72 4 The Semantic Web and Bioinformatics Applications
Figure 4.6 RDF graph for a typical Medline citation. Resource nodes are shown
using ovals, text nodes are shown using rectangles. All links are labeled with the
property. Theml:prefix stands for the Medline ontology.<rdf:Property rdf:ID="DateCreated">
<rdfs:domain rdf:resource="#MedlineCitation"/>
<rdfs:range rdf:resource="#Date"/>
</rdf:Property>Therdf:IDattribute is used fordefininga resource. Each resource is de-
fined exactly once. At this point one can also annotate the resource with
additional property values, as was done above. Therdf:aboutattribute is
used forannotatinga resource. Use this when one is adding property values
to a resource that has been defined elsewhere. Therdf:resourceattribute
is used forreferringto a resource. In terms of statements, userdf:IDand
rdf:aboutfor a resource that is to be the subject of the statement, and use
rdf:resourcewhen the resource is to be the object of the statement. We
leave it as an exercise to the reader to restate the molecule DTD as an RDF
ontology and to write the nitrous oxide molecule document in RDF.
The last important feature that distinguishes RDF from XML is its incor-
poration of built-in inference rules. The most important built-in rule is the
subClass rulebecause this is the rule that implements inheritance and taxo-
nomic classification of concepts. Although there are many notions of hier-
archy, as discussed in section 1.5, the most commonly used is the notion of
taxonomy which is based on the mathematical notion of set containment.