| Today , the activity of network speech has reached an unprecedented level. Due to freedom and open of BBS, BBS has gradually become an important center of information dissemination and distribution, and an influential platform where Internet users express public opinion. The new problem and challenges is emerging when the rumor spreads on the BBS and the public opinion is guided. it is necessary to study the BBS network topology, feature analysis, the relationship between the groups, and Information dissemination. These areas were discussed and analyzed in this paper as follows.First, BBS information network was analyzed and researched by demonstrating in this paper. The BBS raw data used to analyzed came from a domestic well-known websites, which was composed of more than 4,000,000 records of opened and replied topics on the BBS from September 2003 to June 2005 .From the data source, using database technology and data mining techniques ,the empirical BBS network was built with relevant mined information. The network is a directed and weighted network. These nodes are composed of the people who open or replied topics in the network, these edges are composed of are the replied topics in the network. The characteristic of network analysis shows that: the network possesses scale-free and small-world characteristics, in other words, node degree distribution, node weight distribution and edge weight distribution have shown good power-law properties, and has a high-clustering and the small average geodesic length in this model. In addition, members of the network were classified with data mining methods of classification, conjunction and evolution model; and BBS network competition between groups and the dissemination of information and feedback on the BBS are discussed. It is found that those powerful groups who can attract back a lot of topics, frequently reply others'topics or others groups, and these groups whose are the most active members of BBS Network play a key role in the areas of disseminating information and steering public opinion.The second, BBS information network is theoretically analyzed. In order to study the in-depth theory, two BBS theoretical model of information network were presented, one is unweighted, the other is weighted. In the first model, mechanism of dynamic link which the new vertex of the network should have, and mechanism of competition among all network vertexes, are like to real-life BBS network, but the theoretical model can not reflect the weighted relation of real-life BBS network. In the weighted BBS theoretical model, the dynamic weighted mechanisms were introduced. Theoretical analysis and simulation shows that the node degree distribution and node weight distribution and edge weight distribution of the model, possessed the same power-law as the real-life BBS network; and also found that the weight parameters on the theoretical model have improved the performance of the whole network, the cluster coefficient and average path length will be consistent with real-life BBS network by setting the parameters. The study on the real-life BBS network can be more in-depth with these two theoretical model.The third, weighted theory of BBS network was further studied. To descript the characteristic of weight network in the unweighted clustering coefficient, have shown the limitation, because BBS is a weighted network. Based on the realistic need, a weighted coefficient for BBS network was presented, and mathematical expression was given. The weight of edges was thought about detailedly and across-the-board. Clustering coefficient for weighted BBS network have certain advantages by theoretical analysis and simulation proving. |
PDF Full Text Request