OCN: Effectively Utilizing Higher-Order Common Neighbors for Better Link Prediction

TL;DR

This paper introduces Orthogonal Common Neighbor (OCN), a novel link prediction method that reduces redundancy and over-smoothing among higher-order common neighbors, significantly improving prediction accuracy.

Contribution

The paper proposes orthogonalization and normalization techniques to enhance higher-order common neighbor features for better link prediction performance.

Findings

OCN outperforms baselines by 7.7% on benchmarks.

Orthogonalization reduces redundancy among CNs.

Normalization mitigates over-smoothing effects.

Abstract

Common Neighbors (CNs) and their higher-order variants are important pairwise features widely used in state-of-the-art link prediction methods. However, existing methods often struggle with the repetition across different orders of CNs and fail to fully leverage their potential. We identify that these limitations stem from two key issues: redundancy and over-smoothing in high-order common neighbors. To address these challenges, we design orthogonalization to eliminate redundancy between different-order CNs and normalization to mitigate over-smoothing. By combining these two techniques, we propose Orthogonal Common Neighbor (OCN), a novel approach that significantly outperforms the strongest baselines by an average of 7.7\% on popular link prediction benchmarks. A thorough theoretical analysis is provided to support our method. Ablation studies also verify the effectiveness of our orthogonalization and normalization techniques. Code is available at: https://github.com/qingpingmo/OCN.