Research On Information Source Detection Based On Two Layer Coupled Network

With the rapid development of the Internet,people have more and more experience the convenience and fun brought by the Internet.The network has been integrated into people's daily life.The network has become an important part of people's daily life.The fast and convenient update function of online social network can make users obtain information more directly,but the unsubstantiated rumors and false information mislead users through the network.If we can quickly and accurately detect the source of the information,control and minimize the spread of false information,we can greatly reduce its adverse impact on society.Therefore,the research on the source of rumors and false information on the network platform has the key social significance and application value.Based on the affiliation relationship between user network and text similarity network,this thesis constructs the user-text coupled network model,proposes ISD-BFS information source detection algorithm,and explores the problem of information source detection in real social network.Firstly,the user network and text similar network are generated by using the communication relationship between network users and the similar relationship between users publishing microblogs,and the user-text coupled network is constructed based on the affiliation relationship between the two networks.Then,based on the real event data on Sina Weibo platform,this paper analyzes the topological characteristics of user network,text similarity network and user-text coupled network model,verifies the small world characteristics of user-text coupled network,and makes a comparative analysis of network models from the perspective of information source detection.Finally,the advantages of user-text coupled network in information source detection are illustrated..Next,we generate the information spread area network by simulating the information spread on the social network,give the maximum likelihood estimation of the information source,define spread centrality of the node,and prove that the probability that the node is the information source is proportional to the value of its spread centrality.Finally,we propose the ISD-BFS information source detection algorithm.Based on the user-text coupled network,we use ISD-BFS algorithm to compare with other algorithms,and prove the feasibility of ISD-BFS algorithm to detect information source.The sensitivity of ISD-BFS algorithm is analyzed from two aspects of information spread rate and network scale of information spread area network.Finally,we find that DEC,DC and BC are not suitable for information source detection,and ISD-BFS algorithm performs better than UB,RG and DMP algorithm in most conditions.The performance of ISD-BFS algorithm detecting information source is getting better with the increase of information spread rate.