Research On The Method Of Influence Maximization In Social Networks Oriented To Partner Marketing Model

The development of the Internet has led to the rise of social networks,and the research on the impact of social networks has attracted extensive attention from researchers in management,sociology,economics and computer science.Unlike traditional networks,users are responsible for both producing and receiving information and disseminating information in social networks,and the influence of users plays a decisive role in the diffusion of information.In view of this,social network impact analysis has become one of the key issues in social network analysis,and has been applied in many aspects,such as recommendation system,link prediction,community discovery,public opinion control,emergency detection and so on.As a kind of network marketing model,the partner marketing model searches for partners(seed nodes)through online methods and promotes products through partners.This model needs to measure the importance of user nodes on social networks,rank them in order to find the most influential partners,and then maximize the scope of influence under the spread of partners.In the study of the influence ordering problem,the existing algorithms often consider the influencing factors to be too single,and can not measure the importance of the user nodes well,so that the user nodes with greater influence can not be distinguished well.To this end,this paper first proposes an improved node influence neighboring algorithm BNR(Based on Neighbor Relations).Based on the comprehensive consideration of the fit algorithm and the K-nuclear decomposition algorithm,the algorithm uses the entropy weight method to fuse the fit value.The degree value and the k-shell value get the centrality index of the neighboring intimacy,which makes the node influence evaluation more accurate and improves the accuracy of the algorithm.In addition,on the issue of influence maximization of influence,the nodes calculated by the existing algorithms have a small influence range and complicated operations.To this end,this paper proposes an improved influence optimization algorithm TSO(Two Stage Optimization),which is divided into two stages.In the heuristic stage,based on the research of BNR algorithm,the characteristics of dynamic threshold are further integrated,and the evaluation criterion of activation potential is obtained,which can select the seed node with the most influential potential more accurately;the greedy stage uses the mental decomposition network of dynamic programming.Simplify the algorithm to more efficiently select the most influential seed nodes.Finally,by comparing with the original algorithm on the real network dataset,it is verified that the improved BNR algorithm can sort the influence nodes better than the original algorithm.The improved TSO algorithm is better than the original algorithm.It has a better communication effect and a larger range of influence.