Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Self-Reflective Planning with Knowledge Graphs: Enhancing LLM Reasoning Reliability for Question Answering

Jiajun Zhu, Ye Liu, Meikai Bao, Kai Zhang, Yanghai Zhang, Qi Liu

TL;DR

This work tackles LLM hallucinations in knowledge-grounded question answering by introducing Self-Reflective Planning (SRP), a framework that integrates LLMs with knowledge graphs through reference-guided planning, iterative retrieval, and self-reflection. SRP comprises four modules—Reference Searching, Path Planning (relation check and path generation), Knowledge Retrieval, and Reflection and Reasoning (sequence judge, path edit, and answering)—to generate and refine reliable reasoning paths and triplet sequences. Empirical results on WebQSP, CWQ, and GrailQA show SRP achieving state-of-the-art performance on two benchmarks and strong results on the third, with significantly higher reliable answering rates than baselines. The approach demonstrates that structured references and iterative self-correction can substantially improve factual grounding and reasoning reliability in KGQA without relying on fine-tuning. Limitations include higher computational cost due to repeated reasoning cycles and sensitivity to reference quality, suggesting directions for efficiency and robust reference dynamics in future work.

Abstract

Recently, large language models (LLMs) have demonstrated remarkable capabilities in natural language processing tasks, yet they remain prone to hallucinations when reasoning with insufficient internal knowledge. While integrating LLMs with knowledge graphs (KGs) provides access to structured, verifiable information, existing approaches often generate incomplete or factually inconsistent reasoning paths. To this end, we propose Self-Reflective Planning (SRP), a framework that synergizes LLMs with KGs through iterative, reference-guided reasoning. Specifically, given a question and topic entities, SRP first searches for references to guide planning and reflection. In the planning process, it checks initial relations and generates a reasoning path. After retrieving knowledge from KGs through a reasoning path, it implements iterative reflection by judging the retrieval result and editing the reasoning path until the answer is correctly retrieved. Extensive experiments on three public datasets demonstrate that SRP surpasses various strong baselines and further underscore its reliable reasoning ability.

Self-Reflective Planning with Knowledge Graphs: Enhancing LLM Reasoning Reliability for Question Answering

TL;DR

This work tackles LLM hallucinations in knowledge-grounded question answering by introducing Self-Reflective Planning (SRP), a framework that integrates LLMs with knowledge graphs through reference-guided planning, iterative retrieval, and self-reflection. SRP comprises four modules—Reference Searching, Path Planning (relation check and path generation), Knowledge Retrieval, and Reflection and Reasoning (sequence judge, path edit, and answering)—to generate and refine reliable reasoning paths and triplet sequences. Empirical results on WebQSP, CWQ, and GrailQA show SRP achieving state-of-the-art performance on two benchmarks and strong results on the third, with significantly higher reliable answering rates than baselines. The approach demonstrates that structured references and iterative self-correction can substantially improve factual grounding and reasoning reliability in KGQA without relying on fine-tuning. Limitations include higher computational cost due to repeated reasoning cycles and sensitivity to reference quality, suggesting directions for efficiency and robust reference dynamics in future work.

Abstract

Recently, large language models (LLMs) have demonstrated remarkable capabilities in natural language processing tasks, yet they remain prone to hallucinations when reasoning with insufficient internal knowledge. While integrating LLMs with knowledge graphs (KGs) provides access to structured, verifiable information, existing approaches often generate incomplete or factually inconsistent reasoning paths. To this end, we propose Self-Reflective Planning (SRP), a framework that synergizes LLMs with KGs through iterative, reference-guided reasoning. Specifically, given a question and topic entities, SRP first searches for references to guide planning and reflection. In the planning process, it checks initial relations and generates a reasoning path. After retrieving knowledge from KGs through a reasoning path, it implements iterative reflection by judging the retrieval result and editing the reasoning path until the answer is correctly retrieved. Extensive experiments on three public datasets demonstrate that SRP surpasses various strong baselines and further underscore its reliable reasoning ability.

Paper Structure

This paper contains 34 sections, 4 figures, 10 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. LLM Reasoning.
  4. KG-enhanced LLM.
  5. Problem Definition
  6. Reasoning Path.
  7. Triplet Sequence.
  8. Methodology
  9. Overview
  10. Reference Searching
  11. Path Planning
  12. Relation Check.
  13. Path Generation.
  14. Knowledge Retrieval
  15. Reflection and Reasoning
  16. ...and 19 more sections

Figures (4)

  • Figure 1: An illustration of the question answering over knowledge graphs.
  • Figure 2: The framework of our SRP method. It consists of four main parts: (1) Reference Searching, (2) Path Planning, (3) Knowledge Retrieval and (4) Reflection and Reasoning.
  • Figure 3: Reliable answering rate of Readi and SRP on theree datasets.
  • Figure 4: Case study of Readi and SRP. Question 1 comes from WebQSP, and question 2 comes from CWQ.