The Research Of Recommendation Based On Community And Node Role Division In The Social Network

Social networking services(SNS), based on the people's social relations, can provide users with information interaction, friends relationship management, self presentation and other social networking services. Besides, it can contact different people through the Internet and be described as a network structure built by people's relationship. Moreover, with the coming of information society, online social networking services have become one of the Internet's hottest products.Social networking sites, however, due to its huge volume of data, various data structure and complicate data relations, need to be improved on the efficiency of user access to information. As an important service, user recommendation has become a research hotspot in recent years. In addition, online social network is a virtual representation of interpersonal relationship in the real world, so its users have different roles. It is of great significance for the social networks analysis to summary the nodes which play the same or similar role, and divide them into different roles.Therefore, according to the comprehensive summary and analysis of the related research about social networks at home and abroad in recent years, the thesis mainly studied two aspects: user recommendation and node role analysis. The main work and innovations are as follows:1. This thesis presents a user recommendation scheme based on similar community and node role division. First of all, relying on the similarity of node pair, the thesis puts forward a method to calculate the similarity of community pairs from the two perspectives: community structure and attributes of users. Secondly, the thesis divides the user nodes into different roles so that they are at different level in order to recommend friend discriminatively.2. This thesis puts forward a role division scheme based on the social community and clustering according to the influence. In the scheme, nodes are divided into different roles based on the community unit, each node in its different communities may occupy a different place. The implementation of the scheme includes three steps, firstly, the thesis mine social community from the whole network. Secondly, according to the measurement to external and internal tightness of every node in the community, the thesis analyzes the users' social influence. And then the thesis takes the influence of the nodes as their attributes to gather them into clusters, so that they can be divided into different roles.Finally, the thesis selects Sina Weibo data to verify the proposed scheme. The experiment results show that the user recommendation scheme performs well and is suitable for user recommendation where there are communities in the social network and users can be divided into roles of different level. In addition, the node role division scheme can classify nodes in each community into distinct roles effectively.