HG-RAG: Hierarchy-Guided Retrieval-Augmented Generation for Structured Knowledge Graphs

HG-RAG：面向结构化知识图谱的层次引导检索增强生成

Abstract: Retrieval Augmented Generation (RAG) has proven to be a widely successful process at improving the quality of outputs from a Large Language Model (LLM) for wider context. However, RAG systems typically retrieve context from flat document stores, which struggles when queries require hierarchical or relational reasoning across structured knowledge.

摘要： 检索增强生成（RAG）已被证明是一种非常成功的流程，能够通过提供更广泛的上下文来提升大语言模型（LLM）的输出质量。然而，传统的 RAG 系统通常从扁平的文档存储中检索上下文，当查询需要跨结构化知识进行层次化或关系推理时，这种方法往往难以奏效。

I present HG-RAG (Hierarchy-Guided RAG), a framework that performs graph-traversal over a hierarchical knowledge graph to deliver structured context to a language model. My retrieval pipeline resolves a named entity anchor from the query, then expands context upward through parent nodes, laterally through relational neighbors, and downward through child nodes when needed.

我提出了 HG-RAG（层次引导式 RAG），这是一个通过在层次化知识图谱上执行图遍历，从而为语言模型提供结构化上下文的框架。我的检索流水线首先从查询中解析出一个命名实体锚点，然后根据需要向上通过父节点、横向通过关系邻居、向下通过子节点来扩展上下文。

I evaluate HG-RAG against a dense retrieval baseline across three world scales (18-800 nodes) with four query types: local fact, hierarchical, neighborhood, and multi-hop. Results show HG-RAG consistently outperforms the flat baseline on hierarchical, relational, and multi-hop reasoning tasks, while reducing hallucination and maintaining locality coherence.

我在三个不同规模（18-800 个节点）的世界模型中，针对四种查询类型（局部事实、层次化、邻域和多跳查询），将 HG-RAG 与密集检索基线进行了对比评估。结果表明，HG-RAG 在层次化、关系型和多跳推理任务中始终优于扁平基线，同时减少了幻觉并保持了局部连贯性。