Research And Analysis Of Micro-Blog’s False Topic Based On Bayesian Model

Micro-blog, as an important interactive media, has been gradually changing the traditional information dissemination in recent years. Now, it has become the information exchange, publish and display network service platform. Micro-blog, as an instant interactive tool, we can share news by every possible means and on every possible occasion, and it has already incorporated and changed our lives. While we enjoying the great convenience to our lives that the Micro-blog bringing to us, we may forget that it also likely to become a carrier to spread false information. We know that the speed of information dissemination is very fast in the network world, and then it may be provide a condition for false information to spread. In this condition, to identify false blogging topic and prevent to spread are very important.This paper is based on the network relationships to collection and analysis data. It contains three aspects:constructed of the micro-blog data organization structure, the users’emotion computing, as well as the trust network data analysis. The purpose is to get a model to identify the micro-blog’s false topic.First, we analysis the micro-blog’s structural features, propagation characteristics and evolution characteristics, focusing on the micro-blog data organization method. To organization data by main characters and conversation between the users. Then we can get the social and interactive relationship and social connection relationship of all the blog users.Secondly, based on the micro-blog topics, we analysis the blogs’content and the comments, combined with the emotion vocabulary to get the owner’s emotion tendencies, and come to the polarity of the potential influential figures.Thirdly, trying to design characters basic credibility algorithm and get the value of the basic credibility of the characters; to design characters interact directly with the relationship algorithm to obtain the trust value of the direct interaction between the characters, to design characters algorithm to obtain figures indirect interaction between the indirect interaction trust value, and finally considering these factors design trust network construction algorithm to provide reliable evidence for trust reasoning theory. Finally, designing the Naive Bayes trust inference algorithm to identify the influent of the comments to the review and mining the influential figures, then build a multilayer Naive Bayes model to detection false topics of micro-blog.