liveBook · Manning

Advanced rag

As explained in Retrieval Augmented Generation, The Foundational Ideas

Advanced RAG is a retrieval-augmented generation architecture that optimizes retrieval, document processing, and post-retrieval components while retaining the basic pipeline.

Advanced RAG preserves the basic retrieve-and-generate pipeline while systematically improving retrieval, document processing, query handling, and post-retrieval context. It addresses production problems such as vocabulary mismatch, missed relevant passages, redundant context, and limited context windows. section 1.5.2 section 1.6

Typical enhancements include:

  • Preprocess documents with semantically coherent chunking, richer metadata, and handling for images, tables, and structured data. section 1.5.2
  • Enhance queries through expansion, hypothetical document generation, and routing to specialized retrieval strategies. section 1.5.2
  • Improve retrieval with hybrid dense-and-sparse search, ensemble methods, reciprocal rank fusion, and cross-encoder reranking. section 1.5.2
  • Optimize context by compressing results, removing redundancy, and ordering passages for the generator. section 1.5.2

When to upgrade from naive RAG

Consider advanced RAG when retrieval quality limits user satisfaction, vocabulary mismatch exceeds roughly 30%, usage exceeds 1,000 queries daily, or performance differences affect revenue. These figures are calibration starting points rather than measured industry thresholds. section 1.5.5

Figure 1.3 RAG Implementation Decision Tree — Business Decisions guiding RAG choices.

Use the decision tree to compare RAG stages against business constraints before accepting the added infrastructure, latency, and evaluation burden of advanced techniques. section 1.5.5

How Retrieval Augmented Generation, The Foundational Ideas covers it

How deep it goes
Explains component-level optimizations and named techniques, while deferring graph-based modular orchestration and runtime agentic choice.
Examples
Uses intelligent chunking, metadata enrichment, HyDE, RAG-Fusion, reciprocal rank fusion, cross-encoder reranking, compression, and context ordering.
What it assumes
Assumes the basic retrieval-and-generation pipeline and distinguishes it from later modular and agentic architectures.
What you can do after
Apply via query expansion, hybrid retrieval, reranking, compression, and routing while evaluating component failures and operational tradeoffs.
How it connects
Connects naive RAG with hybrid search, query expansion, reranking, context optimization, modular RAG, and agentic RAG.

Also explained in

Compare how 2 books explain advanced rag →

Related concepts

This explainer was generated by Manning's knowledge graph from the text of Retrieval Augmented Generation, The Foundational Ideas. The section links point to the parts of the book it draws on.