search

hbz / lobid-organisations

Transformation, web frontend, and API for lobid-organisations
http://lobid.org/organisations
Eclipse Public License 2.0
13 stars 3 forks source link

Problems with reconciliation endpoint #385

Closed acka47 closed 7 years ago

acka47 commented 7 years ago

Talking with @literarymachine today about an import for CRG, we noticed that the reconciliation endpoint has two problems:

  1. Running reconciliation on the name column of the customer list from EDB, obviously identical strings aren't matched, e.g. "Universitätsbibliothek der RWTH Aachen":

reconciliation-problem

  1. As you can see, no matches are automatically added but have to be confirmed. Usually matches are added automatically in open refine, see e.g. this result of reconciliating with the OEr WOrld Map endpoint.This may have something to do with the match score lobid returns.
fsteeg commented 7 years ago

Deployed to stage, reconcile with http://stage.lobid.org/organisations/reconcile.

literarymachine commented 7 years ago

Running reconciliation on the name column of the customer list from EDB, obviously identical strings aren't matched, e.g. "Universitätsbibliothek der RWTH Aachen"

Resolved, thanks!

As you can see, no matches are automatically added but have to be confirmed. Usually matches are added automatically in open refine, see e.g. this result of reconciliating with the OEr WOrld Map endpoint.This may have something to do with the match score lobid returns.

In case of the automatic matching, still only 25% are reached, but I reckon that this is due to the input data / the OpenRefine probability threshold.

All in all, +1

acka47 commented 7 years ago

Looks much better, +1. @literarymachine, let us know if you see any further problems.

acka47 commented 7 years ago

I just noticed another problem. The reconciliation endpoint also entries deleted from Sigel registry (with "früher: " in their name) which we decided to remove from the regular query, see #360. It would be good if they were also removed from reconicliation.

fsteeg commented 7 years ago

Deployed to production including fix for deleted entries, closing.

Reconcile with http://lobid.org/organisations/reconcile