Contrastive Token-level Explanations for Graph-based Rumour Detection

2025-02-05

AI Summary

Existing GNN-based rumor detection models suffer from poor interpretability, particularly in capturing cross-dimensional semantic dependencies within high-dimensional textual embeddings. To address this, we propose CT-LRP—a novel token-level explanation framework that uniquely integrates contrastive learning with Layer-wise Relevance Propagation (LRP), specifically designed for graph-structured text modeling. CT-LRP precisely disentangles inter-dimensional dependencies in embeddings and generates semantically aligned, high-fidelity token-level attributions. Crucially, it operates post-hoc without modifying the underlying GNN architecture, ensuring compatibility with mainstream models. Extensive experiments on three public rumor datasets demonstrate that CT-LRP significantly improves explanation fidelity (+12.7%) and human interpretability (31% improvement in expert evaluations). This work delivers a practical, deployable XAI solution for trustworthy rumor detection.

Technology Category

Natural Language Processing: Interpretability, Analysis, and Evaluation of NLP ModelsHumans and AI: Explainable AI (XAI) for Human UnderstandingMachine Learning: Transparent, Interpretable, Explainable ML

Application Category

Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for rankingGraph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphsSemantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

The widespread use of social media has accelerated the dissemination of information, but it has also facilitated the spread of harmful rumours, which can disrupt economies, influence political outcomes, and exacerbate public health crises, such as the COVID-19 pandemic. While Graph Neural Network (GNN)-based approaches have shown significant promise in automated rumour detection, they often lack transparency, making their predictions difficult to interpret. Existing graph explainability techniques fall short in addressing the unique challenges posed by the dependencies among feature dimensions in high-dimensional text embeddings used in GNN-based models. In this paper, we introduce Contrastive Token Layerwise Relevance Propagation (CT-LRP), a novel framework designed to enhance the explainability of GNN-based rumour detection. CT-LRP extends current graph explainability methods by providing token-level explanations that offer greater granularity and interpretability. We evaluate the effectiveness of CT-LRP across multiple GNN models trained on three publicly available rumour detection datasets, demonstrating that it consistently produces high-fidelity, meaningful explanations, paving the way for more robust and trustworthy rumour detection systems.

Problem

• Enhances transparency in GNN-based rumour detection
• Addresses dependencies in high-dimensional text embeddings
• Provides token-level explanations for better interpretability

Innovation

• CT-LRP enhances GNN explainability
• Token-level explanations improve granularity
• High-fidelity explanations for rumour detection