Datasets:

rafmacalaba
/

datause-extracted-human473-docs

+---
+configs:
+- config_name: gliner
+  data_files:
+  - split: train
+    path: gliner_train.jsonl
+  - split: val
+    path: gliner_val.jsonl
+  - split: holdout
+    path: gliner_holdout.jsonl
+- config_name: bio
+  data_files:
+  - split: train
+    path: bio_train.jsonl
+  - split: val
+    path: bio_val.jsonl
+  - split: holdout
+    path: bio_holdout.jsonl
+- config_name: gliner2
+  data_files:
+  - split: train
+    path: gliner2_train.jsonl
+  - split: val
+    path: gliner2_val.jsonl
+  - split: holdout
+    path: gliner2_holdout.jsonl
+license: cc-by-4.0
+task_categories:
+- token-classification
+tags:
+- ner
+- span-extraction
+- data-mentions
+- economics
+---
+# datause-extracted-human473-docs
+Every passage of the 162 documents that carry the **473 human-validated
+holdout spans** of the data-use annotation campaign:
+| population | spans | documents |
+|---|---|---|
+| `annotator190` | 190 | 134 |
+| `jdc283` | 283 | 28 |
+| **total** | **473** | **162** |
+It is a **row-for-row subset of [`rafmacalaba/datause-extracted`](https://huggingface.co/datasets/rafmacalaba/datause-extracted)**
+(revision `15812843687e2ec81b261b5895f2019b50a8f97e`): same configs (`gliner`, `bio`, `gliner2`), same split files,
+same columns, same rows — filtered to the documents below. Row counts:
+`gliner/train` 1963, `gliner/val` 414, `gliner/holdout` 342, `bio/train` 1963, `bio/val` 414, `bio/holdout` 342, `gliner2/train` 1963, `gliner2/val` 414, `gliner2/holdout` 342.
+## Selection
+A document is in scope iff it backs at least one of the 473 spans. The span ->
+document join is `corpus_id` + `page` + `chunk` from the judge sheets (exact,
+not title-based); rows are kept by `pdf_url` or `corpus_id`. All passages of a
+selected document are included, so chunks without spans are present too.
+- `gold_spans.jsonl` — the 473 spans with `population`, `surface`,
+  `human_keep`, `corpus_id`, `page`, `chunk`, `title`, `pdf_url`.
+- `source_missing_docs.jsonl` — the 4 documents of the gold set that
+  `datause-extracted` does not contain (`fcv_pads_east_africa:019636`, `fcv_pads_east_africa:019977`, `fcv_pads_east_africa:013470`, `fcv_pads_east_africa:013125`). Their passages come
+  from the upstream extraction meta (`rafmacalaba/fcv-extractions-meta-tiered`), because the source repo has no
+  rows for them; columns there are `corpus_id, doc_id, page, chunk, title,
+  pdf_url, origin, project_id, country, document_type, input_text, population`.
+- `stats.json` — per-config/split row counts and document coverage.
+## Caveats
+- **The source repo's split assignment is not document-disjoint.** Documents of
+  this set appear in more than one of `train`/`val`/`holdout` in
+  `datause-extracted`, so the same document can be in two files here as well.
+  The `split` column is preserved verbatim; the split files are a faithful
+  subset, not a re-split. Treat `train` as neither clean nor a contamination-free
+  reference for these documents.
+- Spans in `datause-extracted` are raw model predictions, not gold. The gold
+  labels live in `gold_spans.jsonl` and in
+  [`rafmacalaba/datause-displacement-reviewed`](https://huggingface.co/datasets/rafmacalaba/datause-displacement-reviewed)
+  (`probe_reviewed`, split `holdout`, kinds `annotator` / `jdc`).
+- Provenance of the document set:
+  `analysis/holdout_documents_annotator190_jdc283.{json,tsv}` in the source
+  project.

bio_holdout.jsonl ADDED Viewed