Batch match organization names to ISIL

[acka47]

@hauschke has a list of ~1100 organizations he wants to get the ISIL for. He would like to do this with open refine but we don't offer an open refine reconciliation service (see also lobid/lodmill#168). I guess , the API already supports this use case if you have the needed programming skills.

We will either have to provide a reconciliation service or create an example script that lets @hauschke execute the matching. He would provide us with a csv list of the names for this...

[acka47]

@hauschke Sent a csv with library names and postalc code and – if existing – DBS ID. You can find it here: https://gist.github.com/acka47/9bdc24359fe811e90026

[acka47]

@fsteeg: Do you want to take this one over?

[fsteeg]

Deployed a first take on a reconciliation service for lobid-organisations:

http://beta.lobid.org/organisations/reconcile

I have not worked with OpenRefine before, so I'm not sure this all makes sense, but here is what I did to test the service (in OpenRefine v2.6-rc1):

• Open bib_ohne_sigel.csv (from https://gist.github.com/acka47/9bdc24359fe811e90026) -> Next -> Create Project
• On the bibliothek column, drop down menu -> Reconcile -> Start Reconciling
• Add Standard Service -> paste http://beta.lobid.org/organisations/reconcile -> Add Service
• Click new lobid-organisations entry to close pane on the left
• Check dbs-id, As Property: a (arbitrary, but needs to be set)
• Check plz, As Property: b (arbitrary, but needs to be set)
• Click Start Reconciling
• For each bibliothek cell OpenRefine now lists the suggested candidates
• Click on the candidates to view their lobid-organisations JSON content
• Deselect the lowest scoring suggestions with the slider on the left (I selected 0.17 - 2.59, resulting in 1015 rows of 1027 total)
• On the bibliothek column drop down menu -> Edit column -> Add column based on this column
• New column name: hbz-id (entries with no Sigel get DBS-<dbs-id> ids in lobid-organisations)
• Expression: cell.recon.best.id -> OK
• On the left, select all again (to get back the rows where we did not add the id)
• In the upper right, Export -> Comma-separated values

The exported CSV file contains all original rows, with 1015 of 1027 now containing a hbz-id. I've uploaded it here: https://gist.github.com/fsteeg/df41a245b9ee404ef036

@hauschke Is this usable for your use case? Anything missing or wrong?

[hauschke]

Thank you very much, it works like a charm! 😄

[acka47]

As @hauschke is satisfied, a +1 from me. I haven't tried it myself, though.

[fsteeg]

Great, happy it works for you @hauschke. Closing.