Research Of The User Influence Evaluation Algorithm Based On Github Social Network

Along with the development of Internet technology, online social networks are growing rapidly and becoming main platforms for people making friends and sharing information on the Internet. On social networks, users can get to know each other, share and disseminate a variety of information. Massive information generated by huge amount of users on online social networks has enormous commercial value aim research value. The influence and information resources each user owns directly shows the tremendous commercial value and spread of potential itself. And finding the influential nodes is of great theoretical significance and application value, because it's important for information propagation, productrecommendation and public opinion analysis. So, how to measure the actual influence of different users effectively, and then tap the potential value in it has become an urgent problem.Firstly, the paper introduces the most well-known features of the social networks, such as rule of 150, six degrees of separation and power law distribution. Second, the paper introduces classical and commonly used ones: degree centrality, betweenness centrality and closeness centrality. These centrality measures capture the importance of nodes in different perspectives. Then the paper analyzes the HITS algorithm based on links analysis. After the analysis we draw some shortcomings of the HITS algorithm when it was applied to social network influence evaluation in detail. As in real life, user not only affected people that it is connected to, but also affected people that it is not connected to. And when evaluating user influence, the HITS algorithm only considers the user that it is connected to, but ignores the user that it is not connected to. Another one was it neglects the reinforcement relationship among different users and the different types of users, which leads to unreasonable results easily.Taken all into consideration, the shortcomings of the HITS algorithm are improved in this paper. The first one is assigning different right-values to the Authority property and the Hub property based on the user's total properties, instead of treating them equally. Another one is, the influence of user is divided into two parts to consider. Therefore, the paper mainly proposed an improved method based on the HITS algorithm and users' properties to evaluate the influence of users in social networks. Finally, validity and rationality of the improved method has been verified by the actual users' data in Github social networks. The paper takes Github website for instance, the paper evaluates the influence of users by using the two algorithms and identified that the improved methods is better than the HITS algorithm.