@DanKornas: Your RAG pipeline doesn’t need to retrieve the same evidence twice. LeanRAG is an open-source RAG framework that uses k…
Summary
LeanRAG is an open-source RAG framework that uses knowledge graphs, semantic aggregation, and hierarchical retrieval to reduce redundancy in retrieval pipelines, providing grounded answers with concise evidence paths.
View Cached Full Text
Cached at: 07/02/26, 08:19 AM
Your RAG pipeline doesn’t need to retrieve the same evidence twice.
LeanRAG is an open-source RAG framework that uses knowledge graphs, semantic aggregation, and hierarchical retrieval to generate grounded answers from structured evidence.
It helps you build less redundant retrieval pipelines by clustering entities into summary nodes, traversing a multi-layer graph, and sending concise evidence paths to the LLM.
Key features:
• Semantic aggregation – clusters low-level entities into coherent summaries with explicit relations • Hierarchical retrieval – anchors queries at detailed entities, then traverses upward through the graph • Redundancy-aware synthesis – streamlines retrieval paths and avoids overlapping content before generation • Two extraction paths – includes CommonKG and GraphRAG-style methods for extracting entities and relations • End-to-end workflow – covers document chunking, graph building, retrieval, and final answer generation
Free public GitHub repo.
Link in the reply
Similar Articles
@tom_doerr: Knowledge graph framework for RAG using semantic aggregation and hierarchical retrieval https://github.com/KnowledgeXLa…
LeanRAG is an open-source framework that enhances Retrieval-Augmented Generation using knowledge graph structures with semantic aggregation and hierarchical retrieval for context-aware, high-fidelity responses. It has been accepted at AAAI-26.
GRACE-RAG: Governed Retrieval Architecture for Canonical Evidence Synthesis, Enabling Lightweight Deployment in Closed-Domain Institutional Settings
This paper introduces GRACE-RAG, a retrieval-governed, graph-augmented RAG architecture that externalizes structural reasoning from generation to a structured retrieval layer, enabling lightweight deployment in closed-domain institutional settings. Experiments show up to 20% quality gains with mid-scale models, reducing computational and latency footprint.
AgenticRAG: Agentic Retrieval for Enterprise Knowledge Bases
This paper introduces AgenticRAG, a framework from Microsoft that enhances enterprise knowledge base retrieval by equipping LLMs with tools for iterative search, document navigation, and analysis. It demonstrates significant improvements in recall and factuality over standard RAG pipelines on multiple benchmarks.
RAG-Anything: All-in-One RAG Framework
RAG-Anything is a new open-source framework that enhances multimodal knowledge retrieval by integrating cross-modal relationships and semantic matching, outperforming existing methods on complex benchmarks.
LatentRAG: Latent Reasoning and Retrieval for Efficient Agentic RAG
LatentRAG is a novel framework that shifts reasoning and retrieval for agentic RAG into continuous latent space, reducing inference latency by approximately 90% while maintaining performance comparable to explicit methods.