Research On The Properties Of The Interactive Network And Searching For Potential Members Of The Organization In Virtual Space

With the rapid rise of Web2.0, various types of virtual community services have sprung up, such as a variety of social networking sites, BBS forums and so on. Since the Internet's own characteristics determine that an individual can establish a large number of "friend relationship" in a short time based on the Web, it will result in that online social networks become more large and complex compared with the actual social networks. The features about online social networks and behavior characteristics of social online network members have been well-focused. At the same time, because of the hidden of membership in a online social networks, how to search for potential members of the organization in aonline social network has become a new reserch hotspot. With the forum of"nuaabbs.xici.net" as a research object, this paper mines the networks of relationships in a large number of real BBS data and searches for potential members of theorganization in online social networks. Our work is as the followings:1. Data about the forum of "nuaabbs.xici.net" is collected by us. This data set contains nearly 90 thousand subject matter from 2005 to 2007, and specifically includes three types of information:①30 thousand BBS user information, such as user nickname, signature, etc.;②nearly 90 thousand thematic information, such as theme name, author of theme, subject matter, release time, etc.;③more than 700 thousand of reply information, such as author of reply, the reply's theme, the contents of reply, reply time, etc. Extensive and real data is the foundation of mining and analysis of the network of BBS members.2. Based on the interactive manner of BBS members, this paper constructs reply network and common reply network of BBS members, and judges whether there is correlation between BBS members by the threshold. Our experiments show that these two kinds of complex networks based on threshold have features of small network and free-scale network. In this paper, the members of BBS are divided into 5 major categories by K-means based on behavior characteristics, then chi-square testing verifies that there is special interactive mode during the 5 different categories of BBS members.3. By analyzing the feature of BBS replys through a large number of thematic content, this paper extracts 30 phrases, which have often been used in BBS replys and could reflect the attitude of the reply author to the theme author. We invite a few of users to assesse the key words, and build confidence between members of the BBS network based on these key words, then discuss and analyze the characteristics in the network of relationships of trust.4 According to the ambiguity and uncertainty of memberships in online social network, this paper makes use of belief propagation to inference or update information about the other members based on the partial information, and uses the extended MPFS algorithm to search for the potential organization members in a complex online social networks.