Upload README.md with huggingface_hub

README.md

@@ -0,0 +1,64 @@
+---
+license: cc-by-4.0
+configs:
+  - config_name: fcv_pads_east_africa
+    data_files:
+      - split: train
+        path: fcv_pads_east_africa.jsonl
+  - config_name: refugee_pads
+    data_files:
+      - split: train
+        path: refugee_pads.jsonl
+  - config_name: jdc_operational
+    data_files:
+      - split: train
+        path: jdc_operational.jsonl
+---
+
+# FCV Extractions Meta
+
+Data-use mention extractions from `rafmacalaba/fcv-extractions`, enriched with the source
+document label metadata from `rafmacalaba/fcv-corpus`. Each extraction row keeps its
+per-page/chunk `input_text`, `entities`, and `has_data_score`, plus the document-level
+labels of its source document (`project_id`, `country`, `document_type`, `language`,
+`disclosure_status`, etc.).
+
+## Configs
+
+One config per source corpus (mirrors `rafmacalaba/fcv-corpus`):
+
+| config | rows | notes |
+|---|---|---|
+| `fcv_pads_east_africa` | 91,267 | WB FCV project documents (East Africa) |
+| `refugee_pads` | 45,046 | World Bank refugee project documents (PADs) |
+| `jdc_operational` | 12,372 | WB operational project documents (JDC operational set) |
+
+## Schema
+
+Common per-config fields:
+
+- `corpus_id` — `{config}:{index:06d}`, matching `rafmacalaba/fcv-corpus` ordering
+- `config` — source corpus config
+- `pdf_url`, `title` — source document
+- document label metadata (fields vary by config; inherited from `fcv-corpus`):
+  `project_id`, `doc_id`, `document_type`, `document_date`, `language`,
+  `disclosure_status`, `country`, `report_no`, `authors`, `disclosure_date`,
+  `major_sector`, `sector`, `topics`, `themes`, `region`, `landing_page_url`, `abstract`
+- `page`, `chunk`, `input_text`, `has_data_score`, `entities` — extraction payload
+  (entities are `{text, label, score, start, end}` spans)
+
+## Usage
+
+```python
+from datasets import load_dataset
+
+fcv    = load_dataset("rafmacalaba/fcv-extractions-meta", "fcv_pads_east_africa")
+refugee = load_dataset("rafmacalaba/fcv-extractions-meta", "refugee_pads")
+jdc     = load_dataset("rafmacalaba/fcv-extractions-meta", "jdc_operational")
+```
+
+## Provenance
+
+Built by joining `rafmacalaba/fcv-extractions` rows to `rafmacalaba/fcv-corpus` documents
+on `config` + `pdf_url` (all 148,685 rows matched). The `fcv_pads_east_asia` config was
+renamed to `fcv_pads_east_africa` to fix the region typo.