RE: Use machine-readable standardized data formats / Use non-proprietary data formats

Annette,

My earlier proposal to limit to tabular data (which I think can be expressed
in CSV and JSON too) was to make it easy to tell what is in scope and what
is not.

More loose definitions like 'structured data' make it again hard to
distinguish. For example, what is the fundamental difference, in terms of
how data is expressed and used, between a 3D model of a sculpture and a 3D
molecular model, between behaviour of liquids in a physics experiment and
traffic flows on a motorway?

The problem I think is when you say "constrain the concept of data so that
it matches our *feeling* for what we are writing BPs about". I am just
wondering if there is a common *feeling* among the members of this group.

Makx.



> -----Original Message-----
> From: Annette Greiner [mailto:amgreiner@lbl.gov]
> Sent: 14 August 2015 00:38
> To: Makx Dekkers <mail@makxdekkers.com>
> Cc: public-dwbp-wg@w3.org
> Subject: Re: Use machine-readable standardized data formats / Use non-
> proprietary data formats
> 
> You're right. We keep coming back to it. When we discuss it, it seems to
me
> we have some sort of general agreement but just haven't figured out how to
> express it. For example, by using them as counterexamples, you clearly
> suggest that structured 3D models of statues or excavation sites and
highly
> structured legislation documents are not what we are after. How can we
> constrain the concept of data so that  it matches our feeling for what we
are
> writing BPs about?
> I think whatever we use should be something that a typical manager or web
> developer would consider a reasonable subset of the many things humans
> call data. Those are the people who will come to the document for
guidance.
> I think we need to avoid trying to *define* data, because it is too vague
a
> concept, and attempts to define it universally, as a dictionary would,
work
> against our need for scoping.
> 
> Okay, I'll go out on a limb and give it a whirl.
> I think we could limit our work to structured sets of symbols (words,
> numbers, or images, still or moving) not intended to convey meaning on
their
> own, but representing observations, facts, or forecasts.
> 
> To be clear, I'm not offering this as a general definition of what data
is; that is
> not what we need. I'm suggesting this as a scope.
> -Annette
> 
> --
> Annette Greiner
> NERSC Data and Analytics Services
> Lawrence Berkeley National Laboratory
> 510-495-2935
> 
> On Aug 13, 2015, at 1:30 PM, Makx Dekkers <mail@makxdekkers.com>
> wrote:
> 
> > I agree we need a better definition of the scope.
> >
> > Limiting to structured data may be a sensible restriction. However,
> > while it would exclude images of paintings, highly structured 3D
> > models of statues or excavation sites would be in scope; it would
> > exclude an unstructured .txt file, but not the highly structured
> > legislation or procurement data that is already getting published as RDF
in
> projects that I know of.
> >
> > The problem we're having is that this discussion pops up every couple
> > of months and then dies out without a clear and citable decision
> > having been taken and recorded.
> >
> > Makx.
> >
> >
> >> -----Original Message-----
> >> From: Annette Greiner [mailto:amgreiner@lbl.gov]
> >> Sent: 13 August 2015 21:48
> >> To: Makx Dekkers <mail@makxdekkers.com>
> >> Cc: public-dwbp-wg@w3.org
> >> Subject: Re: Use machine-readable standardized data formats / Use
> >> non- proprietary data formats
> >>
> >> I think we do need to scope it, but limiting it to tabular data is
> >> too
> > restrictive.
> >> Even CSV and JSON wouldn't qualify. If you meant structured data, I
> >> think that could work.
> >> - Annette
> >>
> >> --
> >> Annette Greiner
> >> NERSC Data and Analytics Services
> >> Lawrence Berkeley National Laboratory
> >> 510-495-2935
> >>
> >

Received on Friday, 14 August 2015 08:18:33 UTC