- From: Drew McDermott <drew.mcdermott@yale.edu>
- Date: Sun, 28 Nov 2004 18:28:32 -0500
- To: public-sws-ig@w3.org
> > [me] > > I think you are still confused about where natural language ends and > > internal representation begins. > > [Bijan Parsia] > "Formal" representations? There's not a real sense in which an OWL > document is *internal* to anything, in most respects. One perenially needs a term for "declarative notation used by a computer that may use mnemonics derived from natural language but is not actually natural." I use the term "internal representation." I guess one could say "formal" instead, but it's a stretch in the case of some representations to believe they're formal in any substantive sense. > > Internally all we have to do is use > > two different symbols, and the two entities are "differentiated." > In OWL, this is necessary but not sufficient. By "differentiated" I meant "not provably equal," not "provably not equal." > You need to assert or > imply a differentFrom in order for them to be required to be distinct > (although they will not be required to be the same absent an asserted > or implied sameAs; in most cases, differentiation of distinctly named > individuals is contingent). > > > (The two names can be made different just by having one be > > fluvial:bank and the other be mercantile:bank.) We can give them > > different properties, but a KR system can believe two entities are > > different even if it doesn't know any properties that differentiate > > them. > > Yep, at least in many, as we have explicit inequality. What we don't > have is the Unique Name Assumption. > > Or we can avail ourselves of a really simple difference; > > for instance, we can declare that economic_entity is disjoint from > > geographic_entity, so all pairs of objects drawn from economic_entity > > x geographic_entity are given distinguishing features, because one has > > the feature "is an economic_entity" and the other has the incompatible > > feature "is a geographic_entity". > [snip] > > or just assert that the two individuals are distinct. If someone wants to split hairs (and there are lots of people hanging around with nothing better to do, I've noticed --- present company excepted), they might observe that asserting not(a=b) _is_ a property that differentiates a and b, to wit, the property of being b, which b has and a apparently does not. -- Drew
Received on Sunday, 28 November 2004 23:28:34 UTC