Database Reference
In-Depth Information
by the Topbraid Composer editing and query tool. Pellet ICV uses an OWL integrity
constraint ontology and translates this automatically into SPARQL queries to vali-
date RDF data. If the query results indicate integrity constraint violations, Pellet ICV
can also provide automatic explanations of why this has happened in order to assist
debugging and data improvement.
8.4 LINKING TO EXTERNAL DATASETS: TYPES OF LINK
In Section 8.5, we outline the process for designing and creating the data links, but
first we look at the big picture of the types of links that can be created and when
they are appropriate to use. There are four different cases under which RDF data can
be linked; three occur when there is correspondence at the instance level, between
resources in the two different datasets, and the other is at the class level, between the
vocabularies describing the two datasets.
8.4.1 c
orreSpondence
betWeen
c
laSSeS
We can link classes or properties in the two RDFS ontologies, which Heath and
Bizer (2011) call
vocabulary links.
If we believe the classes are equivalent—so every
instance in one class will also be an instance in the other (although the two parties
may know about different instances)—we can use the
owl:equivalentClass
relationship. Note that this does not make any statements about instances within the
classes matching at all. Similarly,
owl:equivalentProperty
can be used to
state that a property in one vocabulary is equivalent to the property in another. If we
cannot be sure that the classes or properties match exactly,
rdfs:subClassOf
,
rdfs:subPropertyOf
,
skos:broadMatch
, or
skos:narrowMatch
can be
used to express a looser relationship
.
8.4.2 c
orreSpondence
betWeen
i
nStanceS
: i
identity
Another type of link is the
identity link
, which specifies an equivalence relationship
between two data resources, using the
owl:sameAs
predicate. It is straightforward
to find this kind of link where the data is overlapping or the same but can be dif-
ficult to spot when the data is completely different but describes the same thing. For
example, a Building may be described by its spatial footprint by one person, but by
its address by another person. A third person, who wants to know the location of the
address, will want to match these two types of data and would find the
owl:sameAs
link between the two URIs very useful. We talk more about how to find this kind of
match in Section 8.6.
If Merea Maps is not sure whether anyone else has already minted a URI to iden-
tify the thing in which it is interested, it is quite acceptable for it to create its own
URI and, when ready, to seek out other URIs referring to that entity. These URIs are
known as
URI aliases
. Since Merea Maps' URI contains its namespace, the informa-
tion that it has published about that entity can easily be traced back to Merea Maps
through dereferencing, which is useful for others on the Web who want to know what
Merea Maps is saying about that thing. Multiple URIs about the same thing, linked
