摘要
指代消解是自然语言处理中的重点难点,对信息抽取具有重要意义.指代分有三种形式:代词指代,名词指代及零形回指.其中代词指代和名词指代是汉语中最基础的指代形式,以上两种指代的消解是指代消解研究的基础.为解决这一基本问题,使用决策树方法同时对两种形式指代进行处理.实验结果显示,所提出方法在不牺牲指代消解准确率的基础上,较大提高了指代消解的召回率.
Anaphora resolution is a difficult and important point in Natural Language Processing and important for information retrieval. Anaphora consists of three main forms： pronominal anaphora, nominal anaphora and zero anaphora. In Chinese, pronominal anaphora and nominal anaphora are the most fundamental forms, the resolution of which are the research focus of Coreference Resolution. To solve this problem, we adopt Decision Tree Approach for the treatment of both the anaphora forms. The experimental result shows that the proposed approach can promote the recalling of anaphora resolution without lowering the precision.
出处
《西安文理学院学报（自然科学版）》
2015年第2期65-69,共5页
Journal of Xi’an University(Natural Science Edition)
关键词
信息抽取
指代消解
命名实体
决策树
有限状态机
information retrieval
anaphora resolution
naming entity
decision tree
finite state machine
