基于情感特征的用户聚类分析

摘要

随着互联网行业的发展，各式各样的网站不断兴起，用户在网上发帖留言已经成为一种常态，对用户的留言进行情感分析，分析用户的情感倾向和主观情感已经变得越来越流行。首先，本文用基于情感词典的情感分析方法对用户的情感进行分析，为用户的每一条回帖计算其正向情感分值和负向情感分值得以及计算其正向情感词数目和负向情感词数目，最后以用户为单位，计算均值，得到每个用户的情感特征向量，我们根据用户的情感特征向量，对用户进行聚类分析，探究用户的行为模式有什么不同点和相同点。%With the development of Internet,al kinds of website becomes popular,users like to leave somemessages on it.That is a good resource for us to analysis sentiment for every user. This paper uses a dictionary-based method to analysis users’ sentiment.we calculate thepositive points,negative points,the number of positive word and the number of negative word for every message.then get a mean value of them as a feature vecture for every user.At last,we can use the KMeans algorithe to cluster them and analysis the way when people leave a message.