop", "value": -0.2366331016394389 } }, "eval_metric": "report", "channel": "67", "ts_data_path": "ts_data/L4_T1_Insight_Synthesis_S2_00001.csv" }, { "id": "L3_T2_Contextual_Anomaly_S2_00005", "level": 3, "level_name": "Semantic Reasoning", "category": "Contextual Anomaly", "subtask": "Contextual Anomaly", "question": "Identify the period in channel 762 during 2019 that experienced the most significant extreme surge in flow. (Output format: [YYYY-MM-DD, YYYY-MM-DD])", "answer": "[2019-08-18 03:45:00, 2019-09-27 03:30:00]", "ground_truth": [ "2019-08-18 03:45:00", "2019-09-27 03:30:00" ], "eval_metric": "iou", "channel": "762", "ts_data_path": "ts_data/L3_T2_Contextual_Anomaly_S2_00005.csv" }, { "id": "L3_T1_Composite_Trend_S3_00009", "level": 3, "level_name": "Semantic Reasoning", "category": "Composite Trend", "subtask": "Composite Trend", "question": "Among days in channel HULL during 2017 that exhibit the trend pattern 'rise, then fluctuating stable, then steady stable', identify the top-6 days where the fluctuating stable segment is the most fluctuating. (Output format: a ranked list of dates, e.g., ['YYYY-MM-DD', ...])", "answer": "['2017-01-30', '2017-12-23', '2017-07-25', '2017-12-02', '2017-03-02', '2017-01-23']", "ground_truth": [ "2017-01-30", "2017-12-23", "2017-07-25", "2017-12-02", "2017-03-02", "2017-01-23" ], "eval_metric": "set_f1", "channel": "HULL", "ts_data_path": "ts_data/L3_T1_Composite_Trend_S3_00009.csv" }, { "id": "L3_T3_Causal_Anomaly_S4_00003", "level": 3, "level_name": "Semantic Reasoning", "category": "Causal Anomaly", "subtask": "Causal Anomaly", "question": "Given that channel 625_up_2019 is the upstream source of channel 625_down_2019, identify the time period in 2019 where 625_down_2019 shows a significant causal anomaly, such as an flat line during high activity. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2019-05-02 03:45:00, 2019-05-20 06:15:00]", "ground_truth": [ "2019-05-02 03:45:00", "2019-05-20 06:15:00" ], "eval_metric": "iou", "channel": "625_down_2019", "ts_data_path": "ts_data/L3_T3_Causal_Anomaly_S4_00003.csv" }, { "id": "L4_T1_Insight_Synthesis_S2_00008", "level": 4, "level_name": "Insight Synthesis", "category": "Insight Synthesis", "subtask": "Insight Synthesis", "question": "Analyze the behavior of channel 891 for the period 2022-08.\nPlease use ONLY the following phrases for trend description: rapid rise, gradual rise, rapid fall, gradual fall, steady stable, fluctuating stable.\nProvide a structured report covering:\n1. Trend Segmentation: Describe each stage with precise start/end timestamps (HH:MM:SS) using the phrases above.\n2. Outlier Audit: Identify only significant outliers that deviate sharply from the local trend. Ignore minor background noise.\n(Output format: structured natural language report.)", "answer": "1. Trend Segmentation: from 2022-08-01 00:00:00 to 2022-08-19 14:30:00, the trend showed a rapid fall; from 2022-08-19 14:30:00 to 2022-08-31 23:45:00, the trend showed a gradual fall.\n2. Outlier Audit: A significant spike was detected at 2022-08-18 11:00:00 (value: 17.73).", "ground_truth": { "trend_segments": [ "from 2022-08-01 00:00:00 to 2022-08-19 14:30:00, the trend showed a rapid fall", "from 2022-08-19 14:30:00 to 2022-08-31 23:45:00, the trend showed a gradual fall" ], "segments_meta": [ { "adj": "rapid", "kind": "fall", "start_idx": 0, "end_idx": 1786 }, { "adj": "gradual", "kind": "fall", "start_idx": 1786, "end_idx": 2976 } ], "significant_anomaly": { "timestamp": "2022-08-18 11:00:00", "kind": "significant spike", "value": 17.731442628709956 } }, "eval_metric": "report", "channel": "891", "ts_data_path": "ts_data/L4_T1_Insight_Synthesis_S2_00008.csv" }, { "id": "L3_T2_Contextual_Anomaly_S4_00006", "level": 3, "level_name": "Semantic Reasoning", "category": "Contextual Anomaly", "subtask": "Contextual Anomaly", "question": "Identify the period in channel 441 during 2021 that experienced the most significant extreme surge in flow. (Output format: [YYYY-MM-DD, YYYY-MM-DD])", "answer": "[2021-05-12 03:15:00, 2021-05-26 03:00:00]", "ground_truth": [ "2021-05-12 03:15:00", "2021-05-26 03:00:00" ], "eval_metric": "iou", "channel": "441", "ts_data_path": "ts_data/L3_T2_Contextual_Anomaly_S4_00006.csv" }, { "id": "L2_T2_Periodicity_Detection_S1_00001", "level": 2, "level_name": "Pattern Recognition", "category": "Periodicity Detection", "subtask": "Periodicity Detection", "question": "What is the dominant cycle period (in data points) of channel 123 within [2023-07-26 06:00:00 to 2023-08-01 09:00:00]? (Output format: integer)", "answer": "145", "ground_truth": 145, "eval_metric": "rel_acc", "channel": "123", "ts_data_path": "ts_data/L2_T2_Periodicity_Detection_S1_00001.csv" }, { "id": "L1_T4_Sliding_Window_S0_00002", "level": 1, "level_name": "Basic Operations", "category": "Sliding Window", "subtask": "Sliding Window", "question": "Which 8-day period in 2019-08-25 to 2019-12-16 had the lowest average for channel 