HGOT: Hierarchical Graph of Thoughts for Retrieval-Augmented In-Context Learning in Factuality Evaluation

TL;DR

HGOT introduces a hierarchical graph approach that improves retrieval and factuality in large language models by structured reasoning, citation-based scoring, and enhanced answer selection, outperforming existing methods in multiple benchmarks.

Contribution

The paper presents HGOT, a novel hierarchical graph framework that leverages LLMs' planning abilities and citation metrics to improve factuality and retrieval in in-context learning.

Findings

HGOT outperforms in FEVER by up to 7%.

HGOT matches top models in Open-SQuAD and HotPotQA.

The approach enhances answer credibility through citation-aware scoring.

Abstract

With the widespread adoption of large language models (LLMs) in numerous applications, the challenge of factuality and the propensity for hallucinations has emerged as a significant concern. To address this issue, particularly in retrieval-augmented in-context learning, we introduce the hierarchical graph of thoughts (HGOT), a structured, multi-layered graph approach designed to enhance the retrieval of pertinent passages during in-context learning. The framework utilizes the emergent planning capabilities of LLMs, employing the divide-and-conquer strategy to break down complex queries into manageable sub-queries. It refines self-consistency majority voting for answer selection, which incorporates the recently proposed citation recall and precision metrics to assess the quality of thoughts, linking an answer's credibility intrinsically to the thought's quality. This methodology introduces a weighted system in majority voting, prioritizing answers based on the citation quality of their thoughts. Additionally, we propose a scoring mechanism for evaluating retrieved passages, considering factors such as citation frequency and quality, self-consistency confidence, and the retrieval module's ranking. Experiments indicate that HGOT excels as a versatile approach, outperforming competing models in FEVER by up to $7\%$ and matching leading models such as Retrieve-then-Read in Open-SQuAD, and DSP in HotPotQA, demonstrating its efficacy in enhancing LLMs' factuality.