Event Extraction as Frame-Semantic Parsing

Abstract

Based on the hypothesis that frame-semantic parsing and event extraction are structurally identical tasks, we retrain SEMAFOR, a stateof-the-art frame-semantic parsing system to predict event triggers and arguments. We describe how we change SEMAFOR to be better suited for the new task and show that it performs comparable to one of the best systems in event extraction. We also describe a bias in one of its models and propose a feature factorization which is better suited for this model.