Social Structure Based Research On Knowledge Sharing Network In Virtual Community

Knowledge-sharing in virtual communities is a kind of knowledge exchange between the members. The characters, such as Virtual community’s scale、social structure、strong liquidity among members, increase the complexity of analyzing the structure of knowledge-sharing network in virtual community, and add the difficulty on studying the rules and characteristics of knowledge-sharing in virtual community. Traditional knowledge-sharing network research looks on social network service with static viewpoints, ignored the motility of members and its influence on the transformation of the social structure in the community.Therefore, this article taking the acts of community participants as the key point, analyze knowledge-sharing in virtual community with hierarchical social structure analysis in three steps. Firstly, analyze the role of members. According to "The Law of the Few" and social network analysis, there are "key figures" in the knowledge sharing network, we call them experts、 connectors or presenter. Secondly, divide network structure of virtual community, thus positional addressing of divided communities and connectors can be achieved. However, the existing algorithm of finding "CONNECTORS" has several defects, such as:unclear boundaries which unable to accurately find and ignore the network community structure during the process of searching. This article, using social network analysis to study the meaning of key figures in virtual community, proposes dynamic cosine similarity algorithm. The algorithm is much better than the algorithm of dynamic community discovery based on role assorted thoughts. It measures the distance between the connector and subgroups, using Angle cosine and Euclidean distance formula from two aspects-- similarities and differences. And then, the algorithm successfully discovery strong connected community subgroups and connectors among the subgroups. Thirdly, test the performance of this algorithm. Compare structural constraint algorithm with betweenness centrality on the accuracy and relevance by using Pearson product moment correlation coefficient. Then, the whole cluster coefficient is adopted to measure the coefficient of concentration of subgroups. During the process of searching for "CONNECTORS" in network of virtual community, the algorithm discovery strong connected community subgroups and also reflects the process how to constitute the structure of a virtual community. It provides the research basis of optimization of the structure of social network in virtual community.