Research On Influence Maximization Of Multi-Information Competition And Cooperation Across Social Networks

With the rapid development of the Internet and information technology,online social networks have become important mediums for people to share their lives and obtain information,and the main channels for enterprises to advertise their products.To promote commodities,merchants usually adopt the way of “viral marketing” which aims at selecting certain influential users for a free trial and making more people in the social networks accept the product through the word-of-mouth effect among users.In the context of the research on “viral marketing”,people abstract it as an influence maximization(IM)problem.The classical IM problem mainly focuses on the research of single information propagation in a single network.Nowadays netizens often have various social accounts and when users encounter interesting information,they also tend to forward them between different internet platforms at the same time.This thesis proposes the issue of influence maximization of multi-information competition and cooperation across social networks based on the traditional IM problem in combination with more realistic scenarios of multi-information propagation across social platforms.To address the problem of influence maximization of multi-information competition and cooperation across social networks,the main contents of this thesis are as follows.Firstly,improve the coupling method of social networks and proposes an across social network coupling model by adding cross-network connection based on the content similarity between overlapping users and their followers on another social platform which better characterizes the users' real lives.Then the writer proposes a cooperation and conformity based impact propagation model under competition conditions based on the classic independent cascade model in view of the real scenario of multi-information propagation.Finally,the thesis designs the seed node selection algorithm CCG according to the submodularity and the Three Degrees of Influence Rule.Furthermore,the CCH algorithm is proposed based on the view of reverse reachable set and node avoidance.This thesis uses web crawling to obtain the Twitter-Instagram data set used in the research,which includes users' tweets,pictures,and other information.The writer extracts the topology of across social networks from the Twitter-Instagram data set and carries out the experiments on the across social network coupling model and several algorithms.The experiments reveal that the proposed algorithms have good effectiveness and acceptable operating efficiency.Based on all the above research results,this thesis also designs and implements a prototype system of influence maximization of multi-information competition and cooperation across social networks.The designed prototype system can not only visually display the results of the thesis but also dynamically display the process of multi-information propagation in the coupling network according to the algorithm and number of seed nodes input by the user.