untitled

(ff) #1

82 4 The Semantic Web and Bioinformatics Applications


<owl:ObjectProperty rdf:ID="stores">
<owl:inverseOf rdf:resource="#isStoredIn"/>
</owl:ObjectProperty>
<owl:ObjectProperty rdf:ID="cites">
<owl:inverseOf rdf:resource="#isCitedBy"/>
</owl:ObjectProperty>

There are no other ways to construct a property in OWL. However, there are
some property constraints:


  1. owl:FunctionalProperty.Such a property may relate a subject to at most
    one object. If a particular subject is related to two resources, then those
    two resources must be the same. Mathematically, such a property is a
    partial function.

  2. owl:InverseFunctionalProperty.Such a property is allowed to relate an
    object to at most one subject. This is the same as constraining the inverse
    property to have theowl:FunctionalProperty.

  3. owl:SymmetricProperty.A symmetric property is the same as its inverse
    property.

  4. owl:TransitiveProperty.This imposes the mathematical transitivity con-
    dition on the property.


The semantics of OWL is defined in (Patel-Schneider et al. 2004). The se-
mantics is in terms ofinterpretations. The more commonly used term for an
interpretation is amodel, and that is the term that will usually be used in
this chapter. The detailed definition of an interpretation is complicated, but
it is essentially the same as an RDF graph. An OWL document specifies a
collection of statements. One can also specify that itowl:importsother OWL
documents, and the statements in the imported documents are also regarded
as having been stated. The collection of all the statements forms atheory
about the world. Such a theory is consistent with an infinite collection of
models. These models may be regarded aspossible worlds. As additional
facts become known, the collection of possible worlds gets smaller, as the
new facts eliminate possibilities. This process is analogous to scientific rea-
soning, except that in science one attaches probabilities to possible worlds,
and observations modify these probabilities (using Bayes’ law). This is cov-
ered in chapter 14. Logical theories, by contrast, assign no weights to the
possible worlds.
Free download pdf