面向用户评论的企业竞争情报挖掘研究  

Research on Web Review Mining in Enterprise Competitive Intelligence

作　　者：田雪筠[1,2] 

机构地区：[1]中国科学院文献情报中心 [2]中国科学院大学

出　　处：《图书馆学研究》2014年第19期71-78,共8页Research on Library Science

摘　　要：设计与实现一种面向用户评论的企业竞争情报挖掘算法,首先利用模板抽取评论句中的产品属性词和情感词,获得其搭配短语模式,然后使用面向语义的潜在语义分析(SO-LSA)和面向语义的互信息和信息检索方法 (SO-PM-IR)计算短语的情感值,获得评论的情感特征,找出产品的优缺点。研究表明,该算法能够显著地提高情报分析的准确率和召回率,实现深层次的评论挖掘和情报知识发现。This paper designs and implements a web review mining algorithms in enterprise competttwe intelligence. First, phrase patterns, which represent the product features and their sentiment words syntactic structure of review sentence, are extracted. Then SQ-LSA and SO-PM-IR are applied to calculate the sentiment value of the phrase. Finally, the review＇s sentiment features are obtained, which can find out the pros and cons of the product. Research results show that the algorithm can evidently improve the accuracy and efficiency of intelligence analysis, and implement review mining and intelligence and knowledge discovery at the deep level.

关 键 词：用户评论 企业竞争情报 情报挖掘 情感计算 

分 类 号：G350[文化科学—情报学]