Research On The Method Of Social Network Construction

With the rapid development and widely popularity of computers and the Internet, the interaction between people no longer subject to time and space. People can exchange information through E-mail,instant message,BBS,blog,chat rooms and other means. The real world becomes more and more digitization and network. The social networks can not only entertain, but also create social value, it plays increasingly important role in business and social activities. Using what kind of techniques and methods to mine the social network is a significant research subject in computer field. To sum up, choosing to research the method of social network construction is of profound theoretical and practical significance. Based on related topices of domestic and international research condition, the study does the following work:We classify the web information, divide the social network into autonomous social network, cooperated social network and open social network in the light of the web information which constitute the social network.It proposes a social network construction scheme. Like E-mail, BBS, blog and other autonomous social networks, we build the social network according to the correspondence between friends (E-mail number, message number).Like online theses and other cooperated social networks; we build the social network according to the co-author or the citing author. Like the open social network which is composed of news, profile and so on, we build the social network according to the co-occurrence of names on web pages using a search engine.This paper researches the information extraction technology. For E-mail, we utilize three steps to achieve the extraction of information on the E-mail: extracting the header information, extracting the contents of the message body and analyses message content. For web documents, we extract the information of the web documents through crawling the web page, parsing the page, generating tree structure and getting the information. The experimental results show that the methods in the dissertation are correct and feasible, thus will create theoretical foundation in social network field.This thesis analyses the sample data information from E-mail,blog,post,structures the system of Chinese characteristics and confirms the social network nodes via double transductive support vector machines(DTSVM). Like using a search engine to obtain information, we employ automatic clustering method to determine the social network node.With the FDA (Force-Directed Algorithm) layout algorithm, we use Prefuse visualization tools to achieve social network visualization.Finally,in order to validate the accuracy and feasibility of the algorithm and method proposed in the thesis, the paper tries to construct social network based on E-mail,online thesis and search engine.The Experimental results indicate that this research is advanced and valuable in the field of mining social network.