基于短语模式的情感分析

A Study of Emotion Analysis Based on Phrase Pattern

对文本的情感分析充分利用文本的语义信息,在文本检索领域、网上调查及舆情监测领域是非常有价值的,是目前智能信息处理领域研究的重点和热点。本文针对网民关于话题评论简单,数目众多的特点,应用大连理工大学情感词本体库,利用How Net和word2vec扩展情感词典,利用短语模式计算文章的情感值,最终判断文章的情感倾向。利用这一技术可以在一定程度上解决网络评论信息杂乱的现象,方便用户准确定位所需信息。

The emotional analysis of text uses the semantic information of text. It is very useful in the field of text retrieval, online survey and public opinion monitoring. Considering that the netizens＇ comments are relatively simple and numerous in quantity, our method takes advantage of the phrase pattern. We use word2vec and HowNet to expand the emotional dictionary with borrowing the sentiment word ontology library of Dalian University of Technology. Then the emotional value of a text is computed by phrase pattern, and finally, the emotional tendency can be judged. By this method we can deal with the chaotic information of the network comments to a certain extent, and can make users easier to locate the needed information accurately.