While the web of Linked Data is, in its broadest sense, a set of RDF triples, there are often circumstances in which it is useful to describe a smaller grouping of triples. RDF statements are published to the web as documents conforming to a particular syntax, e.g. RDF/XML, Turtle, or XHTML+RDFa. These documents may be directly inter-linked using See Also relations. To enable user agents to select between links it is useful to indicate the type of document which a link is referencing.

Define a document type describing a conceptual or physical grouping of triples. Indicate where a specific document is of a particular type, including a Topic Relation such as foaf:primaryTopic to relate the document to the resource(s) it is describing.

#document type
foaf:PersonalProfileDocument rdfs:subClassOf foaf:Document.

#specific instance of document, with indication of its topic
<http://www.example.org/doc/john> a foaf:PersonalProfileDocument;
   foaf:primaryTopic <http://www.example.org/doc/john#me>.

<http://www.example.org/doc/john#me> a foaf:Person;
   foaf:name "John".

XML is a document format and XML schemas describe the valid structure of documents. In contrast RDF is defined in terms of sets of triples and schemas are used to support inferencing and description of data structures. It is often useful to describe specific collections of triples. For example within a triple store it is often useful to group triples into Named Graphs. These collections can be usefully annotated in various ways, e.g. indicating their provenance, creation date, origin, etc.

Collections of triples may be published to the web using a variety of syntaxes and mechanisms. It can be useful to partition data into a number of different documents, e.g. to simplify the publishing process or usefully present data to user agents. By annotating the documents to indicate their type we can usefully allow user agents to select specific collections that are more likely to contain information of interest. This can help target crawler behaviour or prioritise documents for de-referencing.

Using document types does not imply that a user agent can make assumptions about the structure or format of the data that will be retrievable. The document may contain information about any number of different resources, or use any RDF syntax.

Two well deployed examples of document typing in use today are RSS 1.0 and FOAF. RSS 1.0 is defined as both an XML and an RDF vocabulary and as such has a strict definition of document that aligns with its use in an XML context. FOAF however is an RDF vocabulary, but has still found it useful to define the notion of a foaf:PersonalProfileDocument which indicates that a document primarily describes a particular person (but may include additional data).

The Document Type pattern is most commonly used in conjunction with the See Also and Annotation patterns. It could also usefully be applied when referencing a Link Base, allowing a user agent to more easily discover Equivalence Links related to a specific resource(s).