Scholarlink

This paper presents FactPromptKG—an initial approach that integrates structured knowledge from Knowledge Graphs into Large Language Models (LLMs) via prompt engineering, targeting the task of fact-checking. Without requiring model retraining, FactPromptKG follows a pipeline consisting of entity extraction, knowledge transformation, triple selection, and knowledge injection into prompts to support inference. Preliminary experiments on a small-scale FactCheck-GPT dataset demonstrate that FactPromptKG outperforms both Zero-shot prompting and Random Knowledge configurations. These results suggest the potential of practical strategies for knowledge incorporation in LLMs.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Towards a Fact-Checking Framework Based on Large Language Models with Structured Knowledge Integration

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Nguyen Huynh Huy Hoang,
  • Dang Duc Chinh,
  • Dinh Hung,
  • Vu Nguyen Hong,
  • Thien Khai Tran

摘要

This paper presents FactPromptKG—an initial approach that integrates structured knowledge from Knowledge Graphs into Large Language Models (LLMs) via prompt engineering, targeting the task of fact-checking. Without requiring model retraining, FactPromptKG follows a pipeline consisting of entity extraction, knowledge transformation, triple selection, and knowledge injection into prompts to support inference. Preliminary experiments on a small-scale FactCheck-GPT dataset demonstrate that FactPromptKG outperforms both Zero-shot prompting and Random Knowledge configurations. These results suggest the potential of practical strategies for knowledge incorporation in LLMs.