Extracting Biological Events from Text Using Simple Syntactic Patterns

摘要

This paper describes a novel approach presented to the BioNLP'11 Shared Task on GENIA event extraction. The approach consists of three steps. First, a dictionary is automatically constructed based on training datasets which is then used to detect candidate triggers and determine their event types. Second, we apply a set of heuristic algorithms which use syntactic patterns and candidate triggers detected in the first step to extract biological events. Finally, a post-processing is used to resolve regulatory events. We achieved an F-score of 43.94% using the online evaluation system.