Community Perceiving And Mining Method On Social Network

In recent years, with the rapid development of social media, a large of useruse social network and mobile terminal to communicating, resource sharing andinformation disseminating. Compared with the era of Web1.0, Social Networkhas many characteristics, such as the strong of subjectivity, the fast ofinformation flows. The relationships can be set up on the Internet. By the end ofthe first quarter of2014, Facebook, Twitter, Sina Weibo users respectively brokethrough1.5billion,600million,800million, and exponential growth. Thus it canbe seen that social media get the attention of people more and more.According to the characteristics of social network on a large scale and largedata, the traditional community perceiving algorithm has some limitations, suchas the high time complexity, the inaccurate of structural division, considering thespecific attribute and behavioral expression of Social Network. In this paper,combined with the actual needs, we will carry out research into the communityperceiving. We propose a novel method for online community perceiving and themethod of mining influence nodes in the information dissemination. The resultsof this study can be applied to network public opinion, advertising benefitmaximization, personalized recommendation and other fields. And this study alsofound the unhealthy users and messages, ensure the security of network.The main contents of this dissertation are expressed as follows:(1) In theaspect of community perceiving, we analyze the characteristics of usersrelationship in Social Network, and research the mechanism of the communityformation. Finally, we propose a novel method for online community perceiving.(2) In the aspect of information transmission, this paper proposes the algorithm ofinformation dissemination tree generating, based on the above analysis results.This algorithm can carry on the back to the information dissemination process,and restore the process of event from initiative stage to decline stage. In addition, we will storage the information transmission to file by the structure of tree.(3) Inthe aspect of mining influence nodes, this paper introduce the three-layermapping model. Firstly, this three-tier analytical mechanism can upgradeinformation-level to relationship-level, and finally identifying the influentialnodes in relationship-level.(4) By merging above improved methods in threeaspects together, social network community analysis system named SNCAS isdesigned and implemented. SNCAS includes multi-source data acquiring module,specific community perceiving module, information dissemination treegenerating module, relationship network building and mining influence nodesmodule. Through experimental evaluation, SNCAS can perceive the specificcommunity and mining the influence nodes effectively on social network.