The Research On Individuals With High Influence And Opposite Influence Maximization In Social Networks

At the beginning of 2020,the novel coronavirus pneumonia has attention to all walks of life and the world is shrouded in tension.In the pneumonia,keywords such as super-spreaders,rumor crackdown,contagious diffusion,and influence prediction were frequently going viral by various media,which made scholars in areas such as the spread of complex network diseases and social network computing spread more important.The influence computing in social networks is one of the important topics.This paper focuses on the algorithms of mining high-influential individuals and the problem of opposite influence maximization,as following:1)The computing of individuals which has a high influence.The existing central methods of high-influential individual research are difficult to determine the appropriate measurement level for individuals.If the measurement level covers the topological information extensively,the calculation cost of the measurement method will increase sharply;if the scope of the measurement level narrowly,the mutual influence of individuals and their next neighbors will be ignored,which directly leads to a large deviation between the measurement result and the actual capacity of influence.This paper according to the Three-Degree Influence Principle that is the statistical conclusion of the complex propagation phenomenon of real networks,three-level influence measurement(TIM)was designed.This method treats indirect neighbors with influence attenuation characteristics as a whole and the second level & the third level neighbors as cumulative calculations.The TIM has significantly improved dimensional accuracy compared with the existing methods and has outstanding performance in indicators such as influence consistency,influence ranking performance,discrimination of measurement value,and suitable for solving the problem of influence maximization.2)The research on the opposite influence maximization.The problem of na?ve influence maximization in social networks is unobstructed and non-intrusive.In actual social networks,the influence among users may form a variety of relationships such as complementary cooperation,oppositional exclusion,or dynamic games.Considering the form of opposite propagation of information,the heat diffusion model was expended to the multi-source heat diffusion model,a random rule for dealing with propagation conflicts was designed,and a pre-selected greedy approximate algorithm(PSGA)was put forward which based on the propagation mechanism of heat model.In the crawler social network data set and the open-source social network data set,simulation experiments were carried out to maximize the opposite influence maximization and rich-club,which verified the effectiveness of the algorithm.