think about James' proposed classification of CDS and use on website

[kohlhase]

On the OM Workshop in Edinburgh we had a discussion on classifying CDs that should take over the old "Official/Exerimental/private/obsolete" classification.

If I remember correctly, James proposed to classify into CDs "Supported/Contributed/External", where

• "Supported" means "supported/maintained" by the OMSoc
• "Contributed" means "contributed to the OMSoc", and hosted there, but not in any form endorsed or maintained.
• "External" means "from an external CD source". And there are potentially multiple of these.
  • For instance all the OMDoc/MMT theories on http://MathHub.info could be made available as OM CDs. This includes the GAP and SageMath CDs used in the SCSCP-based communication in OpenDreamKit. See this paper for a recent account.
  • there were RIACA CDs and SCIEnce CDS (bu they have been folded into "experimental".
  • what am I forgetting?

Internally on the CDs repository we already have the External/Contrib distinction via folders. And we should really be collecting/developing external CD sources and referencing them on the web site.

Actually O/E/P/O is in the OM2 standard, so we can do little to change that, but in the web site, we can group CDs

We should discuss this (the Workshop also decided that "getting more/a credible set of CDs) is of the highest priority for OpenMath in the current year.

[olgacaprotti]

Apologies if what I am about to write has been already discussed during the workshop.

getting more/a credible set of CDs

I am not at all sure that renaming the classification keywords adds anything concrete towards that goal.
What are the concrete issues which make (O/E/P/O) CDs not credible or better not feasible for implementation and actual use?

Recall that the mechanism of CDs came about before the "semantic web" was even a concept - so let us ask the usual question, what would we use now?

Does OpenMath have something to contribute to efforts from linguists like BabelNet (which I just tested to see whether they had a mathematics ontology)? and viceversa, how/where do we integrate their work into ours?

[kohlhase]

Recall that the mechanism of CDs came about before the "semantic web" was even a concept - so let us ask the usual question, what would we use now?

I think that even though CDs predate the semantic web, they are very compatible (we just have to clarify). But they give us something the semantic web does not: normative vocabularies (instead of just generated/suggested ones. So I would do CDs just like we did (modulo syntax, extensions, and a few other niggles).

In essence, CDs are the normative glossary part of OMS (which is the foundation of semantics in OM.

[kohlhase]

But I think that Olga's point (and I think we should discuss this more; just not here) misses what I wanted to discuss here. James' remark about the classification of CDs "Supported/Contributed/External", was not "normative", i.e. "written into the CDs", but administrative for managing and presenting them on the web site.

Now that we have extended CD groups to allow inclusion (see OpenMath/OMSTD#59), we could just use that as an admin mechanism. We could have CD groups supported.cdg, contributed.cdg, and external.cdg.
The first two should be quite simple to write (in fact I will do so to test CDG generation support for OpenMath/OMSTD#59), the last one is very useful. We can use it to document all the CDs and CD groups not hosted on http://openmath.org and give people access to that.

[olgacaprotti]

What are the concrete issues which make (O/E/P/O) CDs not credible or better not feasible for implementation and actual use?

This above was my point and I apologize if it was not the right place to bring it up. I was probably hinting at the fact that for Official CDs we ought to do better if we want the CDs to be used and serve as a proof of concept to attract more CDs -
what does it mean maintained by the OpenMath Society? e.g. do we check that there exists at least one compliant third-party software using the CDs? do we publish access hits to the symbols? do we collect in which mathematical objects the CDs was used? BTW, the official CDs were the ones we used in all the OpenMath projects, so we knew that they were implemented in the software we produced - alas this work was not used directly on the openmath.org site (it did mean to have to maintain several mathematical servers).

In this sense, github is acting as an application supporting every CDs as a renderer and this is a discussion of how to rework, using github, the CD submission workflow.

About using "supported" as a term, I find it is confusing, especially if you read the compliance section where we speak about applications supporting CDs.