Research And Application Of Influence Maximization In Social Networks

Social network analysis is a research hotspot in recent years.The influence maximization and information dissemination prediction are two important research directions of social network.The former aims to find k influential nodes in the social network,so that the diffusion range of their influence can be maximized.The latter is mainly based on the known information dissemination data and combines the influencing factors of information dissemination to predict the future information dissemination situation.In terms of influence maximization,the current research is divided into two directions: the improvement of greedy algorithms and the study of heuristic algorithms.Although the greedy algorithm has been optimized several times,the time complexity is still too high to be suitable for large social networks.However,most heuristic algorithms only consider the influence of nodes themselves,ignoring their environment,so there is also space for improvement.In terms of information dissemination prediction,the current research objects are mainly concentrated in a social application such as Weibo and Twitter.The algorithms for general social networks are often not ideal or the calculation methods are extremely complicated,which is not suitable for real-time prediction of information dissemination.This paper presents an algorithm for maximizing influence based on the shortest path.Firstly,the algorithm takes the sum of the shortest paths from a node to nearby finite nodes as a central index,which not only reduces the time complexity,but also considers the influence of the node in a local area.Secondly,the algorithm takes the maximum value of the network as the number of finite nodes,so that the central index can adapt to the scale and characteristics of different networks.Finally,the algorithm reduces the neighborhood of seed nodes to avoid the phenomenon of rich clubs.We experimentally evaluate our algorithm against the states of the art under several popular diffusion models,using several real social networks.This paper presents a prediction method of information transmission based on the main path network.Considering that in social networks,no matter where the source of information is,information will often spread to the nodes with greater influence and quickly spread to the whole network through them.Meanwhile nodes between seed nodes are more likely to be disseminated preferentially.In this paper,we first use the influence maximization algorithm to find the most influential node set,i.e.seed node set.Then we propose an evaluation criterion for the importance of paths between nodes.According to the evaluation criteria,we find the main paths between these nodes.Finally,we predict the information dissemination according to the main path network composed of the main path.