Temporal Link Prediction Algorithm Based On Local Random Walk

Link prediction is an important part of complex network research.Traditional static link prediction algorithm ignores network evolution over time.But temporal link prediction can use the information of historical network to make better prediction.Firstly,based on local random walk and the static link prediction method,this paper proposes a temporal random walk algorithm to solve the problem of temporal link prediction in un-weighted and undirected networks.The main idea is to use time and network topology information.The algorithm first computes the Markov probability transfer matrix at each time,then combines them into a transformation matrix,and applies the local random walk algorithm to obtain the final prediction result.The experimental results on two real networks show that our algorithm demonstrates better than other algorithms.Secondly,this paper introduces some weighted similarity-based indicators,and applies the modified temporal random walk algorithm to weighted network.Through experiments,this paper compares the performance of the static link prediction algorithm in weighted networks and un-weighted networks,and finds that the performance of some algorithms in weighted network is not as good as that in unweighted network.Therefore,this paper associates the weak link theory and analyzes this theory.In addition,through experiments this paper analyzes the performance of the modified temporal random walk algorithm in weighted network.The experimental results show that the accuracy of the modified algorithm in weighted network is slightly improved.