Event Extraction for Epidemic Prediction

Explore event extraction techniques for epidemic prediction in this informative lecture presented by Tanmay Parekh from UCLA at the USC Information Sciences Institute. Dive into two groundbreaking works that leverage social media data for early epidemic detection and multilingual event extraction. Learn about the development of a framework for extracting and analyzing epidemic-related events from social media posts, including a curated epidemic event ontology and the SPEED dataset focused on COVID-19. Discover how this framework can provide early warnings for epidemics weeks before official declarations. Examine a novel approach called CLaP for zero-shot cross-lingual event extraction, which utilizes instruction-tuned language models for contextual label translation. Gain insights into the speaker's background in multilingual technologies, code-switching, controlled generation, and the application of Large Language Models in Information Extraction across various languages and domains.