Knowledge graph
A knowledge graph is a graph-structured representation of entities and their semantically meaningful relationships, often including associated attributes or facts.
AI-Powered Search
A knowledge graph is a graph-structured representation of entities and the relationships and attributes connecting them.
- How it is used
- Supports domain-aware query interpretation and semantic search by relating entities and concepts beyond keyword matches.
- How deep it goes
- Defines nodes and edges, distinguishes explicit graphs from index-backed SKGs, and works through expansion; defers broader query-time integration patterns.
- Examples
- Uses an ontology with people and food, an Advil query expansion, and indexed job postings and forum discussions.
- What you can do after
- Apply it via search-engine SKG traversal to find related terms and choose weighted query-expansion strategies.
Knowledge Graphs and LLMs in Action
A knowledge graph is an evolving graph data structure of typed entities, attributes, and meaningful named relationships that integrates structured and unstructured domain data.
- How it is used
- Supports domain-specific integration, reasoning, search, recommendation, visualization, machine learning, and grounded intelligent applications.
- How deep it goes
- Explains structure, semantics, integration, construction, querying, and validation, while deferring complete ontology design and unstructured-data extraction details.
- Examples
- Uses healthcare, drug discovery, customer support, clinical decision support, biomedical integration, and archival-text graphs as concrete applications.
- What you can do after
- Build it from scratch using an ontology, mapped data sources, validation, graph queries, visualization, and downstream analysis.
Read how Knowledge Graphs and LLMs in Action explains knowledge graph →
Natural Language Processing in Action, Second Edition
A knowledge graph is a graph-structured representation of entities, concepts, and their relationships.
- How it is used
- Grounds chatbot answers in connected facts and supports questions about relationships between entities.
- How deep it goes
- Develops triples, extraction, queries, and multi-hop inference through examples, while noting that answers require background facts already in the graph.
- Examples
- Uses a military-rank question and Python inspection of NELL entity–relation–value rows to make connected facts concrete.
- What you can do after
- Build it from scratch by extracting relationship triples, then query connected facts to answer questions or check chatbot claims.
Read how Natural Language Processing in Action, Second Edition explains knowledge graph →
Domain-Specific Small Language Models
A structured graph representation of entities and relationships used here to enrich RAG with explicit connections among concepts.
- How it is used
- Augments RAG with explicit entity relationships to support multihop reasoning and broader search across document collections.
- How deep it goes
- Builds graphs through extraction, composition, Leiden community detection, summarization, and global or local search, while deferring alternative graph stores.
- Examples
- Uses document chunks, structured entity-and-relationship JSON, NetworkX graphs, community summaries, and combined intermediate answers.
- What you can do after
- Build and query a GraphRAG prototype with Ollama, Pydantic, NetworkX, Leiden community detection, and generated community summaries.
Read how Domain-Specific Small Language Models explains knowledge graph →
Introduction to Generative AI, Second Edition
A knowledge graph is a conceptual representation of entities as nodes and their relationships as edges.
- How it is used
- Uses a knowledge graph as an analogy for reasoning about language-model knowledge and hallucinations.
- How deep it goes
- Defines nodes and edges and imagines confidence on relationships, but does not map actual model weights into a graph.
- Examples
- Shows a small diagram of entities and relationships and mentions TriviaQA as a question-answering fine-tuning dataset.
- What you can do after
- Evaluate/understand why a model may guess when information is limited or conflicting, without treating the graph analogy as an inspectable map.
Read how Introduction to Generative AI, Second Edition explains knowledge graph →
Essential GraphRAG
A knowledge graph is a structured representation of entities, attributes, and relationships that supports semantic connections among information.
- How it is used
- Improves RAG by connecting structured facts and unstructured context for accurate, explainable, interconnected retrieval.
- How deep it goes
- Explains nodes, relationships, mixed data, extraction, JSON output, and graph import, without detailing a complete graph schema.
- Examples
- Uses employee hierarchies, tasks, articles, legal contracts, payment terms, and active-agreement counts as retrieval examples.
- What you can do after
- Build it from scratch by extracting structured facts from text, modeling relationships, importing JSON into Neo4j, and supporting precise retrieval.
WordPress Power Toolkit: Harness AI to build next-level websites
A knowledge graph is a structured network of entities and labeled relationships that represents interconnected knowledge for querying and analysis.
- How it is used
- Supports AI-oriented optimization by making relationships among content entities machine-readable.
