Open ghost opened 4 years ago
I am sorry I can not help in your issue, but I am wondering how have you added the SPARQL endpoint for the reconciliation.
I am facing the issue with DBpedia Reconciliation service.
Please let me if you have any hint for me.
Using the latest stable versions of open refine and the extension. I created a reconciliation service based on a RDF file. The content look as follows:
In the table I have a column with values e.g. Kerkplein (Begijnendijk) ... These are identical to the rdfs:labels in the rdf file.
Some observations running the reconciliation with auto-match.
The result of the matching is 40%. I expected this to be higher due to the complete similarity between the cell values and rdfs:labels.
When there is no automatic match, the proposed values have a low score e.g. 'Grote Spouwenstraat (Grote-Spouwen)' and are far away from the value looked-up.
Clicking 'Search for a match', gives a list of potential matches with the correct one automatically highlighted.
In one way or another the correct value is known, but is not used previously. Can I influence this?
There is an additional observation that if there are many matches, the correct one doesn't show up due to the limitation of the number of items in. the list shown. There are many Kerkstraten (churchstreets): Kerkstraat (A), Kerstraat (AA), ... A list ttoo long for the current interface.