Bryangmkmk BugMaker-Boyan commited on
Commit
a7e3721
·
0 Parent(s):

Duplicate from HKUSTDial/DataSpace

Browse files

Co-authored-by: Boyan Li <BugMaker-Boyan@users.noreply.huggingface.co>

.gitattributes ADDED
@@ -0,0 +1,60 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ *.7z filter=lfs diff=lfs merge=lfs -text
2
+ *.arrow filter=lfs diff=lfs merge=lfs -text
3
+ *.avro filter=lfs diff=lfs merge=lfs -text
4
+ *.bin filter=lfs diff=lfs merge=lfs -text
5
+ *.bz2 filter=lfs diff=lfs merge=lfs -text
6
+ *.ckpt filter=lfs diff=lfs merge=lfs -text
7
+ *.ftz filter=lfs diff=lfs merge=lfs -text
8
+ *.gz filter=lfs diff=lfs merge=lfs -text
9
+ *.h5 filter=lfs diff=lfs merge=lfs -text
10
+ *.joblib filter=lfs diff=lfs merge=lfs -text
11
+ *.lfs.* filter=lfs diff=lfs merge=lfs -text
12
+ *.lz4 filter=lfs diff=lfs merge=lfs -text
13
+ *.mds filter=lfs diff=lfs merge=lfs -text
14
+ *.mlmodel filter=lfs diff=lfs merge=lfs -text
15
+ *.model filter=lfs diff=lfs merge=lfs -text
16
+ *.msgpack filter=lfs diff=lfs merge=lfs -text
17
+ *.npy filter=lfs diff=lfs merge=lfs -text
18
+ *.npz filter=lfs diff=lfs merge=lfs -text
19
+ *.onnx filter=lfs diff=lfs merge=lfs -text
20
+ *.ot filter=lfs diff=lfs merge=lfs -text
21
+ *.parquet filter=lfs diff=lfs merge=lfs -text
22
+ *.pb filter=lfs diff=lfs merge=lfs -text
23
+ *.pickle filter=lfs diff=lfs merge=lfs -text
24
+ *.pkl filter=lfs diff=lfs merge=lfs -text
25
+ *.pt filter=lfs diff=lfs merge=lfs -text
26
+ *.pth filter=lfs diff=lfs merge=lfs -text
27
+ *.rar filter=lfs diff=lfs merge=lfs -text
28
+ *.safetensors filter=lfs diff=lfs merge=lfs -text
29
+ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
30
+ *.tar.* filter=lfs diff=lfs merge=lfs -text
31
+ *.tar filter=lfs diff=lfs merge=lfs -text
32
+ *.tflite filter=lfs diff=lfs merge=lfs -text
33
+ *.tgz filter=lfs diff=lfs merge=lfs -text
34
+ *.wasm filter=lfs diff=lfs merge=lfs -text
35
+ *.xz filter=lfs diff=lfs merge=lfs -text
36
+ *.zip filter=lfs diff=lfs merge=lfs -text
37
+ *.zst filter=lfs diff=lfs merge=lfs -text
38
+ *tfevents* filter=lfs diff=lfs merge=lfs -text
39
+ # Audio files - uncompressed
40
+ *.pcm filter=lfs diff=lfs merge=lfs -text
41
+ *.sam filter=lfs diff=lfs merge=lfs -text
42
+ *.raw filter=lfs diff=lfs merge=lfs -text
43
+ # Audio files - compressed
44
+ *.aac filter=lfs diff=lfs merge=lfs -text
45
+ *.flac filter=lfs diff=lfs merge=lfs -text
46
+ *.mp3 filter=lfs diff=lfs merge=lfs -text
47
+ *.ogg filter=lfs diff=lfs merge=lfs -text
48
+ *.wav filter=lfs diff=lfs merge=lfs -text
49
+ # Image files - uncompressed
50
+ *.bmp filter=lfs diff=lfs merge=lfs -text
51
+ *.gif filter=lfs diff=lfs merge=lfs -text
52
+ *.png filter=lfs diff=lfs merge=lfs -text
53
+ *.tiff filter=lfs diff=lfs merge=lfs -text
54
+ # Image files - compressed
55
+ *.jpg filter=lfs diff=lfs merge=lfs -text
56
+ *.jpeg filter=lfs diff=lfs merge=lfs -text
57
+ *.webp filter=lfs diff=lfs merge=lfs -text
58
+ # Video files - compressed
59
+ *.mp4 filter=lfs diff=lfs merge=lfs -text
60
+ *.webm filter=lfs diff=lfs merge=lfs -text
CITATION.cff ADDED
@@ -0,0 +1,71 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ cff-version: 1.2.0
2
+ type: dataset
3
+ message: "If you use DataSpace, please cite the accompanying paper."
