KG-TRACES：通过知识图谱约束的轨迹推理与归因监督增强大型语言模型
KG-TRACES: Enhancing Large Language Models with Knowledge Graph-constrained Trajectory Reasoning and Attribution Supervision
AI总结：
提出 KG-TRACES 框架，以知识图谱约束的关系路径、三元组路径和归因过程显式监督 LLM，在 WebQSP、CWQ 及医学领域显著提升复杂推理性能与可解释性。
摘要：
大型语言模型（LLM）已在各类自然语言处理任务中取得显著进展，但它们在复杂推理问题上的表现仍受限于可解释性和可信性不足。这一问题通常表现为幻觉或无法归因的推理过程，限制了其在复杂推理场景中的适用性。为解决该问题，我们提出了 Knowledge Graph-constrained Trajectory Reasoning Attribution and Chain Explanation Supervision（KG-TRACES），这是一个通过对推理路径和过程进行显式监督来增强 LLM 推理能力的新颖框架。KG-TRACES 联合监督模型：（1）预测符号关系路径，（2）预测完整的三元组级推理路径，以及（3）基于推理路径生成具备归因意识的推理过程。在推理阶段，模型可同时适应知识图谱可用和不可用的场景：在可能时从 KG 检索推理路径，在不可用时仅依靠内在知识预测合理的推理路径。该设计使模型能够以可解释且来源可归因的模式进行推理。通过在复杂推理任务上的大量实验，我们证明 KG-TRACES 显著优于现有 SOTA：它在 WebQSP 上将 Hits@1 提高 1.6%、F1 提高 4.7%，并在 CWQ 上实现 Hits@1 提升 4.8%、F1 提升 2.1%。此外，我们展示了其向医学等专业领域的可迁移性。通过可视化推理过程的中间步骤，我们进一步表明，KG-TRACES 引入的显式监督带来了更稳定、更具目标导向的推理过程，并与正确答案紧密对齐。代码可在 https://github.com/Edaizi/KG-TRACES 获取。
英文摘要：
Large language models (LLMs) have made remarkable strides in various natural language processing tasks, but their performance on complex reasoning problems remains hindered by a lack of explainability and trustworthiness. This issue, often manifesting as hallucinations or unattributable reasoning processes, limits their applicability in complex reasoning scenarios. To address this, we propose Knowledge Graph-constrained Trajectory Reasoning Attribution and Chain Explanation Supervision (KG-TRACES), a novel framework that enhances the reasoning ability of LLMs through explicit supervision over reasoning paths and processes. KG-TRACES jointly supervises the model to: (1) predict symbolic relation paths, (2) predict full triple-level reasoning paths, and (3) generate attribution-aware reasoning processes grounded in the reasoning paths. At inference phase, the model adapts to both KG-available and KG-unavailable scenarios, retrieving reasoning paths from a KG when possible or predicting plausible reasoning paths with only intrinsic knowledge when not. This design enables the model to reason in an explainable and source-attributable pattern. Through extensive experiments on complex reasoning tasks, we demonstrate that KG-TRACES significantly outperforms existing SOTA: it improves Hits@1 by 1.6% and F1 by 4.7% on WebQSP, and achieves improvements of 4.8% in Hits@1 and 2.1% in F1 on CWQ. Moreover, we show its transferability to specialized domains such as medicine. By visualizing the intermediate steps of reasoning processes, we further show that the explicit supervision introduced by KG-TRACES leads to more stable and goal-directed reasoning processes, aligning closely with correct answers. Code is available at https://github.com/Edaizi/KG-TRACES.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