154? (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2019-09-21 19:45:00, 2019-09-29 19:30:00]", "ground_truth": [ "2019-09-21 19:45:00", "2019-09-29 19:30:00" ], "eval_metric": "iou", "channel": "154", "ts_data_path": "ts_data/L1_T4_Sliding_Window_S0_00002.csv" }, { "id": "L1_T2_Temporal_Localization_S2_00005", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Temporal Localization", "question": "At what exact timestamp did channel 727 reach its minimum value in 2020-02-01 to 2020-04-22? (Output format: a specific timestamp, e.g., YYYY-MM-DD HH:MM:SS)", "answer": "2020-04-19 10:00:00", "ground_truth": "2020-04-19 10:00:00", "eval_metric": "hit", "channel": "727", "ts_data_path": "ts_data/L1_T2_Temporal_Localization_S2_00005.csv" }, { "id": "L1_T1_Global_Aggregation_S3_00005", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Global Aggregation", "question": "What is the range value of channel 578 in 2020-09-28 to 2020-12-07? (Output format: a single numeric value, rounded to 3 decimal places, e.g., x.xxx)", "answer": "0.947", "ground_truth": 0.947, "eval_metric": "rel_acc", "channel": "578", "ts_data_path": "ts_data/L1_T1_Global_Aggregation_S3_00005.csv" }, { "id": "L2_T2_Periodicity_Detection_S4_00005", "level": 2, "level_name": "Pattern Recognition", "category": "Periodicity Detection", "subtask": "Periodicity Detection", "question": "What is the dominant cycle period (in data points) of channel 589 within [2022-07-22 19:00:00 to 2022-07-30 10:00:00]? (Output format: integer)", "answer": "60", "ground_truth": 60, "eval_metric": "rel_acc", "channel": "589", "ts_data_path": "ts_data/L2_T2_Periodicity_Detection_S4_00005.csv" }, { "id": "L1_T3_Interval_Discovery_S0_00009", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Interval Discovery", "question": "Find the longest period where channel 312 remained above 0.104 in 2020-06. (Output format: a time interval, e.g., [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2020-06-28 16:15:00, 2020-06-30 11:45:00]", "ground_truth": [ "2020-06-28 16:15:00", "2020-06-30 11:45:00" ], "eval_metric": "iou", "channel": "312", "ts_data_path": "ts_data/L1_T3_Interval_Discovery_S0_00009.csv" }, { "id": "L3_T3_Causal_Anomaly_S0_00002", "level": 3, "level_name": "Semantic Reasoning", "category": "Causal Anomaly", "subtask": "Causal Anomaly", "question": "Given that channel 237_up_2021 is the upstream source of channel 237_down_2021, identify the time period in 2021 where 237_down_2021 shows a significant causal anomaly, such as an inverse trend against the source. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2021-03-01 09:15:00, 2021-03-02 22:30:00]", "ground_truth": [ "2021-03-01 09:15:00", "2021-03-02 22:30:00" ], "eval_metric": "iou", "channel": "237_down_2021", "ts_data_path": "ts_data/L3_T3_Causal_Anomaly_S0_00002.csv" }, { "id": "L1_T4_Sliding_Window_S3_00008", "level": 1, "level_name": "Basic Operations", "category": "Sliding Window", "subtask": "Sliding Window", "question": "Which 15-day period in 2022-05-23 to 2023-01-31 had the largest range for channel 595? (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2022-12-09 20:15:00, 2022-12-24 20:00:00]", "ground_truth": [ "2022-12-09 20:15:00", "2022-12-24 20:00:00" ], "eval_metric": "iou", "channel": "595", "ts_data_path": "ts_data/L1_T4_Sliding_Window_S3_00008.csv" }, { "id": "L2_T3_Subsequence_Matching_S2_00007", "level": 2, "level_name": "Pattern Recognition", "category": "Subsequence Matching", "subtask": "Subsequence Matching", "question": "Analyze the reference pattern in [2023-06-27 00:30:00 to 2023-06-27 06:15:00]. Find the time interval where channel 501 exhibits the most similar pattern within the search context [2023-06-29 04:45:00 to 2023-07-06 15:45:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2023-06-30 20:00:00', '2023-07-01 01:45:00']", "ground_truth": [ "2023-06-30 20:00:00", "2023-07-01 01:45:00" ], "eval_metric": "iou", "channel": "501", "ts_data_path": "ts_data/L2_T3_Subsequence_Matching_S2_00007.csv" }, { "id": "L4_T1_Insight_Synthesis_S2_00005", "level": 4, "level_name": "Insight Synthesis", "category": "Insight Synthesis", "subtask": "Insight Synthesis", "question": "Analyze the behavior of channel 728 for the period 2023-01.\nPlease use ONLY the following phrases for trend description: rapid rise, gradual rise, rapid fall, gradual fall, steady stable, fluctuating stable.\nProvide a structured report covering:\n1. Trend Segmentation: Describe each stage with precise start/end timestamps (HH:MM:SS) using the phrases above.\n2. Outlier Audit: Identify only significant outliers that deviate sharply from the local trend. Ignore minor background noise.\n(Output format: structured natural language report.)", "answer": "1. Trend Segmentation: from 2023-01-01 00:00:00 to 2023-01-13 09:30:00, the trend showed a steady stable; from 2023-01-13 09:30:00 to 2023-01-23 17:30:00, the trend showed a rapid rise; from 2023-01-23 17:30:00 to 2023-01-31 23:45:00, the trend showed a gradual fall.