自动内容抽取中的中文事件标注

Chinese Event Annotation in Automatic Content Extraction

事件检测与描述（Event Detection and Characterization,EDC）自2005年作为自动内容抽取（Automatic ContentExtraction,ACE）评测的一个重要子任务出现以来,中文事件的标注、检测与描述越来越成为研究热点。本文就自动内容抽取中的中文事件标注进行详细、系统地研究,主要包括：在ACE会议定义中文事件相关概念的基础上,给出事件标注中事件的可标注内容,包括事件范围及事件触发词等;根据生活中的事件分类在人工事件标注中对EDC的事件进行类别划分及其子类的详细区分,以降低事件检测的复杂度;对每个事件类别（包括子类别）中构成事件的元素进行研究,综合事件类别及其元素信息完成中文事件的标注。本文的研究成果在中文文本信息抽取、自动摘要及主题检测与追踪中得到了很好的应用。

Since Event Detection and Characterization（EDC） was brought forward as an important task of Automatic Content Extraction（ACE）,research on Chinese event annotation and extraction has becoming more and more popular.This paper concentrates on particular and systemic research on Chinese event annotation,which involves in the following contents. First,presents event taggability,which includes event extent and triggers and so on,on the foundation of concepts of event in ACE.Second,classifies event types and subtypes in event annotation of EDC based on real life.Third,analyses event arguments for every event type and subtypes and complete Chinese event annotation integrating event type and its arguments. The research of this paper can well be used in Chinese text processing,automatic text summarization and topic detection and tracking.