Add link to paper and improve metadata

README.md

@@ -1,7 +1,24 @@
 ---
 license: apache-2.0
+task_categories:
+- text-classification
+language:
+- en
+tags:
+- medical
+- biology
+- relation-extraction
 ---
 
+# Drug-ACE
+
 This is the official dataset for the paper **"Applicability Condition Extraction for Therapeutic Drug-Disease Relations"**, accepted by **ACL 2026 Findings**.
 
-Repository: https://github.com/guantingluo98/Drug-ACE
+- **Paper:** [https://huggingface.co/papers/2606.14031](https://huggingface.co/papers/2606.14031)
+- **Repository:** [https://github.com/guantingluo98/Drug-ACE](https://github.com/guantingluo98/Drug-ACE)
+
+## Introduction
+
+Identifying conditions under which a certain drug has a therapeutic effect on a target disease is crucial for clinical decision-making support. Most existing biomedical information extraction methods focus on identifying simple relations between drugs and diseases, overlooking the context-specific conditions. 
+
+Drug-ACE is the first dataset that provides manually annotated triples of drugs, diseases, and applicability conditions extracted from biomedical research literature. It contains 1,119 drug-disease pairs annotated on biomedical paper abstracts.