12.2 Selecting an Ontology Language 287
- OWL Lite. This limited form of OWL was intended more for develop-
ers than for serious use. It allows a developer a first step on the way to
supporting the more substantial OWL-DL and OWL Full languages. - OWL-DL.For applications that fit the description logic approach, this is
a very effective ontology language. Unfortunately, many domains do not
fit. - OWL Full.This is the richest and most flexible of the web-based ontology
languages. It is also the least supported. Inference using OWL Full can be
slow or can fail entirely. This is not a flaw of existing tools, but rather is a
fundamental aspect of this language.
The major ontology languages can be divided into these main groups:- XML DTD and XSD.
- XML Topic Maps.
- RDF and the three OWL languages.
Ontologies within a single group are mostly compatible with one another.
XSD has more features than XML DTD, and it is easy to convert from a DTD
to a schema. Similarly RDF and the OWL languages differ from one another
mainly in what features are supported. Converting ontologies from one of
these language groups to another can be difficult. Converting from the first
group to one of the other two is especially problematic. Topic Maps, RDF,
and OWL require that all relationships be explicit, while XML relationships
are mostly implicit. As noted in the list above, there is an approach that com-
bines the first and third groups. Developing an ontology using this technique
is relatively easy, but it has the disadvantage that one is making no use of the
expressiveness of RDF and OWL.
Note that in the discussion of ontology languages above, the concern was
with conversion of ontologies from one ontology language to another, not
transformation of data from one ontology to another. Data transformation,
which we discussed at length in chapters 9 through 11, can involve trans-
forming data within the same ontology language group as well as between
language groups. Transformation can also involve data that are not web-
based or data that are not based on any formal ontology. While making a
good choice of an ontology language can make the transformation task eas-
ier, developing correct transformation programs can still be difficult.