摘要
网络评论的倾向性识别是一种重要的网络舆情分析技术。针对网络评论的特点,提出了基于句式结构的网络评论倾向性识别方法。该识别方法主要思想是先对已进行倾向性识别的网络评论集进行分词并提取句式结构,形成各种句式结构的基准句集;再对待判别的网络评论提取句式结构,并比较其与相应句式结构的基准句集的相似度,从而判别出待识别的评论的褒贬倾向性。实验表明:该方法的准确率为80%以上,具有良好的实用性。
Orientation identification of web comment is one of the key methods for Internet sentiment analysis.A new method based on sentence structure is presented for orientation identification of Web comment.This method first extracts sentence structure of each comment from existed corpus,and divides the corpus into several clusters that comments in the same cluster have the same sentence structure.Then it computes the similarity between each uncategorized comment and the corresponding cluster,so that the orientation can be identified.The experiment results have proved its effectiveness.
出处
《通信技术》
2011年第2期100-101,111,共3页
Communications Technology
基金
国家自然科学基金资助项目(批准号:60605019
60772098)
国家重大基础研究973计划(2010CB731403/2010CB731406)
关键词
倾向性
网络评论
句式结构
相似度
知网
orientation
Web comment
sentence structure
similarity
HowNet
