动态社会网络中的话题跟踪算法分析-计算机科学与技术专业论文.docxVIP

Abstract Dynamic social network stream can be described as a result of the interaction between time-varying network structure and text documents. A typical example is academic collaboration network, in which the collaboration between authors and papers published by authors are always changing over time. Another example is Microblog system, in which what the users generate and the relationship between users also changes dynamically. One of important tasks in dynamic social network analysis is how to find and track user-interested topic over time. At present, there are many works studying network stream or text document stream separately, however, there are less research which take into account both factors and their interaction. In this paper, in order to track topics over time, we formally define the problem and propose a novel statistical method that models the interplay between textual content and network structures. In order to track the latent topics, we propose a novel statistical model considering the social influence, selection phenomenon, historical status and textual content. O ur model is called Dynamic-social- network based Topic-Tracking model, short for DTT. Specifically, users interest in topics at current time is influenced by his own and his neighbors’ interests, a Gibbs Random Field is defined to model this influence. The link relationships between users are influenced by each users interest in various topics. We adopt a method that is similar to community impact link to model the selection phenomenon. Thereafter a classic topic model is used to describe the generative process of the documents. Finally, in order to ensure the continuity of the topic itself, the dependence of the historical status is also taken into account. Based on the above considerations, the formal definition of the topic tracking problem in dynamic social networks is given, and an effective algorithm for model learning is proposed. Based on the proposed model, experiment