- From: Eric Jain <Eric.Jain@isb-sib.ch>
- Date: Mon, 16 Aug 2004 18:51:28 +0200
- To: "Deborah L. McGuinness" <dlm@ksl.Stanford.EDU>
- CC: public-swbp-wg@w3.org
Deborah L. McGuinness wrote: > While it is true that for some applications it may be useful to minimize > the number of triples, > i do not think that should be an argument that is used to help choose a > modeling scheme in a best practices working note. If you are dealing with several hundred million triples, anything that minimizes the amount of triples is a best practice :-) Also, you could argue that the solution that fullfills your requirements while requiring the fewest triples is in fact the simplest and most natural solution. > I think we want to have modeling solutions that we believe represent > good modeling choices that capture the representation and do not > introduce additional confusion or complication. I agree, but do not understand why the proposed alternative (having classes in the hierarchy be instances of some metaclass) is any more confusing or complicated than any of the other solutions. If there are any practical problems with this approach, please let me know, as I am not very familiar with the theory. > I am not clear that you could claim that this approach does not create > additional maintenance either. By additional maintenance I meant having to keep a set of instances in addition to the classes in the hierarchy, as in approach 2 and 3.
Received on Monday, 16 August 2004 16:51:23 UTC