Knowledge Graphs and LLMs in Action

Combine knowledge graphs with large language models to deliver powerful, reliable, and explainable AI solutions.

Knowledge graphs model relationships between the objects, events, situations, and concepts in your domain so you can readily identify important patterns in your own data and make better decisions. Paired up with large language models, they promise huge potential for working with structured and unstructured enterprise data, building recommendation systems, developing fraud detection mechanisms, delivering customer service chatbots, or more. This book provides tools and techniques for efficiently organizing data, modeling a knowledge graph, and incorporating KGs into the functioning of LLMs—and vice versa.

In Knowledge Graphs and LLMs in Action you will learn how to:

• Model knowledge graphs with an iterative top-down approach based in business needs
• Create a knowledge graph starting from ontologies, taxonomies, and structured data
• Build knowledge graphs from unstructured data sources using LLMs
• Use machine learning algorithms to complete your graphs and derive insights from it
• Reason on the knowledge graph and build KG-powered RAG systems for LLMs

In Knowledge Graphs and LLMs in Action, you’ll discover the theory of knowledge graphs then put them into practice with LLMs to build working intelligence systems. You’ll learn to create KGs from first principles, go hands-on to develop advisor applications for real-world domains like healthcare and finance, build retrieval augmented generation for LLMs, and more.

About the technology

Using knowledge graphs with LLMs reduces hallucinations, enables explainable outputs, and supports better reasoning. By naturally encoding the relationships in your data, knowledge graphs help create AI systems that are more reliable and accurate, even for models that have limited domain knowledge.

About the book

Knowledge Graphs and LLMs in Action shows you how to introduce knowledge graphs constructed from structured and unstructured sources into LLM-powered applications and RAG pipelines. Real-world case studies for domain-specific applications—from healthcare to financial crime detection—illustrate how this powerful pairing works in practice. You’ll especially appreciate the expert insights on knowledge representation and reasoning strategies.

What's inside

• Design knowledge graphs for real-world needs
• Build KGs from structured and unstructured data
• Apply machine learning to enrich, complete, and analyze graphs
• Pair knowledge graphs with RAG systems

About the reader

For ML and AI engineers, data scientists, and data engineers. Examples in Python.

About the author

Alessandro Negro is Chief Scientist at GraphAware and author of Graph-Powered Machine Learning. Vlastimil Kus, Giuseppe Futia, and Fabio Montagna are seasoned ML and AI professionals specializing in Knowledge Graphs, Large Language Models, and Graph Neural Networks.

Table of Contents

Part 1
1 Knowledge graphs and LLMs: A killer combination
2 Intelligent systems: A hybrid approach
Part 2
3 Create your first knowledge graph from ontologies
4 From simple networks to multisource integration
Part 3
5 Extracting domain-specific knowledge from unstructured data
6 Building knowledge graphs with large language models
7 Named entity disambiguation
8 NED with open LLMs and domain ontologies
Part 4
9 Machine learning on knowledge graphs: A primer approach
10 Graph feature engineering: Manual and semiautomated approaches
11 Graph representation learning and graph neural networks
12 Node classification and link prediction with GNNs
Part 5
13 Knowledge graph–powered retrieval-augmented generation
14 Asking a KG questions with natural language
15 Building a QA agent with LangGraph

Dr. Alessandro Negro is the Chief Scientist at GraphAware. Alessandro has been a speaker at many prominent conferences and is the author of the Manning book Graph-Powered Machine Learning and several scientific publications. He is one of the creators of GraphAware Hume, a mission critical knowledge graph platform.

Dr. Vlastimil Kus is the Lead Data Scientist at GraphAware where he contributes to the development of Hume. Over the years he gained significant experience in building and utilizing Knowledge Graphs from unstructured data using NLP and ML techniques in various domains. His current focus is NLP and Graph Machine Learning.

Dr. Giuseppe Futia is Senior Data Scientist at GraphAware and a Fellow at the Nexa Center for Internet & Society. He holds a Ph.D. in computer engineering from the Politecnico di Torino (Italy), where he explored Graph Representation Learning techniques to support the automatic building of Knowledge Graphs.

Fabio Montagna is the Lead Machine Learning Engineer at GraphAware. He holds a master’s degree in software engineering from Unisalento (Italy). As a bridge between science and industry, he assists with moving rapidly from scientific reasoning to product value.