Changing from purl to w3id.org

[hgoerzig]

We discussed to change from purl.org to w3id.org.
Also we want to have the PaN related ontologies in one namespace such as w3id.org/pan-science

We would have the following name spaces:
http://w3id.org/pan-science/panet
http://w3id.org/pan-science/ESRF-ET

http://w3id.org/pan-science/nexusformat/definitions

In w3id.org we need to create an .htaccess document to manage the redirects.
All http namesspaces would redirect to https:// . (all namesace declarations in the PaNET ontology have http and redirect to https)

Further the redirects should go to the documentation if it is a single term:
https://expands-eu.github.io/ExPaNDS-experimental-techniques-ontology/index-en.html#PaNET2012002
https://manual.nexusformat.org/classes/base_classes/NXmonochromator.html#nxmonochromator-energy-field

And http://purl.org/pan-science/PaNET/ redirects to the whole documentation.
http://purl.org/pan-science/PaNET/panet.owl gets you the most recent ontology.

For the new release of the new NeXus version they already want to implement these changes. If you agree I would like to create a corresponding issue in the NeXus github.

[paulmillar]

One question in my mind is whether or not we (a collective "we": PaNET, NeXus and ESRF-ET) should adopt a common pan-science repo in W3ID.

For PaNET, possible PIDs with pan-science repo:

• https://w3id.org/pan-science/PaNET#PaNET01168
• https://w3id.org/pan-science/PaNET#01168 (maybe)
• https://w3id.org/pan-science/PaNET01168

Possible PIDs with a PaNET repo:

• https://w3id.org/PaNET/PaNET#01168
• https://w3id.org/PaNET/PaNET01168

Omitting the pan-science would result in shorter PIDs (both for slash- and hash style IRIs). It would also avoid having to coordinate activities across different groups (NeXus, PaNET, ESRF-ET).

Are there advantages in having a common pan-science repo?

[hgoerzig]

The idea would be to bundle things. We can have an overview page directing directly from https://w3id.org/pan-science/ to the README.md in the w3id.org github explaining the ontologies to community is contributing too.

[paulmillar]

Hi @hgoerzig, I don't really understand your point, here.

Please bear in mind that these are identifiers (i.e., IRIs). For example, if https://w3id.org/pan-science/PaNET/PaNET#01168 is a valid IRI then it doesn't follow that https://w3id.org/pan-science/PaNET/ or https://w3id.org/pan-science/ is a valid IRI. They could be valid IRIs, but it's not guaranteed (and would actually be rather unusual).

In other words, while nothing prevents someone from stripping off path elements from an IRI and putting the resulting URL (e.g., https://w3id.org/pan-science/) in their web-browser, there's also no reasonable expectation that this should "work".

Just to be clear, we may want to have some web presence that describes the family of PaN-focused PID initiatives (ESRFET, PaNET, NeXus, ...) --- I think this would be helpful and something I would support. I just don't see this influencing the choice of W3ID IRIs.

[hgoerzig]

@paulmillar I just want bundle things and want to have an overview somewhere without having anything additional to maintain. This was just a suggestion and lots of people were nodding when I said this. But we don't have to do it this way.

[paulmillar]

@hgoerzig I agree that bundling things together is a good idea: to make ontologies discoverable, if nothing else.

Although I haven't seen any specific statement, I think the w3id.org service is not intended to host content. Rather, it's meant as a "spring board" to content hosted elsewhere. Here's the git repo for w3id.

https://github.com/perma-id/w3id.org/tree/master

I had a look in some repos (more or less at random) and none of them had any content except the .htaccess file and a README.md. Typically, the README.md is not accessible from the w3id.org URL because the rules (in the .htaccess file) will redirect requests to the target page.

In summary: yes, having a webpage listing the PaN ontologies is a good idea, but I think including a common /pan-science/ prefix in IRIs doesn't help us much: we would anyway need to create web content with this content. This is something we should do.

I'm not sure it makes sense to create a PID for this collection of PaN ontologies.

[hgoerzig]

I though using the readme.md exactly as a springboard to link to the different documentations. Having a commom IRI would create a stack of ontologies were users of the ontologies know that these are created by people out of a community and that they somehow belong together. I very often strip down the path of a namesspace in order to see where it leads too.
I also like Wouts suggestion.

[paulmillar]

OK, having looked at a few examples and how w3id.org is used in practice, I think the two options would be PIDs like:

https://w3id.org/pan-ontology/PaNET#01168

or like:

https://w3id.org/PaNET#01168

I'm slightly in favour of the shorter version (I think would be easier to manage), but if people prefer to include the pan-ontology I think that could work, too.

[gkoum]

Regarding the co-existence of ontologies under the same domain of /pan-science, I think it is beneficial and consistent with how communities build sustainable semantic assets, because:

1. Discoverability
   A shared root namespace makes the ecosystem easier to discover and communicate: once someone lands on any ontology, it becomes obvious there is a family of resources under the same umbrella.
2. Interlinking
   Having the ontologies next to each other encourages and highlights formal links (imports, mappings, skos:exactMatch, owl:equivalentClass, bridge ontologies). This is where the real value starts: query expansion, interoperability, and “knowledge fusion” across sources.
3. Longer paths are not a drawback as they add context
   A path like /pan-science/ESRFET/... is slightly longer but provides human- and machine-readable context (“what family does this belong to?”). For users and LLMs, this is often very informative.
4. Common landing page
   We should provide a simple “root” landing page for the namespace (e.g. /pan-science/) that explains:
   • what the namespace contains (PaNET, ESRFET, etc.)
   • the relationships between them (imports, mappings, alignment)
   • pointers to each ontology’s documentation + latest version
   We need to experiment on how to achieve that with a redirect leading to a common page that will also exist in the ontologies files. As Paul said it would be very nice and we just need to see how visible it is.
5. Many maintainers needs to be supported for this to work as Paul also suggested and w3id seems to support that but we need to make sure as well.

I can take an action to prototype this structure and report back with a concrete proposal and an example .htaccess / routing setup that works for PaNET + ESRFET.

[paulmillar]

Hi @gkoum,

Thanks for the feedback, very much appreciated.

There's a separate email discussion on this topic that (surprisingly) you don't seem to be involved. I've added you in CC, so you can be involved there, if you want.

The current thinking is to go for PaNET IDs like:

https://w3id.org/pan/PaNET#01168

This strikes me as a reasonable compromise: it includes the common prefix without making the PID too much longer.