The Resource Description Framework - Linked Data: A Geographic Perspective - page 78

5.6 RDFS
This section addresses how to model RDF data using the RDF Schema language RDFS.
In previous chapters, we looked at data modeling from a Geographic Information
(GI) perspective, but now we take a more computer science-based approach.

5.6.1 C oNCepTs aND i NsTaNCes : i iNsTaNTiaTioN aND H ieRaRCHy iN RDFs

In our examples in this chapter, we have been working with data representing a
real-world object, the Isis Tavern. The data was originally stored in a relational
database, in a table containing data about Topographic Objects. We showed how to
specify using RDF that the Isis Tavern was a Pub, and that Pubs were Topographic
Objects. This introduces two ideas that it is vital to handle when dealing with the
Semantic Web: instantiation and hierarchy . Instantiation simply means that a thing
is a member of a category. In this topic, we tend to refer to these categories as
Concepts, when we are dealing with the GI world, or Classes , when taking a com-
puter science approach. So, we have the class of Pubs, and we say that the Isis
Tavern is an instance of a Pub; that is, it is a member of the class of Pubs. 8 This idea
of instantiation uses the relationship “is a,” which is short for “is an instance of.”
The second idea, hierarchy, may be more familiar to the reader and refers to when
one class is a subclass of another; that is, it takes all the meaning of the superclass
but also has additional, more specialized, properties. For example, Pub is a subclass
of the Topographic Object class. We usually use the relationship “is a kind of” to
describe this hierarchical relationship: “A Pub is a kind of Topographic Object.”
This means that the Pub class inherits all the statements made about Topographic
Object and can have additional descriptions. Note that we have to be careful about
statements that are made about a particular pub, for example, that the Isis Tavern
is next to the River Isis, versus statements that hold true for all pubs, for example,
that Pubs sell Beer. If we state that all Pubs sell Beer, and that the Isis Tavern is
a Pub, then we can infer that the Isis Tavern sells Beer. However, saying that the
Isis Tavern is next to the River Isis, and that the Isis Tavern is a Pub, only tells us
that at least one Pub (the Isis Tavern in this case) sells Beer, not necessarily that all
of them do. We talk more about this sort of logical inference in Chapter 9 regarding
the topic of OWL.

5.6.2 V oCabuLaRies aND o NToLogy

RDF allows us to make statements, in the form of triples, about things in our domain ,
or subject of interest; but to specify which classes we expect in the domain and what
properties can be used with each class, we need something extra: an ontology.
While a vocabulary is the set of terms (class and property names) that can be
used in the domain, an ontology 9 spells out how these vocabulary terms can be used
with each other—it specifies the permitted structure. For example, a vocabulary for
the topographic domain would include the terms “ TopographicObject ,” “ Pub ,”