Improve default entity index browsing

Co-authored-by: Cursor <cursoragent@cursor.com>

README.md

@@ -13,6 +13,8 @@ configs:
   - config_name: entity_index
     default: true
     data_files:
+      - split: curated
+        path: data/entities_curated.jsonl
       - split: entities
         path: data/entities.jsonl
 ---
@@ -27,7 +29,7 @@ configs:
 
 FACTPROP is a factual graph for studying how knowledge updates affect connected facts in large language models. This release includes the original `factprop_graph_v1.pkl` graph checkpoint and the **entity popularity index** used by the project’s browser explorer: 100,015 graph-node records with entity labels, available Wikidata QIDs, and forward-edge object in-degree.
 
-The graph checkpoint includes directed relations and edge attributes for questions, surface statements, and evidence. Experiment outputs and model checkpoints are not included. The dataset viewer displays the entity index; download `factprop_graph_v1.pkl` to access the complete graph.
+The dataset viewer opens on the derived `curated` split so that recognizable, mapped entities appear first. Select the `entities` split for the unchanged complete entity index, including numeric labels and unresolved nodes. Download `factprop_graph_v1.pkl` to access the complete graph with directed relations and edge attributes for questions, surface statements, and evidence.
 
 ## Contents
 
@@ -36,13 +38,16 @@ The graph checkpoint includes directed relations and edge attributes for questio
 | `factprop_graph_v1.pkl` | Original graph checkpoint (125,274,720 bytes), unchanged |
 | `GRAPH.md` | Checkpoint structure and graph loading guide |
 | `load_graph.py` | Download, verify, and load the graph |
-| `data/entities.jsonl` | One object per graph node, suitable for the dataset viewer |
+| `data/entities_curated.jsonl` | Derived browsing view: mapped, readable labels sorted by structural popularity |
+| `data/entities.jsonl` | One unchanged record per graph node, including numeric and unresolved labels |
 | `browser-index.json` | Unchanged index used by the project website |
 | `metadata/graph.json` | Graph counts, thresholds, export date, and source-graph checksum |
 | `metadata/release.json` | Source version and mapping statistics |
 | `SHA256SUMS` | File integrity checksums |
 
-The `entities` split is the complete entity index. It is not a training or test split.
+The `curated` split contains records whose labels include at least one letter and whose QIDs match the standard `Q<number>` form. It is sorted by descending `forward_object_in_degree`. Records and scores are copied without modification, QID duplicates are not merged, and the split is intended only to improve browsing.
+
+The `entities` split remains the complete 100,015-record entity index in its original order. Neither split is a training or test split.
 
 ## Fields
 
@@ -78,6 +83,7 @@ The loader verifies the original file checksum before loading it. Only load pick
 ```python
 from datasets import load_dataset
 
+curated = load_dataset("factprop/FACTPROP", "entity_index", split="curated")
 entities = load_dataset("factprop/FACTPROP", "entity_index", split="entities")
 matches = entities.filter(lambda row: row["qid"] == "Q312")
 print(matches[:])
@@ -101,6 +107,8 @@ with open(path, encoding="utf-8") as stream:
 - Some expected mappings, including the Java and Python programming languages, are absent.
 - A missing mapping does not imply low popularity or a score of zero. A known graph node may independently have a true degree of zero.
 - Multiple nodes with one QID must be reviewed separately. Do not automatically sum their degrees or take the maximum.
+- Numeric labels can represent literal values, identifiers, or extraction artifacts. They are graph nodes, not standalone factual claims, and remain available in the complete `entities` split.
+- The `curated` split is a display filter, not a quality label. Use the complete split and graph-edge context for research analysis.
 - The index supports node-level lookup; use `factprop_graph_v1.pkl` for graph traversal. The graph alone does not include every experiment input or output needed to reproduce all paper results.
 
 ## Provenance and versions

SHA256SUMS

@@ -1,6 +1,7 @@
 8eebe4a8b875d1061c7b55947883b455fa1565fb6a6e39b8056c7160381d7cfc  GRAPH.md
-cc131a7336d16065e8a796d86a10f2db19425d4fae06dc6822d1a34c227c2e33  README.md
+cec445fafa0afc35337bc507a0badba4aedf59b3f581801087f0381ef6368b75  README.md
 fa7d9aec01b9c98c09cebc941265e4351693b362ffb3048d2c86585e84d7800d  browser-index.json
+010beb905ef89b491159a8d8b2b013ea55219d26c8545bed07f64010bfa9ffce  data/entities_curated.jsonl
 2c21f2aefb9cbcc749cd813412dd90fea440050eeb8efb3356de786ebeb501e6  data/entities.jsonl
 437a434260edbb019c85575d65b4775cb2461f145e17966ee3acc8e4625ce7c8  factprop_graph_v1.pkl
 c73d0a4b3f6df2166ba960a64cd5d3522728f46c2597f4f6604cc65533b46888  load_graph.py