Research On Discovering And Tracking Blog Communities

Blog is a platform for information transfer based on RSS. It is a kind of mutual intermediary for author and reader with the style of the daily record. It is a new mutual mode for information diffusion. Compared with conventional webs, links in the blog space are more abundant and conversations between bloggers are more frequent. It brings convenience for users to publish information and discuss on the Internet.Popularity of blogs and the amount of information in the blog space increase fast. So, it is now difficult for Internet users to find information they care about. On the other hand, the blog authors communicate with each other through reading and commenting, it makes to appear lots of hidden community, and how to find these communities and how to analyze the relationship between community members have become a sociological issues to be resolved.In this paper, we proposed this blog community discovery algorithm: event-driven blog community discovery algorithm, through analyzing the blog link feature and content feature. Communities were dynamic, because events were dynamic. Therefore, this paper tracked the evolution of the community based on the discovering of the community. That was the research of tracking community evolution, and proposed a community evolution tracking algorithm based on structure.In order to detect communities, this paper research the blog relation calculating algorithm base on event, and realize a bi-direction awareness calculation model. This model definite the relation between blogs from the aspect of sociological point and calculated the relation strength. Then we research the event-driven blog community discovery algorithm, and adopt a ranking based clustering algorithm, which selected wide influence of the blog as a seed, and then calculate the relation between the blogs and seeds, then do recursive expansion to form the communities of seeds as the center.Event was dynamic, enabling event-related communities were also dynamic and evolution, this paper tracked the evolution of the community based on the discovering of the community, and proposed a community evolution tracking algorithm based on structure. This algorithm calculated core members of the community based on the principle that the core members of the communities are relatively stable. If the intersection of core members is not empty, we determine the evolution relation between the two communities.Based on the above research results, this paper designs and implements an event-driven blog community discovery and evolution tracking experimental prototype system. This system includes data preparation module, blog relational matrix computing module, community discovery module and community evolution tracking module. This system provides a platform in order to carry out experiments and related research.