面向网络评论的观点主题识别研究

Research on the Identification of Opinion Topic Expressed in Web Comments

网络评论的观点分析为及时掌握广大民众的真实观点提供了渠道。观点主题识别作为观点分析的重要组成部分,用以确定观点所指的对象。本文设计了一种领域无关的观点主题识别算法,该算法以网络评论中观点主题产生的方式为依据,采用由内到外的识别过程,分四个部分完成观点主题识别：内部主题词识别、内部主题构建、外部主题识别和主题的组织。算法能够克服分词和短语类主题带来的影响,识别出语义完整的观点主题。对实际网络评论语料进行测试的结果表明,本文的算法能够有效地识别网络评论中的观点主题。

Opinion analysis of network reviews provides a channel to find out the viewpoint of the common people in time, and opinion topic identification,as a significant part of opinion analysis,is aimed at identifying the objects for the expressed opinions.This paper proposes an opinion topic identification algorithm within independent domain.The algorithm,based on the mode of building opinion topic in network reviews,employs the process to identify the topics from inner to outer,and the process is finished through four steps：Inner topic word identification,Inner topic constitution,Outer topic identification and Topic organization.This algorithm can overcome the influence of word-segmentation error and phrase-topic,so it can get opinion topics which have the integrated semantic information.Experiment in real network reviews corpus proves that the algorithm can identify opinion topic in network reviews effectively.