KC-GenRe: A Knowledge-constrained Generative Re-ranking Method Based on Large Language Models for Knowledge Graph Completion

📄 arXiv: 2403.17532v1 📥 PDF

作者: Yilin Wang, Minghao Hu, Zhen Huang, Dongsheng Li, Dong Yang, Xicheng Lu

分类: cs.AI

发布日期: 2024-03-26

备注: This paper has been accepted for publication in the proceedings of LREC-COLING 2024

💡 一句话要点

提出KC-GenRe以解决知识图谱补全中的重排序问题

🎯 匹配领域: 支柱九：具身大模型 (Embodied Foundation Models)

关键词: 知识图谱 补全 生成模型 重排序 交互训练 知识约束 信息检索

📋 核心要点

  1. 现有KGC重排序方法多基于非生成模型，无法充分利用生成模型的知识和能力，导致匹配、排序和遗漏等问题。
  2. KC-GenRe通过将KGC重排序任务转化为候选标识符排序生成问题，并采用知识引导的交互训练方法来提升候选的识别和排序。
  3. 实验结果显示，KC-GenRe在MRR和Hits@1指标上分别比以往方法提升了6.7%和7.7%，展现出其优越性。

📝 摘要（中文）

知识图谱补全（KGC）的目标是预测实体间缺失的事实。以往的KGC重排序方法主要基于非生成语言模型，无法充分利用生成模型的潜力。为此，本文提出KC-GenRe，一种基于大型语言模型的知识约束生成重排序方法。该方法通过将KGC重排序任务形式化为候选标识符排序生成问题，解决了匹配、排序和遗漏等问题。实验结果表明，KC-GenRe在四个数据集上实现了最先进的性能，相较于以往方法，MRR和Hits@1指标分别提升了6.7%和7.7%。

🔬 方法详解

问题定义：本文旨在解决知识图谱补全中的重排序问题，现有方法在候选实体的匹配、排序和遗漏方面存在不足，影响了补全效果。

核心思路：KC-GenRe通过将重排序任务视为候选标识符的排序生成问题，利用生成语言模型的能力来提升排序的准确性和有效性。

技术框架：KC-GenRe的整体架构包括候选生成、知识引导的交互训练和知识增强的约束推理三个主要模块，确保生成的排序结果有效且符合知识约束。

关键创新：该方法的创新在于引入知识约束的生成重排序机制，克服了传统方法的局限性，能够更好地利用预训练知识和生成能力。

关键设计：在设计上，KC-GenRe采用了知识引导的交互训练策略和知识增强的推理方法，确保生成过程中的上下文提示和控制生成，优化了候选的识别和排序。

📊 实验亮点

KC-GenRe在四个数据集上实现了最先进的性能，相较于以往方法，MRR和Hits@1指标分别提升了6.7%和7.7%，而与无重排序的情况相比，提升幅度更是达到9.0%和11.1%。

🎯 应用场景

KC-GenRe在知识图谱补全领域具有广泛的应用潜力，能够提升信息检索、推荐系统和智能问答等任务的准确性和效率。未来，该方法可进一步扩展到其他需要知识推理和生成的场景，推动相关领域的发展。

📄 摘要（原文）

The goal of knowledge graph completion (KGC) is to predict missing facts among entities. Previous methods for KGC re-ranking are mostly built on non-generative language models to obtain the probability of each candidate. Recently, generative large language models (LLMs) have shown outstanding performance on several tasks such as information extraction and dialog systems. Leveraging them for KGC re-ranking is beneficial for leveraging the extensive pre-trained knowledge and powerful generative capabilities. However, it may encounter new problems when accomplishing the task, namely mismatch, misordering and omission. To this end, we introduce KC-GenRe, a knowledge-constrained generative re-ranking method based on LLMs for KGC. To overcome the mismatch issue, we formulate the KGC re-ranking task as a candidate identifier sorting generation problem implemented by generative LLMs. To tackle the misordering issue, we develop a knowledge-guided interactive training method that enhances the identification and ranking of candidates. To address the omission issue, we design a knowledge-augmented constrained inference method that enables contextual prompting and controlled generation, so as to obtain valid rankings. Experimental results show that KG-GenRe achieves state-of-the-art performance on four datasets, with gains of up to 6.7% and 7.7% in the MRR and Hits@1 metric compared to previous methods, and 9.0% and 11.1% compared to that without re-ranking. Extensive analysis demonstrates the effectiveness of components in KG-GenRe.