Research Of Subjective Information Propagation In Social Network

Social networks are important products of Web2.0, as bring the convenience to people's lives, leading to the explosive growth of information. In the cultural life of human, people can not only free to express their opinions in social networks, but also from their own network of relationships in the real world migrate to the Internet, such as Facebook, renren social network, blog networks, etc. Along with the development of the network, the relationship between people is more and more complicated, and the distance between people becomes shorter, all of these leaded to the spreading of information is faster and faster in the network. The sentiment of person would be propagated with the spreading of the topic, the spreading of the sentiment is the show of public opinion. How to monitor public opinion network become the focus of research. In the material life, the e-commerce sites have joined the social networks, which made people can not only rely on the reputation of sellers, also can read comments by other buyer when choosing products. But as the development of the network, the number of the comments is growing day by day, how to mine valuable information from comments is a problem concerned by sellers and buyers.This paper launched research of propagation of subjective information from the spreading of sentiment and trust in the social network. First of all, the spreading of sentiment in social networks is introduced, and we studied the characters of sentiment propagation in hot topics. After analyzed the data come from sina blog, we found that the hot topics had following characters:faster propagation, shorter time duration, and has a peak during propagation. Therefore, this paper added spread time window to the propagation of sentiment in the social network, limiting the influence time of sentiment. After the contrast with the former method, the experimental results show that the prediction accuracy is raised on the method of this paper, and the effectiveness of the method is verified. Secondly, this paper introduces the propagation of trust in the social networks, and after analyzed the relationship between the user similarity and the trust in users on epinion data set, we found that, the similarities between trusted users are greater than distrusted users in general. This paper improved trust propagation model by combined Bayesian theory utilized the relationship between user similarity and user trust degree. After analyzed the result of four groups of experiments, our method is improved in precision, recall, FScore, general precision.After study the propagation of subjective information in sentiment and trust, we proposed the method of study sentiment propagation in hot topics and trust propagation model based on Bayesian theory. The former is the improvement of sentiment propagation in hot topics, the validity of the proposed method is verified by experimental results. The latter improved the trust propagation model by taking advantage of user similarity and combining Bayesian theory, the validity of the proposed method is verified by compare to trust propagation model.