Andrea,
Thank you for bringing this issue back to the forum. My
background, and most of my work history has been in data and
information management systems development, starting with my
joining IBM in 1967. I've been a member of ACM-Sigmod since the
late 70s, and was responsible for one of the first attempts that
I'm aware of to use a commercial data dictionary product to manage
the software application interfaces of a large and complex command
and control database, designed according to the normalization
principles being laid down at the time by Ted Codd, Chris Date,
and Ron Fagin (among many others, I list these only because their
names spring to mind).
I am a dynamicist, which seems to give me a different perspective
on data and its use from what I see in this year's summit. My
perspective is applicable to all stored data whether representing
an ontology, a computer program, or live operational values for an
enterprise. In my view, data, and by extension an ontology,
ontology fragment, ontology module, or a composition of
ontologies, is not a priori a stable entity. This incipient
volatility (Dr. Ted Codd called it 'time-varying') to me is the
essence of the utility function for use/reuse or sharing. In
fact, if the data (ontology) is truly static, frozen, stabilized
or invariant, it's not very interesting to me. All data
represents some abstraction, and the permanence of each
abstraction is always open to question. Of course if the data is
being presented in an historical context (last years tax code) it
mustn't change but I'm assuming for the purposes of Ontology and
ontological commitment the actual semantics must be to some degree
either fluid or so ambiguous as to be useless in an _expression_ in
logic.
So here is how I would differentiate between use/reuse and
sharing:
- When it's made available within a community of interest
(however defined) for other members of that community to use
in some non-prespecified purpose, and the data is changed
after the fact of it's being applied to that purpose, the onus
is upon the supplier to notify those consumers of possible
impacts to them.. That's my view of reuse and reusability; the
intent and commitment of the supplier is that the dynamics of
evolution are accommodated in reuse protocols that allow some
level of cooperation among the participants.
In their use, the Federal and state Tax Codes have many of the
properties of an Ontology, and all users (Accountants,
lawyers, taxpayers, Turbotax...) must be periodically told of
the state of the codes to allow them to properly pay or report
value transactions to the Tax authorities. All of those
affected entities are collaboratively engaged in a cycle of
data reuse. Of course, they share the tax codes with the tax
makers and collectors, but the sharing is on the whole,
incidental
- So, sharing is not a collaborative exercise. Public
libraries are repositories of shared data. The NIH has
specific
rules for sharing of research data (which presumably
applies to BIOMED, OBO, and related health and medicine
ontologies developed under government funding). These rules
are almost exclusively focused on fostering availability of
the data, and imply that the most common form of sharing is
publication in some freely available medium (or document).
The data is shared when anyone else reads it. Collaboration is
not an explicit requirement (though it's acknowledged as a
desirable outcome in some cases). Most importantly, once in
final publication, there is very little effort expended to
correct it, and essentially no effort to coordinate those
corrections with others who may have been impacted by the
distinctions raised in using the pre versus post correction
data.
In summary, Reuse implies collaborative application of shared
knowledge among a community with concomitant communication of
changes to the information, while sharing of an Ontology is
equivalent to publishing it for others to use with no commitment
to coordinate changes or corrigendae, and in general, no
requirement to know who or how the "downstream" community might
employ the Ontology.
I don't know if the ideas I've expressed here have any bearing
on the communique for this year's Summit, but I do appreciate
the opportunity to air my thoughts.
Terry Longstreth - longstreth@xxxxxxx
On 4/17/2014 10:08 PM, Andrea Westerinen wrote: