68 4 The Semantic Web and Bioinformatics Applications
</isStoredIn>
</gene>
</contains>
</locus>
The corresponding RDF graph is shown in figure 4.2. The element names
alternate between names of classes and names of properties, depending on
the “striping” level. Thuslocusis a class,containsis a property,gene
is a class, and so on. Attributes are always names of properties. The nodes
with no label (i.e., the empty ovals in the graph) are calledblankoranonymous
resources. They are important for conveying meaning, but they do not have
explicit URIs. RDF processors generate URIs for blank nodes, but these gen-
erated URIs have no significance. The use of blank nodes in RDF complicates
query processing, compared with XML. However, high-performance graph-
matching systems have been developed that are efficient and scalable. This
will be discussed in section 6.6.
Figure 4.2 RDF graph for an XML document. Resources are represented using
ovals, and rectangles contain data values.
Every link in an RDF graph has three components: the two resources be-
ing linked and the property that links them. Properties are themselves re-
sources, so a link consists of three resources. The two resources being linked
are called thesubjectandobject, while the property that does the linking is