Establish ontology best practices and hygiene checks

[VladimirAlexiev]

Many of the issues labeled ontology here rely on SPARQL queries to find problems.

• And those that do the analysis with unix tools can be amended to use SPARQL
• (Don't forget that rdfs-improvement/README has deeper analysis and more details than the issues)

Say that we fix them, but how can we be sure they won't reoccur? Consider:

• Defining some ontology best practices and conventions
• Implementing some "hygiene" tests/checks to check them periodically

The best advice to follow is EDMC, who have made large long-standing ontology development efforts, in particular FIBO.

Best practices/Conventions

• https://github.com/edmcouncil/fibo/blob/master/ONTOLOGY_GUIDE.md is a very extensive set of good practices. I particularly like
  • "how to write definitions" (and have more advice I can bring in)
  • "how to define more reusable props"
• But I think that CGMES is developed in a pretty strict manner, and props are strictly bound to domain/range
• So we need to cover simpler rules like these (TODO: turn into checklist and link to issues)
  • no HTML in definitions
  • no trailing whitespace in labels/definitions
  • props should start with lowercase
  • URL policy for ontologies, packages. Also classes, props (but these pretty much fall into fixed namespaces cim, nc, eu, dm, md)

Checks

• EDMC "hygiene checks": https://github.com/edmcouncil/fibo/tree/master/etc/testing/hygiene_parameterized (mentioned in #58)
• These docs may be useful, but I haven't read them https://github.com/edmcouncil/fibo/tree/master/etc/testing/test-strategy
• See the checklist I made for ERA: https://github.com/Interoperable-data/ERA-Ontology-3.1.0/issues/82

[Sveino]

We have included this as best we can in the specification. We have also created PROF-AP-CIM, that defines the vocabulary and the constraints for the profile. The plan is to use SHACL to validate if our profile (AP, AP-voc and AP-con) is following the rules.

[VladimirAlexiev]

This issue is about actually implementing the checks, and running them periodically.

• We should start with a list of checks as per https://github.com/Interoperable-data/ERA-Ontology-3.1.0/issues/82
• Then select some tools/approaches to implement them
• Then decide how to run them (Github actions?)
• And how to record the problems for authors to fix

@griddigit-ci @Sveino Do you have a list of checks in the specs? Can you give a link?

Here's a start:

• [ ] no HTML in definitions
• [ ] no leading/trailing whitespace in labels/definitions
• [ ] props should start with lowercase (xx:Package doesn't)
• [ ] URL policy for ontologies, packages.

But we should also:

• add links to issues
• move the checklist in the Description, so github tracks it specially