- From: Graham Klyne <Graham.Klyne@zoo.ox.ac.uk>
- Date: Thu, 29 Mar 2012 15:56:40 +0100
- To: Timothy Lebo <lebot@rpi.edu>
- CC: Stian Soiland-Reyes <soiland-reyes@cs.manchester.ac.uk>, "public-prov-wg@w3.org" <public-prov-wg@w3.org>
Normally, it seems my refrain is to hold back from over-specifying. But in this case, I can't help wondering if sitting on the fence and saying nothing is the worst option. It suggests that in some cases, specializationOf(a,a) may be True, and in others it's False. I'd agree this doesn't need to be covered in the DM (part 1), but for strict interpretations of provenance (-constraints) I think a position probably should be taken. #g -- On 29/03/2012 13:15, Timothy Lebo wrote: > > On Mar 29, 2012, at 6:32 AM, Stian Soiland-Reyes wrote: > >> On Tuesday, March 27, 2012, Graham Klyne wrote: >> >> Personally, I prefer the choice that it is reflexive; i.e. specializationOf(a,a) always holds. As I recall, that seems to simplify some other inferential machinery. >> >> Yes, it solves the turtles-all-the-way problem last highlighted by Tim in this thread, if we also made specializationOf(x,y) imply alternativeOf(x,y), as the unknown top-level y can be specializationOf itself. However I think we dismissed the need for such an inference. >> >> Intuitively it sounds confusing to be an alternative to yourself, or a specialisation of yourself, but as we see above there could be special cases where you would want (a subproperty of) specializationOf/alternativeOf to be reflective, so I would simply say +1 for the conservative say-nothing approach for reflexivity. > > +1 > > -Tim > >> >> -- >> Stian Soiland-Reyes >> >> >> -- >> Stian Soiland-Reyes, myGrid team >> School of Computer Science >> The University of Manchester > >
Received on Thursday, 29 March 2012 14:57:29 UTC