Statistics and Its Interface

Volume 15 (2022)

Number 3

Local neighborhood-based approach of link prediction in networks

Pages: 323 – 334

DOI: https://dx.doi.org/10.4310/21-SII690

Authors

Chunning Wang (School of Mathematics and Statistics, Lanzhou University, Lanzhou, China; and School of Statistics, Lanzhou University of Finance and Economics, Lanzhou, China)

Bingyi Jing (Department of Mathematics, Hong Kong University of Science and Technology, Hong Kong)

Abstract

Network structure has been widely studied in recent decades. One particular usage of the network is to represent the relationship among nodes. Therefore, link prediction plays a crucial role in network analysis. A key issue of link prediction is to estimate the likelihood of potential links between nodes in the network. However, the complex network structure makes such estimation very challenging. In this paper, we propose a link prediction method based on nodes’ local neighborhood (LN), which constructs a local neighborhood for each node and calculates the likelihood of connection between nodes based on their neighbors. Further, we extend the LN method to solve the link prediction problems in a network with node covariates and community structure. Experimental studies on synthetic and real networks demonstrate that the performance of our methods is competitive.

Keywords

Network analysis, Link prediction, Local neighbourhood, Assortative mixing, Disassortative mixing .

2010 Mathematics Subject Classification

Primary 62-07. Secondary 91D30.

This project was supported by National Natural Science Foundation of China (no. 11971214), by the Natural Science Foundation of Gansu Province (no. 20JR5RA204), by the Innovation Ability Promotion Project of Colleges and Universities in Gansu Province (no. 2020A-059), and by the Natural Science Foundation of the Anhui Higher Education Institutions of China (no. KJ2020A0023).

Received 18 August 2020

Accepted 1 July 2021

Published 14 February 2022