Using ontologies

creating semantic patterns that connect

/ en/ article /ontology_en.phpSubjects: rdf/owl, topic maps, xml, xml definitions, controlled vocabularies, designing, information, knowledge sharing, maintaining, ontology, semantic network, standards, taxonomies, thesauri, understanding

In order to enable precise and accurate access to information resources a number of methods for assigning keywords to the resources in a controlled fashion – controlled vocabularies?, taxonomies? and thesauri? - are available today. Since each method has it’s own limitations and drawbacks, the call for a structure that allows for combining the strength of each and at the same time provides the flexibility to be applied in more than one way and for more than just a single purpose.

What is an ontology

An ontology can be envisioned as a kind of ‘semantic network?’ where the subjects (occuring as words and phrases) are related to one another using meaningful relationships that themselves could be subjects as well. An ontology provides a ‘context of understanding’ for information resources that enables people to better interpret the subject a resource is about. Besides this, the clear defined relationships within an ontology provides a mechanism that allows for intuitive navigation through a vast collection of information. The fundamental role of an ontology is to support knowledge sharing and reuse.

Designing an ontology

Choose your domain

An ontology for everything is a real monster, for sure. It’s a better idea to take a more pragmatic approach and start with setting up an ontology for the domain of your core interests of fields of activity (most likely the field of business you’re in). Once the choice for a core domain is made the design and construction can begin.

Quite often the building an ontology is seen as too awkward or as a hill too steep to climb. Mainly because the most common idea is that an ontology is only useful when it is completed (100% correct and huge). This is sometimes true, but most of the time starting off with a limited ontology - and a plan for improving and extending it over time - is useful enough. The added value from a relatively small ontology can already be significant. Most important is that there is a clear understanding what role an ontology is going to play and in what environment. As with most other tools and aids, defining a specification for it’s usage is a reasonable thing to do.

Heer met vogel There are several ways that the building of an ontology can be performed. A few of them will be referred to further on. Please bear in mind that in practise, a specific mix of these methods for building an ontology has to be used.

Getting familiar with the concepts

What ever route for developing an ontology is chosen, it is always useful, even required, to ensure that the organisation is familiar with the concept and the goals for an ontology. One, very interesting, way to do this is by organising design sessions with all the people, the domain experts, involved in this process. During these sessions an actual ontology is conceived in an interactive fashion. Besides getting acquainted with the concept, these sessions will produce a kind of a blueprint where the key-concepts for the envisioned ontology are identified like the types of subjects and relationships that should be a part of this structure.

Extract what is known already

It is not very useful to ignore what is already there. There a quite a number of resources that can be used as a source for the ontology and that can be incorporated without too much trouble. Quite often this can be done in an (semi-) automated fashion. If in previous situations controlled lists of keywords, taxonomies or thesauri were in use, these can be imported within an ontology. Other sources could be existing database schema's, XML definitions? for documents, or even publicly available standardised sets of subjects from the Internet.

Importance of standards

Since an ontology represents a specific domain of knowledge (your domain) and designing and maintaining this knowledge is actually an investment, a reliable format for storage and reuse is of importance. In order to guarantee that there is no dependency with software used and that the ontology is protected from ageing in a technical sense using a well defined open standard for storage and maintenance is clearly a good strategy. There are two options available: Topic Maps? or RDF/OWL?. Both of them use an XML? defined model that enables the exchange of the ontology between applications and / or organisations. Which one to use is dependent on the way the ontology is to be used. If it is required that both standards can be applied for the same ontology, this should be a requirement before the design process is started.

What ...

Designing ...

Importance ...