Criteria: What makes a "good" vocabulary?

[stevieflow]

I've been looking at the practicalities for adding a vocabulary to IATI codelists.

In relation to this, I've started to list what we might expect to see from any external vocabulary in terms of

a) it integrating with IATI codelists markup (at vocabulary level) and b) to assist users to access and adequately de-reference data

These seem to be:

• A vocabulary should have an uri endpoint / web-accessible interface (this is new in 2.02 IATI, but good practice nevertheless)

• Codes/values on the vocabulary should have identifiers

  • These should be unique
  • Ideally, these should be numeric (or not in a specific language / string)
  • it's unclear whether a URI could substitute for a code

• Codes should have a name

  Ideally, codes should also have a description

Additional characteristics we may want to see:

• A vocabulary has a licence to support use/re-use (or - it is not prohibitive / restrictive / only accessed via subscription)

• A vocabulary has a clear change control process and policy

  Ideally, the change history is accessible

• Alongside a human-readable interface, a version of the vocabulary in machine-format may be useful

  Access via API could be an additional bonus

I'd suggest that we formulate a list of tests to help evaluate external vocabularies in relation to this - providing us with selection criteria

[stevieflow]

Additional factor: are publishers engaged and using this vocabulary already? if so, how many / what traction is there?

[stevieflow]

I've also added this as a discussion on the IATI forum: http://discuss.iatistandard.org/t/criteria-for-assessing-external-codelists-vocabularies-registries/498

[mikecastro]

I think it would be helpful if the description had an example of what should be classified under that specific category and as well as example of what shouldn't be classified in that category.

[timgdavies]

Our proposed secondary sector vocabulary AgroVoc meets the criteria here.

It does not fully meet @mikecastro's additional criteria 'out of the box', but as a widely used vocabulary it should be possible to pull back examples of it's application from open sources to help users with categorisation work.

[stevieflow]

@mbacou we discussed this theme - maybe this criteria is useful?