A Method To Evaluate Members' Influence In Social Networks

With the rapid development of communication technology,social networks become bigger and more and more complex,and increasing attentions are paid on them.As a branch of the research on social networks,the research of the evaluation for members' influence in social networks has a high application value.Most of the existing methods for evaluating the members' influence are based on graph theory.However,graph theory mainly studies the general properties of varieties of complex networks,which makes the methods for evaluating the members' influence based on graph theory unsuitable for social networks which have special significance.Therefore,it has important significance to evaluate members' influence combining with the properties of social networks and existing research experiences.Research content of this article is mainly divided into two aspects.First,analyze the special properties of social networks on members' influence evaluation.Second,put forward a method for evaluating members' influence combining with the special properties of social networks and existing research experiences.Firstly,taking advantage of the research experiences of social psychology,this paper analyzes the properties of social networks and members' influence evaluation,and finally finds that a main factor which influences the members' influence evaluation is the network dynamics,and then an index for evaluating members' influence in social networks is put forward.Then,based on the dynamic factors in social networks,a method for evaluating members' influence is put forward.For the problem existed in influence propagation methods that the outputs excessively depend on the inputs,the proposed index is applied in order to make the transition probability matrix exist limit distribution,and the limit distribution equals the stationary distribution.In addition,in view of the weighted graph,the directed graph and the connected graph,this paper gives out the corresponding member influence evaluation algorithm in order to make the algorithm applicable to different structures of social networks.Finally,contrast experiments are carried out on the proposed algorithm and some other classical influence evaluation algorithms,the results show that the proposed algorithm has great resolution for the members' influence in social networks,and it also has a good performance on finding influential members in social networks.