Multi-Sourced, Multi-Agent Evidence Retrieval for Fact-Checking

Researchers propose WKGFC, a new AI system that uses knowledge graphs and multi-agent retrieval to improve fact-checking accuracy. The system addresses limitations of current methods that rely on textual similarity by implementing an automated Markov Decision Process with LLM agents to retrieve and verify evidence from multiple sources.

• →Current fact-checking methods struggle with generalization and rely heavily on textual similarity for evidence retrieval.
• →WKGFC uses authorized open knowledge graphs as core evidence resources for more structured fact verification.
• →The system implements an automatic Markov Decision Process where LLM agents decide retrieval actions based on claims and evidence.
• →Multi-hop semantic relations are better captured through knowledge graph integration compared to traditional methods.
• →The approach combines knowledge graph evidence with web content retrieval for comprehensive fact-checking.