Research And Application On Modeling Of Heterogeneous Social Networks

Research on social network has been growing vigorously in the last century. Leaders in almost any field, respectively, used their knowledge to explore the structural characteristics of social networks, and made some achievements, including mathematics(gragh theory, etc.), physics(human dynamics, etc.) and biology(infectious diseases, etc.) etc. From the beginning of 21 st Century, researchers have been accumulated a mount of research results in the statistical topology, structural modeling and other area on single network. In recent years, the studies of researchers from the single network towards heterogeneous network and other complex networks, from the basic statistical methods to network specific modeling(individual characteristics, triad, structure hole, community, topics, etc.).In this thesis, based on the results of the current study, there are the following innovations:First of all, the core of research is the behavior modeling of the individual level in the social network and base on complex networks, graph theory, statistics, social network structure. According to the individual / group behavior modeling problem, this paper presents a machine learning method based on individual behavior analysis model, the key is the metric of behavior of the individual characteristics and a clustering algorithm based role discovery. The effectiveness of this model has been verified through analyzing the user behavior of tecent micro-blog finally.Secondly, we are trying to promote the model to heterogeneous network from single network. We propose a modeling method based on individual behavior of heterogeneous networks due to the particularity of heterogeneous networks, including the different structure, the disconnected of different networks and the differential feature.Finally, through the research on the above topic, we developed a system to solve the routing process between different data and different algorithms.