GraphRAG Global Map-Reduce Query Implementation
A code-level walkthrough of Global Map-Reduce in graphrag-ts: loading community summaries and descendant-leaf claims, approximate token batching, bounded concurrent Map calls, retries, JSON validation, and usefulness-ranked Reduce packing.
GraphRAG 混合检索与证据溯源实现
基于 graphrag-ts 检索源码解析 local 与 global GraphRAG 查询：实体和图拓扑召回、向量与关键词混合证据、RRF 社区排序、LLM 过滤，以及带 token 预算和失败重试的全局 Map-Reduce。
GraphRAG Global Search Map-Reduce Explained
A detailed guide to Microsoft GraphRAG global search Map-Reduce, explaining how hierarchical community summaries are shuffled and split by token budget, mapped into scored intermediate answers, ranked, and reduced into a corpus-level response.
GraphRAG 里的“社区”概念
深度解读微软 GraphRAG 核心概念“社区（Community）”的理论本质与工程实践，详解基于 Leiden 算法的高内聚子图划分、层级社区架构（顶层/中层/叶子社区）、社区摘要提取与全局答案聚合管道，并附带 TypeScript 伪代码与 Obsidian 概念图谱。
Understanding the Community Concept in GraphRAG
An intuitive and in-depth conceptual guide to the "Community" in Microsoft GraphRAG, explaining graph community detection via Leiden algorithm, hierarchical community trees (root/intermediate/leaf), query-focused community summarization, and candidate answer aggregation with TypeScript examples.
GraphRAG Retrieval Implementation Guide
A complete implementation guide for the GraphRAG query and retrieval phase, detailing query intent parsing, multi-channel entity retrieval with Reciprocal Rank Fusion (RRF), neighbor expansion, multi-dimensional community scoring, LLM filtering, and final global answer synthesis.
GraphRAG 检索实现说明
详解 GraphRAG 索引构建完成后的检索阶段全流程实现，涵盖查询意图解析、多路实体召回与 RRF 倒数排名融合、邻居扩展、社区映射与多维打分排序（语义相似度/实体重叠率/图结构得分），以及 LLM 社区过滤与全局答案生成。
GraphRAG 多文件输入处理架构与目录结构
系统剖析微软 GraphRAG 的多文件输入处理架构与全局索引管道设计，涵盖 TXT/CSV/JSON/Parquet 等多源格式加载、基于句界的分块与跨文档实体去重融合、标准工程目录结构最佳实践及增量更新机制。
GraphRAG PostgreSQL + TypeScript Implementation Design
A comprehensive engineering guide on implementing Microsoft GraphRAG with PostgreSQL and TypeScript, covering relation edge weight calculations, community detection pipelines, hierarchical community summarization, and complete PostgreSQL DDL schemas with TypeScript pseudo-code.