Research On Social Network Structure

In recent years, with the new business such as Web 2.0 putting into practice, social network is becoming more and more popular depending on its properties of authenticity and stability, and it is also playing a very important role in the Web developments. We can see that many social network sites have made tremendous achievements in recent years, such as Myspace which has become one of the America's most popular websites, some statistics show that its total network flow reaches to 4.46% of the total amount; Flickr, the composition of Yahoo's picture sharing service and social network, has already hold 4 million users and 200 million uploaded pictures; Facebook, as the America's second-largest social networking site, has reached a staggering amount of 15 billion dollars market value during the short 3 years since its foundation; and in China "xiaoneiwang", as the uppermost communicating platform of college students, has also hold 22 million users who registered with real information.Social network is a network platform of real interpersonal ties. As the online assistant communicating tool of the real-world, it has a great difference with the traditional virtual network. The traditional virtual-network, whose structure is relatively disordered, is divorced from the social reality; it is also difficult to make academic analyses on it. On the other hand, social network, as a subset of the network of real-world interpersonal ties, possess some similar properties of interpersonal ties. So I think it is feasible to do some study on social network, and there will be significant academic-import to promote the future development of social network.This paper takes the social network as the platform to make in-depth study and analysis about the interpersonal ties which is based on the graph theory of computer science. Based on this we bring forward a new algorithm of community-identification which is founded on the theory of satellite graph structure. Our algorithm makes full use of the structure property that there exits a great amount of satellite sub-graphs on the network to improve the efficiency of current identification-algorithms. Otherwise, this paper also brings a new idea of user-influence evaluation system named UserRank in the field of social network for the first time. The UserRank system admits that the interpersonal structure of the social network determines the user-influence. By analyzing and calculating of the interpersonal structure, we can evaluate the user-influence properly. The UserRank model is based on the famous theory PageRank, and has made some necessary improvements according to the special structure properties of social network. We also made the model verification on the "xiaoneiwang" platform through the calculation of all Zhejiang University users' influences. The result shows that the system has strong accuracy and feasibility.