Hi Calle,
I don't clearly understand your point of view.
It is very hard to build a classification, since it needs to be immediately
complete and accurate.
It is easier to build an ontology, since you can start with 2 words and a
link, and end up with 100 000 words and millions of links.
The only problem is the transition between both steps. And the energy it
requires.
I have been able to build a 35 000 terms Lexique ; I could give it to you,
but unfortunately it is in French. If you know a way we can translate it,
say English, German and Spanish, the open source community would have a very
strong starting point and work all together on the links.
Regards,
Philippe
----- Original Message -----
From: "Calle Hedberg" <[EMAIL PROTECTED]>
To: <[EMAIL PROTECTED]>
Sent: Monday, July 09, 2001 5:35 AM
Subject: Re: Ontologies and Linguistics
> Hi,
>
> Andrew wrote that
>
> > The debate, therefore, is between those who believe one should setup
> > committees and spend millions to formulate a comprehensive set of
> > "reserved words" and mandate exclusive use of the resulting "word-list".
> >
> > Vs. starting with a small and incomplete set but allow users to build as
> > they go.
>
> I agree 100% with the latter approach with regard to most health info
> systems.
> A similar strategy has been followed in the development of the Internet,
by
> the way (start small and lean, use it in practice, allow
> decentralised growth).
>
> I prefer not to use the term "building" or even "development" for such
> prototyping
> processes, though - I prefer the term "cultivation" because of its complex
> socio-cultural aspects. Every step users take are usually mired in
> contradictions, conflicts, discussions, negotiations and compromises.
(I've
> usually found a dialectic perspective to fit these processes well, but
> that's another discussion.)
>
> Another crucial issue, though, is that there must be structures in place
> that allow (read: compel) users to share what they have built/cultivated
> with other users, and that such shared information gradually is turned
into
> 'toolboxes' and standards. In other words, innovation rooted in actual use
> AND standardisation must go hand in hand.
>
> I have a good example of this from South Africa: The National Health
> Information Systems / SA committee decided already in 1994, with WHO
> support, to develop a National Data Dictionary along the lines of
> Australia's.
>
> The problem was that most dominant players believed that a "real" Data
> Dictionary must consist of at least 4-5,000 items - otherwise it was not a
> "real" Data Dictionary - and that this therefore had to be developed by a
> national
> "authorised" workgroup. End result: The Data Dictionary workgroup
struggled
> for years,
> consulted with every conceivable party on every item big and small (most
> managers regrettably perceived it as a
> technical project of little interest to them), and managed to define only
> 30-40
> data elements before it disintegrated. These defined elements were also
not
> used
> in practice since there was no vehicle in place to bring them out to
users.
>
> Last year the Health Information Systems Programme developed a simple Data
> Dictionary tool as part of our District Health Information Software, and
> initially included around 700-800 data elements from the
national/provinical
> Essential Data Sets. Since the Data Dictionary despite its implementation
in
> a rather primitive prototype now was firmly linked to a piece of software
in
> daily use countrywide AND based on the provinces own data sets, we have
seen
> a rapid growth in the number of people working on the definitions etc of
> THEIR OWN Essential data elements.
>
> Final point: In my opinion, most large-scale "building" of ontologies and
> coding schemes are either outright proprietary or they are turned into
> somebody's "private" domain. That's something is "open source" in the
> standard meaning of the words does not hinder active gatekeeping in
> practice.
>
> I have lately, for instance, had discussions with various WHO managers
about
> getting access to the source code of their "free" software, or in some
cases
> tried to just get a copy of the compiled versions. I've had very little
> success, which again just shows that WHO is no different from other UN
> Organisations: They are happy to provide support to developing countries,
> but only on their own terms.
>
> Regards
>
> calle
>
> *********************************************
> Calle Hedberg
> 3 Pillans Road,
> 7700 Rosebank, SOUTH AFRICA
> Tel/fax (home): +27-21-685-6472; Cell: +27-82-853-5352
> Email: [EMAIL PROTECTED]
> *********************************************
>
