Measurement And Analysis Of Social Network Structure

In recent years, with the development of Internet technologies, especially mobile Internet popularization, social network as a more real and stable dating patterns becoming more and more popular in community. In particular, Twitter, facebook, SINA weibo represented such an explosion in online social network development. SINA weibo, official figures showed, ending March2013, registered users of Sina weibo has more than530million.Social networks as a online expansion of real relationship in real life, therefore have a general social network features, like relative order and structure of certain laws. Based on this, we think that is feasible to study the structure of the social network, also is significant for promoting the further development of social networks.Based on complex network theory the knowledge of graph theory, the paper give an in-depth study on the structure of social network. First a stable, high-performance data crawl platform is built. On the platform, crawler module is designed using parallel computing framework Hadoop as a core technology, and a weighted breadth-first algorithm is proposed in order to achieve a random sample. In the distributed storage modules, consistent hashing algorithm is used as a data device mapping algorithm. Separation technology of read and write to the database, as well as the replica set technology are using to improve the stability and responsiveness of the database cluster. Secondly, we do a structural analysis between original Sina weibo network and the processed network by removing human intervention. Degree distribution which is estimated by Maximum Likelihood method and Kolmogorov-Smirnov(KS) test is used to analysis network of no scale characteristics. Average shortest path which is calculated by parallel dijkstra algorithm based on Hadoop is used to analysis network of small world characteristics. Also reciprocity is computed to network of symmetric, Pearson coefficient is used to measured network assortativity. Finally, Joint degree distribution is calculated to analysis the correlation between network in degree distribution and out degree distribution.Through analysis of SINA weibo networks, we find that real networks is more of an information dissemination, communication platform. Therefore the network is more like classic social networking structurally after removing human intervention...