What it is
In the W3C's Resource Description Framework, a graph is a set of triples, each with a subject, a predicate and an object. SPARQL is the W3C family of specifications for querying such graphs.
A second tradition is the property graph, in which nodes and relationships carry properties. Its query language, GQL, is reported as published as an international standard, ISO/IEC 39075, in April 2024.
It is different from a semantic layer. A semantic layer defines metrics and dimensions so that business figures are calculated the same way everywhere. A knowledge graph models entities and how they relate. The two can be used together, but they answer different questions.
Why it matters
A graph makes relationships explicit, which tabular models leave implicit. That helps when the question is about connections: which records describe the same customer, or what depends on what.
The standards are mature. DII has read no evidence of how widely enterprises use knowledge graphs, and makes no claim about a trend.
The problem it addresses
DII editorial view
Tabular models leave relationships implicit: that two records describe the same customer, or that one thing depends on another, has to be worked out each time. A knowledge graph stores entities and their relationships directly, so those questions can be asked of the data.
Business and organisational relevance
DII editorial view
A knowledge graph is a long-lived modelling investment. It pays back where the questions that matter are about how things connect, and it needs people with the skill to maintain the model.
Practical use cases
- Connecting records about the same entity across systems
- A shared vocabulary of business concepts
- Supplying structured context to AI applications
Benefits and opportunities
- Relationships are first-class and queryable
- Open standards for exchange
Risks, limitations and dependencies
- Modelling a domain as a graph takes specialist skill and sustained effort
- Two families of technology (RDF and property graphs) with different standards and tools
- Value depends on the quality of the underlying identifiers and definitions
Prerequisites for adoption
DII editorial view
- Reliable identifiers that say when two records are the same thing
- People with the skill to model and maintain the graph
- A narrow first domain
Implementation considerations
- Choose the family deliberately. RDF and property graphs have different standards, tools and strengths.
- Start with a narrow domain. Modelling takes specialist skill and sustained effort; scope decides whether it is finished.
- Settle identifiers first. A graph is only as good as the keys that say two records are the same thing.
The DII Radar assessment
Catalogue status
Not yet assessed. DII has not published a Radar assessment for Knowledge Graphs. It is listed in the catalogue so that it can be explained. No relevance, direction, strategic impact or enterprise adoption is given for it, and it is not plotted on the Radar.
What to consider next
DII editorial view
- Choose between RDF and property graphs deliberately
- Start with a narrow domain
- Settle identifiers first
Evidence and references
The sources DII read for this explainer. Each entry gives what the record holds; nothing is added.
- What is a data fabric? (IBM Think) A vendor description, cited as such. View source for reference 1 (external site)
- RDF 1.1 Concepts and Abstract Syntax View source for reference 2 (external site)
- SPARQL 1.1 Overview View source for reference 3 (external site)
- GQL Standards The ISO catalogue page refused the request and was not read, so the standard is cited through this site. View source for reference 4 (external site)