Database Reference
In-Depth Information
Appendix A
OWL Species
OWL (Web Ontology Language) is not a simple language; it has distinct dialects
or species and subspecies. Figure A.1 nicely shows this linguistic complexity.
However, things are not quite as bad as they may first appear. There have been two
versions of OWL standardized: OWL 1 became a World Wide Web Consortium
(W3C) Recommendation in 2004 (Dean and Schreiber, 2004) and consists of three
“species”: OWL Full, OWL DL, and OWL Lite. This was followed by OWL 2,
which was standardized in 2009 (W3C 2009); OWL 2 is an evolution of OWL 1.
As well as OWL 2 DL and OWL 2 Full, there are three sublanguages of OWL 2 DL
offered: OWL 2 QL, OWL 2 EL, and OWL 2 RL. So, what are the differences and
when should each be used?
OWL DL
1
is the most widely used version of OWL by far; indeed, often refer-
ences to “OWL” are actually referring to OWL DL. All OWL languages are based on
first-order logic; the
DL
in OWL DL stands for Description Logic, which is the name
given to the subset of first-order logic that OWL DL uses. OWL DL was designed to
balance expressivity (being able to express complex knowledge structures) against
computational completeness (any statement that can be made in the OWL DL lan-
guage is either true in the ontology or false) and decidability (if the truth or falsehood
of a new statement can be determined based only on the set of statements provided
in the ontology). It also took into account which practical reasoning algorithms were
available at the time the language was designed to compute the logical consequences
of the ontology statements. That is, OWL DL was designed to be able to say as
much as possible and make complicated, detailed statements about the world while
knowing that the reasoning calculations would actually come back with an answer
within a finite time. Every ontology language has to take a stance on this issue of
balancing expressivity against efficiency of reasoning: how much can be said versus
how quickly, whether one or every answer can be reached by the reasoner, or indeed
whether an answer can be reached at all.
OWL Lite is a sublanguage and a subset of OWL DL that was specified for
the benefit of early tool builders who wanted to get started. Now, it is rarely used,
and you are unlikely to come across any ontologies that specify themselves to be
“OWL Lite.” (On a side note, many ontologies will not need to use every single type
of expression that is possible in OWL DL, but their “expressivity” (which level of
complexity of expression they use) will in practice be determined by the content
of the domain, not by the mathematics of the various types of logic. Hence, it is
nearly impossible to find a domain that fits precisely into the OWL Lite subset.)
OWL Full offers more expressiveness than OWL DL by forgoing any computational
guarantees; that is, no answer may ever be reached. It uses different semantics than
241
