Abstract: Aiming at the single group of communicators in the existing opinion dissemination model and the lack of research on the influence of opinion leaders with different attitudes,the S3IR opinion dissemination model with three different attitudes of positive,negative and general neutral communicators is proposed.Based on this model,mean field equations are established and the propagation dynamics are analyzed.Real social networks are established using real datasets from Face-book and LastFM Asia,and the Monte Carlo algorithm is used to simulate the opinion propagation process,and the feasibility of S3IR model is verified in two real social networks,a comparative study between the S3IR model and the SIR model is carried out.By controlling the number of opinion leaders and comparing the propagation process of public opinion in networks with different degrees of agglomeration,the effects of the number of opinion leaders and the degree of network agglomeration on the propagation of public opinion are ex-plored.By controlling the infection rate of positive and negative communicators influenced by opinion leaders,the effects of the infectious power of opinion leaders on the propagation of public opinion are explored.Experiments show that the S3IR model is feasible and effective,the greater the number of positive or negative opinion leaders,the stronger the infectious force,the greater the degree of social network agglomeration,the more likely to lead to an increase in the number of communicators with the same viewpoints,and accelerate the dissemination of public opinion.