HeRE data release: gold, silver (agree3 + all signals), HeRE-KG, dataset card

README.md

@@ -0,0 +1,81 @@
+---
+license: cc-by-sa-4.0
+language:
+  - he
+task_categories:
+  - text-classification
+task_ids:
+  - natural-language-inference
+pretty_name: HeRE
+size_categories:
+  - 1M<n<10M
+configs:
+  - config_name: gold
+    data_files:
+      - split: validation
+        path: gold/validation.parquet
+      - split: test
+        path: gold/test.parquet
+  - config_name: silver
+    data_files:
+      - split: train
+        path: silver/here_silver.parquet
+  - config_name: silver_all_signals
+    data_files:
+      - split: train
+        path: silver/silver_all_signals.parquet
+---
+
+# HeRE: Hebrew Relation Extraction benchmark
+
+Data release for **HeRE: A Novel Benchmark for Hebrew Relation Extraction via LLM-Guided Denoising of Knowledge Graph Alignments** (Ron Keinan, Amir David Nissan Cohen, Reut Tsarfaty; AACL 2026). Code: https://github.com/Ronke21/HeRE.
+
+Every row is a Hebrew Wikipedia passage aligned by distant supervision (CROCODILE) to a Wikidata triple (subject, predicate, object) whose two entities are mentioned in the passage. The label answers one question: **does the passage express the stated relation between the subject and the object?**
+
+## Configs
+
+| Config | Split | Rows | Label |
+|---|---|---|---|
+| `gold` | `validation` | 500 | human, dual-annotated and adjudicated; the set on which the denoising pipeline was selected |
+| `gold` | `test` | 500 | human, dual-annotated and adjudicated; never used for selection (203 unseen-predicate rows, 97 depth rows, 200 silver-drawn rows) |
+| `silver` | `train` | 1,491,071 | `agree3`: unanimous vote of Gemma-4-31B-it (few-shot), Gemma-3-27B-it (few-shot) and DictaLM-3.0-24B (HebNLI-fine-tuned NLI); 1,080,606 positive / 410,465 negative |
+| `silver_all_signals` | `train` | 2,564,035 | every silver row with the raw score and hard prediction of all ten deployed signals and four consensus rules; `agree3_label` is -1 where the three models disagree |
+
+`kg/here_kg_edges.tsv.gz` (1,080,606 rows) is **HeRE-KG**: the positive `agree3` edges as `subject, predicate, object, docid, n_signals_pos, mean_score`, i.e. a Hebrew knowledge graph in which every edge points to the passage (`docid`, join on `silver`) that expresses it. Statistics in `kg/here_kg_stats.md`.
+
+## Columns
+
+Gold: `item_id, docid, title, text, subject, predicate, object, label, annotator2_label, notes, word_count` (+ `selection_reason` in test). `label` is the adjudicated label used throughout the paper; `annotator2_label` is the second annotator's blind label. Inter-annotator agreement before adjudication: Cohen's kappa 0.809 (validation), 0.757 (test).
+
+Silver: `docid, title, text, subject, predicate, object, label` (or `agree3_label`), then for each signal `<family>__<model>__score` (float) and `<family>__<model>__pred` (0/1; -5 or null where the signal abstains) for
+
+| Column stem | Strategy | Model |
+|---|---|---|
+| `opensource_llm__gemma4_31b` | 4, few-shot | Gemma-4-31B-it |
+| `opensource_llm__gemma3_27b` | 4, few-shot | Gemma-3-27B-it |
+| `opensource_llm__dictalm3` | 4, few-shot | DictaLM-3.0-24B-Base |
+| `finetuned_nli__dictalm24b` | 2, LoRA NLI | DictaLM-3.0-24B |
+| `finetuned_nli__aya32b` | 2, LoRA NLI (threshold 0.9) | Aya-Expanse-32b |
+| `finetuned_nli__gemma2_9b` | 2, LoRA NLI (threshold 0.9) | Gemma-2-9B |
+| `encoder_nli__neodictabert` | 1, encoder NLI | NeoDictaBERT |
+| `encoder_nli__xlmroberta` | 1, encoder NLI | XLM-RoBERTa-large |
+| `cross_rc__neodictabert_k3` | 3, cross-trained | NeoDictaBERT, K=3 |
+| `cross_rc__mmbert_k5` | 3, cross-trained | mmBERT, K=5 |
+
+Consensus columns: `vote_all__pred`, `vote7__pred`, `agree4__pred`, `agree2__pred` (-5 = abstain).
+
+## Quality
+
+Measured on the held-out gold test set: the best single deployed signal (Gemma-4-31B-it) reaches F1 0.844 (0.944 on predicates seen during development, 0.776 on unseen ones). The published `agree3` label covers 66.8% of the test rows and on them reaches precision 0.899, recall 0.934, F1 0.916 (95% CI 0.886 to 0.944). Use the per-signal scores to choose a different precision/recall operating point.
+
+## Contamination note
+
+The 500 test rows were drawn from the silver corpus; the 499 that match a silver row are **withheld from every silver and KG file** here. Do not add them back when training.
+
+## Sources and licenses
+
+Text: Hebrew Wikipedia (2025 dump), CC BY-SA 4.0, which this release inherits. Triples: Wikidata, CC0. Alignment: CROCODILE (Huguet Cabot and Navigli, 2021). Text was cleaned of wiki markup; entity fields were normalised.
+
+## Citation
+
+Keinan, R., Cohen, A. D. N., and Tsarfaty, R. (2026). HeRE: A Novel Benchmark for Hebrew Relation Extraction via LLM-Guided Denoising of Knowledge Graph Alignments. In *Proceedings of AACL 2026*.

gold/test.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e1ee175eb7db3456f823ed4dda021aef8c6fef17f68c2ed134ab3f9f7a47a402
+size 1177897

gold/validation.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3905bd5971bc5835682192bf0fa920fd22d3bbafab36fb311a59f95f1b22e2ea
+size 229492

kg/here_kg_edges.tsv.gz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:483780d1b3082337ff2e64a4e95ec4829d3617175df92d9d6f1c9f334f9081ab
+size 17847466

kg/here_kg_stats.md

@@ -0,0 +1,15 @@
+# HeRE-KG statistics (v3, agree3-positive, gold test rows withheld)
+
+| Statistic | Value |
+|---|---|
+| Edges (evidence-linked mentions) | 1,080,606 |
+| Unique triples (s,p,o) | 860,991 |
+| Entities | 424,216 |
+| Relation types | 917 |
+| Relations with >=100 edges | 306 |
+| Source documents | 212,332 |
+| Mean / median entity degree | 5.1 / 2 |
+| Max entity degree | 20,559 |
+| Connected components | 7,248 |
+| Largest component (share of entities) | 405,407 (95.6%) |
+| Top-5 relations | מדינה (123,037), עיסוק (93,769), מופע של (76,933), אזרחות (34,001), מקום לידה (30,569) |

silver/here_silver.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cd93782a6eca8d74da6a880b974db4cfcfb96d86aebd606331361c0953160daf
+size 836465411

silver/silver_all_signals.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1a3b00aab94fb47a8ed422486be41628d5b012e12b016184e56b91ac0e0a2fb2
+size 1277679951