4
+ title: "DataSpace: Benchmarking Data Agents for Verifiable Analytics over Heterogeneous Workspaces"
5
+ authors:
6
+ - family-names: Li
7
+ given-names: Boyan
8
+ - family-names: Liang
9
+ given-names: Zhuowen
10
+ - family-names: Xie
11
+ given-names: Yupeng
12
+ - family-names: Lin
13
+ given-names: Xiaotian
14
+ - family-names: Luo
15
+ given-names: Tianqi
16
+ - family-names: Liu
17
+ given-names: Xinyu
18
+ - family-names: Zhu
19
+ given-names: Yizhang
20
+ - family-names: Peng
21
+ given-names: Zhangyang
22
+ - family-names: Li
23
+ given-names: Yuan
24
+ - family-names: Zhang
25
+ given-names: Zhengxuan
26
+ - family-names: Zhang
27
+ given-names: Jiayi
28
+ - family-names: Tang
29
+ given-names: Nan
30
+ - family-names: Li
31
+ given-names: Guoliang
32
+ - family-names: Luo
33
+ given-names: Yuyu
34
+ date-released: 2026-07-27
35
+ license: MIT
36
+ url: "https://huggingface.co/datasets/HKUSTDial/DataSpace"
37
+ repository-code: "https://github.com/HKUSTDial/DataSpace"
38
+ preferred-citation:
39
+ type: article
40
+ title: "DataSpace: Benchmarking Data Agents for Verifiable Analytics over Heterogeneous Workspaces"
41
+ authors:
42
+ - family-names: Li
43
+ given-names: Boyan
44
+ - family-names: Liang
45
+ given-names: Zhuowen
46
+ - family-names: Xie
47
+ given-names: Yupeng
48
+ - family-names: Lin
49
+ given-names: Xiaotian
50
+ - family-names: Luo
51
+ given-names: Tianqi
52
+ - family-names: Liu
53
+ given-names: Xinyu
54
+ - family-names: Zhu
55
+ given-names: Yizhang
56
+ - family-names: Peng
57
+ given-names: Zhangyang
58
+ - family-names: Li
59
+ given-names: Yuan
60
+ - family-names: Zhang
61
+ given-names: Zhengxuan
62
+ - family-names: Zhang
63
+ given-names: Jiayi
64
+ - family-names: Tang
65
+ given-names: Nan
66
+ - family-names: Li
67
+ given-names: Guoliang
68
+ - family-names: Luo
69
+ given-names: Yuyu
70
+ year: 2026
71
+ url: "https://arxiv.org/abs/2608.03451"
LICENSE ADDED
@@ -0,0 +1,21 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ MIT License
2
+
3
+ Copyright (c) 2026 DataSpace Authors
4
+
5
+ Permission is hereby granted, free of charge, to any person obtaining a copy
6
+ of this dataset and associated documentation files (the "Dataset"), to deal
7
+ in the Dataset without restriction, including without limitation the rights
8
+ to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
9
+ copies of the Dataset, and to permit persons to whom the Dataset is furnished
10
+ to do so, subject to the following conditions:
11
+
12
+ The above copyright notice and this permission notice shall be included in all
13
+ copies or substantial portions of the Dataset.
14
+
15
+ THE DATASET IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
16
+ IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
17
+ FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
18
+ AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
19
+ LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
20
+ OUT OF OR IN CONNECTION WITH THE DATASET OR THE USE OR OTHER DEALINGS IN THE
21
+ DATASET.
