Research On Group Structure Of Micro-Blogging Network

Vith the development of Internet technology, micro-blogging suddenly became a new form of the network. The formation and propagation of topics on micro-blogging network have a phenomenon of group. Micro-blogging group structures have characteristics of community structure in complex networks. The research to the group structure of micro-blogging network can help to understand the propagation law of topic. Therefore, it has theoretical and practical significance.Firstly, this paper studies the relevant theoretical knowledge in the field of complex network, analyzes the classical algorithms of community discovery, and describes the basic concepts, contents, research methods, theoretical knowledge of the social network analysis.Secondly, this paper presents an approach of interest-based similarity to mine the micro-blogging community structure. Find the key words of micro-blogging text by means of word segmentation and statistical analysis of the word frequency and use key words to establish user's interest model. Then calculate the similarity of user's interest. Find and determine the group structures of micro-blogging by similar relationship among users'interest. Experiments show that this algorithm could accurately calculate micro-blogging users'group structure.The actual data analysis shows that micro-blogging group structure has scale-free feature and core-periphery structure feature. And this paper presents an algorithm to identify three kinds of nodes in the core-periphery structure. Experimental results show that the algorithm can effectively identify the core nodes, intermediate nodes and edge nodes in the group structure.Finally, this paper researches the application of group structure discovery to mine the hot topics of micro-blogging. By group structure mining algorithms we could get network group structure and the largest sub-group. The topic which sub-group members discuss in micro-blogging can be viewed as a hot topic.