*On OpenRefine platform:*
I do think that getting free text feedback is a good idea.
The most effective feedback we have ever had was always through one-on-one
interactions with users themselves and listening through the datasets they
bring or interested in.
So putting OpenRefine in the hands of others and recording their habits is
quite useful.
I have found that bringing a unique perspective is sometimes surfaced by
giving users unusual datasets to try to reconcile, instead of just allowing
them to reconcile their own datasets.
Hosting a Reconciliation Day is a good option to gather feedback and record
it. Perhaps getting someone to sponsor it with Beer and Pizza will open
more mouths to speak as well :-)
*Regarding matching by External identifiers:*
OpenRefine's earlier version with Freebase Recon had the option to match
solely against MQL ID's. OpenRefine lost this ability in newer versions.
The current version of OpenRefine (screenshots) has the ability
technically, but it is not so evident to inexperienced users and also does
not give effective results currently.
This could be an improvement on both sides, where the UI exposes an option
to "Reconcile against a Recon Entity ID" and the Recon Service performs
lookups exclusively against Entity ID's in its service offering.
If Multiple types of Entity ID's are exposed by a Recon service offering,
then those should be exposed in the UI to the user, perhaps in a similar
dropdown box that is enumerated only with Identifier Types.
In the case of Wikidata, I think there are 5 Entity Identifier Types now,
maybe more with Lexemes (L), Forms (F) and Senses (S)?
[image: Annotation 2019-07-20 141855.png]
[image: Annotation 2019-07-20 141915.png]
Thad
https://www.linkedin.com/in/thadguidry/