\n2. Outlier Audit: A significant drop was detected at 2023-01-26 15:45:00 (value: -526.18).", "ground_truth": { "trend_segments": [ "from 2023-01-01 00:00:00 to 2023-01-13 09:30:00, the trend showed a steady stable", "from 2023-01-13 09:30:00 to 2023-01-23 17:30:00, the trend showed a rapid rise", "from 2023-01-23 17:30:00 to 2023-01-31 23:45:00, the trend showed a gradual fall" ], "segments_meta": [ { "adj": "steady", "kind": "stable", "start_idx": 0, "end_idx": 1190 }, { "adj": "rapid", "kind": "rise", "start_idx": 1190, "end_idx": 2182 }, { "adj": "gradual", "kind": "fall", "start_idx": 2182, "end_idx": 2976 } ], "significant_anomaly": { "timestamp": "2023-01-26 15:45:00", "kind": "significant drop", "value": -526.177448842067 } }, "eval_metric": "report", "channel": "728", "ts_data_path": "ts_data/L4_T1_Insight_Synthesis_S2_00005.csv" }, { "id": "L3_T3_Causal_Anomaly_S3_00004", "level": 3, "level_name": "Semantic Reasoning", "category": "Causal Anomaly", "subtask": "Causal Anomaly", "question": "Given that channel 496_up_2022 is the upstream source of channel 496_down_2022, identify the time period in 2022 where 496_down_2022 shows a significant causal anomaly, such as an inverse trend against the source. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2022-02-11 04:15:00, 2022-02-14 17:45:00]", "ground_truth": [ "2022-02-11 04:15:00", "2022-02-14 17:45:00" ], "eval_metric": "iou", "channel": "496_down_2022", "ts_data_path": "ts_data/L3_T3_Causal_Anomaly_S3_00004.csv" }, { "id": "L3_T1_Composite_Trend_S2_00003", "level": 3, "level_name": "Semantic Reasoning", "category": "Composite Trend", "subtask": "Composite Trend", "question": "Among days in channel 495 during 2020 that exhibit the trend pattern 'slow rise, then rapid fall, then steady stable', identify the top-4 days where the slow rise segment is the slowest. (Output format: a ranked list of dates, e.g., ['YYYY-MM-DD', ...])", "answer": "['2020-01-20', '2020-03-20', '2020-10-17', '2020-03-11']", "ground_truth": [ "2020-01-20", "2020-03-20", "2020-10-17", "2020-03-11" ], "eval_metric": "set_f1", "channel": "495", "ts_data_path": "ts_data/L3_T1_Composite_Trend_S2_00003.csv" }, { "id": "L2_T2_Periodicity_Detection_S2_00002", "level": 2, "level_name": "Pattern Recognition", "category": "Periodicity Detection", "subtask": "Periodicity Detection", "question": "What is the dominant cycle period (in data points) of channel 237 within [2022-05-02 10:30:00 to 2022-05-09 03:45:00]? (Output format: integer)", "answer": "46", "ground_truth": 46, "eval_metric": "rel_acc", "channel": "237", "ts_data_path": "ts_data/L2_T2_Periodicity_Detection_S2_00002.csv" }, { "id": "L3_T1_Composite_Trend_S4_00009", "level": 3, "level_name": "Semantic Reasoning", "category": "Composite Trend", "subtask": "Composite Trend", "question": "Among days in channel 169 during 2023 that exhibit the trend pattern 'rapid fall, then fall, then fluctuating stable, then rapid rise', identify the top-8 days where the rapid fall segment is the fastest. (Output format: a ranked list of dates, e.g., ['YYYY-MM-DD', ...])", "answer": "['2023-04-16', '2023-05-22', '2023-12-02', '2023-06-28', '2023-04-07', '2023-03-18', '2023-03-26', '2023-08-01']", "ground_truth": [ "2023-04-16", "2023-05-22", "2023-12-02", "2023-06-28", "2023-04-07", "2023-03-18", "2023-03-26", "2023-08-01" ], "eval_metric": "set_f1", "channel": "169", "ts_data_path": "ts_data/L3_T1_Composite_Trend_S4_00009.csv" }, { "id": "L1_T2_Temporal_Localization_S4_00001", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Temporal Localization", "question": "At what exact timestamp did channel 146 reach its minimum value in 2021-06-18 to 2021-11-04? (Output format: a specific timestamp, e.g., YYYY-MM-DD HH:MM:SS)", "answer": "2021-08-18 05:00:00", "ground_truth": "2021-08-18 05:00:00", "eval_metric": "hit", "channel": "146", "ts_data_path": "ts_data/L1_T2_Temporal_Localization_S4_00001.csv" }, { "id": "L1_T3_Interval_Discovery_S2_00009", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Interval Discovery", "question": "Find the longest period where channel LULL remained above 1.218 in 2017-06-26 to 2018-02-06. (Output format: a time interval, e.g., [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2017-08-20 12:00:00, 2017-08-23 09:00:00]", "ground_truth": [ "2017-08-20 12:00:00", "2017-08-23 09:00:00" ], "eval_metric": "iou", "channel": "LULL", "ts_data_path": "ts_data/L1_T3_Interval_Discovery_S2_00009.csv" }, { "id": "L3_T2_Contextual_Anomaly_S3_00005", "level": 3, "level_name": "Semantic Reasoning", "category": "Contextual Anomaly", "subtask": "Contextual Anomaly", "question": "Identify the period in channel 626 during 2020 that experienced the most significant historically high water level. (Output format: [YYYY-MM-DD, YYYY-MM-DD])", "answer": "[2020-10-18 03:00:00, 2020-11-12 02:45:00]", "ground_truth": [ "2020-10-18 03:00:00", "2020-11-12 02:45:00" ], "eval_metric": "iou", "channel": "626", "ts_data_path": "ts_data/L3_T2_Contextual_Anomaly_S3_00005.csv" }, { "id": "L1_T4_Sliding_Window_S3_00002", "level": 