一种基于介词用法的灾难事件信息抽取方法

A Disaster Event Information Extraction Method Based on Preposition's Usages

为提高事件信息抽取技术的准确率,通过对文本中介词用法进行识别,采用一种条件随机场统计模型和介词用法信息相结合的事件信息抽取方法,完成抽取系统的编写,以达到理想的抽取结果。本文采用《人民日报》作为实验语料,通过大量实验,结果准确率达到90.33%,比不含介词用法信息的方法提高9.14%,证明了此方法的有效性。

To improve the efficiency of the event information extraction,through the recognition of preposition＇s usages in text,an approach based on preposition＇s usages and Conditional Random Fields（ CRFs） is proposed for event information extracting,and the paper completes the coding of extraction system,in order to achieve the better extraction result. ＂People＇s Daily＂is used as experiment corpus,the approach achieves a better performance,the precision rate reaches 90. 33%,increases 9. 14% than the method without preposition＇s usage.