LAG: Logic-Augmented Generation from a Cartesian Perspective

TL;DR

LAG introduces a logic-based framework for large language models that decomposes complex questions into logical sub-questions, improving reasoning accuracy and reducing hallucinations by systematically organizing knowledge and reasoning steps.

Contribution

This paper proposes a novel logic-augmented generation paradigm that enhances LLM reasoning by question decomposition, logical dependency ordering, and structured knowledge integration.

Findings

LAG outperforms existing methods on four benchmarks.

LAG significantly reduces hallucinations in complex reasoning tasks.

LAG improves accuracy in knowledge-intensive tasks.

Abstract

Large language models (LLMs) have demonstrated remarkable capabilities across a wide range of tasks, yet exhibit critical limitations in knowledge-intensive tasks, often generating hallucinations when faced with questions requiring specialized expertise. While retrieval-augmented generation (RAG) mitigates this by integrating external knowledge, it struggles with complex reasoning scenarios due to its reliance on direct semantic retrieval and lack of structured logical organization. Inspired by Cartesian principles from \textit{Discours de la m\'ethode}, this paper introduces Logic-Augmented Generation (LAG), a novel paradigm that reframes knowledge augmentation through systematic question decomposition, atomic memory bank and logic-aware reasoning. Specifically, LAG first decomposes complex questions into atomic sub-questions ordered by logical dependencies. It then resolves these sequentially, using prior answers to guide context retrieval for subsequent sub-questions, ensuring stepwise grounding in the logical chain. Experiments on four benchmarks demonstrate that LAG significantly improves accuracy and reduces hallucination over existing methods.