We saw in the last edition about the Semantic Web how ontologies were databases or knowledge bases about an area of human knowledge. In this edition, we are going to explore ontologies a bit more. An ontology itself may be divided into several components. First is the taxonomy. A taxonomy is basically just a collection of concepts or abstract ideas. For example, in the financial world, a taxonomy might consist of bonds, stocks, interest rates, and so on.

The next element of an ontology consists of relationships between the different concepts in the taxonomy. For example, in the financial example, the taxonomy might include interest rate swaps and currencies, and there might be a relationship between interest rate swaps such that a given swap pays in a given currency. Next there are facts which give us specific information about various objects in the taxonomy. Facts are quite similar to relationships although they tend to be a bit more static-a fact might be that a stock is a type of financial instrument-unlikely to change. Finally, a rule is something telling us to take some form of action based on the taxonomy, relationships, and facts.

So what does an ontology buy us? After all, it has been possible for years to build these components of an ontology using standard object oriented technology (C++, Java, etc.)? What is new in the concept of an ontology? The basic advantage is that an ontology is portable. It contains information about a particular sphere of activity that doesn't change depending upon the organization that is using it. Thus, for example, in the past when a financial services firm wanted to build any type of financial application, there would first need to be an enormous amount of effort spent between the IT staff and the finance people trying to understand the language that each community spoke. Only when an agreement had been reached on some form of mutually understood "language" for representing financial concepts in IT would it then be possible to begin the design of the system. But, unfortunately, once the system design progressed to any significant degree, the understanding that existed between the technical and business people would essentially be buried in the project specifications. If the project was successful-if enough key people on both sides stuck around until it was done-then the basic momentum established at the beginning might be enough to get a good system working. However, there would be no standard, documented language for communication between the IT people and the business people, so as soon as a new project started, all that painful effort to get the two groups talking would have to be repeated.

The idea of an ontology is that there exists, from the beginning of a project, a language of communication which technical and business people agree on, and which can be used as a starting point for the project. This doesn't mean that there isn't still a lot of work involved. Goldman Sachs and Charles Schwab may start with the same ontology but they are going to want to use it differently. But the ontology provides a very good starting point for communication between the different groups. In a sense, every successful IT project involves a certain amount of AI because a computer system is built which does something intelligent-which solves a problem better than it had previously been solved. Yet up to now there has been no way for the AI aspects of projects to be modularized in such a way that they could be repeated from one project to the next. Ontologies provide a good framework for changing this.

Ontologies exist in a large variety of different human domains at the present time, including finance, taxation, e-commerce, insurance, and medicine. They are no doubt going to improve in quality and number in the months and years ahead. How do we go about developing an ontology? We'll talk about this a little more in upcoming editions, but one important tool for developing ontologies is the Resource Description Framework (RDF). RDF is an XML-based markup language for developing ontologies and metadata.

More information about ontologies may be found at Tom Gruber's What is an Ontology? page.

In the next edition, we will talk more about the Resource Description Framework (RDF).


Home: Ramalila.NET



All copyrights are maintained by respective contributors and may not be reused without permission. Graphics and scripts may not be directly linked to. Site assets copyright © 2000 RamaLila.com and respective authors.
By using this site, you agree to relinquish all liabilities and claims financial or otherwise against RamaLila and its contributors. Visit this site at your own risk.