Research On Node Role Discovery Algorithm In Complex Network

Nowadays with the development of internet, the research on social media and social networks becomes a hot spot. As people are able to collect and analyze large-scale data of human behavior and discover the behavior patterns of individuals and groups, the birth of "computing sociology" marks a new area of research. The driving force of research on social networks not only comes from sociologists, economists, but also many broader interdisciplinary field scholars. This broad interdisciplinary research interests has important significance to social networks study both in theory and practical application.This paper focuses on the issues of nodes role analysis in complex networks. There are four aspects:1. An empirical study of the group’s behavior in a complex networks. In the early stage of human behavior studies, empirical study has practical significance, is the basic of user behavior model. This section provides an overview of the human behavior dynamics related methods and techniques, and an empirical study on telecommunications networks and BBS forum datasets.2. Node role analysis algorithm based on directed topological potential. Firstly, we give the definition of directed topology potential, then, from both the qualitative and quantitative two aspects, we propose the nodes role analysis algorithm and a method for temporal networks evolution detection.3. Role discovery based on sociology attributes clustering in online social network. In this paper, for online social networks (sina micro-blog), we put forward a series of user characteristics attribute metrics from sociological view, select non-negative matrix factorization and MDL model selection method for user roles research.4. Network evolution analysis based on node role. Based on tensor model, we propose a user influential role evolution analysis method, and do experiment for typical telecommunications networks, and do comparative study with the based on matrix decomposition algorithm.In conclusion, in this paper we do research on empirical study of group behavior, node role analysis, and the network evolution related method, and do experiment analysis in a communication network, online social networks, organizational networks.