A comparative analysis of local network similarity measurements: application to author citation networks

TL;DR

This study compares 10 local network similarity measures for predicting future author citation links, revealing that Jaccard performs well while weighted extensions and neural networks offer limited improvements.

Contribution

It provides a comparative analysis of local similarity measures in temporal author citation networks, highlighting the effectiveness of Jaccard and limitations of weighted and neural approaches.

Findings

Jaccard coefficient is among the most effective similarity measures.

Weighted extensions of local measures do not significantly improve prediction.

Neural network combining multiple measures does not outperform individual measures.

Abstract

Understanding the evolution of paper and author citations is of paramount importance for the design of research policies and evaluation criteria that can promote and accelerate scientific discoveries. Recently many studies on the evolution of science have been conducted in the context of the emergent science of science field. While many studies have probed the link problem in citation networks, only a few works have analyzed the temporal nature of link prediction in author citation networks. In this study we compared the performance of 10 well-known local network similarity measurements to predict future links in author citations networks. Differently from traditional link prediction methods, the temporal nature of the predict links is relevant for our approach. Our analysis revealed interesting results. The Jaccard coefficient was found to be among the most relevant measurements. The preferential attachment measurement, conversely, displayed the worst performance. We also found that the extension of local measurements to their weighted version do not significantly improved the performance of predicting citations. Finally, we also found that a neural network approach summarizing the information from all 10 considered similarity measurements was not able to provide the highest prediction performance.