Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

NeSTR: A Neuro-Symbolic Abductive Framework for Temporal Reasoning in Large Language Models

Feng Liang, Weixin Zeng, Runhao Zhao, Xiang Zhao

TL;DR

The paper tackles temporal reasoning in large language models by marrying structured symbolic representations of time with neural abductive reasoning in a zero-shot setting. NeSTR encodes temporal facts as predicates, then uses a five-stage prompting strategy—symbolic representation, neural-symbolic inference, consistency verification, abductive reflection, and final answer extraction—to enforce temporal coherence. Experiments on TimeQA and TempReason show state-of-the-art zero-shot performance across diverse baselines and model sizes, with ablations confirming the contribution of each component. The work demonstrates the effectiveness and interpretability of neuro-symbolic temporal reasoning and points to broader applicability beyond temporal QA.

Abstract

Large Language Models (LLMs) have demonstrated remarkable performance across a wide range of natural language processing tasks. However, temporal reasoning, particularly under complex temporal constraints, remains a major challenge. To this end, existing approaches have explored symbolic methods, which encode temporal structure explicitly, and reflective mechanisms, which revise reasoning errors through multi-step inference. Nonetheless, symbolic approaches often underutilize the reasoning capabilities of LLMs, while reflective methods typically lack structured temporal representations, which can result in inconsistent or hallucinated reasoning. As a result, even when the correct temporal context is available, LLMs may still misinterpret or misapply time-related information, leading to incomplete or inaccurate answers. To address these limitations, in this work, we propose Neuro-Symbolic Temporal Reasoning (NeSTR), a novel framework that integrates structured symbolic representations with hybrid reflective reasoning to enhance the temporal sensitivity of LLM inference. NeSTR preserves explicit temporal relations through symbolic encoding, enforces logical consistency via verification, and corrects flawed inferences using abductive reflection. Extensive experiments on diverse temporal question answering benchmarks demonstrate that NeSTR achieves superior zero-shot performance and consistently improves temporal reasoning without any fine-tuning, showcasing the advantage of neuro-symbolic integration in enhancing temporal understanding in large language models.

NeSTR: A Neuro-Symbolic Abductive Framework for Temporal Reasoning in Large Language Models

TL;DR

The paper tackles temporal reasoning in large language models by marrying structured symbolic representations of time with neural abductive reasoning in a zero-shot setting. NeSTR encodes temporal facts as predicates, then uses a five-stage prompting strategy—symbolic representation, neural-symbolic inference, consistency verification, abductive reflection, and final answer extraction—to enforce temporal coherence. Experiments on TimeQA and TempReason show state-of-the-art zero-shot performance across diverse baselines and model sizes, with ablations confirming the contribution of each component. The work demonstrates the effectiveness and interpretability of neuro-symbolic temporal reasoning and points to broader applicability beyond temporal QA.

Abstract

Large Language Models (LLMs) have demonstrated remarkable performance across a wide range of natural language processing tasks. However, temporal reasoning, particularly under complex temporal constraints, remains a major challenge. To this end, existing approaches have explored symbolic methods, which encode temporal structure explicitly, and reflective mechanisms, which revise reasoning errors through multi-step inference. Nonetheless, symbolic approaches often underutilize the reasoning capabilities of LLMs, while reflective methods typically lack structured temporal representations, which can result in inconsistent or hallucinated reasoning. As a result, even when the correct temporal context is available, LLMs may still misinterpret or misapply time-related information, leading to incomplete or inaccurate answers. To address these limitations, in this work, we propose Neuro-Symbolic Temporal Reasoning (NeSTR), a novel framework that integrates structured symbolic representations with hybrid reflective reasoning to enhance the temporal sensitivity of LLM inference. NeSTR preserves explicit temporal relations through symbolic encoding, enforces logical consistency via verification, and corrects flawed inferences using abductive reflection. Extensive experiments on diverse temporal question answering benchmarks demonstrate that NeSTR achieves superior zero-shot performance and consistently improves temporal reasoning without any fine-tuning, showcasing the advantage of neuro-symbolic integration in enhancing temporal understanding in large language models.

Paper Structure

This paper contains 21 sections, 6 equations, 1 figure, 3 tables, 1 algorithm.

Table of Contents

  1. Introduction
  2. NeSTR
  3. Problem Setting
  4. NeSTR Architecture
  5. Symbolic Representation
  6. Neural-Symbolic Inference
  7. Consistency Verification
  8. Abductive Reflection and Correction
  9. Final Answer Extraction
  10. Experiments
  11. Experimental Setup
  12. Datasets.
  13. Evaluation Metrics.
  14. Baselines.
  15. Implementation Details.
  16. ...and 6 more sections

Figures (1)

  • Figure 1: Overview of NeSTR compared to vanilla inference on a temporal question. Given a question and temporal contexts (left), vanilla models often fail due to inadequate temporal reasoning. In contrast, NeSTR explicitly decomposes the reasoning process into symbolic representation, inference, consistency checking, and reflection, achieving accurate and temporally consistent answers.