Home

OpenRefine Reconciliation API

Documentation

The OpenRefine Reconciliation API does not specify the implementation of the matching algorithm. Looking at the reconciliation APIs listed on the wiki and in the documentation, the following match using query only and not properties:

• Nomenklatura
• FundRef
• JournalTOCs (Table of Contents)
• VIAF (Virtual International Authority File) (source code)
• FAST (Faceted Application of Subject Terminology)
• Ordnance Survey
• redmine-reconcile
• reconcile-demo
• reconciliation_service_skeleton
• phyloinformatics (except GBIF) (source code)

The following use properties to filter results in a binary fashion: that is, only results that exactly match the properties are then matched on the query.

• Sunlight Labs: Influence Explorer
• Grano
• Grano
• Open Knowledge: Helmut
• Open Knowledge: Helmut
• phyloinformatics (only GBIF)

OpenCorporates changes the scoring algorithm based on the jurisdiction_code and uses sample_date for scoring.

RDF Refine lets you customize how you match, but SPARQL doesn't support scoring/ranking of results.

The listed Library of Congress Subject Headings uses RDF Refine.

The following have not been audited:

• DBpedia (Java)
• open-reconcile (Java)
• reconcile-csv (Clojure)
• VIVO documentation unclear