微博负向情感热点话题发现模型

微博中热点话题,尤其负面情感热点话题对舆情的发现起到了重要作用,本文从情感的角度出发,提出了一个面向微博负向情感的热点事件发现模型。首先,在数据预处理阶段除了对微博文本中含有＂@＂＂#＂的博文进行过滤,并引入户信息对休眠用户及僵尸用户进行了剔除;其次构造情感分类器,对博文进行情感分类,筛选出负向情感博文;然后根据词频和词语增长速度对主题词进行评价;接着根据词意相似度以及共现度对话题进行聚类;最后通过计算话题负向情感值对负向情感热点话题进行细粒度划分。

Microblog detection model of the negative emotion hot topics

Hot topics in microblog,especially hot topics in negative emotion,which has a important effect on the detection of public opinion.Form the perspective of emotion,this article proposes a detection model of hot topics in negative emotion.Firstly,in the stage of data preprocessing,introducing the user information to filter out the dormancy users and zombie users,except that the posts contain＂@＂＂#＂are filtered.Secondly,constructing emotion classifiers,which can classify the posts on emotion and screen negative emotional posts.Then we evaluate the subject heading based on the word frequency,the growth rate of words and the value of the words.Subsequently,we clustering the subject heading by word similarity and co-occurrence.Finally,to the hot topics with a fine-grained partition by computing negative emotion value.