Retrieval-Augmented Generation (RAG)

Architectures, Agents, and LLMs from Fundamentals to Production
Erscheinungsjahr 2027
ISBN: 978-3-032-46194-0
Verlag: Springer

Buch, Englisch, Format (B × H): 155 mm x 235 mm

This book provides a comprehensive and production-oriented treatment of retrieval-augmented generation (RAG) systems, guiding readers from the foundations of information retrieval and vector search to advanced agentic architectures, knowledge graphs, evaluation, governance, and enterprise deployment. Rather than presenting RAG as a collection of isolated techniques, it explains the evolutionary logic behind each architectural choice, showing how chunking, metadata, retrieval, reranking, graph-based reasoning, memory, and tool use emerged to address specific limitations of earlier approaches. Emphasizing durable principles over transient tools, this book combines conceptual rigor with practical implementation guidance through extensive hands-on labs and real-world case studies. By integrating reliability, evaluation, security, governance, and knowledge representation into a unified architectural perspective, this book equips researchers, practitioners, and advanced students to design, assess, and deploy trustworthy AI assistants in complex technical environments.This book features a supplementary guide for further learning and each chapter contains hands-on labs.

Zielgruppe

Professional/practitioner

Autoren/Hrsg.

Ros, Frédéric

Ros, Yann

Fachgebiete

• Mathematik | Informatik EDV | Informatik Informatik Künstliche Intelligenz
• Interdisziplinäres Wissenschaften Wissenschaften: Forschung und Information Kybernetik, Systemtheorie, Komplexe Systeme
• Technische Wissenschaften Elektronik | Nachrichtentechnik Nachrichten- und Kommunikationstechnik
• Technische Wissenschaften Elektronik | Nachrichtentechnik Elektronik

Inhaltsverzeichnis

Introduction.- Information retrieval before LLMs.- Part I — Fundamentals of RAG.- Introduction to RAG.- Detailed architecture of a RAG system.- Vector representation of text. - Part II — Methodology: a RAG chatbot on your own data.- The enterprise roadmap.- The challenge of real documents.- The AI-ready document: designing for humans and machines.- Text & office documents.- Tables & structured data.- Scanned documents & multimodal parsing.- Ingesting the web & heterogeneous sources.- Part III — Chunking, metadata & structuring.- Chunking strategies.- Chunking from canonical JSON.- Documentary hierarchy & Parent-Child.- Metadata: the underrated lever.- Structuring knowledge.- Advanced chunking.- Retrieval techniques.- Searching fast at scale.- Re-ranking.- Query transformation.- Part V — Advanced architectures: from pipeline to agent.- The seven failure points- Retrieval becomes a tool call.- Self-RAG, CRAG & Adaptive-RAG.- Agentic RAG & Multi-Agent RAG.- GraphRAG: implementation.- Beyond the flat graph: hypergraphs & hierarchical navigation.- Taxonomies, ontologies, and business graphs.- Part VI — Going to production: reliability, security, evaluation.- Evaluation.- Conversational RAG.- Security & data governance.- The user experience of a RAG.- Case studies by domain.- Part VII — Beyond retrieval: business calibration.- The paradox: retrieving without choosing like an expert.- Business calibration.- Automatable annotations.- Conclusion.

Über Autor(innen)

Frédéric Ros holds a Ph.D. from ENGREF (AgroParisTech, Paris) and an Habilitation à Diriger des Recherches (HDR) from the University of Orléans. He began his career at INRAE in image analysis for robotics and artificial systems before leading computer vision and artificial intelligence activities within major international technology companies, including Schlumberger and GEMALTO (now THALES). His expertise spans machine vision, artificial intelligence, data analysis, fuzzy systems, machine learning, digital watermarking, and evolutionary algorithms, with a long-standing interest in adaptive and self-tuning systems for complex industrial environments. He has also supervised several doctoral and research projects at the interface between academia and industry. Since 2008, he has been Professor at the University of Orléans and an associate researcher at the PRISME Laboratory, where his research focuses on pattern recognition, unsupervised learning, knowledge integration, and pseudo-labeling. 

Yann Ros is a Cybersecurity Engineer specializing in cyber defense, security automation, and Security Operations Centers (SOC). Trained in mathematics and computer science, he holds a master's degree in Cryptology and Cybersecurity from the University of Bordeaux. He began his career in cybersecurity at Airbus, as a consultant with Sopra Steria, working on detection engineering, SOAR orchestration, incident response, and security automation. He currently serves as a CERT Analyst for the Centre-Val de Loire regional Computer Emergency Response Team (CERT), where he supports organizations across the region in incident response and threat analysis. His technical expertise extends to low-level security: reverse engineering of binaries, assembly language, and malware analysis, combining static disassembly with dynamic behavioral analysis. He also teaches cybersecurity and reverse engineering, with a hands-on approach focused on understanding how software and attacks actually work. His interests include secure software engineering, cyber defense, automation, and the integration of artificial intelligence into operational cybersecurity, including the security of AI-based systems themselves.