The Research And Implement Of Tibetan Web Network Community Division

There are massive relationships in the Tibetan pages.By the basic view, these pages engender communities by some hyperlinks. Firstly, we introduce processing technology of Web links, and then analyze of a variety of complex network models. The sticking point on the establishment of model include that there are charset GB in these pages, so we carry out a detailed analysis to solve it and design the storage structure of pages. Secondly, we list the three models, for example, Karate club network、Dolphin network、Football network, and analyze them with Tibetan web network model. Finally, according to the known details of community, different division algorithms are suitable for different situations, so there isn’t a good evaluation system to integrate them, and then we put forward the concept of noise nodes by the comparative analysis of the background of models.The structure of the web hyperlink was directed graphs, but lots of community division algorithms were designed for undirected graphs. We discussed the generating process of various network models after we put the undirected graphs model to do some modifications to make it more suitable for directed graphs, and then we proposed a web community division algorithm based on hub-and-spoke topology. Because the GN division algorithm requires human intervention operation, we will stop it at the specified number of community, and then calculate their noise nodes for comparison. The algorithm works well, and we use a graphical way to show it.The innovations of this paper include the following aspects, for example, the Tibetan Web recognition using Java, partitioning algorithm based on the scale-free network directed graph, and use noise nodes for evaluating quality. More work in the future will consider the appropriate recursive rewriting and optimization of multi-core algorithm.