12.8 Validating and Modifying the Ontology 315
- Dynamics.Can a property value change? This tests whether the property
is intrinsic or extrinsic and can affect the class hierarchy as discussed in
subsection 12.7.1. For example, can a patient change his or her name?
Consistency checking is the only validation activity that can currently be
done using automated tools. Obviously, consistency is fundamental to any
formal system that supports logical inference. If a formal system is inconsis-
tent, then every statement can be proven true (and also proven false, since
true = false in an inconsistent formal system). Nearly all XML parsers have
the ability to check consistency, but one must usually request that this be
done. XML parsers refer to consistency checking asvalidation, even though
this term more properly refers to much more than just consistency checking.
There are many RDF and OWL tools that can be used for checking consis-
tency:
- ConsVISor(Kokar et al. 2001) web service
- Euler(Euler 2003) downloadable software
- F-OWL(FOWL 2003) downloadable software
- Pellet(Pellet 2003) web service
- vOWLidator(vOWLidator 2003) web service
- WonderWeb(WonderWeb 2004) web service
Consistency checkers vary with respect to how they explain the problems
that are found. In addition to finding inconsistencies, most of the tools also
give advice about situations that are not inconsistencies but which could be
indicative of an error. Such a situation is called asymptomby analogy with
the medical notion of a symptom of a disease (Baclawski et al. 2004). The
ConsVISor consistency checker is unique in having the ability to produce
output that itself conforms to an ontology.
When flaws in the ontology design are revealed during validation, the on-
tology must be modified. Ontologies are also modified after they are pub-
lished. This can happen because new concepts have been introduced, exist-
ing concepts change their meaning, or concepts can be related in new ways.
When concepts and relationships change, it is tempting to modify the ontol-
ogy to reflect those changes. However, the danger is that programs and data
that depend on the ontology will no longer be compatible. Ontology modifi-
cation is also called ontology evolution. Certain modifications are relatively