Research On The Intluence Maximization Problem In Online Social Networks

Recently,with the rapid development of the online social platforms,people prefer to get and share information with each other through the online social platforms.Inspired by this phenomenon,"social marketing" finds opportunities.They select a small subset of the most influential users to endorse their products on the social network to maximize the number of people accepting the products,which is influence maximization problem.At present,the context problem has become a research topic and achieve some successful results.However,current work sometimes has slow running time or inaccurate influence spread.In addition,most of the research work is based on static social networks,ignoring the dynamic of real social networks,which may lead to poor scalability of the influence maximization algorithm.Therefore,to deal with the shortcomings of the context influence maximization problem,this paper provides novel research methods and conducts the following research:(1)The influence maximization algorithm based on the static social network.In order to deal with the unbalanced problem between the running time and influence spread,this paper proposes a static social network influence maximization algorithm based on community structure.Firstly,it utilizes the community detection algorithm for community division.And then,Then,the degree index is improved,and a new evaluation,propagation degree,is presented.In addition,combining propagation degree and K core to select the most influential nodes in the each community as candidate seeds.Meanwhile,the nodes that cross the most number of different community join to the candidate seeds.And finally,it uses the greedy algorithm to ensure the final optimal seed set.Experiments are performed on four different types of real data sets,which verify that our proposed algorithm can effectively balance the influence spread and running time.(2)The influence maximization algorithm based on the dynamic social network.On the real social network,the structure of social network is constantly changing and the relationship between users is also constantly updating.Considering the dynamic of the real social network,we propose a novel influence maximization problem based on the dynamic social network.Firstly,in order to describe the dynamic of social network,this paper starts with adding and deleting nodes,adding,deleting and reconnecting edges.Then,combining propagation degree and K core to select the most influential nodes in the each community as candidate seeds.Meanwhile,the nodes that cross the most number of different community join to the candidate seeds.In addition,as to a dynamically changing node,this article firstly matches it with a similar community,and then matches it to a similar node belonging to the context community,which is used to ensure whether this changing node is candidate node.And finally,utilizing the greedy algorithm to find the final seed set.Furthermore,this article has carried out experimental verification under the premise of conforming to the dynamics of the network.The experimental results show that this method performs better than these traditional heuristic algorithms in the aspect of influence spread and running time.In addition,comparing with the greedy algorithm,the similar the influence spread is obtained.In a conclusion,the algorithm has a better extension for the influence maximization problem of social networks.