Error in Wikidata Reconciliation

I am trying to reconcile some simple Wikidata (i.e. cities, countries) and some less simple Wikidata (names of publishers). Each time I try this, I get the attached error message when I open the reconciliation window. Then each reconciliation comes back with no results at all. Can anyone explain or troubleshoot this for me?

image

Hi @Lucy_Hinnie

That looks like its pointed to the older wikidata reconcilation service which is no longer used. I think if you remove the Wikidata reconciliation service you have installed and then add a new one using the URL:
https://wikidata.reconci.link/en/api

Then you should find that resolves the issue (I hope!)

Best wishes,

Owen

Thanks Owen. I tried this and still seem to get this:

image

Have I got something wrong at my end I wonder? :thinking: Could it be just too many to reconcile? (Have c. 500 rows at the moment)

The number won’t be an issue - and the Guess Types query doesn’t use the whole set of data anyway.
This is now the correct URL so we’ve ruled that out :slight_smile:

Are you connecting from behind a firewall, or is there a proxy server in place? I think we’ve seen this before where the request is not being routed via a corporate / organisational proxy server and so being blocked at the local firewall

Oh now there’s a thought! I am onsite today - I was using it yesterday at home without difficulty but have had no success today. Is there any workaround for this or do I just need to do it on offsite days?

There’s definitely some things to try - I just need to dig out some details (@grahamjevon might also have some hints for this)

1 Like

What platform (windows/mac/linux) are you running on and how do you start OpenRefine?

@Lucy_Hinnie it could well be that you need to enter some port details into the OR config file. I seem to remember (pre-Covid) that I had to do that before I could use any reconciliation services. I’m pretty sure that is something that Technology will have to do as I think you need admin privileges to edit that file.

Hi @ostephens and @grahamjevon, sorry for the late response - I was not at my desk on Thurs-Fri. I am currently using Windows and I start OpenRefine through the icon on my desktop. I hope this helps?

I found a solution! I was plugged into the wired network, but it went down today so I moved over the wireless - lo and behold, OpenRefine was able to reconcile again. A simple solution but I’m glad to see it.

@Lucy_Hinnie glad you resolved it - and that seems to confirm it’s something about the local network setup :slight_smile:

The reason I asked about how you were starting OpenRefine is that you can add options to tell OpenRefine to use any proxy that is configured locally but the way this is done on Windows is different depending on how you start the system.

So just in case it is helpful you can see

Leaving some extra notes here in case anyone finds this issue in the future …

If you are starting via the icon on your desktop, that’s using “openrefine.exe” which means the options used on starting OpenRefine are the ones in the openrefine.l4j.ini file. The other option for starting OpenRefine on Windows is via refine.bat (Running OpenRefine | OpenRefine) and in that case the settings are taken from refine.ini instead.

For proxy use there are a few relevant settings:

-Djava.net.useSystemProxies=<true or false>
-Dhttp.proxyHost=<local proxy address>
-Dhttp.proxyPort=<local proxy port>
-Dhttp.nonProxyHosts=<any hosts to exclude from the proxy requests>

I can see that there have been various issues with the use of proxies with OpenRefine and a fix was contributed for using proxy servers with reconciliation which is included in the new (at time of writing) 3.7 beta release (see https://github.com/OpenRefine/OpenRefine/issues/3919 for the relevant issue description on GitHub)