[profgui] Definitions needed for progress

All,

I seem to have forgotten to make the minutes of yesterday's profgui
meeting [1] public (I've asked DaveR to help), but there are many notes
in the Google doc that we were working on.[2] Refreshing my memory of
the discussion, there are a number of concepts that we need to clarify
so that we can use them consistently throughout the document.

I am recording here what we discussed. It is lengthy and complex so we
will need to break it down into individual topics for discussion. Some
link logically to already active github issues, so I will move these
comments to those. Others may require us to open new issues. Most of the
concepts in question are not limited to the profiles guidance document
so we need to consider them in the context of the entire suite of
deliverables.

If you know of issues or discussions that are relevant but have not been
listed here, please add them.

1. Constraint (also constrain) [3]
The two main uses seem to be
a. - strict (or "hard"): to constrain means to define something that can
be validated or measured. A constraint example would be: propertyX is
mandatory
b. - non-strict (or "soft"): to constrain means to make statements or
declare something. This example would be: propertyX is included in the
profile but is optional; no other rules are applied.

The group tended to favor the "soft" definition
Related issues:
https://github.com/w3c/dxwg/issues/721


2. "profile of"
What conditions entail that one thing is a profile of another thing?
a. - any use of elements (classes, properties), even if minimal (e.g.
just use one DC term)[4 - Tom Bakers response]
b. - use of all of the elements from a "base specification" and possibly
add elements from another another specification and/or other constraints.
c. - use of a "significant portion" of a base specification or other
profile.
d. - something is a profile of something else if the creator of the
profile declares it as such

Note that if there is to be any formal inheritance then some of these
definitions would not be usable.

Related issues:
https://github.com/w3c/dxwg/issues/486

3. Are "profile of" and "base specification" redundant?
It appears that every "profile of" has a base specification, so it seems
to be sufficient to simply talk about profiles. We have reduced the use
of "base specification" in this document.

4. "inheritance"
There are three types of inheritance:
a. inheritance of a "statement of conformance" which is the outcome of
the reuse of defined terms (and their constraints) (possibly to be
called reuse or a hierarchy)
b. the mechanism for handling individual constraints (axioms, algorithms)
c. inheritance of profiles or specifications, where implementation
requires accessing and including "upstream" elements in the target profile.

related issues:
https://github.com/w3c/dxwg/issues/643
https://github.com/w3c/dxwg/issues/642
https://github.com/w3c/dxwg/issues/748

5. term use v. profiling
Is use of terms from a vocabulary itself a profile or is something more
needed? a or b:
a. - something is a profile if it is a use of terms from other
namespaces even if it does not constrain
b. - a list of used terms is not a profile unless it defines or adds
constraints
c. - can something be a profile if it has a mix of constrained and
un-constrained terms?

This last brought up an interesting question: is DCAT a profile?

kc
[1] https://www.w3.org/2019/02/28-profgui-minutes.html
[2]
https://docs.google.com/document/d/1Y4jP4SGZMnt63EpjTX11-hW6-3mxlaq1i-Lbiw4tx1M/edit#heading=h.u65kvtxp5hmg
[3] See note here:
https://docs.google.com/document/d/1Y4jP4SGZMnt63EpjTX11-hW6-3mxlaq1i-Lbiw4tx1M/edit#heading=h.rmh835lzn3bp
[4]
https://lists.w3.org/Archives/Public/public-dxwg-comments/2019Feb/0005.html
-- 
Karen Coyle
kcoyle@kcoyle.net http://kcoyle.net
m: 1-510-435-8234 (Signal)
skype: kcoylenet/+1-510-984-3600

Received on Friday, 1 March 2019 19:42:16 UTC