基于BootStrapping的中文事件元素抽取系统设计与实现

采用基于BootStrapping的方法实现中文事件元素抽取系统。其中,将事件元素抽取定义为一个模式匹配问题。针对这一问题,首先构建了初始种子集,然后创新性地引入了BootStrapping方法构建模板集,并使用模式匹配的方法进行事件元素抽取。在模板构造过程中,提出了基于BestMatch的模板泛化算法1]。对任意两个事件实例模板2]进行匹配,计算其匹配代价并泛化,提高了模板的覆盖能力。所实现的系统在ACE 2005语料测试中取得了不错结果。

Design and Implementation of Event Arguments Extraction System based on BootStrapping

This paper implements a system of event arguments extraction by using pattern matching ways.In this paper,the task of event arguments extraction is defined as a pattern matching problem.To solve this problem,the paper first builds the initial seed set,and then uses originally BootStrapping to build a set of templates,simultaneously uses pattern matching method for extracting event arguments.In the pattern construction process,this paper proposes BestMatch based pattern generalization algorithm.According to the costs defined above,the Soft-Pattern Learner is able to find the best generalization of any two instance patterns.Event arguments extraction has a preferable results on the ACE corpus.