Research On Wikipedia-based Social Network Analysis Technique

In recent years, the rapid development of internet makes up a huge knowledge system, which provides information and platforms for the study of social network analysis. At the same time, computer technology continues to improve and update, which provides a variety of tools and means. In this paper, we use Wikipedia as a research object to tap the similarity relationships between entities. Three different networks, network of technology-technology, network of technology scientists, and network of scientists-scientists are built and analyzed. This work is mainly reflected in the following respects.Firstly, this paper briefly analyzes and summarizes the most commonly used tools for social network analysis and methods of Wikipedia's data extraction. In using of these tools and methods, we extracted real people information from Wikipedia, and built the dictionaries of the network technology and scientists. Several related networks are also built for deeper analysis. Concentrating on the core members'finding in the scientists'network, we designed a new algorithm, which performs better to find the core subnets and core members.Secondly, we built a network of research fields in computer network to demonstrate their relations. By using of degree centrality analysis, betweenness centrality analysis and k-means cluster analysis, we obtained the "hot spots" and "core" scientists in the research field of computer network technology.Thirdly, we built the network of scientists, analyzed three aspects of their interrelations, including the betweenness centrality, k-core and core-periphery structure, and got the corresponding subnets. Then we designed a social network mining algorithm, which was used to obtain the core subnet networks and select the core members of the subnet. In this way, we could achieve the purpose of finding leaders of the network. We also compared the differences between the method of using social network mining algorithm to obtain core subnets and the way of using core subnets-edge structure analysis to get core groups.Fourthly, Meta-Network can be used to describe all the types of relations between various entities and expose the rich correlations and inherent rules of the objects systematically and completely. We built the meta-matrix of computer-network related Wikipedia, and selected the 2-mode networks that reflected the relations between computer network technology and scientists, and between scientists and their organizations for further study. We made social network analysis on these two networks by centrality computing and core identification, and draw some useful information such as the hot technology and core research group. We obtained the 2-mode matrix by calculating the above two matrixes, and avoided the raw data gathering to build the relation network between computer network technology and organizations. The analysis on the new matrix can lead to some useful conclusions such as which research group might be the leader in a certain field. Furthermore, we discussed the comprehensive application of meta-network analysis, and the examples are comparisons between the centrality attributes for all the important relation networks and synthetic query for interesting information.