Research On Influence Maximization In Social Networks Based On Trust-relationship

Social networks are the basis of communication and influence spread,Social relationship is the basic element of social network.Via strong social relationship such as friends,colleagues and classmates,rapid cascade spread of influence could be achieved in local network;on the other hand,the certainty and ordering property of these relationship are also conducive to pre-evaluating the trajectory and range of influence diffusion.In the current social network modeling,most of the work focuses on the modeling and analysis of positive trust relationship,while to some extent ignores negative trust relationship,such as neglect,hostility,aversion and other common social relationships,which are also of research value.Inspired by real datasets of online social networks and different types of trust relationships in social networks,we propose a new model of influence diffusion based on trust relationship.According to the structure and propagation characteristics of the model,we study the relevant strategies to measure the ability of influence propagation,designs and then validates effective solutions to the influence maximization program.Specifically as follows:1)Based on the analysis of typical scenarios,the hypothesis of trust-relationship-based model is put forward,and the characteristics of trust relationship and influence diffusion are qualitatively analyzed with real social network datasets.Based on the above work,the influence diffusion model based on trust relationship(LT-TR model)is proposed.2)According to the LT-TR model,different types of trust relationship and substructures and their impact on the diffusion process are discussed.Upon deducing the theoretical calculation formula and practical application formula of node structure benefit,the influence evaluation strategy based on structure benefit is put forward.3)We study the existing typical influence maximization algorithms of social networks,summarize their design ideas and application scenarios,and propose three influence maximization algorithms based on different scenarios and influence evaluation strategies of previous studies.4)Based on the real dataset,the simulation experiments are designed,and the existing classical influence maximization algorithm and the proposed algorithm are compared horizontally.The performance of the algorithm in running time,coverage and other dimensions is verified,and the application value of the algorithm is proved.Experiments show that the influence diffusion model can better simulate the restricted propagation scenarios in the symbolic network.Besides,the greedy algorithm based on structural benefit has some comprehensive performance advantages over the classical heuristic and greedy influence maximization algorithm,and could be better applied to real social network scenarios.