Research On Microblog Rumors Detection Pattern Based On Sentiment Analysis

With the development of Internet technology, social network service is affecting people's life gradually. Micro-blog as a very important part of social networking services, which brings convenient information, also contains rumors at the same time. It can not only affect the individual, but also can affect the society to some extent. So how to automatically and effectively identify micro-blog rumors, causing widespread concern in the community and the relevant researchers.In traditional research of microblog rumors detection, researchers mostly model the problem as a binary classification problem in a supervised learning process, hence the primary focus has been feature selection composed of content features, propagation features and user features. All of these features are shallow. This ignores the mining of deeper features such as sentiment tendency feature of the comment and microblog propagation structure, so they have not achieved very good results. In this paper, we take Sina Weibo as an example, which is the largest microblog platform in China. Based on the previous research results, this paper puts forward the sentiment tendency feature of the comment into the characteristic attribute of the micro-blog rumor recognition. At the same time, the micro-blog repost process is modeled as a propagation tree structure, the similarity of the propagation tree is calculated by the graph kernel function. And then put forward a sentiment-based hybrid kernel SVM classifier, we use this classifier to complete the identification of microblog rumors. The main work of this paper includes:First of all, this paper completes preprocessing operations, such as micro-blog data spam filtering, word segmentation. Then we use the emotion dictionary to analyze the emotion tendency of microblog comments to get the emotion characteristic. Then the hybrid kernel function is constructed to support vector machine classification based on the graph kernel. Finally, we use the Sentiment-based hybrid kernel SVM(SHSVM) which is proposed in this paper to classify the experiment.Experimental results show that the classification model proposed in this paper is better than the previous research model. And it has certain practical application value.