Paper: A fast rule-based approach for biomedical event extraction

ACL ID W13-2014
Title A fast rule-based approach for biomedical event extraction
Venue Proceedings of the BioNLP Shared Task 2013 Workshop
Session
Year 2013
Authors

In this paper we present a biomedical event extraction system for the BioNLP 2013 event extraction task. Our system consists of two phases. In the learning phase, a dictionary and patterns are generated automatically from annotated events. In the extraction phase, the dictionary and obtained patterns are applied to extract events from input text. When evaluated on the GENIA event extraction task of the BioNLP 2013 shared task, the system obtained the best results on strict matching and the third best on approximate span and recursive matching, with F-scores of 48.92 and 50.68, respectively. Moreover, it has excellent performance in terms of speed.