Research On The Impact Maximization Algorithm In Dynamic Complex Networks

In recent years,the rapid development of the social network provides abundant services for users to make fr-iends,exchange information,as an important medium for merchandising.Influence maximization is a problem of finding a small subset of influential users in social networks that occupies an important position in the field of social network analysis,and having widely practical application,for example making and advertising,so it has very high value in research and application.However,because the large number of users of social networks and complex network topology bring difficult to users selecting,and expansion and evolution of network will result more difficult.How to select influential users quickly and accurately in the ever-changing network is an urgent challenge.Information in social network have abundant types and users have different preferences which will affect the information spread in the network,and will also decide the selected seeds whether to achieve the best results in the specific topic.At the same time,the social networking site will give users some message in the rich form,so the social networking site itself is an implicit influencer.How to rational use the site's influence in the social network based topic-aware will become the new problem of influence maximization.In view of the above challenges and problems,this paper conduct research from the following two aspects:(1)Because the existing static algorithms can not adapt to the continuous evolution of social networks,resulting in network changes and needing to recalculate cause the time-consuming problem.In this paper,DIM algorithm is proposed based on the influence localization feature.The algorithm includes two stages:initial seed acquisition algorithm Init Seed and incremental update algorithm Inc-Seed.The initial seed acquisition algorithm can determine the influence path set and the initial seed,and prepare for the next stage.Incremental update algorithm Inc_Seed can quickly determine the influence range of the change,and only recalculate the influence of the node influenced by the change to reduce the time overhead.And this paper further propose two pruning strategies:influence value increment pruning strategies and degree pruning strategy for incremental update stage to further reduce the calculation,and design the Opt-DIM algorithm,which ensure the influence in the process,significantly improve the efficiency at the same time.(2)In view of the user's preference for the topic and the influence of the website,existing algorithms have done a lot of research on the topic-aware,but have not considered the influence of the website to influence spread.The website is considered as external factor,and put forward the influence maximization algorithm TIP based on topic-aware recommendation social network.Combined with the actual factors determine the influence probability between users and the probability from site to users,influence path as unit to calculate the influence of users.According to the activation,selected seed nodes adaptively to achieve the greatest comment influence of the seed and the site.And put forward two optimization strategies:influence optimization strategy and degree ranking optimization strategy to further improve the execution efficiency of the algorithm.