Research On Nodes And Relationship In Social Network

With the rapid development of online social networks, the way of life and thinking has been impacted and changed. In the invisible, people have become a member of various social networks. People publish their things of life and express their own story in some communica-tive social networks such as QQ and micro-blog. Not only can they express their life, but also can observe people and things around. The research on social network has important theoreti-cal and practical significance. For example, it has a definite object in view of recommend products or friends or to lead some opinion, if a targeted study on a certain user on social network has been conducted. The theme in this article is the nodes and the relationship be-tween nodes of a social network.The nodes and the edges stand for different practical meanings in different social net-work structure. For example, the nodes represent users and the edges the relationship of friends or forwards in the social network of Facebook. This paper takes different characteris-tics of different social networks into consideration, and the study spreads as the two aspects.First, the research on nodes in social network has been conducted. In social networks, the importance of every nodes is different, so as the influence. The method of how to measure the significance of nodes helps the users or the merchant find nodes having the greatest influence. The paper puts forward an algorithm to evaluate the influence of users in social networks. Compared with traditional methods, the paper thinks over the active of the users and the qual-ity of the micro-blogs. The experiments on both public data sets and real data show that the method is feasible, and it can reflect the real influence of users better compared with the tradi-tional methods.Second, the research on the relationship of nodes in social networks has been conducted. The study develops from the two aspects:the attributes of nodes and the topological structure of the network. After a conclusion to traditional methods on node proximity and topological structure, a method based on AdaBoost to predict unknown links has been raised. Experi-ments on public co-authorship data sets prove that the strong classifier based on AdaBoost has more accurate to improve the accuracy of the unknown links, compared with the traditional weak classifiers based on traditional node attribute and path.Last, induction and summary on social networks has been conducted, and future work is put forward. The study on social network can be developed from different points, and every algorithm has its own weakness. The research on social networks has a guiding role in actual application.