Back to feed
AR
arXiv CS.AI
6/29/2026
ToE: A Hierarchical and Explainable Claim Verification Framework with Dynamic Multi-source Evidence Retrieval and Aggregation

ToE: A Hierarchical and Explainable Claim Verification Framework with Dynamic Multi-source Evidence Retrieval and Aggregation

Short summary

ToE proposes a hierarchical fact-checking framework using reinforcement learning to retrieve and aggregate evidence from multiple sources, addressing misinformation from adversarially crafted AI-generated content. The framework demonstrates 4-24% improvements over baselines with formal error bounds. Tested across multiple datasets and LLMs with strong performance on poisoned inputs.

  • •Hierarchical fact-checking framework combining RL-driven retrieval with evidence aggregation
  • •4-24% improvements over baselines with theoretical error bounds on optimal policy convergence
  • •Addresses misinformation from adversarially crafted AI-generated content (GEO poisoning)

Generated with AI, which can make mistakes.

#research-breakthrough#ai-tools#ai-agents
Read full article at arXiv CS.AI

Is this a good recommendation for you?

Comments

Failed to load comments. Please try again.

Explore more

Adversarial Social Epistemology for Assemblies of Humans and Large Language Models
arXiv CS.AI
AgentKGV: Agentic LLM-RAG Framework with Two-Stage Training for the Fact Verification of Knowledge Graphs
arXiv cs.CL
Interventional Grounding Audits: Black-Box Premise-Dependency Tests for LLM Chain-of-Thought via Predicate Substitution
arXiv CS.AI
Ontology-Amplified Distillation and Contextuality Auditing for Sovereign Enterprise Language Models: A Combined Proof-of-Mechanism and Negative-Results Method Study
arXiv CS.AI