Forecasting Networks Links with Laplace Characteristic and Geographical Information in Complex Networks

Abstract

Forecasting links in a network is a crucial task in various applications such as social networks, internet traffic management, and data mining. Many studies on forecasting links in social networks and on other networks have been conducted over the last decade. In this paper, we propose a novel method based on graph Laplacian eigenmaps for predicting the geographic location of nodes in complex networks. Our method utilizes the adjacency matrix of the network and generates a scoring matrix that captures the similarity between nodes in terms of their geographic location. By transforming the distance matrices into score matrices using exponential decay, we show that the method achieves consistently high performance across various real-world datasets, surpassing other state-of-the-art methods. Our experiments on real-world networks demonstrate that The LCG method proposed in this study exhibits consistently high performance across most of the evaluated datasets, with an average score of 0.95%, surpassing the other methods.