[sc34wg3] RE: [tm-pubsubj-comment] RE: paradigmatic PSIs

[Mason, James David (MXM)]

Thanks, Bernard.

This discussion, when I first saw it, was just going around in private
messages to half a dozen people, not even on tm-pubsubj-comment. I thought
it  needed wider discussion. It also kept getting assertions that the ISO
standard was broken, which meant it should have been on the SC34/WG3 list.
So I tossed it out to list-world and accepted that I would get four copies
back, just trying to get it to the right people. 

I fully agree that we need wider addressability, as you suggest below. And
some of the mechanisms you point to, such as the OASIS registry, are part of
the right solution.

I jumped into this because so much of it was being carried out as an
abstract moan that the whole Topic Map world was going to come crashing down
if we didn't change everything and do it right now. That's absurd. We can do
a lot of things with TMs as they are currently specified and implemented.
Other things may have to wait, but that doesn't mean that the basic
mechanism is worthless.

What I wanted was for the participants to realize that there are a lot of
components in a larger solution, such as figuring out the most practical
means of registering and announcing PSIs. Not all of the components belong
in any one place, and certainly not all of them belong in ISO/IEC 13250.
Some components may belong in SC34, though not in 13250. Some of them
definitely don't belong in SC34. You have a committee in OASIS to work on
some components. For others, we'll just have to depend on help from the W3C
and others. 

I've yet to see a standard that fixes everything. All the workable standards
I've seen in the past 20 years have been compromises, and I'd hate to see
one that attempted to be a total solution. (Actually, I did see an attempt
at such. It was called ODA, and SC34's predecessors wasted entirely too much
time defending open, extensible standards like SGML and HyTime from it. Lots
of other people poured lots of  resources into ODA, all of which came to
nought. If I keep telling the Topic Map crew in SC34 that some things are
outside their job, it's because I don't want them to fall into the ODA
trap.)

'Nuff said. If I say any more, I'll sound like the moaners.

Jim


> -----Original Message-----
> From:	Bernard Vatant [SMTP:bernard.vatant@mondeca.com]
> Sent:	Wednesday, April 10, 2002 4:54 AM
> To:	tm-pubsubj-comment
> Subject:	Re: [tm-pubsubj-comment] RE: paradigmatic PSIs
> 
> Jim
> 
> First a technical request. Please let's stop replication of this thread on
> four lists +
> private copies. This debate is about PSIs, so its natural forum is
> tm-pubsubj-comment, and
> I assume people really interested in PSIs are subscribed to that list. If
> they are not,
> they should ;-)
> 
> I would like to put some elements on the table that IMO tend to show that
> your position is
> not sustainable any more against Steve's one, that I agree upon basically.
> Please believe
> I feel quite arrogant here, knowing the accumulated experience and wisdom
> you have in
> standards matters, that should push the relative freshman I am in those
> matters  to bow
> with respect and retire :))
> 
> But - I generally don't like those Big Words - we are maybe in the middle
> of a paradigm
> shift (said it) where the notion of PSI is bound to play a central role.
> I'll just put
> together some signs.
> 
> -- Any document, regulation, standard, concept ... that is not somehow
> addressable through
> the open Web is bound to be more and more confidential and eventually
> disappear to the
> profit of whatever is addressable. To the question "What do you speak
> about?", the answer
> is practically, and will be more and more: "See that URL". Both for humans
> and for
> computers. And this is not specific to topic maps. Topic maps have simply
> identified and
> now attempt to regulate a practice that has spread out as quickly as the
> Web. Because it
> is so simple, so evident. RDF uses it, namespaces use it. Like them or
> not, namespaces are
> here to stay, and widely be used, there again by both humans and machines.
> And there are
> more and more people saying that non-resolvable namespaces are bad
> practice, and that
> namespaces should resolve to some resource explaining to humans what they
> are about. So I
> think we are bound to a convergence between namespaces and PSIs.
> 
> -- W3C Web Ontology Working Group - Requirements for Ontology Web Language
> http://www.w3.org/TR/webont-req/#section-requirements
> "Unambiguous term referencing with URIs :
> Two terms in different ontologies must have distinct absolute identifiers
> (although they
> may have identical relative identifiers). It must be possible to uniquely
> identify a term
> in an ontology using a URI reference."
> 
> BTW see another message about a formal liaison kick-off
> between W3C WebOnt and OASIS PubSubj.
> 
> -- OASIS Standards Registry initiative http://www.oasis-open.org/stdsreg/.
> Among the objectives in the Charter:
> "3. A recommendation for how standards registries, both current and
> future, can be used together to allow common searching across the
> registries hosted by various organizations."
> 
> I figure "searching" there means both human searching and engine searching
> - that's what
> proposed standard metadata are about.
> And I figure "hosted" mean "published on the Web"
> 
> -- Last point. Having the notion of "role" for example defined by a PSI is
> maybe pointless
> if you consider only topic map applications. But if you look further to
> the big picture,
> other applications (whatever semantic application you can imagine) should
> be able to
> identify and disambiguate the concept of "role in topic maps" among other
> similar
> concepts. Imagine that the Common Logic effort
> http://cl.tamu.edu/minutes/stanford-minutes.html or some equivalent effort
> goes to the
> point of setting a common logical or mathematical ground for KIF,
> Conceptual Graphs, RDF,
> Topic Maps, UML ... How will one express that the concept of "role in
> topic maps" is
> mapped to whatever expression in KIF or CG or RDF is we have not PSIs for
> all of them?
> 
> This is a real challenge, of which proliferation of standards is one
> aspect, and for which
> PSIs will be critical tools. Interoperability is the general objective,
> although I agree
> with you that interoperability of Opera with Nuclear Security does not
> make much sense.
> But interoperability of topic maps with RDF and KIF and CG,
> interoperability of ontology
> languages, interoperability of standard concepts and terms does make sense
> and is
> critical. That's why I think we need those "paradigmatic" PSIs
> 
> In that context, I feel like concepts in ISO standards being not defined
> by PSIs being a
> major issue. This is not specific to ISO 13250, but a general remark about
> ISO process.
> 
> Regards
> 
> Bernard
> 
> 
> 
> 
> ----------------------------------------------------------------
> To subscribe or unsubscribe from this elist use the subscription
> manager: <http://lists.oasis-open.org/ob/adm.pl>