A OpenRefine reconciliation service that works from CSV files.

Someone just handed you some datasets they claim are related and you should be able to mesh them up easily - no problem you're the data wizard. Except there is no clear unique identifier you could use to join the dataset. All you know is that if several columns are the same, it should be the same. Enter typos and misspellings and here you go: A long night ahead.

Reconcile-CSV aims to ease your pain: It acts as a OpenRefine reconciliation service and performs fuzzy matching to identify related entries.

Create a column with Unique-ID's you will use to match.

Pre-compiled:

java -Xmx2g -jar reconcile-csv-1.0.1-SNAPSHOT-standalone.jar <file> <primary search column> <column with id's>

With Leiningen:

lein run <file> <primary search column> <column with id's>

Then add http://localhost:8000/reconcile as a reconciliation service to refine. You can add more columns through the reconcile-interface in Refine.

Reconcile away!

Then use:

cell.recon.match.id

to get the ID from the match.

Copyright © 2013 Michael Bauer, Open Knowledge Foundation

Distributed under the BSD-2 Clause license. See LICENSE for details