A Self-Explainable Heterogeneous GNN for Relational Deep Learning

2024-11-30

arXiv.org

AI Summary

Existing heterogeneous graph neural networks (HGNNs) struggle to efficiently and interpretably mine discriminative meta-paths from multi-table relational databases: conventional approaches rely on expert priors or exhaustive enumeration, suffering poor scalability; recent unsupervised methods erroneously assume class labels depend solely on meta-path *existence*, ignoring multiplicity and semantics. This paper proposes the first self-explaining HGNN framework supporting *multi-instance meta-path aggregation*. It integrates differentiable meta-path discovery, attention-driven importance scoring, and an aggregation-based semantic reasoning module—enabling node classification decisions to explicitly depend on the joint semantic aggregation of multiple meta-paths. Evaluated on synthetic and real-world datasets, our method significantly outperforms state-of-the-art baselines, achieving up to a 12.6% accuracy gain. Moreover, it automatically identifies highly discriminative meta-paths that faithfully reflect the model’s reasoning logic, enhancing both performance and interpretability.

Technology Category

• Search and Optimization: Metareasoning and Metaheuristics
• Data Mining & Knowledge Management: Graph Mining, Social Network Analysis & Community
• Knowledge Representation and Reasoning: Computational Complexity of Reasoning

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG

Abstract

Recently, significant attention has been given to the idea of viewing relational databases as heterogeneous graphs, enabling the application of graph neural network (GNN) technology for predictive tasks. However, existing GNN methods struggle with the complexity of the heterogeneous graphs induced by databases with numerous tables and relations. Traditional approaches either consider all possible relational meta-paths, thus failing to scale with the number of relations, or rely on domain experts to identify relevant meta-paths. A recent solution does manage to learn informative meta-paths without expert supervision, but assumes that a node's class depends solely on the existence of a meta-path occurrence. In this work, we present a self-explainable heterogeneous GNN for relational data, that supports models in which class membership depends on aggregate information obtained from multiple occurrences of a meta-path. Experimental results show that in the context of relational databases, our approach effectively identifies informative meta-paths that faithfully capture the model's reasoning mechanisms. It significantly outperforms existing methods in both synthetic and real-world scenario.

Problem

• addresses complexity in heterogeneous graphs
• learns meta-paths without expert supervision
• improves relational database predictive accuracy

Innovation

• Self-explainable heterogeneous GNN
• Aggregate meta-path information
• Automated informative meta-path identification