GraphCheck: Breaking Long-Term Text Barriers with Extracted Knowledge Graph-Powered Fact-Checking

2025-02-23

AI Summary

Large language models (LLMs) frequently generate implicit factual errors in long-text generation—especially critical in high-stakes domains like medicine—while existing document-retrieval-based fact-checking methods struggle with multi-hop logical reasoning and incur prohibitive computational overhead. Method: We propose the first single-pass fact-checking framework that injects an extractive knowledge graph (KG) as soft prompts into LLMs. Our approach integrates automated KG construction, graph neural network (GNN)-based representation learning, and structured reasoning-enhanced soft prompt tuning to enable end-to-end multi-hop relational modeling. Contribution/Results: Evaluated on seven general-domain and medical benchmarks, our method achieves an average 6.1% improvement over strong baselines. It matches the performance of DeepSeek-V3 and OpenAI-o1 while using significantly fewer parameters, thereby transcending conventional pairwise discriminative paradigms.

Technology Category

• Natural Language Processing: Fact-Checking / Misinformation Detection (NLP Focus)
• Machine Learning: Large Multimodal Models (LMMs)
• Data Mining & Knowledge Management: Linked Open Data, Knowledge Graphs & KB Completion

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Search and Retrieval-Augmented AI: Search Tool Learning with LLM: Teaching LLMs to invoke search and make use of retrieved information

Abstract

Large language models (LLMs) are widely used, but they often generate subtle factual errors, especially in long-form text. These errors are fatal in some specialized domains such as medicine. Existing fact-checking with grounding documents methods face two main challenges: (1) they struggle to understand complex multihop relations in long documents, often overlooking subtle factual errors; (2) most specialized methods rely on pairwise comparisons, requiring multiple model calls, leading to high resource and computational costs. To address these challenges, we propose extbf{ extit{GraphCheck}}, a fact-checking framework that uses extracted knowledge graphs to enhance text representation. Graph Neural Networks further process these graphs as a soft prompt, enabling LLMs to incorporate structured knowledge more effectively. Enhanced with graph-based reasoning, GraphCheck captures multihop reasoning chains which are often overlooked by existing methods, enabling precise and efficient fact-checking in a single inference call. Experimental results on seven benchmarks spanning both general and medical domains demonstrate a 6.1% overall improvement over baseline models. Notably, GraphCheck outperforms existing specialized fact-checkers and achieves comparable performance with state-of-the-art LLMs, such as DeepSeek-V3 and OpenAI-o1, with significantly fewer parameters.

Problem

• Detect subtle factual errors in long-form text
• Improve multihop relation understanding in complex documents
• Reduce computational costs in specialized fact-checking

Innovation

• Knowledge graph-enhanced text representation
• Graph Neural Networks for soft prompts
• Single inference call for efficient fact-checking