Directed network link prediction method based on linear programming

Abstract

The invention belongs to the technical field of complex network link prediction, and particularly relates to a directed network link prediction method based on linear programming, which comprises thefollowing steps of: firstly, introducing adjustable parameters, distinguishing influence weights of three types of neighbors, and then taking the contribution degree of the influence weights to the formation of connected edges as an unknown quantity; an optimization function is established through structural analysis of the directed network, a linear programming problem about a contribution degreematrix is constructed, an optimal solution of the contribution degree matrix is solved, and finally a link prediction index is constructed in combination with the optimal solution and used for directed network link prediction. According to the invention, the specific local structure of the directed network is considered, the influence of three types of neighbor nodes on the connection edge formation is distinguished by using adjustable parameters, and the optimal contribution matrix of the neighbor nodes is solved by using a linear programming method, so that compared with the traditional method, the method is more consistent with network structure characteristics, and the result is more universal; and the method has robust prediction performance in various types of networks.

Description

technical field [0001] The invention belongs to the technical field of complex network link prediction, and in particular relates to a directed network link prediction method based on linear programming. Background technique [0002] As the basic research object of network science, complex network can be used to abstract modeling of various complex systems and nonlinear processes, which brings great convenience to the study of its internal structure mechanism and dynamic characteristics. As a basic problem in complex network research, link prediction aims to predict unknown or future links in the network by using observed network structure, attributes and other information. Its essence is a large-scale graph data mining problem, and it has shown great application value in many fields such as recommendation systems, transportation planning, and biological research. [0003] Since it was first proposed in 2003, link prediction has been researched for nearly 20 years, and rela...

AI Technical Summary

Problems solved by technology

Such methods are based on a priori assumptions about the network structure and are not universally applicable to different types of directed networks.

Images