SUO: Ontology case study
Folks,
Bill Andersen and I were speaking a moment ago and he pointed out that I
hadn't related an application of ontology that we did a while ago. My hope
is that this case should point out one simple, concrete application of an
ontology, in a deployed business environment.
During the dot-com boom we worked with a Internet-based real-estate
company to solve a database integration task. I was surprised to find out
as we started the project that what consumers know as the Multiple Listing
Service database that realtors use is actually a collection of some 30,000
locally-developed databases that record information about homes for sale in
a particular geographic area. All the databases cover the same basic
information, but all the table names, field names, and symbols may be
different from one database to the next.
We created an ontology of real estate in first order logic, using an
upper ontology. We then "compiled" the ontology by hand into a relational
database. We then wrote scripts, again by hand, to map the contents of
each MLS database into our common database. The ontology was robust and
comprehensive enough such that after the first couple databases, and we
eventually mapped several dozen of them, there were no changes to the ontology.
The company's web site went live for several months, using our
ontology-based database, and then they exhausted their funding and went out
of business.
Of course this is just an anecdote. I doubt it's going to change
anyone's mind who already has a strong opinion about the usefulness of a
single ontology for supporting integration. But at the very least, it's a
concrete instance of the productive use of a particular ontology on one
commercial integration task. I've provided references in previous messages
to other government research projects where we've used ontologies for
integration as well.
One impact of an ontology that I find undeniable is that at the very
least, a formal ontology can serve as a more precise set of comments about
the meaning of database tables and fields than informal English.
This also points out some fruitful areas of research for tool
builders. It would be nice to compile the ontology to an SQL DB schema. I
believe Bill's company is working on that. Another extremely valuable tool
would be one that aids in doing the mapping from disparate databases to a
common database. I'm doubtful about the prospects for doing that
completely automatically, but hopefully that tools could be developed that
would help.
I hope this is helpful.
Adam
Adam Pease
Teknowledge
(650) 424-0500 x571