Rethinking Link Prediction for Directed Graphs

2025-02-08

AI Summary

This paper addresses two critical challenges in directed link prediction: insufficient expressive power of graph embeddings and the absence of standardized evaluation benchmarks. To tackle these, we propose a unified expressivity analysis framework and introduce DirLinkBench—the first standardized, rigorous benchmark for directed link prediction. Theoretically, we establish that DiGAE is equivalent to a GCN operating on an associated bipartite graph, and identify the pivotal roles of dual-embedding representations and decoder design in predictive performance. Methodologically, we propose the Spectral-Domain Directed Graph Autoencoder (SDGAE), which achieves significant gains over existing state-of-the-art models on DirLinkBench. Comprehensive empirical evaluation reveals that mainstream models remain far from performance saturation under rigorous assessment. Our work pinpoints the core challenge as the joint optimization of directional modeling and structural awareness, and provides the community with a reproducible, extensible evaluation paradigm and principled insights for model design.

Technology Category

• Machine Learning: Graph-based Machine Learning
• Data Mining & Knowledge Management: Linked Open Data, Knowledge Graphs & KB Completion
• Knowledge Representation and Reasoning: Computational Complexity of Reasoning

Application Category

• Graph Algorithms and Modeling for the Web: Graph embeddings and representation learning for Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

Link prediction for directed graphs is a crucial task with diverse real-world applications. Recent advances in embedding methods and Graph Neural Networks (GNNs) have shown promising improvements. However, these methods often lack a thorough analysis of embedding expressiveness and suffer from ineffective benchmarks for a fair evaluation. In this paper, we propose a unified framework to assess the expressiveness of existing methods, highlighting the impact of dual embeddings and decoder design on performance. To address limitations in current experimental setups, we introduce DirLinkBench, a robust new benchmark with comprehensive coverage and standardized evaluation. The results show that current methods struggle to achieve strong performance on the new benchmark, while DiGAE outperforms others overall. We further revisit DiGAE theoretically, showing its graph convolution aligns with GCN on an undirected bipartite graph. Inspired by these insights, we propose a novel spectral directed graph auto-encoder SDGAE that achieves SOTA results on DirLinkBench. Finally, we analyze key factors influencing directed link prediction and highlight open challenges.

Problem

• Assess expressiveness of directed graph methods
• Introduce robust benchmark DirLinkBench
• Propose novel spectral directed graph auto-encoder

Innovation

• Unified framework for embedding expressiveness
• Introduction of DirLinkBench benchmark
• Novel spectral directed graph auto-encoder