A Multi-Agent Approach for Claim Verification from Tabular Data Documents

arXiv cs.CL / 4/21/2026

📰 NewsIdeas & Deep AnalysisModels & Research

共有:

Key Points

The paper introduces MACE, a multi-agent framework for verifying claims extracted from tabular data documents.
Instead of relying on complex pretraining or decomposition alone, MACE uses three specialized agents—Planner, Executor, and Verifier—to produce clear, interpretable verification traces.
Each agent runs in a zero-shot Chain-of-Thought setting, with the Planner outlining reasoning strategies, the Executor detailing computations, and the Verifier checking logic.
Experiments show MACE reaches state-of-the-art results on two datasets and matches top models on two others, while achieving 80–100% of best performance using smaller models (27–92B parameters vs 235B).

Abstract

We present a novel approach for claim verification from tabular data documents. Recent LLM-based approaches either employ complex pretraining/fine-tuning or decompose verification into subtasks, often lacking comprehensive explanations and generalizability. To address these limitations, we propose a Multi-Agentic framework for Claim verification (MACE) consisting of three specialized agents: Planner, Executor, and Verifier. Instead of elaborate finetuning, each agent employs a zero-shot Chain-of-Thought setup to perform its tasks. MACE produces interpretable verification traces, with the Planner generating explicit reasoning strategies, the Executor providing detailed computation steps, and the Verifier validating the logic. Experiments demonstrate that MACE achieves state-of-the-art (SOTA) performance on two datasets and performs on par with the best models on two others, while achieving 80--100\% of best performance with substantially smaller models: 27--92B parameters versus 235B. This combination of competitive performance, memory efficiency, and transparent reasoning highlights our framework's effectiveness.