foreword
In Essential GraphRAG, Tomaž and Oskar demonstrate how to implement a GraphRAG system from scratch, without relying on existing frameworks. They pull back the curtain, revealing the code behind contemporary AI applications. The book covers major GraphRAG innovations through worked examples you can code and run. Exercises explore nuances and alternatives, with references to primary sources on arXiv. Starting with simple RAG patterns, chapters progress through GraphRAG techniques to agentic workflows.
By working through these coding examples, reading referenced articles, and solving exercises, you’ll learn
- How RAG improves large language model accuracy by retrieving external data
- How knowledge graphs extend RAG for more structured and precise information retrieval
- How to use query rewriting techniques and strategies for embedding and document chunking, adapted for various use cases
- How to build agentic systems for complex scenarios
At every step, Tomaž and Oskar guide you on improving retrieval accuracy, structuring responses, and evaluating results, helping you understand the tradeoffs of mixing and matching approaches for your specific needs.