- How deep it goes
- Defines the purpose of structured relationship markup without detailing a graph model or implementation.
- Examples
- Illustrates relationships among entities, topics, and events with a visual relationship diagram.
- What you can do after
- Apply it by structuring relationships among content entities to facilitate AI analysis.
Read how WordPress Power Toolkit: Harness AI to build next-level websites explains knowledge graph →
LLMs in Production
A knowledge graph represents entities and the relationships between them as a graph of nodes and edges.
- How it is used
- Improves RAG retrieval when questions require traversing relationships among linked entities.
- How deep it goes
- Defines nodes and edges, compares graph and vector retrieval, and identifies query-generation and document-conversion hurdles.
- Examples
- Uses Gal Gadot’s relationship to Jaron Varsano and his occupation to illustrate multipart retrieval.
- What you can do after
- Evaluate when relationship-aware graph retrieval outweighs the simplicity of vector-based document search.
The Art of AI Product Development
An organized graph of entities and explicitly represented relationships that can be queried for facts or context.
- How it is used
- Improves RAG retrieval when entity relationships and connected context matter more than keyword or text similarity alone.
- How deep it goes
- Explains graph construction, subgraph traversal, hybrid retrieval, and schema tailoring, but defers deeper implementation details.
- Examples
- Uses companies, products, acquisitions, competition, regulations, and market events as graph entities and relationships.
- What you can do after
- Build it from scratch by modeling domain entities, defining relationships, traversing subgraphs, and integrating graph retrieval into RAG.
Read how The Art of AI Product Development explains knowledge graph →
AI Agents and Applications
A knowledge graph is a connected representation of entities, relationships, and associated information intended to support querying and reasoning.
- How it is used
- Supports relationship-aware retrieval, complex querying, reasoning, inference, and explainable answers in LLM applications.
- How deep it goes
- Defines nodes, edges, triples, extraction, querying, and GraphRAG, but stops short of full database implementation.
- Examples
- Uses Roberto, Inter Milan, Serie A, RDF triples, a KG-RAG architecture, and a schema-constrained Cypher prompt.
- What you can do after
- Build it from extracted entities and relationships, query it with SPARQL or Cypher, and synthesize answers from retrieved graph data.
Read how AI Agents and Applications explains knowledge graph →
Graph Neural Networks in Action
A knowledge graph is a heterogeneous graph that represents entities and relationships with explicit semantic context, metadata, and schema information.
- How it is used
- Supports querying, pattern identification, connection discovery, and relationship inference over richly contextualized heterogeneous data.
- How deep it goes
- Defines semantic entities, relationships, metadata, and schemas, then illustrates them through an academic network without implementation details.
- Examples
- Uses a university physics research network containing professors, students, papers, topics, departments, and typed relationships.
- What you can do after
- Evaluate how explicit semantics and metadata enable more targeted graph queries and analysis.
Read how Graph Neural Networks in Action explains knowledge graph →
AI Agents in Action, Second Edition
A knowledge graph is a graph-structured representation of entities, entity types, observations, and relationships among them.
- How it is used
- Supports relationship-aware retrieval and persistent long-term memory for agents.
- How deep it goes
- Explains nodes, edges, observations, traversal, and memory workflows, while deferring graph database internals.
- Examples
- Uses a Back to the Future script, Micheal-to-Calgary relationship, and agent experiences as concrete graph examples.
- What you can do after
- Operate/configure a local MCP memory server and structure agent experiences for retrieval, recording, and relation building.
Read how AI Agents in Action, Second Edition explains knowledge graph →
Related concepts
- GraphRAGused by · GraphRAG is a retrieval-augmented generation approach that uses graph-structured knowledge and relationships to retrieve context for language-model responses.
- heterogeneous graphkind of · A heterogeneous graph is a graph containing multiple types of nodes, relationships, or both.
- graphkind of · A graph is a mathematical structure consisting of vertices connected by edges that represent relationships or connections.
- edgehas part · An edge is a graph connection between two vertices that represents a relationship and may be directed, labeled, or weighted.
- language modelused by · A language model is a computational model that represents patterns in linguistic tokens to estimate sequence probabilities and predict or generate language from context.
- nodehas part · A node is a discrete element in a structured system, such as a graph, tree, network, or distributed system, representing an entity, state, component, or computational unit and potentially connected to other elements.
These guides are generated from the text of Manning books by Manning's knowledge graph and link to the sections they draw on. The full books are available on liveBook and at manning.com.