计算机应用 ›› 2010, Vol. 30 ›› Issue (11): 2937-2940.

• 数据库与数据挖掘 • 上一篇    下一篇

网络评论倾向性分析

丁建立1,慈祥2,黄剑雄1   

  1. 1.
    2. 中国民航大学
  • 收稿日期:2010-05-24 修回日期:2010-07-12 发布日期:2010-11-05 出版日期:2010-11-01
  • 通讯作者: 慈祥
  • 基金资助:
    基于服务架构的民航公众信息服务平台;基于生物免疫机制的枢纽机场航班延误状态智能监测和预警方法研究

Orientation analysis of Web reviews

  • Received:2010-05-24 Revised:2010-07-12 Online:2010-11-05 Published:2010-11-01
  • Contact: ci xiang

摘要: Web2.0的兴起使得包括新闻评论、产品评论在内的各种网络评论大量涌现,针对评论信息的监管和利用中的问题多种多样,重点研究其中的网络评论倾向性分析。以知网为基本的语义字典,提出一种改进的词汇相似度计算方法,在此基础上融合同义词词林对词汇的倾向性计算做出改进,进而利用相关语言学知识实现了从细粒度的词汇到粗粒度的评论的倾向性判断。实验表明,该方法对于真实网络环境下的网络评论倾向性分析具有较高的准确率。

关键词: 知网, 同义词词林, 网络评论, 倾向性分析

Abstract: The rise of Web 2.0 leads a lot of Web reviews including news review and product review. Problems during monitoring and using these review information are various, and the orientation analysis of Web reviews was emphasized. HowNet was used as the basic semantic dictionary and an improved method to calculate the word similarity was proposed, then the calculation of word orientation was improved by combining Tongyici Cilin, after that the orientation discrimination was realized from low granularity' word to high granularity' sentence by using language knowledge. The experimental results show that the method has high accuracy to the orientation analysis of Web reviews in the real Internet environment.

Key words: hownet, Tongyici Cilin, web review, orientation analysis