Members' Behavior Analysis Of Social Network Based On Data

Social network theory originated in 1930 s, and has been rapidly developed in recent decades. It developed from a single subject to a new scientific research paradigm. It has been gradually improved and formed systematic theories, methods and systems in sociology, anthropology, mathematics, physics, economics and other fields.The concept of "social network" is first proposed by a famous British anthropologist R. Brown. He discussed how culture affected the behavior of internal members of the community(such as villages, tribes, etc.) to explore the concept of network. Since then, from the 1930 s to the 60 s, more and more scholars began to construct the concept of social structure and social network in various kinds of fields, for example, in psychology, F. Heider studied social attitudes and social equilibrium, in anthropology, Mayo et al researched on the Hawthorne plant, in mathematics, Euler pioneered the graph theory. With the coming of the era of Web2.0, the development of the Internet which as an opportunity gave birth to a large number of online social networks. No matter from the network structure to evolution mechanism or to the behavior characteristics, more and more research fields and directions carried out toward deeper level.All the social networks can be abstracted to the network which nodes represent network members and links represent some connections between members. The behavior of members can be studied from the structure and evolution of the network. In this paper, we analyzed social network member behavior from both linking behavior among members and spontaneous action of members, of which the former regarded members as nodes in order to study the connection mode, and the latter regarded members as entities to study the behavior characteristics. The three classic connection modes: completely randomized connection, preferential attachment and hybrid connection only considered new members can establish links to existing members, ignoring the existing members can also create new connections. Therefore, this paper optimized the connection model and put forwarded a new connection mode model, and finally get more comprehensive and accurate network connection mode. For the spontaneous behavior analysis of members, it is based on the definition of attributes to study the online social network member behavior. Then divided members to groups based on their interests, and judge the validity of the connection is established according to their interests.In this paper, the following work has been done:(1) By studying the structure characteristics and fundamental theory of social network, we analyzed and compared the structure characteristics and basic index of different real networks.(2) Collected data and visualized three networks: “Radio operators” network, “Friends” network, and “GR-QC” network. This work laid the foundation for the latter portion.(3) Analyzed the linking behavior among members: calculated the connection modes of three networks and analyzed them combined with their degree distribution. The result showed that for the “GR-QC” network, the description of its connection mode has unreasonable factors, and needed to be improved.(4) We added the situation which new links can be established between existed members to optimize the connection mode, and then proposed the new connection model. Data fitting and analysis of the “GR-QC” network have been done under the new connection model, and the result showed that this model is more consistent with the growth and evolution rules of network.(5) On the research of spontaneous behaviors of members: we took Ren Ren network as an example, and represented each member as a one-dimensional vector with 4-lenth by defining the attributes of members. The members were divided into 6 groups after clustering, and then we analyzed the behavior characteristics of each group.(6) Based on the user sharing attribute, the members were divided according to their interests, which laid foundation for judging whether connection is established according to the interests. After doing above judgment, we established a connection model which based on interest, and then we calculated the validity of the establishment of the connection of Ren Ren.The main innovation of this paper:(1) We calculated the connection mode of real networks and proposed new connection model. It described the structure evolution process of social network on a deeper level.(2) This paper defined the member attributes of Ren Ren network, and clustered user behavior and divided user interests. It also judged whether connection is established according to the interest of members.The judgment in the last portion is one combining point of linking behavior and spontaneous behavior of members. We just made simple analysis and calculation, so whether there are other relationships between the two aspects of member behaviors so as to well explain the operating mechanism of social network haven't researched in this paper, and it needed to be improved.