Information Diffusion Analysis In Social Network Based On Entity-Level Sentiment Analysis

With the rapid development of social network,micrblog application like Sina Weibo and Twitter,social application like Facebook,photo sharing application like Instagram have been an important part in people's real life.Events in real world lead to a series of discussions in social network,and public opinion in social networks also can affect the development of real world.A large number of social applications and lots of users of social network generate a huge amount of information.On the one hand,this information provides convenience for users,on the other hand also gives users information noise problems.Emotional information is undoubtedly one of the most importmant information in social network.It has important theoretical and practical significance for the research of emotion information to understand the user's intention and the guide information diffusion.The research and utilization of emotion information in social network is facing great challenge.Different from traditional text information,social network has short text,has more noise,more complex semantics and less contextual information and so on,so there are many differences between tradicitonal methods and methods using emotion information.Especially for the entity-level sentiment analysis,there is a huge challenge.Based on the outstanding research,this paper studies the method of persona with entity-level emotional information and information diffusion process based on entity-level emotional information.The main research and achievements are as follows:(1)In the case of portrait,for the current protrait,they didn't consider emotion information,so we use entity-level emotion information to rich it.In this paper,based on the existing methods,we use Wikipedia as an external knowledge base to construct the entity linking and category tree.At the same time,we use syntactic structure information to do entity-level sentiment analysis,then binds the emotional information to the correspinding entity in the category tree.Finally,we show the user protrait with a web application.(2)In the case of entity-based information diffusion,we note that existing research on retweeting behavior didn't take entity-level emotion information as a factor that may affect user's retweeting behavior.Therefore,based on the above-mentioned portraits,we make assumptions pf three kinds of emotion diffusion models:”The Same”,”The Different” and ”User Bias”.We use the outstanding feature set and the emotion diffusion models above to predict users' retweeting behavior,and the result shows that with emotion information the classification has higher accuracy,then with each feature's ROC,we explain the emotional information can affect users' retweeting behavior.In summary,this paper studies the influence of entity-level emotional information in social media on user portraiture and information duffision.It has an important theoretical and practical significance.