1, "level_name": "Basic Operations", "category": "Sliding Window", "subtask": "Sliding Window", "question": "Which 75-day period in 2021-02-11 to 2021-11-15 had the largest range for channel 237? (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2021-02-11 19:15:00, 2021-04-25 21:00:00]", "ground_truth": [ "2021-02-11 19:15:00", "2021-04-25 21:00:00" ], "eval_metric": "iou", "channel": "237", "ts_data_path": "ts_data/L1_T4_Sliding_Window_S3_00002.csv" }, { "id": "L3_T1_Composite_Trend_S4_00004", "level": 3, "level_name": "Semantic Reasoning", "category": "Composite Trend", "subtask": "Composite Trend", "question": "Among days in channel 501 during 2023 that exhibit the trend pattern 'slow rise, then rapid fall, then slow fall, then slow rise', identify the top-8 days where the slow fall segment is the slowest. (Output format: a ranked list of dates, e.g., ['YYYY-MM-DD', ...])", "answer": "['2023-07-13', '2023-07-06', '2023-09-13', '2023-04-26', '2023-01-07', '2023-09-02', '2023-10-15', '2023-10-09']", "ground_truth": [ "2023-07-13", "2023-07-06", "2023-09-13", "2023-04-26", "2023-01-07", "2023-09-02", "2023-10-15", "2023-10-09" ], "eval_metric": "set_f1", "channel": "501", "ts_data_path": "ts_data/L3_T1_Composite_Trend_S4_00004.csv" }, { "id": "L2_T3_Subsequence_Matching_S2_00004", "level": 2, "level_name": "Pattern Recognition", "category": "Subsequence Matching", "subtask": "Subsequence Matching", "question": "Analyze the reference pattern in [2016-08-05 17:15:00 to 2016-08-05 23:00:00]. Find the time interval where channel MUFL exhibits the most similar pattern within the search context [2016-08-07 07:00:00 to 2016-08-12 08:00:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2016-08-07 09:30:00', '2016-08-07 15:15:00']", "ground_truth": [ "2016-08-07 09:30:00", "2016-08-07 15:15:00" ], "eval_metric": "iou", "channel": "MUFL", "ts_data_path": "ts_data/L2_T3_Subsequence_Matching_S2_00004.csv" }, { "id": "L1_T4_Sliding_Window_S3_00007", "level": 1, "level_name": "Basic Operations", "category": "Sliding Window", "subtask": "Sliding Window", "question": "Which 101-day period in 2021-11-05 to 2022-10-02 had the highest variance for channel 430? (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2022-02-06 18:15:00, 2022-05-18 18:00:00]", "ground_truth": [ "2022-02-06 18:15:00", "2022-05-18 18:00:00" ], "eval_metric": "iou", "channel": "430", "ts_data_path": "ts_data/L1_T4_Sliding_Window_S3_00007.csv" }, { "id": "L1_T4_Sliding_Window_S4_00006", "level": 1, "level_name": "Basic Operations", "category": "Sliding Window", "subtask": "Sliding Window", "question": "Which 22-day period in 2017-10-05 to 2018-01-21 had the largest range for channel HUFL? (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2017-11-20 07:45:00, 2017-12-12 07:30:00]", "ground_truth": [ "2017-11-20 07:45:00", "2017-12-12 07:30:00" ], "eval_metric": "iou", "channel": "HUFL", "ts_data_path": "ts_data/L1_T4_Sliding_Window_S4_00006.csv" }, { "id": "L3_T1_Composite_Trend_S2_00009", "level": 3, "level_name": "Semantic Reasoning", "category": "Composite Trend", "subtask": "Composite Trend", "question": "Among days in channel 71 during 2022 that exhibit the trend pattern 'slow rise, then rapid fall, then fall', identify the top-4 days where the rapid fall segment is the fastest. (Output format: a ranked list of dates, e.g., ['YYYY-MM-DD', ...])", "answer": "['2022-02-02', '2022-07-24', '2022-01-06', '2022-03-05']", "ground_truth": [ "2022-02-02", "2022-07-24", "2022-01-06", "2022-03-05" ], "eval_metric": "set_f1", "channel": "71", "ts_data_path": "ts_data/L3_T1_Composite_Trend_S2_00009.csv" }, { "id": "L1_T3_Interval_Discovery_S0_00006", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Interval Discovery", "question": "Find the longest period where channel 895 remained above 0.909 in 2020-08. (Output format: a time interval, e.g., [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2020-08-26 08:15:00, 2020-08-31 17:00:00]", "ground_truth": [ "2020-08-26 08:15:00", "2020-08-31 17:00:00" ], "eval_metric": "iou", "channel": "895", "ts_data_path": "ts_data/L1_T3_Interval_Discovery_S0_00006.csv" }, { "id": "L4_T1_Insight_Synthesis_S4_00002", "level": 4, "level_name": "Insight Synthesis", "category": "Insight Synthesis", "subtask": "Insight Synthesis", "question": "Analyze the behavior of channel 170 for the period 2019-11.\nPlease use ONLY the following phrases for trend description: rapid rise, gradual rise, rapid fall, gradual fall, steady stable, fluctuating stable.\nProvide a structured report covering:\n1. Trend Segmentation: Describe each stage with precise start/end timestamps (HH:MM:SS) using the phrases above.\n2. Outlier Audit: Identify only significant outliers that deviate sharply from the local trend. Ignore minor background noise.