Research On Attributes Of Users And Sentiments Analysis Of Events Based On Microblog Data

With the development and popularization of mass media, the Internet has become an important platform for users to obtain information, express their views, and exchange ideas. Official information published by authority news website forms flow of public opinion through discussion of folk social network, which impacts the spread and evolution of web event. However, the interaction mechanism between the official public opinion field and social network public opinion field is not clear, which makes it hard to control negative emotions when emergencies occur. Therefore, the impact of public opinion on the event has attracted a lot of researchers’ attention. Behaviors of users in the web are one kind of mapping of the behaviors of users in the physical world, in this way, traits of users impact the reaction to web events. This paper contributes to mining patterns of users’ behavior from expressions of users in the web, furthermore to analyze its role in the event propagation and evolution process. In social network, ways such as forward, review and praise facilitate the propagation of event. Meanwhile, if users add personal views or emotions when they propagate information, then they promote the evolution of the event. So, this paper makes effects to study on the basic features of web users, traits, as well as the information transmission mechanism using interdisciplinary methods. Specifically, we focus on user model and event propagation model including background knowledge and emotion, in other words we study the mechanism of information propagation from the micro and macro aspects. The main work and innovations of this paper are as follows:1. Based on the analysis of social psychology for sentiment, we put forward calculation method of the multi-granularity emotion based on fuzzy logic. This method calculates emotions of word depends on specific context and find emotion of popular language adaptively. In addition, based on the fuzzy logic, the membership degree of word to each dimension is arranged to [0, 1], breaking the limitation that sum of values of all dimensions have to be one, which makes the fuzzy computation of emotions become more suitable in the cognitive process of human beings.2. Based on data of social network, we built user model. The user model consider the basic characteristics of the user, including the user’s gender, age, education level, and complete the basic features of the missing user information. In addition this user model also includes background knowledge, user interest, emotional background, personal character, etc. This model mine personalization features of user on receiving and spreading information.3. Based on the official and folk field of public opinion, we put forward one method to calculate the interaction mechanism of sentiment. The method measures content and emotion of event and the emotion, and extracts the induced relationship between two public opinion fields. Moreover it can put forward suggestions for controlling events.4. Based on theory of network dynamics, we study the cascading phenomenon of information transmission. Within the community where users have close relations, it is easy to develop cascading phenomenon; among communities, where links are loose, it is easy for information to evolve. Once cascading phenomenon is formed, the original information of events occurs is missing, then "go with the flow" happens.