Explaining Temporal Graph Neural Networks via Feature-induced Information Flow

2026-06-25

AI Summary

This work addresses the limited interpretability of existing Event Temporal Graph Neural Networks (ETGNNs), which often fail to fully capture information flow from event-induced variables and struggle with long-range temporal dependencies. To overcome this, we propose an attribution method based on a Normalized Relevance Metric (NRM) that, for the first time, comprehensively traces the complete information pathways between event embeddings and induced variables. Our approach employs modular decomposition to accommodate complex ETGNN architectures, enabling cross-layer variable comparison and higher-order interaction analysis among events. This significantly enhances both the completeness and readability of model explanations. Empirical evaluations on datasets spanning epidemic tracking, social dynamics, and political events demonstrate that our method consistently outperforms state-of-the-art techniques in both explanation fidelity and human interpretability.

Abstract

Event-based Temporal Graph Neural Networks (ETGNNs) have demonstrated strong performance across a wide range of applications, including social network analysis, epidemic tracing, recommender systems, and political event forecasting. However, their increasing complexity poses significant challenges for explainability. Existing explanation methods focus only on a subset of the information flow within ETGNNs, typically tracing contributions from the event-related embeddings to the output. Consequently, they overlook the important pathways through event-induced variables, which mediate interactions between nodes and thereby play a central role in capturing long-range temporal dependencies. To overcome this limitation, we propose a novel attribution method that analyzes the \emph{entire} information flow through all event-associated variables. Our method is built upon the recent Normalized Relevance Measure (NRM) framework, which enables explicit quantification of information flow originating from event embeddings as well as information flow passing through event-induced variables. It also ensures comparability of latent variables across layers, and supports higher-order analysis of interactions between events. To handle the architectural complexity of ETGNNs, we extend the NRM framework with a modular decomposition procedure that facilitates the systematic construction of relevance structure for complex neural architectures. We evaluate our approach on two synthetic datasets for epidemic tracing and social dynamics, as well as a real-world dataset of political event networks. Our qualitative and quantitative experiments show that our method consistently outperforms existing explanation approaches while producing more human-interpretable explanations.

Problem

• Temporal Graph Neural Networks
• Explainability
• Information Flow
• Event-induced Variables
• Attribution

Innovation

• Temporal Graph Neural Networks
• Explainability
• Information Flow
• Event-induced Variables
• Normalized Relevance Measure