[topicmapmail] PSIs - alternatives: comment

[Alexander Sigel]

Simon
(and topicmapmail list subscribers),

> Have people considered the idea of not trying to have 
> unique PSIs, but instead having a group of PSIs refer to each 
> other, when their sponsors consider that they are equivalent?

The ideal of the collocation objective demands that everything on the same subject should be accessible
from one virtual access point (here: starting from a known published subject).
Hence, arbitrary proliferation of Published Subjects is to be avoided.

However, it is perfectly possible that for the same subject there exist several published subjects,
minted by different issuers, even using different identifiers (PRIs, public resource identifiers).

Consider e.g. two publishers of classification systems wishing to retain their proprietary class names and numbers,
even if they agree that both stand for the same subject.

Now anyone wishing to do so (including the publishers) can publish a mapping between both published subjects
in topic map form, saying e.g. they are identical. Such a mapping statement does not even have to be true.

Citing Steve Pepper's example, I could e.g. say (in LTM syntax) that GM, the country code for Germany as issued by the CIA,
is the same as 276, as issued by ISO 3166:
---
[@"http://psi.oasis-open.org/iso3166/#276"
@"http://psi.example.org/cia/countries/GM"]
---

This mapping is only valid for the issuer, not for all people using arbitrary topic maps. So if you use it,
you could either accept it as true, or scope it e.g. by the issuer (or not use it at all).

This way, your proposal to have several PSIs referring to each other is possible.

> There could be a impartial host for PSIs - maybe 
> topicmaps.org? - respected by the entire community. That 
> might work, if it were open for everyone to put them in 
> freely or for a small fee. But if that doesn't happen, if two 
> or more organisations decide they need PSIs on the same 
> topics, how can anyone decide between them rationally?

Everyone can publish PSIs, and everyone can host PSI registries, and everyone can issue, scope and use mappings.
One does not need impartial hosts, and no such decision is needed, IMHO. The user community will decide
by issuing most trusted mappings.

> A "democratic" alternative would be to have PSIs resolving to 
> a file that named other, equivalent, PSIs, along with a 
> description of the subject as seen by that authority.
This should rather be done with mappings in topic map form,
since they are better automatically processable and have a more explicit semantics than text.

> Maybe near-equivalents could also be included as such, but not just 
> any related subjects.
Yes, one can use an arsenal of semantic relations from knowledge organization theory to model
typically needed relation types.
In a paper submitted to TMRA06 such a relation set is suggested.

In other words: It is already possible to do what you propose, and this is perfectly democratic.

> This should result in a highly 
> resilient set of equivalent PSIs which would be proof against 
> any one of them going down.
The issue of protecting against peers going down is a different one from ensuring
a democratic approach. Using PSI caching and archives like archive.org, a published subject,
once published, can neither be retracted nor be lost, only superceded.

> It would be up to the user community to decide which ones to use in their topic maps, 
> but it would not matter. In a declaration in a topic map one 
> could name a few of the equivalents together.
Once again, this is already perfectly possible now, only not much used in practice.

As the currently best papers on this subject, I highly recommend:

* Pepper, S. (2006a). A Proposal for Public Resource Identifiers. Paper submitted to TMRA 2006.
* Pepper, S. (2006b). Towards the Semantic Superhighway. A Manifesto for Published Subjects.
  Paper presented at the WWW2006 Workshop: Identity, Reference, and the Web (IRW2006).
  Retrieved 2006-06-06, from
http://www.ibiblio.org/hhalpin/irw2006/spepper.pdf http://www.ibiblio.org/hhalpin/irw2006/spepper.html http://www.ontopia.net/topicmaps/materials/The_Case_for_Published_Subjects.pdf 
http://www.ibiblio.org/hhalpin/irw2006/presentations/Public_Resource_Identifiers.ppt (Slides)
http://www.ontopia.net/topicmaps/materials/Public_Resource_Identifiers.ppt (Slides).

Best regards
alex
---
Alexander Sigel, M.A., Researcher in Semantic Knowledge Networking
Univ. of Cologne, Dept. of Information Systems & Information Management
http://www.wim.uni-koeln.de/19.0.html
sigel at wim.uni-koeln.de, +49 221 470-5322

> -----Ursprüngliche Nachricht-----
> Von: topicmapmail-bounces at infoloom.com 
> [mailto:topicmapmail-bounces at infoloom.com] Im Auftrag von Simon Grant
> Gesendet: Dienstag, 20. Juni 2006 12:59
> An: TopicMapMail
> Betreff: [topicmapmail] PSIs - alternatives