Database Reference
In-Depth Information
object. Objects may in turn be classes of things or individuals or may be represented
by a value (also known as a literal).
A very significant advantage of the triple data structure is that one of the current
bugbears of data integration, that of discovery and manipulating diverse data struc-
tures that are often poorly documented, simply goes away; all data on the Linked
Data Web has the same structure: the triple.
Linked Data is also built on well-known and proven technologies such as HTTP
(Hypertext Transfer Protocol). Everything is identified using HTTP URIs, mean-
ing that it is possible to mint identifiers such that their uniqueness can be ensured.
By making the URIs dereferenceable, they can be used to point to the data that
underlies the concept, individual, or property in question. Hence, the Linked Data
Web can be seen as a network of nodes made up of URIs representing classes and
individuals, which are linked by URIs relating to properties. Publishers are able to
build links between their data and the data published by others simply by adding
triples that establish the links.
Data on the Linked Data Web can be described using ontologies; these may be
simple RDFS (RDF Schema) ontologies that do little more than specify vocabu-
laries used by the Linked Data, or they may be more complex and expressed in
OWL (Web Ontology Language), thus enabling sophisticated inferences to be made
over the data. An advantage of describing Linked Data using ontologies is that they
remain independent of the application code, so the meaning of the data is much more
visible than in conventional solutions, where much of the structure is buried in the
application code.
The development of OWL ontologies can be quite a complex process, and a
number of methodologies exist to assist the process. As OWL is based on first-order
logic, it can be quite difficult for nonlogicians to become familiar with the more
subtle aspects. It is also easy to confuse OWL classes with classes from the
object-oriented (OO) paradigm as they look very similar on the surface.
The following are points to remember:
•
The basic and universal structure of Linked Data is the triple, which repre-
sents the relationship between a subject and object associated by a predicate
and expressed as
subject predicate object
.
•
The Linked Data Web uses well-proven technologies such as HTTP and the
URI scheme.
•
Everything is identified using URIs.
•
Publishers are able to build links between their data and the data published
by others simply by adding triples that establish the links.
•
All things other than data values are identified by URIs, and these URIs
should deference to return the data that is associated with the URIs.
•
Linked Data may be described using ontologies. The ontologies themselves
may be very simple (often expressed using RDFS) and may be little more
than specifications of the vocabularies used to describe the Linked Data, or
they may be more complex, expressed using OWL, to enable inferences to
be made from the Linked Data.