\n(Output format: structured natural language report.)", "answer": "1. Trend Segmentation: from 2019-11-01 00:00:00 to 2019-11-10 09:00:00, the trend showed a fluctuating stable; from 2019-11-10 09:00:00 to 2019-11-14 03:00:00, the trend showed a rapid rise; from 2019-11-14 03:00:00 to 2019-11-25 09:00:00, the trend showed a gradual fall; from 2019-11-25 09:00:00 to 2019-11-30 23:45:00, the trend showed a steady stable.\n2. Outlier Audit: A significant spike was detected at 2019-11-22 02:45:00 (value: 3692.89).", "ground_truth": { "trend_segments": [ "from 2019-11-01 00:00:00 to 2019-11-10 09:00:00, the trend showed a fluctuating stable", "from 2019-11-10 09:00:00 to 2019-11-14 03:00:00, the trend showed a rapid rise", "from 2019-11-14 03:00:00 to 2019-11-25 09:00:00, the trend showed a gradual fall", "from 2019-11-25 09:00:00 to 2019-11-30 23:45:00, the trend showed a steady stable" ], "segments_meta": [ { "adj": "fluctuating", "kind": "stable", "start_idx": 0, "end_idx": 900 }, { "adj": "rapid", "kind": "rise", "start_idx": 900, "end_idx": 1260 }, { "adj": "gradual", "kind": "fall", "start_idx": 1260, "end_idx": 2340 }, { "adj": "steady", "kind": "stable", "start_idx": 2340, "end_idx": 2880 } ], "significant_anomaly": { "timestamp": "2019-11-22 02:45:00", "kind": "significant spike", "value": 3692.885508312672 } }, "eval_metric": "report", "channel": "170", "ts_data_path": "ts_data/L4_T1_Insight_Synthesis_S4_00002.csv" }, { "id": "L1_T1_Global_Aggregation_S1_00002", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Global Aggregation", "question": "What is the average value of channel 71 in 2023-07-13 to 2023-07-28? (Output format: a single numeric value, rounded to 3 decimal places, e.g., x.xxx)", "answer": "0.062", "ground_truth": 0.062, "eval_metric": "rel_acc", "channel": "71", "ts_data_path": "ts_data/L1_T1_Global_Aggregation_S1_00002.csv" }, { "id": "L3_T2_Contextual_Anomaly_S2_00008", "level": 3, "level_name": "Semantic Reasoning", "category": "Contextual Anomaly", "subtask": "Contextual Anomaly", "question": "Identify the period in channel 123 during 2021 that experienced the most significant severe drought. (Output format: [YYYY-MM-DD, YYYY-MM-DD])", "answer": "[2021-04-05 19:45:00, 2021-05-24 19:30:00]", "ground_truth": [ "2021-04-05 19:45:00", "2021-05-24 19:30:00" ], "eval_metric": "iou", "channel": "123", "ts_data_path": "ts_data/L3_T2_Contextual_Anomaly_S2_00008.csv" }, { "id": "L1_T1_Global_Aggregation_S2_00008", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Global Aggregation", "question": "What is the average value of channel 729 in 2020-01? (Output format: a single numeric value, rounded to 3 decimal places, e.g., x.xxx)", "answer": "46.193", "ground_truth": 46.193, "eval_metric": "rel_acc", "channel": "729", "ts_data_path": "ts_data/L1_T1_Global_Aggregation_S2_00008.csv" }, { "id": "L1_T4_Sliding_Window_S0_00009", "level": 1, "level_name": "Basic Operations", "category": "Sliding Window", "subtask": "Sliding Window", "question": "Which 2-day period in 2022-09-26 to 2022-11-26 had the highest variance for channel 627? (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2022-10-18 07:45:00, 2022-10-20 07:30:00]", "ground_truth": [ "2022-10-18 07:45:00", "2022-10-20 07:30:00" ], "eval_metric": "iou", "channel": "627", "ts_data_path": "ts_data/L1_T4_Sliding_Window_S0_00009.csv" }, { "id": "L2_T3_Subsequence_Matching_S3_00006", "level": 2, "level_name": "Pattern Recognition", "category": "Subsequence Matching", "subtask": "Subsequence Matching", "question": "Analyze the reference pattern in [2020-04-22 03:00:00 to 2020-04-22 12:45:00]. Find the time interval where channel 895 exhibits the most similar pattern within the search context [2020-04-23 22:00:00 to 2020-04-29 00:00:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2020-04-28 06:00:00', '2020-04-28 15:45:00']", "ground_truth": [ "2020-04-28 06:00:00", "2020-04-28 15:45:00" ], "eval_metric": "iou", "channel": "895", "ts_data_path": "ts_data/L2_T3_Subsequence_Matching_S3_00006.csv" }, { "id": "L2_T1_Shape_Identification_S1_00002", "level": 2, "level_name": "Pattern Recognition", "category": "Shape Identification", "subtask": "Shape Identification", "question": "Identify the time range of the deepest deep valley in channel 430 within [2019-08-09 15:30:00 to 2019-08-22 01:00:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2019-08-12 01:00:00', '2019-08-13 21:15:00']", "ground_truth": [ "2019-08-12 01:00:00", "2019-08-13 21:15:00" ], "eval_metric": "iou", "channel": "430", "ts_data_path": "ts_data/L2_T1_Shape_Identification_S1_00002.csv" }, { "id": "L3_T3_Causal_Anomaly_S4_00001", "level": 3, "level_name": "Semantic Reasoning", "category": "Causal Anomaly", "subtask": "Causal Anomaly", "question": "Given that channel 154_up_2022 is the upstream source of channel 154_down_2022, identify the time period in 2022 where 154_down_2022 shows a significant causal anomaly, such as an inverse trend against the source. