What is OWL

This is a general introduction to ontologies and OWL. There are a number of general resources available that also cover these same topic, notably ontogenesis, so this introduction is fairly brief.

What is an ontology?

The struggle with defining what an ontology is has been fairly long and traumatic; here, we define it as a representation of knowledge in a domain. In a sense, this is true of any data model, so more specifically, ontologies normally consist of a set of concepts with properties or relationships between them; most ontologies are either frame based, that is have a set of defined slots into which relationships fit, or have use a description logic -- a form of logic built for the purpose.

For a longer description please see the article on ontogenesis.

Why use an ontology?

Ontologies are most useful when a complex domain needs to be represented computationally; see ontogenesis for a more detailed description.

In general, the knowledge needs to be fairly complex to make use of an ontology worthwhile; ontologies largely describe things categorically, so are not useful, for instance, where there is a lot of probabilistic or numerical data; this is where a statistical model will work better.

To a programmer, an ontology looks somewhat like a type system, although ontology languages are generally more expressive, but do not interact with an underlying programming language.

Probably the most heavy use of ontologies has been in biomedicine, where very large ontologies have been used to annotate data.

What is OWL?

OWL is a W3C standard language that can be used to build ontologies. OWL has two main selling points; firstly, as a W3C standard, it integrates with all the other W3C standard, builds on RDF and XML and is fully web capable; and, second, it maps to a well defined description logic, the practical upshot of which is that it is possible to perform computational reasoning over OWL statements, draw inferences and detect inconsistencies.

Further Resources

The OWL pizza tutorial provides a good overview for building ontologies in OWL, although it uses Protege rather than tawny.

The Good OD tutorial provides a good description of both the philosophical and computational issues of ontology building.

tawny-pizza is a port of the Pizza ontology to use tawny and provides a nice example ontology.

tawny-simple is a literately-programmed use of tawny describing the inferences that can be made on the basic of OWL statements.