Evaluating link prediction: New perspectives and recommendations

TL;DR

This paper critically examines how various factors influence link prediction performance, proposing a comprehensive evaluation framework and offering best practice recommendations for more rigorous and meaningful assessments in network science.

Contribution

It introduces a controlled experimental setup for evaluating link prediction methods considering multiple data and application factors, and provides practical guidelines for better evaluation practices.

Findings

Network type and problem specifics significantly affect LP performance.

Evaluation metrics and class imbalance impact early retrieval effectiveness.

Guidelines improve the rigor and relevance of LP method assessments.

Abstract

Link prediction (LP) is an important problem in network science and machine learning research. The state-of-the-art LP methods are usually evaluated in a uniform setup, ignoring several factors associated with the data and application specific needs. We identify a number of such factors, such as, network-type, problem-type, geodesic distance between the end nodes and its distribution over the classes, nature and applicability of LP methods, class imbalance and its impact on early retrieval, evaluation metric, etc., and present an experimental setup which allows us to evaluate LP methods in a rigorous and controlled manner. We perform extensive experiments with a variety of LP methods over real network datasets in this controlled setup, and gather valuable insights on the interactions of these factors with the performance of LP through an array of carefully designed hypotheses. Following the insights, we provide recommendations to be followed as best practice for evaluating LP methods.