book

Knowledge Graphs and LLMs in Action

Name: Knowledge Graphs and LLMs in Action
ISBN: 9781633439894

by Fabio Montagna, Giuseppe Futia, Alessandro Negro, Vlastimil Kus

October 2025

Intermediate to advanced

472 pages

16h 25m

English

Manning Publications

Audiobook available

Read now

Unlock full access

Knowledge Graphs and LLMs in Action
copyright
contents
forewords
dedication
preface
acknowledgments
about this book
about the authors
about the cover illustration

Part 1 Foundations of hybrid intelligent systems
1 Knowledge graphs and LLMs: A killer combination
1.1 Knowledge graphs1.2 Large language models1.3 KGs and LLMs: Stronger together1.4 The paradigm shift in data-driven applications1.4.1 The four pillars of knowledge graphs1.5 Building data-driven applications using KGs and LLMs1.5.1 Example use case: Drug discovery and development1.5.2 Example use case: Conversational AI for customer support1.5.3 Deciding whether to use a KG1.6 Knowledge graph technologies1.6.1 Taxonomies and ontologies1.7 How do we teach KGs and LLMs?
2 Intelligent systems: A hybrid approach
2.1 What is intelligence?2.2 Designing an intelligent system2.2.1 What is an intelligent system?2.2.2 Categories of intelligent systems2.2.3 Characteristics of an intelligent system2.3 Knowledge acquisition and representation2.4 Reasoning2.5 Reasoning engines2.5.1 Limitations of a pure deductive reasoning engine2.5.2 Using inductive reasoning and ML2.5.3 The role of LLMs in the reasoning engine2.6 A KG approach to IASs
Part 2 Building knowledge graphs from structured data sources
3 Create your first knowledge graph from ontologies
3.1 Knowledge graph building: Warmup3.1.1 Business and domain understanding3.1.2 Data understanding3.2 Understanding knowledge graph technologies3.2.1 RDF or LPG? A goal-driven discussion3.2.2 Representing edge properties with RDF and LPG3.3 Building a knowledge graph3.3.1 Ontology ingestion and processing with neosemantics3.3.2 Annotation ingestion and processing3.4 Querying the data3.5 Reasoning over the KG
4 From simple networks to multisource integration
4.1 Biomedical knowledge graphs and applications4.2 Multi-omic applications of KGs4.2.1 Creating a KG from the PPI and protein-disease networks4.2.2 High-level analysis of the resulting KGs4.2.3 Domain-specific analysis of the PPI and disease KG4.3 Pharmaceutical applications of KGs4.3.1 Deep analysis of the Hetionet knowledge graph4.3.2 LLM-assisted interpretation of pathway analysis results4.4 Clinical applications of KGs4.4.1 LLM-guided clinical decision support analysis
Part 3 Building knowledge graphs from text
5 Extracting domain-specific knowledge from unstructured data
5.1 The archives challenge5.2 Key concepts of knowledge extraction5.2.1 Recognizing named entities5.2.2 Extracting relations5.3 Building KGs with large language models5.3.1 Using LLMs5.3.2 Prompt engineering examples5.3.3 Prompt engineering guidelines5.3.4 KG building: Traditional NLP or LLMs?
6 Building knowledge graphs with large language models
6.1 Transforming an archive to a KG6.1.1 Graph modeling6.1.2 Creating a metagraph6.1.3 Normalization and cleansing6.1.4 Graph-based entity resolution6.2 Intellectual network analysis: The value of graphs6.3 Next steps in the Rockefeller Archive Center project6.4 The value of knowledge graphs in the LLM era
7 Named entity disambiguation
7.1 From recognition to disambiguation7.2 Understanding named entity disambiguation7.3 Domain-based NED and LLMs7.4 Business and domain understanding7.4.1 Context7.4.2 Use case definition7.5 Understanding the data7.5.1 Unstructured data7.5.2 Domain ontologies7.6 Building a SoHO knowledge graph7.6.1 Defining the schema7.6.2 Processing and ingesting documents7.6.3 Disambiguating and ingesting medical entities7.6.4 Processing, loading, and mapping ontologies7.6.5 Generating entity co-occurrences7.7 KG-based use cases7.7.1 Conceptual search7.7.2 Structured knowledge-based search7.7.3 KG-based interpretability and discovery7.7.4 Uncovering new knowledge
8 NED with open LLMs and domain ontologies
8.1 Understanding limitations of traditional NED systems8.2 Ingesting the domain ontology8.3 Setting up the model with Ollama and Llama 3.1 8B8.4 End-to-end NED process8.4.1 Named entity recognition8.4.2 Candidate selection8.4.3 Candidate disambiguation8.5 Conclusions
Part 4 Machine learning on knowledge graphs
9 Machine learning on knowledge graphs: A primer approach
9.1 Machine learning on graphs: Why?9.2 Machine learning on graphs: What?9.2.1 Node classification9.2.2 Link prediction (a.k.a. relationship prediction)9.2.3 Clustering and community detection9.2.4 Graph classification9.3 Machine learning on graphs: How?9.3.1 Node classification and link prediction9.3.2 Graph classification9.3.3 Graph clustering
10 Graph feature engineering: Manual and semiautomated approaches
