- From: Irene Polikoff <irene@topquadrant.com>
- Date: Fri, 13 Nov 2015 13:45:55 -0500
- To: Arthur Ryman <arthur.ryman@gmail.com>
- CC: "public-data-shapes-wg@w3.org" <public-data-shapes-wg@w3.org>
Arthur, Yes, this helps to clarify the question. Personally, the self-consistency of the design paradigm used is probably the most important factor in making it easier for me to understand it. Meta-classes or no meta-classes, in itself, makes no difference to me as long as I see a clear design pattern. Irene On 11/13/15, 1:30 PM, "Arthur Ryman" <arthur.ryman@gmail.com> wrote: >Irene, > >I'll clarify. > >Holger is proposing a model that involves meta-classes. I am proposing >a model that does not involve meta-classes. > >Personally, I find models that use meta-classes to be harder to >understand than those that do not use meta-classes. Do you find models >the use meta-classes to be easier to understand? > >-- Arthur > > > >On Fri, Nov 13, 2015 at 12:02 PM, Irene Polikoff <irene@topquadrant.com> >wrote: >> Arthur, >> >> I think it was Einstein who is credited with saying that "everything >> should be as simple as it can be, but not simpler.¡± As many, I find the >> statement insightful and agree with it. So, I think in principle we are >>in >> agreement on this, but then there is a matter of judgement and agreement >> over what different people consider to be as simple as it can be. >> >> As for DL, I don¡¯t really know what it means in our context to ¡°keep >> within bounds of DL¡±. I see DL as a pretty complex topic and don¡¯t >> understand its relevance here. Are there applications for which DL is >>the >> simplest way to go and what are they? I don't have the definitive >>answer, >> but I doubt that data validation or UI description are it. >> >> If you are saying that for ease of understanding DL chosen not to have >> instance to also be a class and this precedence proves that such >>approach >> is "as simple as it can be, but not simpler¡±, I don¡¯t quite follow the >> reasoning. First, I am not certain that ease of understanding was the >> motivation. As I heard it, this had to do with some limitations of >>tableau >> algorithms and concerns about decidability. Second, in trying to use OWL >> while staying (for whatever reason) within DL, many people found this >> separation too limiting for their modeling. They asked for it to be >> removed. Further work on the algorithms found that this limitation was >>not >> necessary and it was removed. >> >> >> Irene >> >> >> >> >> >> On 11/13/15, 7:54 AM, "Arthur Ryman" <arthur.ryman@gmail.com> wrote: >> >>>Irene, >>> >>>As a general rule, I think we should keep the SHACL model as simple as >>>possible to make life easier for our target users. I think we can keep >>>within the bounds of DL. What is your opinion? >>> >>>-- Arthur >>> >>>On Thu, Nov 12, 2015 at 2:44 PM, Irene Polikoff <irene@topquadrant.com> >>>wrote: >>>> >>>> Arthur, >>>> >>>> Prior to OWL 2, OWL DL indeed had a strict limitation regarding >>>> disjointness of classes and individuals, but this limitation was >>>>removed >>>> in OWL 2 even for DL. Users wanted to be able to have the same thing >>>>as >>>>a >>>> class and an individual and further work on the tableau algorithms for >>>>DL >>>> revealed that they can cope with this. At least, this is my >>>>understanding >>>> of where things stand today. >>>> >>>> Best, >>>> >>>> Irene Polikoff >>>> >>>> >>>> >>>> >>>> On 11/12/15, 2:24 PM, "Arthur Ryman" <arthur.ryman@gmail.com> wrote: >>>> >>>>>Irene, >>>>> >>>>>I am referring to OWL DL. The partitioning of things into classes, >>>>>properties, and individuals allows you to express description logics >>>>>in OWL. This is a restricted style of modelling which is simpler to >>>>>understand and makes certain computations more tractable. >>>>> >>>>>-- Arthur >>>>> >>>>>On Thu, Nov 12, 2015 at 2:07 PM, Irene Polikoff >>>>><irene@topquadrant.com> >>>>>wrote: >>>>>> I may have mentioned this before, but in case I didn©öt, I do not >>>>>>believe >>>>>> it is correct to say that the idea behind OWL is not to allow >>>>>>meta-classes >>>>>> and to have classes, properties and individuals to be disjoint. >>>>>> >>>>>> Irene Polikoff >>>>>> >>>>>> >>>>>> >>>>>> >>>>>> >>>>>> On 11/11/15, 11:24 PM, "Arthur Ryman" <arthur.ryman@gmail.com> >>>>>>wrote: >>>>>> >>>>>>>I feel that people have an easier time understanding models in which >>>>>>>meta-classes are absent. This is the idea behind OWL and description >>>>>>>logic in which things are either classes, properties, or >>>>>>>individuals. >>>>>> >>>>>> >>>> >>>> >> >>
Received on Friday, 13 November 2015 18:46:35 UTC