Namespaces -- the final decision

Message
Author
mstrand
Posts: 5
Joined: Sun Jan 06, 2013 12:03 am

Re: Namespaces -- the final decision

#11 Post by mstrand »

OnnoPaap wrote: [...] Personally I also think it is a good practice to never use a namespace that is in fact a resolvable URL. Because that situation is usually temporary, dependent of the existence or death of the domain name.
But we are talking about the namespace of the data model here? ISO 15926 Part 2 ? Surely that is a pretty permanent thing, and should have a permanent URI. Having unresolveable URIs is in direct opposition to the Linked Data idea. An idea that is getting more and more traction, not least in the W3C community.
Morten R. Strand
Det Norske Veritas (DNV)

OnnoPaap
Posts: 189
Joined: Sun Jan 22, 2012 9:14 pm

Re: Namespaces -- the final decision

#12 Post by OnnoPaap »

For the proposal on the JORD namespace this namespace seems ok:

Code: Select all

http://posccaesar.org/rdl
I propose that namespace-fractionID should be chosen such that there is the least chance on having to change it.
Two examples:
  1. A class is proposed in the JORD RDL which is candidate for standardization in ISO 15926-4. Its namespace then is:

    Code: Select all

    http://standards.tc184-sc4.org/iso/15926/tech/reference-data
    
    (see http://standards.iso.org/iso/15926/-4 for the spreadsheets).
  2. A specialized class is created in a construction project triple store of company Xyz-ltd. The client of the project is OilCo. The class gets the namespace of OilCo, because that's where the class will be handed over to in the future.

In both examples the chosen namespace is that of the eventual owner of the data. Not that of the owner at that time. That gives the least risk of having to change the namespace.

vvagr
Posts: 282
Joined: Mon Feb 27, 2012 11:01 pm
Location: Moscow, Russia
Contact:

Re: Namespaces -- the final decision

#13 Post by vvagr »

Onno, what should be done with existing RD items with such approach? All data should be partitioned in two categories, and resources which are candidate for Part 4 should be assigned URIs in new namespace?

What about URIs of relationships? If both ends are in ISO namespace, then relationship should be in the same namespace, and if only one end - then relationship remains in http://posccaesar.org/rdl ?

OnnoPaap
Posts: 189
Joined: Sun Jan 22, 2012 9:14 pm

Re: Namespaces -- the final decision

#14 Post by OnnoPaap »

vvagr wrote:Onno, what should be done with existing RD items with such approach? All data should be partitioned in two categories, and resources which are candidate for Part 4 should be assigned URIs in new namespace?
As far as I know the existing classes coming from the PCA RDS don't have a namespace yet. We need to assign it for the first time. If you find IDs with namespaces, they were assigned by an automatic tool.

In the RDS there are also classes from another engineering standard. E.g. ASTM. That is also worth thinking over, to give those an appropriate namespace.
vvagr wrote: If both ends are in ISO namespace, then relationship should be in the same namespace
Why?
mstrand wrote: Having unresolveable URIs is in direct opposition to the Linked Data idea. An idea that is getting more and more traction, not least in the W3C community.
Yes they should be resolvable. That's not the same as using URIs that are not URLs.

That's a technical problem. See http://15926.org/viewtopic.php?f=25&t=50

vvagr
Posts: 282
Joined: Mon Feb 27, 2012 11:01 pm
Location: Moscow, Russia
Contact:

Re: Namespaces -- the final decision

#15 Post by vvagr »

As far as I know the existing classes coming from the PCA RDS don't have a namespace yet.
In what sense they "don't have a namespace"? If you query JORD endpoint http://posccaesar.org/endpoint/ - all resources are identified by URIs formed from RDS-numbers in http://posccaesar.org/rdl/ namespace.

PUMP is http://posccaesar.org/rdl/RDS327239 .

What should be a proper URI for PUMP if your system is implemented?

ASTM A 240/A 240M - S32304 class has URI http://posccaesar.org/rdl/RDS640889

What its URI should be?
vvagr wrote:
If both ends are in ISO namespace, then relationship should be in the same namespace
Why?
If PUMP is a member of ROTATING MECHANICAL EQUIPMENT CLASS, and both PUMP and ROTATING MECHANICAL EQUIPMENT CLASS have URIs in ISO namespace, then classification relationship between them also should have an URI in ISO namespace.

On the other hand if ASME B73.2M TYPE VB CENTRIFUGAL PUMP class has URI in PCA namespace (or even ASME namespace?) - then specialisation connecting it to PUMP is also recorded in this namespace, not in ISO namespace.

vvagr
Posts: 282
Joined: Mon Feb 27, 2012 11:01 pm
Location: Moscow, Russia
Contact:

Re: Namespaces -- the final decision

#16 Post by vvagr »

Onno,

PURL technology is still somewhat unclear to me. Please answer my questions in the thread you're referencing above.

As far as I understand, PURL usage for SPARQL query redirection is not the same concept of dereferenceability Morten is talking about.

Semantic Web requires that an attempt to point your browser to http://standards.tc184-sc4.org/iso/1592 ... /RDS327239 - brings you to the page dedicated to the PUMP. This page is currently at http://posccaesar.org/rdl/page/RDS327239 (notice that it is NOT a PUMP URI http://posccaesar.org/rdl/RDS327239, but your browser is redirected from URI to page URL).

It is impossible to achieve without collaboration of ISO system administrators, with PURL or otherwise.

Post Reply