Research On Multi-factor Aware Advertising Recommendation For Online Social Networks

With the rapid development of computer science and mobile computing,online social networks,in which users share information,establish and maintain social connection with others,have become the most popular platform for people of all ages.Data generated by social platforms in different fields,such as the sociology,behavioral psychology and information science,is more valuable than platforms themselves.Business enterprises are paying more and more attention to those platforms,and want to enhance the effect of product promotion or advertising recommendation with the help of influential users in social platforms.Therefore,how to identify and then select the most influential users(the problem of influence maximization),as well as explore the mode of advertising dissemination in social networks are two challenges faced by researchers who work on social network-oriented advertising recommendation.In this dissertation,we conduct research on influence maximization and advertisement recommendation mechanism in social networks.The main work includes:(1)Taking cohesiveness and similarity into account,we proposed an efficient advertisement recommendation system to spread advertisement across social networks and the EPR algorithm to select seed nodes with high influence.Based on classical Page Rank algorithm,EPR can identify individuals who are highly central and then select seed nodes.The advertisement recommendation system leverage the “worth-of mouth” wisdom to reveal the potential relationship and design the opinion update rule to imitate information propagation in a social network,and it assumed that the user accept advertisements or not follows the Bayesian Probability Theory.The extensive experimental results on real datasets have verified the superiority of our proposed algorithm over benchmark algorithms in terms of infected users,the number of iterations and running time.(2)Considering the positive and negative connections in signed social networks,we propose a new framework to characterize the advertisement propagation process and the SPR algorithm to solve the problem of influence maximization.To achieve influence maximization in signed social networks,a novel SPR algorithm have been proposed to select the initial seed nodes by jointly considering their positive and negative connections.The information propagation process can simulate the advertisement spreading in signed networks,which models the dynamics of individuals' opinions and attitudes towards the advertisement based on recommendations from both positive and negative neighbors.The extensive experimental results confirm that our proposed SPR algorithm can effectively improve the rate and range of advertisement recommendation than benchmark algorithms in both synthetic and real datasets.