Social Network Relationship Prediction Method Based On Cognitive Model

In recent years,with the development of Facebook?Weibo?Renren and other social network platforms,social networks had gradually developed from simple social networks of dozens nodes into complex social networks of tens of thousands nodes.The researchers gradually found that the traditional data sampling method unable to meet the development needs of the complex social network structures.However,in social networks,the topology prediction was easily affected by response time,packet loss,individual behavior and other factors.To resolve these problems,this paper had presented two new methods of topology prediction based on the cognitive model:BTCS(Binary chop Threshold based Cognitive Social structures)prediction algorithm and CCS(Community Cognitive Social structures)prediction algorithm.And this paper analyzed the prediction accuracy and predicted time with two methods.The main works are as follows:Firstly,the thesis introduced the relationship prediction technology of social networks and described the structure of social relations cognitive model,then applied the social cognitive model into the relationship prediction technology in social networks.Secondly,a new social relationship prediction algorithm based on cognitive model,namely BTCS,was developed to solve the low accuracy problem by virtue of traditional prediction method under low sampling rate.On basis of the cognitive ability of single node on the whole network,this new method could acquire all the information on all nodes in the network creatively by sampling partly and randomly.Thus,the prediction of social relationship among all the nodes under low sampling would be obtained by extrapolation from that of sampled nodes.In order to analyze the performance of this algorithm,several control experiments in different networks were conducted to compare the difference between this method and traditional methods.Results showed that the new algorithm improved predicting accuracy and reduced prediction time under low sampling rate.Finally,according to nodes within the same community of higher cognition,the thesis proposed a new predicted method in community,namely CCS to resolve the problem on the poor stability and slow predicted time in random sampling prediction algorithm.It combined cognitive characteristics and community property,and applied to the relationship prediction.The algorithm implements were a community-based relationship prediction.Compared with the other random sampling prediction algorithm,simulation results demonstrated that the proposed algorithm improves the stability and reduces the time overhead.In conclusion,this paper puts forward a new method to improve the update efficiency of Prediction,which has a certain theoretical significance and application value.