Research On Rumors Detection And Rumors Spreading

With the rapid development of China's Internet,a variety of social media appears in our lives.As China's largest microblogging social media and information exchange platform,Sina Weibo not only provides users with a lot of information,but also promotes the further spread of rumors,which results in many negative effects.Rumors detection in Sina Weibo can help to reduce damages caused by the spread of rumors,and control the influence of the rumors.Thus,this work is of much importance to our society and government.The major work of this paper is to detect rumors based on LSTM,and analysis the feature of rumors spreading in the network.In the research of microblogging rumor detection,firstly we regard every word in the Weibo sentence as an independent word vector against each other,with no semantic relationship.A sentence-level vector was generated simply by summing up theses word vectors.Then its representation power was tested through some machine learning classifier.The experimental results showed that due to a lack of the semantic information,using these sentence vectors will lead to a low accuracy of rumor detection.In order to reach higher accuracy of rumor detection,we use deep learning method which can extract advanced features to construct the rumor recognition model.Different types of neural network structure were used for comparative experiments.We found that deep learning models can effectively identify rumors,in which the bidirectional LSTM model has reached the best results.So,to study the influence of Weibo propagation characteristics on rumor detection,a sentence-level vector generated by LSTM,containing propagation characteristics was proposed.The experiment results showed that the propagation characteristics can help to better identify rumors.In the research of rumors spreading,we built a basic game model as well as a responsibility-based game model considering the Weibo user's selection strategy.In order to simulate the process of rumor spreading,we combined the game model with the network structure of the Weibo rumor spreading.Finally,the effectiveness of our method was confirmed.The game model can stimulate the process of rumor spreading to a certain extent,which was influenced by the network structure.The neighboring users will have a great impact on the current user.When a lot of neighboring users send the rumor,the current user will tend to follow them.In general,this paper not only detects the content of microblogging rumors,but also establishes the game model to simulate the process of rumors spreading in the real network.We evaluated the influence scope of rumors in the network and verified the validity of our method.