This repository has been archived by the owner on Nov 10, 2022. It is now read-only.
-
Notifications
You must be signed in to change notification settings - Fork 24
Type constraints should be smoother #30
Labels
Comments
Check with Magnus on this. His function to populate the Classification panel will probably be useful to look at. Usually the last 3 subclasses on the chain are the ones that your looking for in terms of domain "proximity" from what I have found. https://tools.wmflabs.org/reasonator/?&q=3918 |
The other way to do this would be through "relatedness" via properties, rather than classes. https://www.wikidata.org/wiki/Wikidata:SPARQL_query_service/queries#Properties_registered_as_properties_for_this_type_.28P1963.29_for_a_class_or_its_super-classes |
Sign up for free
to subscribe to this conversation on GitHub.
Already have an account?
Sign in.
Currently, candidates whose type does not match the target type (via subclasses) are only included if no other candidate of the right type was found, and with a score divided by two. But some types are arguably closer than others, like "university" and "school" are close to each other while not being subclasses of each other. "school" and "tombstone" are further apart.
Figure out a way to return penalized candidates whose type is "close" to the target type (for an appropriate notion of proximity).
The text was updated successfully, but these errors were encountered: