Research On The Maximization Of Positive Influence Of Signed Social Networks

In social networks,users play the role of not only recipients,but also producers and communicators of information,which allows information to spread quickly on online social networks,affecting many people in a short period of time.Nowadays,online social communication is becoming part of people's life increasingly with the increasing number of online social network users.In the analysis of social network research,the Influence Maximization(IM)problem is one of the hot research topics.The IM problem of social networks,which has important research significance in real life,is usually applied to viral marketing.The IM problem is to find the most influential set of user nodes in the social network.These node sets allow information to be propagated in a model with the maximum range of influence.At present,the classical propagation models include independent cascade model and linear threshold model.It has been confirmed that propagation under these propagation models is an NP-Hard problem.However,the social networks studied at present are basically unsigned networks,in which users' attitudes are not taken into account.Therefore,the spread of information in unsigned networks lead to the impact maximization overestimated easily.In recent years,researchers have begun to focus on the issue of Positive Influence Maximization(PIM)for information dissemination in signed networks.Aiming at the PIM problem,this paper studies the propagation model optimization and application algorithm.In terms of model,this paper proposes that the existing ICP model can be further improved and a new model is presented.In terms of algorithm,this paper takes the advantages of intelligent algorithms and proposes new algorithms to try to solve the PIM problem.Finally,the advantages of the proposed algorithm are proved by experiments.The detailed contents of the research are as follows:1.Through the in-depth analysis of the propagation process of the Polar Independent Cascading Model(IC-P),it is considered that the user does not have to activate his neighbor after being activated.In order to reflect this process,this paper adds the user's willingness to spread on the polarity independent cascade model and proposes the AIC-P model.It is then proven that the model has monotonicity and submodulus.2.The application algorithms to solve the PIM problem are mainly greedy and heuristic,but both of them have shortcomings.The advantage of greedy algorithm lies in its high accuracy,but the running time is too long.Heuristic algorithm,on the other hand,has high efficiency but less accuracy.By carefully analyzing the existing research work and shortcomings,this paper will introduce the differential evolution(DE)algorithm and propose the DE algorithm(PWDE)based on the node search strategy of propaganda willingness to solve the PIM problem.3.Through the simulation experiments on two real-world network datasets of different scales,It is proven that the advantage of PWDE algorithm is several times shorter than the greedy algorithm in terms of running time,and it can guarantee the proximity to greedy algorithm in terms of precision.