Abstract
An empirical analysis of 30.8K health-related Community Notes on X finds a median delay of 17.6 hours before a note receives a helpfulness status. The paper proposes CrowdNotes+, an LLM-based framework that combines evidence-grounded note augmentation with utility-guided note automation, assessed through a hierarchical evaluation of relevance, correctness, and helpfulness on HealthNotes, a benchmark of 1.2K annotated health notes.
Publication
Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)

Staff Alumni (Jul ‘24–‘26); Research Fellow (Jul ‘24–‘26)
Postdoctoral Research Fellow at WING & NUS CTIC

Staff Alumni (Aug ‘25–‘26); Research Assistant (Aug ‘25–‘26)
Research Assistant

CSC Visiting Student Alumnus (Sep ‘25–Sep ‘26)
Visiting student; interests include Multimodal Misinformation, Large Vision-Language Models.

Associate Professor
WING lead; interests include Digital Libraries, Information Retrieval and Natural Language Processing.