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2022-10-13 05:00:00, 2022-11-01 18:30:00]", "ground_truth": [ "2022-10-13 05:00:00", "2022-11-01 18:30:00" ], "eval_metric": "iou", "channel": "154_down_2022", "ts_data_path": "ts_data/L3_T3_Causal_Anomaly_S4_00001.csv" }, { "id": "L3_T2_Contextual_Anomaly_S1_00002", "level": 3, "level_name": "Semantic Reasoning", "category": "Contextual Anomaly", "subtask": "Contextual Anomaly", "question": "Identify the period in channel 154 during 2022 that experienced the most significant historically high water level. (Output format: [YYYY-MM-DD, YYYY-MM-DD])", "answer": "[2022-07-12 05:30:00, 2022-08-18 05:15:00]", "ground_truth": [ "2022-07-12 05:30:00", "2022-08-18 05:15:00" ], "eval_metric": "iou", "channel": "154", "ts_data_path": "ts_data/L3_T2_Contextual_Anomaly_S1_00002.csv" }, { "id": "L2_T2_Periodicity_Detection_S0_00001", "level": 2, "level_name": "Pattern Recognition", "category": "Periodicity Detection", "subtask": "Periodicity Detection", "question": "What is the dominant cycle period (in data points) of channel 155 within [2021-08-01 18:00:00 to 2021-08-06 23:00:00]? (Output format: integer)", "answer": "16", "ground_truth": 16, "eval_metric": "rel_acc", "channel": "155", "ts_data_path": "ts_data/L2_T2_Periodicity_Detection_S0_00001.csv" }, { "id": "L3_T3_Causal_Anomaly_S1_00005", "level": 3, "level_name": "Semantic Reasoning", "category": "Causal Anomaly", "subtask": "Causal Anomaly", "question": "Given that channel 71_up_2020 is the upstream source of channel 71_down_2020, identify the time period in 2020 where 71_down_2020 shows a significant causal anomaly, such as an inverse trend against the source. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2020-05-14 10:30:00, 2020-05-17 11:15:00]", "ground_truth": [ "2020-05-14 10:30:00", "2020-05-17 11:15:00" ], "eval_metric": "iou", "channel": "71_down_2020", "ts_data_path": "ts_data/L3_T3_Causal_Anomaly_S1_00005.csv" }, { "id": "L3_T3_Causal_Anomaly_S1_00007", "level": 3, "level_name": "Semantic Reasoning", "category": "Causal Anomaly", "subtask": "Causal Anomaly", "question": "Given that channel 151_up_2019 is the upstream source of channel 151_down_2019, identify the time period in 2019 where 151_down_2019 shows a significant causal anomaly, such as an flat line during high activity. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2019-06-30 08:00:00, 2019-07-01 20:30:00]", "ground_truth": [ "2019-06-30 08:00:00", "2019-07-01 20:30:00" ], "eval_metric": "iou", "channel": "151_down_2019", "ts_data_path": "ts_data/L3_T3_Causal_Anomaly_S1_00007.csv" }, { "id": "L3_T3_Causal_Anomaly_S3_00006", "level": 3, "level_name": "Semantic Reasoning", "category": "Causal Anomaly", "subtask": "Causal Anomaly", "question": "Given that channel 683_up_2023 is the upstream source of channel 683_down_2023, identify the time period in 2023 where 683_down_2023 shows a significant causal anomaly, such as an inverse trend against the source. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2023-01-07 13:00:00, 2023-01-11 00:00:00]", "ground_truth": [ "2023-01-07 13:00:00", "2023-01-11 00:00:00" ], "eval_metric": "iou", "channel": "683_down_2023", "ts_data_path": "ts_data/L3_T3_Causal_Anomaly_S3_00006.csv" }, { "id": "L4_T1_Insight_Synthesis_S0_00006", "level": 4, "level_name": "Insight Synthesis", "category": "Insight Synthesis", "subtask": "Insight Synthesis", "question": "Analyze the behavior of channel 894 for the period 2022-02.\nPlease use ONLY the following phrases for trend description: rapid rise, gradual rise, rapid fall, gradual fall, steady stable, fluctuating stable.\nProvide a structured report covering:\n1. Trend Segmentation: Describe each stage with precise start/end timestamps (HH:MM:SS) using the phrases above.\n2. Outlier Audit: Identify only significant outliers that deviate sharply from the local trend. Ignore minor background noise.\n(Output format: structured natural language report.)", "answer": "1. Trend Segmentation: from 2022-02-01 00:00:00 to 2022-02-04 12:00:00, the trend showed a gradual fall; from 2022-02-04 12:00:00 to 2022-02-11 12:00:00, the trend showed a rapid fall; from 2022-02-11 12:00:00 to 2022-02-15 00:00:00, the trend showed a steady stable; from 2022-02-15 00:00:00 to 2022-02-23 18:00:00, the trend showed a rapid rise; from 2022-02-23 18:00:00 to 2022-02-28 23:45:00, the trend showed a gradual fall.\n2. Outlier Audit: A significant spike was detected at 2022-02-06 18:30:00 (value: 44.24).", "ground_truth": { "trend_segments": [ "from 2022-02-01 00:00:00 to 2022-02-04 12:00:00, the trend showed a gradual fall", "from 2022-02-04 12:00:00 to 2022-02-11 12:00:00, the trend showed a rapid fall", "from 2022-02-11 12:00:00 to 2022-02-15 00:00:00, the trend showed a steady stable", "from 2022-02-15 00:00:00 to 2022-02-23 18:00:00, the trend showed a rapid rise", "from 2022-02-23 18:00:00 to 2022-02-28 23:45:00, the trend showed a gradual fall" ], "segments_meta": [ { "adj": "gradual", "kind": "fall", "start_idx": 0, "end_idx": 336 }, { "adj": "rapid", "kind": "fall", "start_idx": 336, "end_idx": 1008 }, { "adj": "steady", "kind": "stable", "start_idx": 1008, "end_idx": 