Dependency Parsing Based Event Recognition
基于依存分析的事件识别

Event Extraction is an important part of information extraction. As the basis of Event Extraction, Event Recognition directly affects the results of Event Extraction. Machine learning based Event Recognition needs to find more features in words. For the deficiency of present Event Recognition method, this paper presented a novel method of Depen-dency Parsing based Event Recognition (DPER). Dependency parsing was used to find the syntactic relation among triggers and other words. As one of features, this relation was used to event classification on SVM and then to event recognition. The experiments show DPER has better performance than traditional method, and Event Recognition integrating multi-features improves F-measure to 69.3 %.