Extraction of Diagnostic Information from Social Media

Sichang Tu

Date: 2021-10-29 / 3:00 ~ 4:00 PM
Location: https://emory.zoom.us/j/99364825782


This is a collaborative project between Emory NLP and Real Life Science, aiming to recognize entities types (e.g. patient, treatment) and extract their relations (e.g. diagnose, side effect) in documents collected from social media such as Reddit and Twitter. This presentation includes mainly two parts: (1) data and annotation scheme; (2) NLP models for the task. I will first present the annotated examples and illustrate the basic annotation scheme. I will then review the recent development in similar settings.