Direct Evaluation of Chain-of-Thought in Multi-hop Reasoning with Knowledge Graphs

TL;DR

This paper evaluates the reasoning process of large language models in multi-hop question answering using knowledge graphs, revealing that models often produce correct answers through flawed reasoning despite understanding the task.

Contribution

It introduces a novel evaluation paradigm for assessing both reasoning correctness and answer accuracy of LLMs in multi-hop QA with knowledge graphs.

Findings

LLMs have sufficient reasoning knowledge.

Significant gap between answer accuracy and reasoning faithfulness.

Models often produce correct answers via incorrect reasoning.

Abstract

Large language models (LLMs) demonstrate strong reasoning abilities when prompted to generate chain-of-thought (CoT) explanations alongside answers. However, previous research on evaluating LLMs has solely focused on answer accuracy, neglecting the correctness of the generated CoT. In this paper, we delve deeper into the CoT reasoning capabilities of LLMs in multi-hop question answering by utilizing knowledge graphs (KGs). We propose a novel discriminative and generative CoT evaluation paradigm to assess LLMs' knowledge of reasoning and the accuracy of the generated CoT. Through experiments conducted on 5 different families of LLMs across 2 multi-hop question-answering datasets, we find that LLMs possess sufficient knowledge to perform reasoning. However, there exists a significant disparity between answer accuracy and faithfulness of the CoT reasoning generated by LLMs, indicating that they often arrive at correct answers through incorrect reasoning.