A Study Of Influence Blocking Maximization Algorithm Based On Spontaneous Rebuttal Behavior

One of the reasons why online social networks have become an important platform for people to communicate and get news is that social networks can facilitate the rapid spread of information.However,this also provides conditions for the rapid spread of negative and false information.The existence of following and recommendation mechanisms in social network platforms has led to the possibility that users may spread negative information in social networks due to unintentional or special motives.Negative information has become a major threat in social networks,and how to minimize the spread of negative information is significant to maintain the civilization and health of social networks and the harmony and stability of society.By selecting some seed users in social networks to spread positive information and compete with negative information to block the spread of negative information has received a lot of attention from researchers.However,there are some users in real social networks who,after seeing negative information,will refute the negative information based on their own expertise and spread the refutation information to other users,and this phenomenon is called spontaneous rebuttal behavior of users in this paper.Questions about the role of users’spontaneous rebuttal behavior in weakening the impact of negative information and how to select seed users for spreading positive information while considering users’ rebuttal behavior have not yet attracted attention.This paper focuses on the problem of how to select seed users who spread positive information to minimize the influence of negative information when considering the existence of spontaneous rebuttal behavior of users in social networks.In this paper,two competitive independent cascade models considering users’ rebuttal behaviors are first proposed and the positive influence spread maximization problem is defined,and finally a randomized algorithm satisfying the 1-1/e-ε approximation is designed based on reverse influence sampling.Specifically,first,two competitive independent cascade models considering users’ rebuttal behaviors are proposed in this paper,and discrete-time and continuous-time models are further proposed for each model considering the delayed and asynchronous nature of information propagation in social networks.Second,the positive influence spread of the set is proposed in this paper to measure the ability of a given set to weaken the influence of negative information,and the positive influence spread maximization problem is defined as finding the optimal seed set that maximizes the positive influence spread.It is shown that the problem has NP-hard complexity in the model proposed in this paper.Third,for each model proposed in this paper,it is shown that the positive influence spread function all satisfy monotonic submodularity,and then a randomized algorithm satisfying the 1-1/e-ε approximation is proposed based on the idea of reverse influence sampling and the correctness and time complexity of the algorithm is proved.Fourth,the effectiveness and efficiency of the randomized algorithm proposed in this paper is evaluated in the experiment section for each model.The experiments on six data sets show that the output solution of the randomized algorithm based on reverse influence sampling proposed in this paper can effectively reduce the number of nodes affected by negative information compared with the benchmark algorithm,and the algorithm also runs more efficiently than the greedy algorithm based on Monte Carlo simulation.