From Local to Global: A Graph RAG Approach to Query-Focused Summarization

By Darren Edge, Ha Trinh, Newman Cheng, Joshua Bradley, A. Chao, Apurva N. Mody, Steven Truitt, Jonathan Larson. arXiv.org, 2024

1,774 citations recorded by the source corpus.

The use of retrieval-augmented generation (RAG) to retrieve relevant information from an external knowledge source enables large language models (LLMs) to answer questions over private and/or previously unseen document collections. However, RAG fails on global questions directed at an entire text corpus, such as"What are the main themes in the dataset?", since this is inherently a query-focused summarization (QFS) task, rather than an explicit retrieval task. Prior QFS methods, meanwhile, do not scale to the quantities of text indexed by typical RAG systems. To combine the strengths of these contrasting methods, we propose GraphRAG, a graph-based approach to question answering over private text corpora that scales with both the generality of user questions and the quantity of source text. Our approach uses an LLM to build a graph index in two stages: first, to derive an entity knowledge graph from the source documents, then to pregenerate community summaries for all groups of closely related entities. Given a question, each community summary is used to generate a partial response, before all partial responses are again summarized in a final response to the user. For a class of global sensemaking questions over datasets in the 1 million token range, we show that GraphRAG leads to substantial improvements over a conventional RAG baseline for both the comprehensiveness and diversity of generated answers.

View Paper: https://www.semanticscholar.org/paper/c1799bf28d1ae93e1631be5b59196ee1e568f538

Citations in the corpus

• Search-o1: Agentic Search-Enhanced Large Reasoning Models (2025, unknown): “Furthermore, some studies have explored end-to-end model training to implement RAG systems (Li et al., 2024, 2025b) and knowledge-graph-based RAG systems (Edge et al., 2024).”
• LightRAG: Simple and Fast Retrieval-Augmented Generation (2024, unknown): “By accessing both specific details and overarching themes, LightRAG adeptly responds to complex queries involving interconnected topics, providing contex-tually relevant answers of high quality.”
• A-MEM: Agentic Memory for LLM Agents (2025, unknown): “While graph databases provide structured organization for memory systems, their reliance on predefined schemas and relationships fundamentally limits their adaptability.”
• Graph Retrieval-Augmented Generation: A Survey (2024, unknown): “The development of Large Language Models like GPT-4 [ 116 ],”
• ReMindRAG: Low-Cost LLM-Guided Knowledge Graph Traversal for Efficient RAG (2025, unknown): “It outperforms advanced RAG methods (e.g., GraphRAG [9]) and long-context LLMs in both QA and non-QA tasks, with balanced inference efficiency.”
• A Matryoshka Hierarchical RAG for Efficient Multi-Hop Question Answering (2026, unknown): “The remaining baselines covered graph-based and reasoning-augmented approaches; they are: GraphRAG [4], LightRAG [6], HippoRAG2 [8], KGP [39], and ToG [32].”
• From Semantic Retrieval to Conversational Agent: A Web-Based RAG Architecture for Interactive System Dynamics Modeling (2026, unknown): “Furthermore, in highly structured domains, these conversational approaches increasingly intersect with graph-based retrieval mechanisms [22].”
• Constrained Dominant Sets for Multimodal Document Question Answering (2026, unknown): “Bold marks the best per-column score systems: GPT-5 [44], Deepseek-OCR [42], RAGAnything [8], MA-RAG [9], GraphRAG [29], LightRAG [7], MM-GraphRAG [6], VisDoMRAG [4], and ViDoRAG [5].”
• PDE-Agents: An LLM-Orchestrated Multi-Agent Framework for Automated Finite Element Simulations with Knowledge Graph-Augmented Reasoning (2026, unknown): “The graph-structured variant, GraphRAG [10], further enables multi-hop reasoning over relational knowledge—particularly relevant in engineering contexts where material properties, mesh 7.”
• Min-Cost Flow Routing for Evidence Assembly in Long Multimodal Documents (2026, unknown): “GraphRAG (Edge et al., 2024) and LightRAG (Guo et al., 2024) construct entity–relation graphs, HippoRAG (Gutiérrez et al., 2024) applies personalised PageRank to a knowledge graph, and MMGraphRAG (Wan & Yu, 2025) incorporates multimodal nodes.”
• MM-BizRAG: Rethinking Multimodal Retrieval-Augmented Generation for General Purpose Enterprise Q&A (2026, unknown): “Modern RAG systems have evolved beyond text-only inputs to incorporate multiple modalities including images, videos, and complex document graphs (Abootorabi et al., 2025; Mei et al., 2025; Gao et al., 2025; Edge et al., 2024), enabling retrieval and reasoning across diverse data types.”
• GraphRAG on Consumer Hardware: Benchmarking Local LLMs for Healthcare EHR Schema Retrieval (2026, unknown): “3.0 [2] and implement a six-stage pipeline.”
• ASTRA-QA: A Benchmark for Abstract Question Answering over Documents (2026, unknown): “Recent RAG methods [2–4, 7, 24, 25] have moved further toward long-form and document-level synthesis through hierarchical retrieval, graph-structured indexing, and long-context reading.”
• XGRAG: A Graph-Native Framework for Explaining KG-based Retrieval-Augmented Generation (2026, unknown): “Our framework is built upon a GraphRAG (Edge et al., 2024) backbone, which performs three key tasks: (1) constructing a global knowledge graph, G = ( V, E ) , from source documents during indexing; (2) retrieving a query-relevant subgraph, G ret , for a given query during retrieval; and (3)…”
• MG2-RAG: Multi-Granularity Graph for Multimodal Retrieval-Augmented Generation (2026, unknown): “Graph RAG.”
• MDKeyChunker: What Does One LLM Call per Chunk Buy for Markdown Retrieval? (2026, unknown): “GraphRAG [Edge et al., 2024] splits text into fixed-size token units, uses an LLM to extract entities, relationships, and claims from each unit (with optional additional extraction rounds), builds a graph, partitions it into a hierarchy of communities, and summarizes each community; it targets…”
• Mindscape-Aware Retrieval Augmented Generation for Improved Long Context Understanding (2025, unknown): “Such supervision is essential for training long-context retrievers, whether evidence is represented as text chunks in standard RAG or as graph nodes in GraphRAG.”