README.md ADDED
@@ -0,0 +1,180 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ pretty_name: DataSpace
3
+ license: mit
4
+ language:
5
+ - en
6
+ - zh
7
+ task_categories:
8
+ - table-question-answering
9
+ size_categories:
10
+ - n<1K
11
+ tags:
12
+ - data-agent
13
+ - benchmark
14
+ - heterogeneous-data
15
+ - multimodal
16
+ - tabular
17
+ - document
18
+ - video
19
+ configs:
20
+ - config_name: default
21
+ data_files:
22
+ - split: test
23
+ path: data/tasks.jsonl
24
+ ---
25
+
26
+ # DataSpace
27
+
28
+ **[Paper](https://arxiv.org/abs/2608.03451) ·
29
+ [Code](https://github.com/HKUSTDial/DataSpace) ·
30
+ [Leaderboard](https://dataspace-bench.github.io/) ·
31
+ [KDD Cup 2026](https://dataagent.top)**
32
+
33
+ DataSpace is a benchmark for data agents that perform verifiable analytics over
34
+ heterogeneous, task-local workspaces. Each task provides a natural-language
35
+ question and a workspace containing a combination of CSV, JSON, SQLite,
36
+ Markdown, PDF, and video artifacts. The required output is a complete tabular
37
+ result.
38
+
39
+ DataSpace is also the official benchmark of the
40
+ [KDD Cup 2026 Data Agent Track](https://dataagent.top).
41
+
42
+ ## Release policy
43
+
44
+ This release makes all 410 task inputs public. It includes `gold.csv` and the
45
+ frozen evaluation configuration for a representative set of 60 tasks, allowing
46
+ end-to-end local development and validation. References for the other 350 tasks
47
+ are withheld for official full-benchmark evaluation.
48
+
49
+ The public-reference set was selected by deterministic constrained coverage
50
+ optimization over source domains, question--workspace language combinations,
51
+ evidence-carrier combinations, required modalities, analytical operations,
52
+ answer shapes, numeric comparison modes, and ordering behavior. It covers all
53
+ six modalities, all annotated analytical-operation families, and all
54
+ evidence-carrier groups without repeated source samples.
55
+
56
+ ## Benchmark summary
57
+
58
+ | Property | Value |
59
+ | --- | ---: |
60
+ | Task inputs | 410 |
61
+ | Public reference packages | 60 |
62
+ | Context files | 7,439 |
63
+ | Uncompressed context size | 15.0 GB |
64
+ | CSV files | 2,809 |
65
+ | JSON files | 1,927 |
66
+ | SQLite databases | 402 |
67
+ | Markdown documents | 1,024 |
68
+ | PDF documents | 1,088 |
69
+ | Videos | 189 |
70
+
71
+ Chinese and English may occur across both the question and workspace artifacts.
72
+ Tasks may require evidence discovery, structured extraction, joins across
73
+ representations, filtering, aggregation, ranking, temporal reasoning, and
74
+ document or video understanding.
75
+
76
+ ## Repository layout
77
+
78
+ ```text
79
+ .
80
+ ├── README.md
81
+ ├── CITATION.cff
82
+ ├── LICENSE
83
+ ├── data/
84
+ │ ├── tasks.jsonl
85
+ │ └── public_reference_tasks.json
86
+ └── release/
87
+ ├── DataSpace-Benchmark.zip
88
+ └── DataSpace-Benchmark.zip.sha256
89
+ ```
90
+
91
+ `data/tasks.jsonl` is a lightweight index for the Hugging Face Dataset Viewer.
92
+ It contains questions, workspace paths, available modality composition, file
93
+ inventories, and a public-reference availability flag. It does not expose
94
+ answer or solution-path metadata for withheld tasks.
95
+
96
+ ## Download
97
+
98
+ Install the Hugging Face Hub client and download the complete release:
99
+
100
+ ```bash
101
+ pip install -U huggingface_hub
102
+ hf download HKUSTDial/DataSpace \
103
+ release/DataSpace-Benchmark.zip \
104
+ release/DataSpace-Benchmark.zip.sha256 \
105
+ --repo-type dataset \
106
+ --local-dir .
107
+ ```
108
+
109
+ Verify and extract it:
110
+
111
+ ```bash
112
+ cd release
113
+ sha256sum --check DataSpace-Benchmark.zip.sha256
114
+ unzip DataSpace-Benchmark.zip
115
+ ```
116
+
117
+ For task `task_N`, the agent receives:
118
+
119
+ ```text
120
+ DataSpace-Benchmark/input/task_N/task.json
121
+ DataSpace-Benchmark/input/task_N/context/
122
+ ```
123
+
124
+ and should write:
125
+
126
+ ```text
127
+ predictions/task_N/prediction.csv
128
+ ```
129
+
130
+ ## Loading the task index
131
+
132
+ ```python
133
+ from datasets import load_dataset
134
+
135
+ tasks = load_dataset("HKUSTDial/DataSpace", split="test")
136
+ task = tasks[0]
137
+ print(task["question"])
138
+ print(task["workspace_path"])
139
+ ```
140
+
141
+ Resolve `workspace_path` relative to the extracted
142
+ `DataSpace-Benchmark/` directory.
