中文事件抽取研究及实现  

Research on Chinese Event Extraction and Realization

作　　者：魏新宇 秦颖[1] 

机构地区：[1]北京外国语大学计算机系,北京100089

出　　处：《电脑编程技巧与维护》2014年第10期31-34,共4页Computer Programming Skills & Maintenance

基　　金：校级科研专项基金项目"基于平行语料库的学生译文自动评价研究与实现"(编号:2009JJ056)

摘　　要：在信息抽取越来越重要的今天,作为其重要研究方向,事件抽取也备受关注。选择了基于事件实例的方式,提出在对事件非事件句子分类的基础上,进行事件关键元素提取的研究。在句子分类上抓取句子特征进行分类,随后对包含事件句子中词进行特征抓取分类,选出特定事件元素。选取了SVM多元分类算法,并取得了一定成果。Nowadays, the information extraction is becoming more and more important. As a major research direction of it, event extraction has drawn a lot of attentions. Without adopting the event extraction based on event triggers, we choose the event instance. This paper proposes a research on the event argument recognition based on the filtered sentences. We have two main steps： 1） choose the characters of sentences and select event sentences 2） choose the characters of nouns and se- lect the key event arguments. We decide to use the SVM algorithm to train our data, and get an effective result.

关 键 词：中文事件提取 事件元素识别 事件实例驱动 

分 类 号：TP391.1[自动化与计算机技术—计算机应用技术]