Grants talk:Project/Rapid/Abián/Study on Wikidata property constraints

Constraint reports not mentioned

Oddly d:Wikidata:Database reports/Constraint violations aren't mentioned. Original development of constraints was by Ivan A. Krestinin based on input from the Wikidata community. The extension merely incorporated this in Wikibase (different GUI). Krbot regularly updates these reports while the extension now has "live" checks.

The main difference we have seen between the database reports and the extension is that users of the extension tend to have a bottom up approach aiming at zero violations (even for non-mandatory constraints or mere suggestions) compared to the top-down approach that handles hundreds if not thousands of items at once based on the database reports. --Jura1 (talk) 12:15, 24 September 2019 (UTC)

Thanks for your comment! I have removed the context on the origin of the constraints, which was unnecessary and could indeed be misleading. This study would analyze the use and definitions of the constraints, which are shared by the MediaWiki extension and all its alternative solutions; the study does not depend on specific implementations or user interfaces. --abián 13:44, 24 September 2019 (UTC)

Looks better. Especially as there are three ways constraint violations are made available and one of those three is currently not finalized.Jura1 (talk) 10:59, 25 September 2019 (UTC)

Interesting point. When I read and endorsed this proposal, the reference to Ivan's constraint violations reports was already removed, and I was indeed wondering whether they can just be ignored here. For many Wikidata users, they are still the original and most important method to use the constraints system; unfortunately, the reports have always been incomplete, as Ivan does not support all Wikidata functionality (he ignores/ignored qualifiers and references, ranks, and some "official" constraint types, and he has/had his own interpretation of some constraint types which differs from the "official" one). Consequence is that we hesitate to implement some constraint types properly, in order not to break the reports for the properties which would carry them. I would recommend to keep this in mind when counting actual constraint definition use… —MisterSynergy (talk) 21:16, 25 September 2019 (UTC)

The proposal excludes user interface questions (at least as I read it) .. so none of the three ways is being studied (the gadget, database reports, query server). Maybe one could also mention d:Special:ConstraintReport, but that is identical to the gadget.

As it excludes violations, their handling and the change over time .. the scope of the proposed study is somewhat limited.

As the purpose of the constraints is to improve content, it's unclear if that should be dissociated from their definition. Jura1 (talk) 08:29, 26 September 2019 (UTC)