Resarch On Node Influence Measurement And Influence Maximization In Social Networks

Social network continues to thrive and has become an indispensable part of people's lives in the recent years.It should now be considered as an important platform for people to obtain and exchange information.User influence and influence propagation are the prominent features of social networks.Thus,social influence research in social network has an crucial role in information dissemination,personalized recommendation,product marketing,expert discovery,rumor containment and other fields.Social network influence research is carried out in two aspects: Node influence measurement and influence maximization.The main work is as follows.Firstly,this paper introduces the research for the node influence measure and the influence maximization,describes the related concepts of social network,then describes the social influence issues.Secondly,in terms of social network node influence measurement,only one evaluation feature is considered for the existing measurement methods,we propose new suitable method for the Weibo-like social network.This method comprehensively considered the social network individual attribute characteristics and network topology features of Weibo users.Furthermore,multi-attribute decision-making method is utilized for better measurement accuracy of user node influence.Then,social influence maximization is studied by analyzing the characteristics of the existing influence propagation model,we propose an improved maximization algorithm based on the known issues of traditional greedy and hybrid greedy algorithms.In the heuristic stage,the thresholds of two-level neighbors and their own nodes to optimize the selection of seed nodes are considered.I the greedy stage,simple optimization strategy is proposed.Therefore,the problems of influence maximization can be solved in a more accurate and efficient way.Finally,the coding is implemented.The node influence measurement method and the maximization algorithm proposed are respectively tested on the real social network dataset.Meanwhile,the proposed algorithm is compared with the existing algorithms to verify the effectiveness.