The Research Of Link Prediction Algorithm Based On Spectral Analysis Method

Many of the real systems in life can be regarded as a complex network system,link prediction as a complex network is very important research topic,its important role is to study the evolution of these networks or development direction,for example,how to know the interaction between different proteins in the protein network,and whether the two social network nodes are friends.Therefore,the link prediction research has important practical significance.Link prediction is one of the most important research directions of data mining in complex networks.The research of link prediction has been widely concerned.Most of the algorithms are used to analyze the topology of the network,and to judge whether there is any connection between nodes,by calculating the similarity of nodes.However,in the recent years,for most link prediction,the new social network link prediction algorithm needs to know the additional attribute information of the nodes itself,but it is very difficult to get this additional information,and to get the extra information of the network.In this thesis,a link prediction algorithm is proposed for link prediction.The spectral analysis is introduced into the algorithm.The eigenvalues and eigenvectors of the Laplacian matrix are used to map the nodes to the two-dimensional space.The similarity calculation of the nodes is directly transformed into the classification prediction of the edges.In this thesis,a test data set and six real network data sets are used to prove the feasibility of the algorithm,and the results of the existing algorithms are compared with previous algorithms to prove the effectiveness of the algorithm.On the basis of above results,the thesis introduces a new similarity calculation method in order to improve the prediction effect of the algorithm and consider the influence of the community structure on the link prediction results(the connection probability of the nodes in the same community is greater than that between the societies),that is based on the angular distance spectrum analysis method.It is proved that a better community structure can be detected by using the similarity calculation method.Finally,the improved algorithm that introduces the angular distance is verified by the same test data set and six real network datasets which proves that the improved algorithm can improve the link prediction effect.On the basis of improving the forecasting effect,it can be found that the ability to improve the ability to predict the structure of the community.