TEGRA: Text Encoding With Graph and Retrieval Augmentation for Misinformation Detection

TL;DR

TEG addresses misinformation detection by fusing an OpenIE graph with the text to create a hybrid representation that facilitates explicit knowledge integration. TEGRA extends TEG with retrieval-based external knowledge, building two enriched graphs and using a triple-selection mechanism to filter information. Across four datasets, TEG and especially TEGRA yield improvements over strong baselines, highlighting the value of structured knowledge grounding for detection tasks. The approach offers a principled, interpretable framework that can generalize to other knowledge-intensive problems beyond misinformation detection.

Abstract

Misinformation detection is a critical task that can benefit significantly from the integration of external knowledge, much like manual fact-checking. In this work, we propose a novel method for representing textual documents that facilitates the incorporation of information from a knowledge base. Our approach, Text Encoding with Graph (TEG), processes documents by extracting structured information in the form of a graph and encoding both the text and the graph for classification purposes. Through extensive experiments, we demonstrate that this hybrid representation enhances misinformation detection performance compared to using language models alone. Furthermore, we introduce TEGRA, an extension of our framework that integrates domain-specific knowledge, further enhancing classification accuracy in most cases.

Figures (2)

• Figure 1: The global processing architecture of TEG. Each component can be swapped with a similar component with the same function.
• Figure A.1: Datasets statistics regarding lengths, and graph sizes.