W3C home > Mailing lists > Public > public-xg-eiif@w3.org > January 2009

Re: Universal Core?

From: C H <craighubleyca@yahoo.com>
Date: Wed, 14 Jan 2009 23:18:28 -0800 (PST)
To: public-xg-eiif <public-xg-eiif@w3.org>, Renato Iannella <renato@nicta.com.au>
Message-ID: <538988.20127.qm@web51407.mail.re2.yahoo.com>

Unfortunately the history of such "universal" data warehousing / ontology projects is very discouraging.  Someone who doesn't know this history has probably sold someone who doesn't know much about philosophy on the idea. 
Or else it's just a joke.

People have been searching for such "universally understood concepts of who, what, when, and where" for thousands of years.  At least they left how and why off the list, but that reduces it to maybe a problem several hundred years in duration.  So I'm not surprised you see blank pages...

Anyone seriously interested in such a problem would have to address some tough foundations problems, of which I'll illustrate just three quickly:

- Capital assets are the most important "what" and any ontology of these
  must deal effectively with the fact that they come at least in natural/
  ecological, manufactured/infrastructural, financial/contractual/formal
  and social/reputational/informal forms, and that's not even counting the
  individual capabilities/skills or instructions/protocols that we employ,
  which personally I believe are two more types of capital asset (making
  six)

  How far will such a "universal" get if it confuses such basic ontology
  categories as a person's body vs. their relationships vs. instructions
  they carry out well but which in theory any other person can carry out?
  Yet almost all "human capital" or "human resource" models do confuse it.

  Without such a model, planning logistics and maintenance is *very* hard
  and so given a bad model you can expect splintering away from the "core".

- "Who" has obvious problems of which the worst is assuming that it will
  always and immediately resolve to a specific living person by name.  In
  fact we deal with a lot of abstraction regarding "who" in daily life and
  most of our characterizations of people are quite faceted and fragmented
  by design.  For privacy and to avoid biases, we might have to divide our
  perceptions of people into financial, medical, legal or governmental and
  professional accreditation, social service or victim status, and so on...
  all without the person providing the service knowing the other facets...

  One of the worst problems in government is the legitimate division of
  all data about us into silos to protect us, which doesn't protect us or
  the public at key moments when the data really must all come together.

- While "when" and "where" seem objective to people used to coordinate and
  scheduling systems, the complex conditionals we deal with in daily life 
  are well beyond most ontologies.  We have proposed already to deal with
  trajectories/routes/schedules to allow rendezvous en route or to allow a
  heuristic approach to finding a person who is not where they are thought
  to be or belong.  But we fell short of really acknowledging that this is
  a path not just through spacetime but through a series of intents/choices
  that we would never be able to characterize adequately except maybe as a
  textual note regarding each choice along the path.  Those applications
  that are not safety-critical are going to have a rough time dealing with
  this generality and where they don't need to, the representation must
  thus collapse to a simple list of coordinates or a single spacetime blob.

  While this problem is actually solvable and must be solved as part of
  the EIIF project to avoid the horror of dozens of scheduling/routing/
  GPS conventions to represent trajectory/route, it's not obvious to all
  that it's necessary.  Heck some people still think dates can exist with
  no time zone, but of course they don't, and time zones are also problems
  when you cross them if your representation assumed that all times along
  the path would be in the same zone, i.e. the zone was assigned per path
  or (disastrous) per application via some kind of "local" assumption.;-(

  And let's not even start on the many ways people represent "where".  The
  GPS helps immensely but even if we all had them, they are not quite yet
  accurate enough to guide us to the correct office door or bed in a dorm
  or refugee tent, nor do we necessarily expect all those to be RF-tagged.
  Add in the relative location issue (it's the third door on the left of
  the aft cabin in hallway C on deck 3, which moves with the ship... so it
  can't be a GPS coordinate...) and you have something still very unsolved.

If someone does intend to solve all these problems, I would like to help, but I do not frankly believe that the project is more than an idea yet...

--- On Wed, 1/14/09, Renato Iannella <renato@nicta.com.au> wrote:
> Universal Core (UCore) is a federal information sharing
> initiative that supports the National Information Sharing
> Strategy and all associated Departmental / Agency
> strategies. UCore enables information sharing by defining an
> implementable specification (XML Schema) containing agreed
> upon representations for the most commonly shared and
> universally understood concepts of who, what, when, and where.
> 
> https://ucore.gov/
> =====
> 
> Seems relevant (to say the least!) - but the web site is
> empty....does anyone have any insights?
> 
> 
> Cheers...  Renato Iannella
> NICTA


      
Received on Thursday, 15 January 2009 07:19:09 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Thursday, 15 January 2009 07:19:09 GMT