Research On Maximizing Information Diffusion In Ultra-large-scale Network

With the rapid development of the Internet and social networks,information dissemination has become more rapid,extensive,and complex.However,the spread of false information,viral marketing,and other problems also occurs on the internet.Therefore,research on the information transmission process in social networks is becoming increasingly important.It not only provides a decision-making basis for information control and disease prevention and control but also guides product marketing and trends.In the study of information dissemination,the influence maximization algorithm of propagation is one of the important topics of influence research.However,as networks grow larger,many networks have become large-scale,resulting in noticeable overlap of influence.The dissemination of information in such networks can lead to significant information redundancy,which causes a waste of influence and hinders effective marketing promotion of products.Therefore,this thesis presents two algorithms for maximizing influence in large-scale social networks,namely the Layered Voting(LV)and Local Isolated Centrality(LIC)algorithms.These algorithms are optimized to address the issue of influence overlap and have low time complexity,enabling them to quickly obtain results while ensuring the effective spread of influence.Additionally,the algorithms are integrated into distributed graph calculations,allowing for efficient implementation and demonstrating high execution efficiency and extensibility through experiments.In addition to the study of diffusion maximization,this thesis also models two common phenomena: external diffusion and incomplete immunity.It proposes the Incomplete Immune Propagation Model with External Diffusion(SIA),which is based on the SIR and SIS models.This model takes into account the actual situation of external diffusion and repeated propagation.To validate the model’s effectiveness,the thesis discusses diffusion simulations using large social network data and varying parameters.The experimental results indicate that external diffusion has a minimal effect on the propagation threshold,mainly promoting the acceleration of transmission.The probability of propagation is identified as the main factor affecting the spread of information.The research presented in this thesis includes algorithms for influence maximization in large-scale social networks,a new information diffusion model called SIA,and the algorithms to address the issue of influence overlap.These studies hold broad application prospects in solving information dissemination problems and combating false information.