Research On Influence Maximization Of Social Networks

In this era of rapid development in all aspects,competition is everywhere.Social networks are no exception,with a lot of product information or different speeches,and there is competition for the dissemination of this information.Most of the current research on social network influence maximization is directed to the propagation of a certain piece of information,but in actual social networks,more information is spread.Therefore,research on the issue of maximizing influence based on competition is proposed.For the problem of maximizing competitive influence,the competitive linear threshold model was first proposed.The model considers the competition in information dissemination,and the independent cascading model can not well reflect the cumulative activation effect of the neighbor node of the node,and combines the cumulative characteristics of the linear threshold model.In this model,when two pieces of information are propagated,the set of seed nodes of one piece of information is known,and the set of seed nodes of the other piece of information is propagated in the network.Secondly,for the problem of maximizing competition impact,a DAG-based competition impact maximization algorithm is proposed.The algorithm combines the local accumulation characteristics of the competitive linear threshold model proposed earlier,and utilizes the fast impact calculation of local directed acyclic graph DAG.The algorithm uses each node in the network as the root node v to construct a local DAG,and excludes seed nodes with known information,and only considers the influence of nodes in the local DAG on node v.Then calculate the influence of all the nodes,and update the influence of the nodes in the process of selecting the seed nodes,sort according to the influence of the nodes,and select the top k nodes in sequence as the seed nodes.Finally,on two data sets,a comparative experiment is conducted to verify the reliability of the competitive linear threshold model for use in the subsequent comparative experiment based on the DAG-based competitive influence maximization algorithm,and then through four real network data Set,the seed nodes selected by the algorithm and the classic influence maximization algorithm are simulated and propagated in the competitive linear threshold model to obtain the propagation range and running time.Experiments show that the experiment has better performance in terms of propagation range and time efficiency.