143
+
144
+ ## Evaluation
145
+
146
+ The official evaluator and baseline implementations are maintained in the
147
+ [DataSpace GitHub repository](https://github.com/HKUSTDial/DataSpace).
148
+ Local evaluation is available for the 60 public-reference tasks. Official
149
+ full-benchmark submissions are scored once over all 410 tasks, and missing
150
+ predictions count as incorrect. Submission information and verified results
151
+ are maintained on the [DataSpace Leaderboard](https://dataspace-bench.github.io/).
152
+
153
+ ## Construction and quality control
154
+
155
+ DataSpace is constructed from EHRSQL and BULL Text-to-SQL source instances.
156
+ The construction pipeline applies cross-language transformation,
157
+ constraint-aware data sampling, modality routing and artifact rendering, and
158
+ human review and data repair. Eleven domain experts cross-reviewed task
159
+ solvability, question clarity, workspace evidence, reference answers, and
160
+ evaluation configurations.
161
+
162
+ ## License
163
+
164
+ DataSpace is released under the MIT License. See [LICENSE](LICENSE).
165
+
166
+ ## Citation
167
+
168
+ ```bibtex
169
+ @article{li2026dataspace,
170
+ title = {DataSpace: Benchmarking Data Agents for Verifiable Analytics
171
+ over Heterogeneous Workspaces},
172
+ author = {Boyan Li and Zhuowen Liang and Yupeng Xie and Xiaotian Lin and
173
+ Tianqi Luo and Xinyu Liu and Yizhang Zhu and Zhangyang Peng and
174
+ Yuan Li and Zhengxuan Zhang and Jiayi Zhang and Nan Tang and
175
+ Guoliang Li and Yuyu Luo},
176
+ journal = {arXiv preprint arXiv:2608.03451},
177
+ year = {2026},
178
+ url = {https://arxiv.org/abs/2608.03451}
179
+ }
180
+ ```
data/public_reference_tasks.json ADDED
@@ -0,0 +1,66 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "selection_method": "deterministic constrained coverage optimization over benchmark composition, evidence paths, analytical operations, answer shapes, and evaluator behaviors",
3
+ "task_count": 60,
4
+ "task_ids": [
5
+ "task_10",
6
+ "task_18",
7
+ "task_23",
8
+ "task_24",
9
+ "task_33",
10
+ "task_35",
11
+ "task_43",
12
+ "task_44",
13
+ "task_48",
14
+ "task_58",
15
+ "task_85",
16
+ "task_86",
17
+ "task_87",
18
+ "task_102",
19
+ "task_103",
20
+ "task_111",
21
+ "task_123",
22
+ "task_127",
23
+ "task_128",
24
+ "task_129",
25
+ "task_133",
26
+ "task_140",
27
+ "task_159",
28
+ "task_166",
29
+ "task_168",
30
+ "task_182",
31
+ "task_184",
32
+ "task_196",
33
+ "task_199",
34
+ "task_200",
35
+ "task_208",
36
+ "task_215",
37
+ "task_257",
38
+ "task_259",
39
+ "task_263",
40
+ "task_264",
41
+ "task_267",
42
+ "task_269",
43
+ "task_279",
44
+ "task_285",
45
+ "task_298",
46
+ "task_300",
47
+ "task_312",
48
+ "task_322",
49
+ "task_326",
50
+ "task_328",
51
+ "task_329",
52
+ "task_336",
53
+ "task_337",
54
+ "task_361",
55
+ "task_362",
56
+ "task_371",
57
+ "task_378",
58
+ "task_380",
59
+ "task_382",
60
+ "task_389",
61
+ "task_391",
62
+ "task_398",
63
+ "task_402",
64
+ "task_409"
65
+ ]
66
+ }
data/tasks.jsonl ADDED
The diff for this file is too large to render. See raw diff
 
release/DataSpace-Benchmark.zip ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:68f221fafcf0b50396eca55f8298af1ad73d586a6cac6995916afca535a560af
3
+ size 2878651222
release/DataSpace-Benchmark.zip.sha256 ADDED
@@ -0,0 +1 @@
 
 
1
+ 68f221fafcf0b50396eca55f8298af1ad73d586a6cac6995916afca535a560af DataSpace-Benchmark.zip