Research On Influence Maximization Method Based On Community Discovery And Independent Cascade Model

With the rapid development of Internet technology,the proportion of various online social media in the lives of netizens has increased year by year.Internet-based social media is subtly changing people's lives.Compared with daily life,users are often more inclined to Post your opinions and opinions on social media such as Weibo,WeChat,Twitter,etc.A large number of current events,user reviews,and user attention have formed a large amount of user information in the virtual social network,which has attracted many scholars to carry out research and derived many research directions.The issue of maximizing influence is one of them.The problem of maximizing influence is how to find some people with higher influence in social networks,and then use these people as the starting point for influence transmission,and try to spread the in-formation to as many people as possible.It has a wide range of applications in disease control,and cascading failure assessment of complex systems.The key to maximizing influence is how to find the small number of key users who can spread the influence to more users on social networks.How to select this small number of key users,the current research directions are mainly divided into two types:methods based on greedy strategies and methods based on heuristics.The former can obtain higher accuracy,but requires a lot of Monte-Carlo simulations,so it is less efficient.The latter mainly uses the node characteristics of social networks to quickly select nodes,but the accuracy is lost a lot.The main research contents and innovations of this article are:1.A new social network-based influence maximization method is proposed,which reason-ably uses the social characteristics of social networks and combines the advantages of greedy strategy and heuristic strategy method;2.The propagation model has been improved.In the improved independent cascading model,this article assigns different weights to the nodes based on their relative importance in the social network,and at the same time realizes the difference in the impact probability between nodes,so that the impact probability It changes with the number of interactions between nodes and the weight of nodes,which realizes the difference in the impact probability between nodes;3.For the divided communities,this paper proposes a method to select potentially high-impact nodes RSRW,which can quickly find high-impact nodes in the community and join the candidate node library;4.Based on four real data sets,the proposed method is compared with other excellent meth-ods.The experimental results show that the proposed method achieves the goal of ensuring the accuracy of the algorithm as much as possible while reducing the running time.