Re: Rationales for OWL 2 new features

2009/4/29 Peter F. Patel-Schneider <pfps@research.bell-labs.com>:
> Several new features that were not in the initial list of new features
> have been added to NF&R recently to make NF&R comprehensive.
> Unfortunately, this highlights the fact that these new features do not
> have the same level of rationale as the other new features.
>
> These new features are:
> - data range boolean combinations
> - datatype definitions
> - annotation property axioms (subproperty, domain, range)
> - top and bottom properties
> - anonymous individuals
> - inverse properties
>
> None of these features have the same level of rationale as the initial
> list of new features but some of them have some rationale in NF&R.  The
> WG should probably ensure that each of these features does have an
> adequate rationale in NF&R.
>
> My suggestion is that whoever was behind each of these features (You
> should remember who you are!) should be responsible for determining
> whether the rationale in NF&R is adequate and producing a message to the
> WG so indicating or producing some rationale to put in NF&R.  It would
> be best if these rationales were not technical rationales.
>
> peter
>
> PS:  Here are my initial thoughts on which of these features have
> adequate rationale:

Absolutely true. This reflects the reality:

(A1)>  data range boolean combinations        no - no rationale or example
(A2)>  datatype definitions                   no - no rationale or example
(A3)>  annotation property axioms             no - except for subproperty
(A4)>  top and bottom properties              no - no rationale or example
(A5)>  anonymous individuals                  maybe - technical argument
(A6) >  inverse properties                     maybe - technical argument

All, A1 to A6, as opposed to other (initial) new features, numbered
from F1 to F15 in NF&R, are not features based on requirements,
examples from real UCs described somewhere => they miss rationale,
examples, UCs in NF&R

- A5 and A6 have been migrated from SS&FS to NF&R (at last minute before LC1)
The only reply about rationale I got, was: technical reasons.

- A1 to A4 have been added to OWL 2 after (quite) late WG discussions.
They have been recently added to NF&R to make it comprehensive in
response to Peter's review and QRG authors query  (to provide links!).
I do not remember the "champions" and clear rationale behind them.
It's obviously easy to invent family-style examples,  like in SS&FS or
the Primer, (e.g. for datatype definitions, to define AdultAge) and
even to relate some them  to an existing UC of the Appendix. But for
the purpose of a "Requirements" document, which seems to be the
favorite of some members, it would be prefered to provide examples
from *real* applications, real UCs, and better rationale.

-- 
Christine

Received on Thursday, 30 April 2009 06:20:54 UTC