10.1 Manual node features10.1.1 Degree10.1.2 Triangles10.1.3 Density10.1.4 Geodesic (or shortest) path10.1.5 Closeness10.1.6 Betweenness10.1.7 PageRank10.1.8 Prediction10.2 Manual relationship features10.2.1 Node-based representation10.2.2 Path-based features10.3 Semiautomated feature extraction10.3.1 Performing ReFeX manually10.3.2 Performing ReFeX automatically with code
11 Graph representation learning and graph neural networks
11.1 Embeddings in graph representation learning11.1.1 Understanding graph embeddings: From discrete to continuous11.1.2 Real-world applications and examples11.2 The encoder–decoder model11.2.1 The encoder: Converting graph structure to vectors11.2.2 The decoder: Reconstructing graph properties11.2.3 The power of the framework11.2.4 Node2Vec: An example of an encoder–decoder framework11.3 Shallow embeddings: A first approach to graph representation11.3.1 Understanding shallow embeddings11.3.2 Limitations of shallow embeddings11.4 Embeddings in knowledge graphs11.4.1 Loss function11.4.2 Multirelationship decoder11.5 Message passing and graph neural networks11.5.1 The message-passing framework: A neural conversation11.5.2 Motivation and intuition: Why message passing works11.5.3 The basic GNN model11.5.4 Message passing with self-loops11.6 Generalized aggregation and update methods11.6.1 Neighborhood normalization11.6.2 Neighborhood attention11.6.3 Multihead attention and transformer connections11.6.4 Generalized update methods11.7 The synergy of GNNs and LLMs
12 Node classification and link prediction with GNNs
12.1 Node classification for anti-money laundering applications12.1.1 Input data12.1.2 Graph processor: Data preparation12.1.3 Graph processor: Homogeneous PyG graph12.1.4 Encoder–decoder architecture12.1.5 Evaluation and analysis12.2 Link prediction for movie recommendations12.2.1 Input data12.2.2 Graph processor: Data preparation12.2.3 Graph processor: Heterogeneous PyG graph12.2.4 Encoder–decoder architecture12.2.5 Evaluation and analysis
Part 5 Information retrieval with knowledge graphs and LLMs
13 Knowledge graph–powered retrieval-augmented generation
13.1 AI agents13.2 Chatting with the LLM13.3 Challenges in the production environment13.4 Chatting with the AI about private data13.4.1 Retrieval-augmented generation13.4.2 Vector-based RAG limitations13.4.3 Graph RAG13.4.4 Reasoning agents13.4.5 Let’s chat with our KG
14 Asking a KG questions with natural language
14.1 Querying a knowledge graph in the policing domain14.1.1 Enabling domain experts with knowledge graphs14.2 RAG for KG querying: Capabilities and challenges14.2.1 RAG effectiveness with complete context14.2.2 RAG fragility with incomplete retrieval14.3 Schema-based approach for querying KGs14.3.1 Understanding and using graph schemas14.4 Think like an expert: Using metadata for enhanced querying14.5 Intent detection: Understanding user expectations14.5.1 Classifying by visualization type14.5.2 Is it data, documentation, or just complaining?14.6 From schema to LLM-ready context14.6.1 Schema extraction and representation14.6.2 Enriching schemas with descriptive annotations14.6.3 A practical approach to schema representation14.7 It’s time to think: Understanding LLM reasoning14.7.1 The order matters: Answer first vs. reasoning first14.7.2 Thinking in queries: From text to Cypher14.7.3 Structuring output for reliable query generation14.8 Response summarization: From results to insights
15 Building a QA agent with LangGraph
15.1 Building the LangGraph pipeline15.1.1 System architecture overview15.1.2 Configuring pipeline components15.1.3 Schema translation service15.1.4 State management design15.1.5 Pipeline agent implementation15.1.6 Pipeline integration layer15.2 Streamlit application15.2.1 Application overview15.2.2 LangGraph integration15.3 Expert-emulating investigation15.3.1 Identifying the initial case15.3.2 Spatial analysis of surveillance coverage15.3.3 Vehicle pattern detection15.3.4 Context-aware request refinement15.3.5 Historical record analysis15.4 Future directions and enhancements15.4.1 Learning from use15.4.2 Enhancing core capabilities15.4.3 Advanced evolution paths
appendix A Introduction to graphs
A.1 What is a graph?A.2 Graphs as models of networksA.3 Representing graphs
appendix B Neo4j
B.1 Introduction to Neo4jB.2 Installing Neo4jB.2.1 Installing a Neo4j serverB.2.2 Neo4j Desktop installationB.3 CypherB.4 Installing pluginsB.4.1 Installing APOC CoreB.4.2 GDS installationB.5 Cleaning
appendix C Building knowledge graphs from structured sources
C.1 MicroRNA–disease association: WarmupC.1.1 Key conceptsC.1.2 Business understandingC.1.3 Data understandingC.2 Building the miRNA knowledge graphC.2.1 Importing known miRNA–disease connectionsC.2.2 Importing the disease ontologyC.2.3 Importing miRNA informationC.3 Exploring and analyzing the miRNA KG
references

