Large Language Models (LLMs) have shown potential in reasoning over structured environments, e.g., knowledge graph and table. Such tasks typically require multi-hop reasoning, i.e., match natural language utterance with instances in the environment. Previous methods leverage LLMs to incrementally build a reasoning path, where the LLMs either invoke tools or pick up schemas by step-by-step interacting with the environment. We propose Reasoning-Path-Editing (Readi), a novel framework where LLMs can efficiently and faithfully reason over structured environments. In Readi, LLMs initially generate a reasoning path given a query, and edit the path only when necessary. We instantiate the path on structured environments and provide feedback to edit the path if anything goes wrong. Experimental results on three KGQA and two TableQA datasets show the effectiveness of Readi, significantly surpassing previous LLM-based methods (by 9.1% Hit@1 on WebQSP, 12.4% on MQA-3H and 9.5% on WTQ), comparable with state-of-the-art fine-tuned methods (67% on CWQ and 74.7% on WebQSP) and substantially boosting the vanilla LLMs (by 14.9% on CWQ). Our code will be available on https://aka.ms/readi.
@article{arxiv.2403.08593,
title = {Call Me When Necessary: LLMs can Efficiently and Faithfully Reason over Structured Environments},
author = {Sitao Cheng and Ziyuan Zhuang and Yong Xu and Fangkai Yang and Chaoyun Zhang and Xiaoting Qin and Xiang Huang and Ling Chen and Qingwei Lin and Dongmei Zhang and Saravan Rajmohan and Qi Zhang},
journal= {arXiv preprint arXiv:2403.08593},
year = {2024}
}