1344 }, { "adj": "rapid", "kind": "rise", "start_idx": 1344, "end_idx": 2184 }, { "adj": "gradual", "kind": "fall", "start_idx": 2184, "end_idx": 2688 } ], "significant_anomaly": { "timestamp": "2022-02-06 18:30:00", "kind": "significant spike", "value": 44.24379596122765 } }, "eval_metric": "report", "channel": "894", "ts_data_path": "ts_data/L4_T1_Insight_Synthesis_S0_00006.csv" }, { "id": "L3_T1_Composite_Trend_S1_00007", "level": 3, "level_name": "Semantic Reasoning", "category": "Composite Trend", "subtask": "Composite Trend", "question": "Among days in channel 166 during 2021 that exhibit the trend pattern 'fall, then fluctuating stable', identify the top-2 days where the fluctuating stable segment is the most fluctuating. (Output format: a ranked list of dates, e.g., ['YYYY-MM-DD', ...])", "answer": "['2021-09-12', '2021-07-18']", "ground_truth": [ "2021-09-12", "2021-07-18" ], "eval_metric": "set_f1", "channel": "166", "ts_data_path": "ts_data/L3_T1_Composite_Trend_S1_00007.csv" }, { "id": "L4_T1_Insight_Synthesis_S1_00004", "level": 4, "level_name": "Insight Synthesis", "category": "Insight Synthesis", "subtask": "Insight Synthesis", "question": "Analyze the behavior of channel 591 for the period 2022-04.\nPlease use ONLY the following phrases for trend description: rapid rise, gradual rise, rapid fall, gradual fall, steady stable, fluctuating stable.\nProvide a structured report covering:\n1. Trend Segmentation: Describe each stage with precise start/end timestamps (HH:MM:SS) using the phrases above.\n2. Outlier Audit: Identify only significant outliers that deviate sharply from the local trend. Ignore minor background noise.\n(Output format: structured natural language report.)", "answer": "1. Trend Segmentation: from 2022-04-01 00:00:00 to 2022-04-08 12:00:00, the trend showed a rapid fall; from 2022-04-08 12:00:00 to 2022-04-16 00:00:00, the trend showed a gradual fall; from 2022-04-16 00:00:00 to 2022-04-18 12:00:00, the trend showed a steady stable; from 2022-04-18 12:00:00 to 2022-04-24 18:00:00, the trend showed a gradual rise; from 2022-04-24 18:00:00 to 2022-04-30 23:45:00, the trend showed a fluctuating stable.\n2. Outlier Audit: A significant drop was detected at 2022-04-10 08:45:00 (value: -899.30).", "ground_truth": { "trend_segments": [ "from 2022-04-01 00:00:00 to 2022-04-08 12:00:00, the trend showed a rapid fall", "from 2022-04-08 12:00:00 to 2022-04-16 00:00:00, the trend showed a gradual fall", "from 2022-04-16 00:00:00 to 2022-04-18 12:00:00, the trend showed a steady stable", "from 2022-04-18 12:00:00 to 2022-04-24 18:00:00, the trend showed a gradual rise", "from 2022-04-24 18:00:00 to 2022-04-30 23:45:00, the trend showed a fluctuating stable" ], "segments_meta": [ { "adj": "rapid", "kind": "fall", "start_idx": 0, "end_idx": 720 }, { "adj": "gradual", "kind": "fall", "start_idx": 720, "end_idx": 1440 }, { "adj": "steady", "kind": "stable", "start_idx": 1440, "end_idx": 1680 }, { "adj": "gradual", "kind": "rise", "start_idx": 1680, "end_idx": 2280 }, { "adj": "fluctuating", "kind": "stable", "start_idx": 2280, "end_idx": 2880 } ], "significant_anomaly": { "timestamp": "2022-04-10 08:45:00", "kind": "significant drop", "value": -899.3039064179695 } }, "eval_metric": "report", "channel": "591", "ts_data_path": "ts_data/L4_T1_Insight_Synthesis_S1_00004.csv" }, { "id": "L3_T1_Composite_Trend_S0_00009", "level": 3, "level_name": "Semantic Reasoning", "category": "Composite Trend", "subtask": "Composite Trend", "question": "Among days in channel LUFL during 2017 that exhibit the trend pattern 'steady stable, then rise, then slow fall, then fluctuating stable', identify the top-3 days where the fluctuating stable segment is the most fluctuating. (Output format: a ranked list of dates, e.g., ['YYYY-MM-DD', ...])", "answer": "['2017-01-07', '2017-06-01', '2017-06-11']", "ground_truth": [ "2017-01-07", "2017-06-01", "2017-06-11" ], "eval_metric": "set_f1", "channel": "LUFL", "ts_data_path": "ts_data/L3_T1_Composite_Trend_S0_00009.csv" }, { "id": "L3_T3_Causal_Anomaly_S3_00005", "level": 3, "level_name": "Semantic Reasoning", "category": "Causal Anomaly", "subtask": "Causal Anomaly", "question": "Given that channel 680_up_2019 is the upstream source of channel 680_down_2019, identify the time period in 2019 where 680_down_2019 shows a significant causal anomaly, such as an flat line during high activity. