RE: Global concept identification and reference: Published Subjects and decentrally provided identification points for notions

Hello Andy


> I just finished reading: "Published Subjects: Introduction and Basic
> Requirements".

Good. Hope everyone speaking about it has done the same :))

> The argument made by the PSI folks that anyone can
> issue PSI's is fundamentally flawed.

I like this kind of frontal attack ...

> This statement doesn't take
> into account the realities of copyright law.

Actually we thought a lot about it during TC work, even if it does not show in the first
deliverable. If you look at the set of PSIs for ISO 639 languages at
http://psi.oasis-open.org/iso/639/, you can see an example of a non-authorized publisher
(the OASIS GeoLang TC) "hi-jacking" ISO 639 standard (intellectual property of ISO). We
have always thought this set only as an example of what could be done, and that
eventually, those PSIs should be published by the original publishing authority (ISO)
under its own namespace.

> Since a PSI requires
> you to publish a human readable description of the concept, any
> non-authorized third party description will be in violation of the
> KOS publisher's copyright.  Therefore PSI's can only *legally* be
> issued by the publisher of the KOS.

I acknowledge indeed that intellectual property on subject indicators is a complex issue.
OTOH I would say that in the spirit of Topic Maps specification (which was at the origin
of the PSI concept, even if the scope of PSI application is far wider than Topic Maps),
like in the spirit of the Semantic Web, a *subject* (indicated by a PSI) belongs to no
one. Everyone can say and publish anything about anything, and indicate subjects through
PSI mechanism. Which would be in violation of intellectual property would be to copy whole
or part of material under copyright, and publish it under any namespace without agreement
of the original source. But this is not particular to PSIs ...

> The document initially has the
> focus that PSI's are for KOS publishers, but then they go off into
> the wind without thinking about the implications of suggesting
> *anyone* can create them.

Could you explain what makes for you the difference between a "KOS publisher", and
"anyone" publishing a KOS? If it's a difference in quality, users will tell in the long
run. Good PSIs will be the ones more used on the WEB, hence more referenced and more
visible etc ... Trust, for PSIs like for anything else on the Web, is a bootstrapping
build-up process that eventually will make the difference. Since Alexander has quoted the
collective book on Topic Maps (Park & Hunting: XML Topic Maps. Creating and Using Topic
Maps for the Web. Addison-Wesley 2002/2003) may I add that I already expressed that
bootstrapping viewpoint in Chapter 5 on PSIs. See in particular p.78 the proposal for an
"emptiness" subject indicator ...

> Certainly, once published, anyone can use them since the requirement
> is that the URI must resolve to a human readable resource that is a
> description of the concept in the KOS.  This implies that a PSI
> cannot be behind authentication schemes, but it's not strictly
> forbidden and I'm assuming it's not forbidden to allow KOS
> publishers to use PSI's in a private service oriented context.

Indeed. This is written also in the PubSubj recommendation. "Published" does not mean
necessarily "Public". Actually we have used in several Mondeca ITM applications for
customers the PSI mechanism for subjects strictly limited to intranet environments.

> Some KOS publishers will not be willing to use PSI's due to
> intellectual property concerns, but might be willing to provide
> opaque URI's to concepts in their KOS.

This is something I have hard time to understand, actually. How do you figure folks will
use identifiers defined in your namespace if you don't provide clear, non-ambiguous
descriptions of the subjects those identifiers identify (read : subject indicators). The
risk in doing so is, either to see users turn towards other identifiers, maybe less
trustable, but for which subject indicators are available, or to see less reliable third
parties provide second-hand hacked subject indicators for your own identifiers. Both are
bad scenarios to consider IMO.

But maybe you think we have to live in a world with "proprietary concepts" and
"open-source concepts", with the same arguments about quality as in any other proprietary
vs open-source debate.

Bernard

**********************************************************************************

Bernard Vatant
Senior Consultant
Knowledge Engineering
bernard.vatant@mondeca.com

"Making Sense of Content" :  http://www.mondeca.com
"Everything is a Subject" :  http://universimmedia.blogspot.com

**********************************************************************************

Received on Friday, 12 November 2004 14:53:59 UTC