摘要
事件抽取是信息抽取领域一个重要的研究方向,事件抽取模式的定义和获取是其中的一个关键问题。提出了一种基于动词论元结构层次模型,将事件元素与动词的语义角色相对应,在实体、词性、关键词层次对事件元素进行语义约束的事件抽取模式定义方法。另外,为减轻模式建设的代价,提出了一种从标注语料中自动归纳事件抽取模式的方法。在此基础上,以发布事件为实例构建了实验系统,实验结果表明该方法的F指数达到71.7%。
Event extraction is an important research point in the area of information extraction ,and the definition and generation of patterns is a key issue in it.This paper brings forward a new pattern definition method which is based on the level of verb argument structure.It maps the event argument to the semantic role,and it provides an event extraction pattern based on the semantic constraints of event argument in the levels of entities, part of speech, and keywords Finally it proposes a method to generate extraction patterns from the corpus in order to alleviate the cost of pattern building.An issuance event information extraction system is conducted using that method and the experimental results show that the F-Measure achieves 71.7%.
出处
《微计算机信息》
2010年第9期187-189,共3页
Control & Automation
关键词
事件抽取
事件模式
语义角色标注
event extraction
event patterns
semantic role labeling