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "[2019-06-09 08:00:00, 2019-06-13 19:00:00]", "ground_truth": [ "2019-06-09 08:00:00", "2019-06-13 19:00:00" ], "eval_metric": "iou", "channel": "680_down_2019", "ts_data_path": "ts_data/L3_T3_Causal_Anomaly_S3_00005.csv" }, { "id": "L2_T2_Periodicity_Detection_S4_00009", "level": 2, "level_name": "Pattern Recognition", "category": "Periodicity Detection", "subtask": "Periodicity Detection", "question": "What is the dominant cycle period (in data points) of channel 578 within [2019-11-27 18:00:00 to 2019-12-10 21:15:00]? (Output format: integer)", "answer": "85", "ground_truth": 85, "eval_metric": "rel_acc", "channel": "578", "ts_data_path": "ts_data/L2_T2_Periodicity_Detection_S4_00009.csv" }, { "id": "L1_T2_Temporal_Localization_S0_00006", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Temporal Localization", "question": "At what exact timestamp did channel 626 reach its maximum value in 2020-11? (Output format: a specific timestamp, e.g., YYYY-MM-DD HH:MM:SS)", "answer": "2020-11-01 03:15:00", "ground_truth": "2020-11-01 03:15:00", "eval_metric": "hit", "channel": "626", "ts_data_path": "ts_data/L1_T2_Temporal_Localization_S0_00006.csv" }, { "id": "L1_T2_Temporal_Localization_S4_00006", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Temporal Localization", "question": "At what exact timestamp did channel 67 reach its minimum value in 2022-11-16 to 2023-04-02? (Output format: a specific timestamp, e.g., YYYY-MM-DD HH:MM:SS)", "answer": "2023-02-14 17:15:00", "ground_truth": "2023-02-14 17:15:00", "eval_metric": "hit", "channel": "67", "ts_data_path": "ts_data/L1_T2_Temporal_Localization_S4_00006.csv" }, { "id": "L2_T2_Periodicity_Detection_S0_00004", "level": 2, "level_name": "Pattern Recognition", "category": "Periodicity Detection", "subtask": "Periodicity Detection", "question": "What is the dominant cycle period (in data points) of channel 580 within [2020-12-07 16:00:00 to 2020-12-11 20:15:00]? (Output format: integer)", "answer": "134", "ground_truth": 134, "eval_metric": "rel_acc", "channel": "580", "ts_data_path": "ts_data/L2_T2_Periodicity_Detection_S0_00004.csv" }, { "id": "L4_T1_Insight_Synthesis_S0_00007", "level": 4, "level_name": "Insight Synthesis", "category": "Insight Synthesis", "subtask": "Insight Synthesis", "question": "Analyze the behavior of channel HULL for the period 2017-02.\nPlease use ONLY the following phrases for trend description: rapid rise, gradual rise, rapid fall, gradual fall, steady stable, fluctuating stable.\nProvide a structured report covering:\n1. Trend Segmentation: Describe each stage with precise start/end timestamps (HH:MM:SS) using the phrases above.\n2. Outlier Audit: Identify only significant outliers that deviate sharply from the local trend. Ignore minor background noise.\n(Output format: structured natural language report.)", "answer": "1. Trend Segmentation: from 2017-02-01 00:00:00 to 2017-02-12 04:45:00, the trend showed a gradual rise; from 2017-02-12 04:45:00 to 2017-02-19 16:00:00, the trend showed a fluctuating stable; from 2017-02-19 16:00:00 to 2017-02-28 23:45:00, the trend showed a steady stable.\n2. Outlier Audit: A significant spike was detected at 2017-02-07 08:15:00 (value: 27.80).", "ground_truth": { "trend_segments": [ "from 2017-02-01 00:00:00 to 2017-02-12 04:45:00, the trend showed a gradual rise", "from 2017-02-12 04:45:00 to 2017-02-19 16:00:00, the trend showed a fluctuating stable", "from 2017-02-19 16:00:00 to 2017-02-28 23:45:00, the trend showed a steady stable" ], "segments_meta": [ { "adj": "gradual", "kind": "rise", "start_idx": 0, "end_idx": 1075 }, { "adj": "fluctuating", "kind": "stable", "start_idx": 1075, "end_idx": 1792 }, { "adj": "steady", "kind": "stable", "start_idx": 1792, "end_idx": 2688 } ], "significant_anomaly": { "timestamp": "2017-02-07 08:15:00", "kind": "significant spike", "value": 27.802805143325482 } }, "eval_metric": "report", "channel": "HULL", "ts_data_path": "ts_data/L4_T1_Insight_Synthesis_S0_00007.csv" }, { "id": "L1_T2_Temporal_Localization_S0_00001", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Temporal Localization", "question": "At what exact timestamp did channel 67 first rise above 0.218 in 2019-06-08 to 2019-07-02? (Output format: a specific timestamp, e.g., YYYY-MM-DD HH:MM:SS)", "answer": "2019-06-08 13:00:00", "ground_truth": "2019-06-08 13:00:00", "eval_metric": "hit", "channel": "67", "ts_data_path": "ts_data/L1_T2_Temporal_Localization_S0_00001.csv" }, { "id": "L1_T2_Temporal_Localization_S1_00006", "level": 1, "level_name": "Basic Operations", "category": "Atomic Retrieval", "subtask": "Temporal Localization", "question": "At what exact timestamp did channel 627 first rise above 0.378 in 2020-01? (Output format: a specific timestamp, e.g., YYYY-MM-DD HH:MM:SS)", "answer": "2020-01-05 07:45:00", "ground_truth": "2020-01-05 07:45:00", "eval_metric": "hit", "channel": "627", "ts_data_path": "ts_data/L1_T2_Temporal_Localization_S1_00006.csv" }, { "id": "L2_T1_Shape_Identification_S4_00001", "level": 2, "level_name": "Pattern Recognition", "category": "Shape Identification", "subtask": "Shape Identification", "question": "Identify the time range of the largest step ascent in channel 71 within [2022-08-03 10:30:00 to 2022-08-06 10:45:00]. (Output format: [YYYY-MM-DD HH:MM:SS, YYYY-MM-DD HH:MM:SS])", "answer": "['2022-08-03 18:15:00', '2022-08-04 04:45:00']", "ground_truth": [ "2022-08-03 18:15:00", "2022-08-04 04:45:00" ], "eval_metric": "iou", "channel": "71", "ts_data_path": "ts_data/L2_T1_Shape_Identification_S4_00001.csv" } ]