基于词向量模型的情感分析

Sentiment Analysis Based on Word Vector Model

随着移动互联网的发展,以商品评论等带有主观性的短文本信息急剧增加.海量的文本信息使得人工管理越来越困难.本文以商品评论为研究对象进行情感分析.针对商品评论为短文本的特点,本文在词向量的基础上提出了词向量叠加方法和加权词向量方法进行文本特征的提取,从而更深层次的提取短文本特征.在进行评论情感分析模型性能的比较中,说明了本文所提方法的有效性.基于情感分析技术可以解决人工难以胜任的海量商品评论的分类,方便用户快速获取有效信息.

With the development of Internet, text information, such as product review, increases rapidly. The mass text information makes it more difficult to make artificial management. Considering that product reviews are short text, this paper comes up with the method of word vector superposition and weighted word vector. In the result of sentiment analysis, the method is proved effective. Emotional analysis technology can solve the difficulty of artificial classification in the mass of product review, and help users to get information quickly.