Wikidata reconciliation: latest changes

27 views
Skip to first unread message

Antonin Delpeuch (lists)

unread,
Apr 20, 2017, 4:37:31 AM4/20/17
to openr...@googlegroups.com
Hi,

Here is a quick review of the new features in the reconciliation interface:

- Disjunction in SPARQL property paths is now supported. This is useful
when the column you want to refine on is not always represented with the
same property in Wikidata.
Documentation:
https://github.com/OpenRefine/OpenRefine/wiki/Reconciliation#advanced-usage-property-paths

- Strict matching for identifiers: properties marked as identifiers in
Wikidata will not be fuzzy-matched anymore.

- Reconciliation by unique identifiers is supported. If some of your
rows contain some unique identifiers that could also be present on
Wikidata, you can forward them to the reconciliation interface.
All items with matching identifiers will get a perfect matching score
regardless of their other attributes (even the name is ignored).
Documentation:
https://github.com/OpenRefine/OpenRefine/wiki/Reconciliation#reconciling-via-unique-identifiers

Any feedback on these features is welcome. The interface should still
not be considered stable as I intend to tweak the scoring and optimize
for speed.

Cheers,
Antonin

qi cui

unread,
Apr 21, 2017, 10:25:25 PM4/21/17
to OpenRefine, li...@antonin.delpeuch.eu
Cool. Not sure whether 3 rows per sec is a little bit slow. Is this constrained by the server capability or from OpenRefine? 

Antonin Delpeuch (lists)

unread,
Apr 22, 2017, 4:24:30 AM4/22/17
to qi cui, OpenRefine
The query rate can definitely be improved server side, by reducing the
number of API calls made for each query.

Antonin
Reply all
Reply to author
Forward
0 new messages