摘要
【目的/意义】事件角色填充是浅层语义分析的一种方法,文章研究的目的是将句中与核心词有依存关系的句法成分进行标注,确定其在事件框架中充当的角色类型,为语义检索提供依据。【方法/过程】通过对语义框架和语法知识的介绍,提出了基于句法依存分析的角色填充思路和技术。【结果/结论】通过"灾难场景"框架下的"森林火灾"事件为例,用最大熵算法对填充过程进行了说明,例证了方法的有效性。
[ Purpose/significance ] Event Role annotation is a method of shallow semantic analysis. This paper aims at anno- tating the syntactic components which have dependency relationship with the core words in order to determine the role types they play in the event frame and provide the basis for semantic retrieval. [ Method/process ] The article introduces the semantie framework and grammar knowledge, gives the idea and method of role annotation based on syntactic dependency analysis. [Result/conclusion] At last, the effectiveness of the method is proved by the maximum entropy algorithm to de- scribe the annotation process through the example of "forest fire" event frame under "disaster scene".
出处
《情报科学》
CSSCI
北大核心
2017年第7期65-69,共5页
Information Science
基金
在线教育研究基金(2016YB119)
国家社科基金(13CTQ030)
关键词
依存分析
语义标注
事件抽取
角色填充
dependency analysis
semantic annotation
event extraction
role annotation