Re: Place of UFO in ontology development

I would agree that it would be nice to have good UFOs. They would  
help us think about modeling problems and make our graphs more easily  
mergeable. But the point we are making  is that it is not because  
UFOs are impossible, difficult or not ready, or because there is not  
going to be a single UFO, that nothing can be done on the SemWeb.

Henry

On 5 Apr 2006, at 21:02, David Price wrote:

> The Semantic Web seems to me to be between a rock and a hard place.  
> The rock seems to be the view that UFOs are problematic. The hard  
> place seems to be the idea that people around the globe are going  
> to develop ontologies in isolation, then import ontologies  
> developed by others into their application and magically usefully  
> reuse them. While that approach might work for a few simple cases  
> like FOAF, to my mind it presents a serious problem for the  
> Semantic Web for anything even slightly more complex.
>
> Given the hard place, it's unclear to me why anyone would insist on  
> throwing the UFO baby out with the bath water. NIST hosted a  
> workshop a couple of weeks ago on UFOs so everyone has not yet  
> given up (see http://ontolog.cim3.net/cgi-bin/wiki.pl? 
> UpperOntologySummit ). Practically speaking, some level of UFO that  
> forces a rigourous analysis of the concepts in an ontology seems  
> quite useful to me. That there are incompatible UFOs is no reason  
> to throw them out. Almost all ontologies developed independently  
> are incompatible so how can the import and reuse approach work  
> either? In my 20 odd years experience with IT practitioners I've  
> found wildly varying degrees of model quality, believe everyone  
> would benefit from a good analysis approach and think a UFO can  
> play that role. The question seems to me to be just how "U" the  
> UFOs need to be and in what situations they solve a real-world  
> problem. The UFO approach does not have to be perfect, just produce  
> better results than the alternative.
>
> I've worked for more than 15 years in ISO/OMG on standards for  
> engineering modeling and so do have a bias in that direction. I'm  
> also part of a project trying to make practical use of ISO 15926-2  
> which was presented at the NIST summit. After seeing hundreds of  
> models/databases designed to support engineering applications I  
> also doubt that a statistical/structural analysis can determine the  
> intended semantics behind them but will try and keep an open mind.  
> It seems to me that useful ontologies are far more likely to be  
> engineered into existence so I'll end my comments and go back to  
> trying to do just that.
>
> Cheers,
> David
[snip]

Received on Wednesday, 5 April 2006 19:20:57 UTC