Re: shapes and classes: different

On 2/5/15 7:32 AM, Richard Cyganiak wrote:
> You make it sound like attaching shapes to classes is something that
> one would only want to do in closed systems. I disagree with this. I
> expect that ontologies designed for use in open systems will also
> include shapes in their definitions. I know I would have liked to
> include some in several ontologies that I’ve worked on.
>
>>> - […] we should not force future systems to couple shapes with
>>> classes when they don't need to be coupled.
> We should also not force future systems to decouple shapes and
> classes when they don’t need to be decoupled.

The way that the Dublin Core community sees this working is through
published application profiles. For data which I want to share, and
which I want to be shared widely even with folks outside of my immediate
community, I want two things:

1) a minimally constrained ontology that is designed for re-use
("minimally constrained semantics" - T Gruber). This doesn't mean "no 
shapes" but it may well mean "insufficient for closed world validation 
needs"

2) a profile through which I express (in a machine-actionable form) MY
view of MY data (which may not be how others see or choose to use the
data I provide). This profile is what I validate against.

This serves both those who may wish to make use of my data, but who are
not likely to engage in a data exchange agreement with me, as well as my
community partners, who wish to do some alignment of their data with
mine for inter-system linking.

The ability for different users to have different "views" of the data is
key to the success of a heterogeneous community of users. Even within
closed or semi-closed systems different views must be possible because 
there are different materials being described and a variety of users and 
uses. It seems obvious that the same data may be re-used even within 
closed systems, and I think this is what Peter P-S was referring to in 
his response to Holger's example. And, btw, we've been doing this in 
databases for decades, haven't we?

I promised to find some documentation. This article has some diagrams
that may suffice, even if folks don't have time to read the text:

      http://dlib.org/dlib/january10/hillmann/01hillmann.html

I don't know that this is the best way to accomplish this, but it is one
idea that is circulating.

kc

-- 
Karen Coyle
kcoyle@kcoyle.net http://kcoyle.net
m: 1-510-435-8234
skype: kcoylenet/+1-510-984-3600

Received on Thursday, 5 February 2015 16:17:27 UTC