What Is a Knowledge Graph?
A knowledge graph represents entities, their attributes, and the relationships between them. Entities can be people, organizations, products, places, or other things relevant to a domain. By making their connections explicit, a knowledge graph helps people and software interpret information in context.
For example, a knowledge graph might show that Acme Corporation owns Account 123, holds Contract 456, and purchases Product A. Instead of treating these as isolated records, it represents how they connect. A user or application can then follow those connections to answer questions such as “Which contracts and products are associated with this customer?”
An ontology can define the concepts and relationship types used in a knowledge graph. It might specify what a Customer or Contract is and what it means for a customer to hold a contract. Not every knowledge graph uses a formal ontology, however, and the degree of structure depends on the use case.
What Problems Does a Knowledge Graph Address?
Enterprise information often sits in separate systems, with relationships left implicit or buried in application logic. A CRM may identify a customer, a billing system may record its accounts, and a contract system may describe its agreements. Consumers must work out which records refer to the same organization and how those records relate.
A knowledge graph makes those connections explicit and available for reuse. It can support:
- Connected discovery – Help users find related information, such as the products, agreements, and accounts associated with a customer.
- Understanding dependencies – Show how business or technical assets connect, supporting impact analysis and investigation.
- Shared context – Preserve identified relationships so each new application does not have to reconstruct them independently.
- AI retrieval – Help AI applications identify relevant entities and follow relationships to retrieve information for a task.
As organizations add more data sources and consumers, the value of understanding connections grows. A knowledge graph can help reduce fragmented views of the business, provided that its entities and relationships are accurately identified and maintained.
How a Knowledge Graph Works
Entities are commonly represented as nodes and relationships as edges. Attributes describe details such as an account identifier or a contract status. For example, Acme Corporation and Contract 456 can be nodes connected by a “holds contract” relationship.
Building a useful graph requires more than drawing connections. Teams must identify entities across sources, define what relationships mean, retain information about their origin, and maintain them as source information changes. Knowledge graphs can be stored in graph databases or constructed through other approaches, including virtual access to distributed data.
A graph is not automatically current, complete, or governed. Its usefulness depends on the quality of the source information, the accuracy of the relationships, and the access and update processes supporting it.
How Knowledge Graphs Relate to Other Data Concepts
The following distinctions use the same customer example to show the role of each concept.
| Concept | Question it Answers | Customer Example |
|---|---|---|
| Data Semantics | What does this data mean? | “Active customer” means a customer with at least one current contract. |
| Ontology | What kinds of things exist and how can they relate? | Defines Customer, Account, and Contract, and specifies relationships such as a Customer owns an Account. |
| Knowledge Graph | Which specific things are connected? | Represents Acme Corporation as owning Account 123 and holding Contract 456. |
| Semantic Layer | What is the standard way to represent shared entities? | Makes such entities as the Active Customer definition and Contract Value calculation reusable across dashboards, applications, and AI agents. |
An ontology describes the domain model; a knowledge graph represents connected information using that model or another defined structure. A semantic layer makes business definitions and rules usable by consumers. These approaches can work together, but none should be treated as a substitute for all the others.
How a Knowledge Graph Differs from a Graph Database
A graph database is a technology for storing and querying connected data. A knowledge graph is a representation of information and its meaning. A graph database can support a knowledge graph, but storing data as nodes and edges does not by itself establish trusted business definitions, provenance, or governance.
Where Denodo Fits
Knowledge graphs can represent business entities, such as customers and contracts, or the data ecosystem itself, including datasets, definitions, applications, and governance controls. Denodo’s Data Marketplace addresses this second need by connecting data products with the business and technical context around them.
Users can explore relationships among data products, glossaries, data dictionaries, applications, lineage, and governance controls. This enterprise knowledge graph helps users understand how data is defined, governed, used, and connected. Denodo’s universal semantic layer complements that relationship visibility by making business definitions, metrics, and relationships reusable across distributed data.
Denodo’s AI data layer combines semantic trust with connectivity, live delivery, and governed access to create active context. Relationship visibility contributes to this broader outcome, while the architecture supplies current enterprise information and applies policies at the point of access. Together, these capabilities support governed, reusable enterprise data products for analytics, applications, and AI.