汉语框架网络知识库的语义角色特征识别

Characteristic Recognition of the Semantic Roles in CFN

以汉语框架网络数据库为研究对象,从短语类型、主控范畴、句法树路径、位置、中心词、语态等特征方面综合考虑,达到识别各种框架语义角色的目的,并结合盗窃框架中的词元"偷",对一系列语义角色特征的规律进行总结说明,进而对CFN框架网络语义层自动识别理论进行探索。

Chinese FrameNet database describe the words of semantic framework and study the chinese texts as a reference of Framenet. Rich semantic resources of CFN are the basis to realization of the machine semantics understanding. As a primary method of semantics understanding, automatic semantic tagging has become the focus and difficulties of the study, the paper comprehensively study the sematic roles such as phrase type, control areas, syntax tree path, location, head, voice characteristics, etc in order to identify all kinds of frame semantic roles. We summarize the law of the characteristic of semantic roles as the example of lexical ＂steal＂ and explore the theory of identification of automatic semantic layers in CFN.