Upload README.md with huggingface_hub

README.md

@@ -1,58 +1,63 @@
 ---
 dataset_info:
-  - config_name: queries
-    features:
-      - name: text
-        dtype: string
-      - name: query_type
-        dtype: string
-  - config_name: full_dataset
+  - config_name: paw_edits
     features:
       - name: text
         dtype: string
-      - name: aw_section
-        dtype: bool
       - name: page
         dtype: string
       - name: section_name
         dtype: string
-  - config_name: paired_dataset
+      - name: is_paw_edit
+        dtype: bool
+  - config_name: queries
     features:
-      - name: text
+      - name: query
+        dtype: string
+      - name: type
         dtype: string
 license: cc-by-4.0
 task_categories:
   - text-classification
 tags:
-  - data-attribution
-  - wikipedia
   - animal-welfare
-  - influence-functions
+  - wikipedia
+  - data-attribution
+  - llm-training
 ---
 
-# PAW Wikipedia Editing Attribution Dataset
+# PAW Wikipedia Attribution Dataset
 
-Dataset for "Small edits, large models: How Wikipedia advocacy shapes LLM values" (Navas, Brazilek, Gnauck, 2026).
+Data for the paper "Small edits, large models: How Wikipedia advocacy shapes LLM values" (Navas, Brazilek & Gnauck, 2026).
 
-## Contents
+## Files
 
-### queries.jsonl (170 queries)
-- **80 animal welfare queries**: e.g., "Aldi animal welfare", "What is McDonald's chicken welfare policy?"
-- **90 general queries**: e.g., "Aldi", "How many McDonald's locations are there?"
-- Used to probe whether attribution signals are topic-specific
+### paw_edits.csv (236 rows)
+Wikipedia sections used in training data attribution experiments. Each row is a text chunk from Wikipedia.
 
-### full_dataset_236.jsonl (236 documents)
-- **118 animal welfare sections** from 108 PAW-edited Wikipedia pages
-- **118 control sections** randomly sampled from WikiText-103 (seed 42)
-- Used for MAGIC training influence estimation and fine-tuning ablation
+| Column | Description |
+|--------|-------------|
+| `text` | The Wikipedia section text |
+| `page` | Wikipedia article title |
+| `section_name` | Section heading |
+| `is_paw_edit` | Whether this section was added/edited by Pro-Animal Wikipedians |
 
-### paired_dataset_72.jsonl (72 documents)
-- **36 animal welfare sections** paired with **36 non-AW control sections** from the same Wikipedia pages
-- Used for TrackStar retrieval attribution (within-article pairing controls for article popularity, writing quality, and topic domain)
+118 PAW-edited sections (animal welfare content added by advocacy editors) paired with 118 control sections from the same Wikipedia articles.
+
+### queries.csv (170 rows)
+Queries used to measure training data influence.
+
+| Column | Description |
+|--------|-------------|
+| `query` | The query text |
+| `type` | `animal_welfare` (80 queries) or `general` (90 queries) |
+
+Animal welfare queries ask about animal welfare at specific companies/by specific politicians. General queries ask about the same entities without mentioning animal welfare.
 
 ## Citation
+
 ```bibtex
-@article{navas2026smalledits,
+@article{navas2026small,
   title={Small edits, large models: How Wikipedia advocacy shapes LLM values},
   author={Navas, M and Brazilek, J and Gnauck, A},
   year={2026}