Community Mining Technology Research In Heterogeneous Network

Internet has become more and more popular with the progress of information technology. With the increasing number of Internet users, a large number of various types of web sites on-line, a significant increase in daily or updated web pages, and many social networks which are similar to reality are emerging gradually in the Internet. In social network analysis society is usually treated as a social network or graph, graph node is the object in society, the edges between nodes are the social relations between objects. This graph is often very complex, because society objects have complex properties, and the corresponding social relations between the objects are variant.Community mining is a research direction in social network analysis field. Mining natural, useful, relatively stable community in a network is valuable for network information collection and mining, information recommendations, and predicting the evolution of network. Not only that, the community mining technology also has great commercial value, community mining technology can find information hidden behind the webpages, divide the huge cyberspace into many web communities, community mining technology achieve the customer market segments, commercial companies could then implement more targeted marketing strategy.In most existing community mining strategy the social network is modeled as a homogeneous network, but in reality, there exists a large number of heterogeneous web objects, heterogeneous web relationships between objects in a social network, therefore, community mining strategy based on homogeneous network will inevitably be unable to tap the precise distinction between various types of web objects and object relationships. This paper aims at the current status of mining based on homogeneous networks, for presenting community structure more accurately and more comprehensively, presents a web community mining strategy based on a heterogeneous network by observing and studying the web community structure, and defines the content of this paper– opinion leaders community. This paper proposes a web forum opinion leaders community mining algorithm"FOLCM"(Forum Opinion Leaders Community Mining) for mining community of opinion leaders who could lead public opinion by learning web page authority concept from the classic web structure mining algorithm HITS and voting mechanism from PageRank, FOLCM algorithm could quantify the influences of forum users. We design and implement a web community mining system for mining opinion leaders group who have high ability of public opinion based on FOLCM algorithm. Experiments with a lot of real web data prove that the proposed community mining strategy is able to accurately mine the opinion leaders community in the Internet forums.