Accurate Link Prediction for Edge-Incomplete Graphs via PU Learning

TL;DR

This paper introduces PULL, a novel link prediction method for edge-incomplete graphs that leverages PU learning to improve accuracy by effectively handling missing links.

Contribution

The paper proposes PULL, a PU-learning-based approach that models latent variables for edges, outperforming existing methods in predicting missing links in incomplete graphs.

Findings

PULL outperforms baseline methods on five real-world datasets.

The approach effectively models unobserved links using latent variables.

Extensive experiments demonstrate improved link prediction accuracy.

Abstract

Given an edge-incomplete graph, how can we accurately find the missing links? The link prediction in edge-incomplete graphs aims to discover the missing relations between entities when their relationships are represented as a graph. Edge-incomplete graphs are prevalent in real-world due to practical limitations, such as not checking all users when adding friends in a social network. Addressing the problem is crucial for various tasks, including recommending friends in social networks and finding references in citation networks. However, previous approaches rely heavily on the given edge-incomplete (observed) graph, making it challenging to consider the missing (unobserved) links during training. In this paper, we propose PULL (PU-Learning-based Link predictor), an accurate link prediction method based on the positive-unlabeled (PU) learning. PULL treats the observed edges in the training graph as positive examples, and the unconnected node pairs as unlabeled ones. PULL effectively prevents the link predictor from overfitting to the observed graph by proposing latent variables for every edge, and leveraging the expected graph structure with respect to the variables. Extensive experiments on five real-world datasets show that PULL consistently outperforms the baselines for predicting links in edge-incomplete graphs.