3

Has anyone been experiencing problems with reconciliation in OpenRefine? I've imported a list of American universities and colleges, selected 50 rows, and tried Freebase, DBpedia, OpenCorporates reconciliation services. I've previously had multiple successes with DBpedia (for colleges and universities), but right now, none of these are working. (I went through every service listed, too.) I've trimmed leading and trailing spaces, checked for duplicates, etc. Things were working fine only about two months ago, and I have not updated OpenRefine during that time. [UPDATED]: I was able to get one reconciliation service to work, but not with the ontology I want. So either I'm forgetting some key bit of info, or the services that reconciliation university/ college names no longer function.

Dominic Sayers
  • 1,783
  • 2
  • 20
  • 26
ultrageek
  • 637
  • 1
  • 6
  • 13
  • 1
    Freebase is shutting down and reconciliation API no longer works. And I had no success with [RDF extension](http://refine.deri.ie/) to reconcile against SPARQL endpoints — it never returned any data for me. Tried multiple versions. – kolen Jun 09 '15 at 10:17
  • Yeah, I was worried about freebase, though maybe there'll be a Wikidata equivalent. I tried SPARQL and it used to work (late last year). But I've had success with DBpedia reconciliation in OpenRefine very recently. Now it's not working either. Used to be a huge time saver. I'm dreading having to manually fill in the fields, with hundreds of records – ultrageek Jun 09 '15 at 10:32
  • 1
    You will have more answer by asking OpenRefine mailing list: https://groups.google.com/forum/?fromgroups#!forum/openrefine – magdmartin Jun 09 '15 at 12:18
  • Thanks, @magdmartin. I didn't realize any of the Google URLs for Refine were still active. I kept seeing links over to Github for code/docs, so it didn't occur to me that there was a Google Groups forum. – ultrageek Jun 09 '15 at 18:41
  • What is the exact error you see? When I try reconciliation, I get a continuous 'Working' pop up. This is true of all the end-points I've tried (DBpedia, factforge, geosparql). – kurious Aug 09 '15 at 17:15
  • @kurious I cannot recall at this point. The service has been working. However, on weekends, sometimes when the dbpedia.org site is down for maintainenance, I see a spinning indicator in OpenRefine. I've made it a habit to visit dbpedia.org in my browser to verify, if OpenRefine is not working. I'm not sure about factforge and geosparql endpoints. I've not tried them unfortunately. Could be another dbpedia code change issue. – ultrageek Aug 14 '15 at 22:53

3 Answers3

5

The OpenCorporates Refine service (https://opencorporates.com/reconcile) is currently working, but note you need to use https - http did work at some point in the past but no longer.

Using the RDF Refine extension (http://refine.deri.ie) and trying a SPARQL based reconciliation against DBPedia I'm finding problems at the moment, but using the same approach for other SPARQL services I have no problems. I don't know what the underlying cause for this is, but it seems likely this is due to some issue or change at DBPedia rather than OpenRefine or the RDF Refine extension.

Any more information about how you are setting up the Reconciliation services and any extensions you are using might help in further diagnosis

Owen Stephens
  • 1,550
  • 1
  • 8
  • 10
  • Thanks for the response. I set up the service months ago -- probably early 2014, I think - when I first installed OpenRefine. So I don't recall, but I have not changed anything recently. I *have* had DBpedia, OpenCorporates and several other services work fine for different projects -- and particularly DBpedia worked for past sets of American colleges and universities. – ultrageek Jun 09 '15 at 18:16
  • 1
    I tested the DBPedia SPARQL endpoint at http://dbpedia.org/sparql when you first reported this problem and found the same issue as you. I've now tested again and it works for me fine. Nothing has changed at my end between the two tests. The most likely explanation (to me anyway) is the issue was caused be either a temporary or intermittent issue with the DBPedia SPARQL endpoint. – Owen Stephens Jun 17 '15 at 09:44
  • Thanks, Owen. Filip mentioned above that the DBpedia devs fixed the problem a few days back. It was working for me last I checked, too. Would be nice to know what happened though. – ultrageek Jun 17 '15 at 10:25
  • Is it still working for you? I am getting the same problem again, its keep running adding reconciliation service but there is no update. – Gautam Shahi Mar 29 '20 at 22:54
  • This question was asked in 2015 and seemed to be a temporary problem at that point, so it seems unlikely you are experiencing a problem for the same reason. I’d suggest asking on the OpenRefine google group ( http://groups.google.com/group/openrefine/ ) as a first step to resolving your problems giving as much detail as you can. – Owen Stephens Mar 29 '20 at 23:00
  • Thanks, already posted in the group, not getting any specific error, that's why did not create a separate issue. – Gautam Shahi Mar 29 '20 at 23:04
4

I have been having the same problem with reconciliation using the RDF extension and DBpedia.org. However, if you try some other service (e.g., your local file or Spanish DBpedia (http://es.dbpedia.org/sparql)) it works very well.

Like Owen already mentioned, it is likely that the it is due to DBpedia. Looks like the problem is with XML 1.1. and the recent update of DBpedia.org, take a look here https://github.com/openlink/virtuoso-opensource/issues/405 If you see OpenRefine log file or the console, this is exactly the same exception that we get when doing reconciliation against the DBpedia.org.

Hopes this give you some idea.

Cheers

Filip Radulovic
  • 56
  • 1
  • Thanks for the response. I can understand some Spanish, so I could give that a try -- but I'm not sure I will get a full listing of the U.S. universities/colleges. Maybe I need to update my OpenRefine installation. – ultrageek Jun 09 '15 at 18:14
  • Okay, so the Spanish DBpedia reconciliation service shows ontology options (the way the English service used to, up until recently), although the service seems slower than for English. But the automatic matching to best candidate was only 25 out of 50. I'm pretty sure the English service was more accurate. – ultrageek Jun 09 '15 at 19:05
  • So the Spanish DBpedia recon service is also having problems. Only half the fairly well-known universities matched, and when I try to manually match the rest, I see a "something went wrong - try again later" type of message. I've seen that in the English service in the past, but usually briefly. – ultrageek Jun 09 '15 at 20:57
  • 1
    The problem with the DBpedia.org is that it does not work at all. Spanish one can do a job at least until the problem with the other one is resolved, and possibly the performance will be less good. I doubt however that updating OpenRefine installation will solve the issue, since most probably the problem is with DBpedia. Nevertheless, if you notice some improvements, please share. Cheers. – Filip Radulovic Jun 10 '15 at 09:31
  • Good point, Filip. That makes sense. No improvements. I ended up having to figure out a SPARQL query to get a list of U.S. universities, imported that into OpenRefine, tweaked the wiki URLs to DBpedia URLs, then modified for the JSON URL. That was the simplest way around the problem that I could figure out. – ultrageek Jun 10 '15 at 22:12
  • I am informed that people in charge of DBpedia.org fixed the problem. Try it, it works for me now. – Filip Radulovic Jun 11 '15 at 07:53
  • Cool. It's slow to startup but it appears to be working. Thanks for the update. The workaround I used wasted two days and required so much data massage. – ultrageek Jun 11 '15 at 18:27
0

At the moment there exists a possibility to create "database" (actually, owl\rdf ontology) and operation json for reconcilliation "on-the-fly" from OpenRefine facets and operation json. Please refer to https://stdgont.uk.to for details.

Vladimir S.
  • 1