Content preview from Knowledge Graphs and LLMs in Action

10 Graph feature engineering: Manual and semiautomated approaches

This chapter covers

Manual feature engineering techniques for nodes and relationships in graphs
Combining domain expertise with semiautomated extraction in a graph representation
Real-world applications of feature engineering

The success of machine learning (ML) on graphs depends on a fundamental challenge: how to effectively represent graph elements (nodes, relationships, and entire graphs) as vectors that ML algorithms can process. This representation step, often called vectorization or featurization, determines how well our models can learn and make predictions.

Although modern ML algorithms—from traditional approaches like logistic regression and random forests to ...

Become an O’Reilly member and get unlimited access to this title plus top books and audiobooks from O’Reilly and nearly 200 top publishers, thousands of courses curated by job role, 150+ live events each month,
and much more.

Read now

Unlock full access

More than 5,000 organizations count on O’Reilly

O’Reilly covers everything we've got, with content to help us build a world-class technology community, upgrade the capabilities and competencies of our teams, and improve overall team performance as well as their engagement.

Julian F.

Head of Cybersecurity

I wanted to learn C and C++, but it didn't click for me until I picked up an O'Reilly book. When I went on the O’Reilly platform, I was astonished to find all the books there, plus live events and sandboxes so you could play around with the technology.

Addison B.

Field Engineer

I’ve been on the O’Reilly platform for more than eight years. I use a couple of learning platforms, but I'm on O'Reilly more than anybody else. When you're there, you start learning. I'm never disappointed.

Amir M.

Data Platform Tech Lead

I'm always learning. So when I got on to O'Reilly, I was like a kid in a candy store. There are playlists. There are answers. There's on-demand training. It's worth its weight in gold, in terms of what it allows me to do.

Mark W.

Embedded Software Engineer

Publisher Resources

ISBN: 9781633439894Publisher Support Publisher Website

Cloud Computing

Data Engineering

Data Science

AI & ML

Programming Languages

Software Architecture

IT/Ops

Security

Design

Business

Soft Skills

Knowledge Graphs and LLMs in Action

by Fabio Montagna, Giuseppe Futia, Alessandro Negro, Vlastimil Kus