JanZerebecki added a subscriber: JanZerebecki.
JanZerebecki closed this task as "Declined".
JanZerebecki claimed this task.
JanZerebecki added a comment.

I understand this is certainly alluring, but this won't work for multiple 
reasons:

1. The constraint violations are not a state of only one item. At least they 
are a state of the property, the statement on that property that states the 
constraint and the item. Often more items and properties come into this.
2. The constraints may change at any time so an item may gain or loose 
violations without it changing, while the change stream is immutable.
3. Count of constraint violations don't easily relate to what a human might see 
as quality. A version with more violations may be of higher quality than one 
with less.

Together this means that we can only automatically infer anything when the 
whole of Wikidata (items, properties and constraints) has no violations, that 
is we know of no problems according to our rules. As soon as something changes 
and that creates at least 1 violation we can not automatically say if that is 
worse or better than before, it only means we have at least one problem in our 
constraints or in our data, but the problem may have existed before.


TASK DETAIL
  https://phabricator.wikimedia.org/T105875

EMAIL PREFERENCES
  https://phabricator.wikimedia.org/settings/panel/emailpreferences/

To: JanZerebecki
Cc: JanZerebecki, Aklapper, Karima, Wikidata-bugs, aude, Krenair, Malyacko, 
P.Copp



_______________________________________________
Wikidata-bugs mailing list
Wikidata-bugs@lists.wikimedia.org
https://lists.wikimedia.org/mailman/listinfo/wikidata-bugs

Reply via email to