Just to add to Egon's comment IRI's can also be added to Wikidata using the Property P2888 - Exact Match. On Wed, May 17, 2017 at 4:14 PM, Egon Willighagen < egon.willighagen(a)gmail.com&gt; wrote:

Hi, Thanks to Egon I now understand that I can change the RDF URI template. The old template was pointing to the Linked Data browser (i.e., HTML documents). I've changed this to the URI template of the identifiers for Dutch municipalities. When I visit one of the Dutch municipalities, it still has the old URI value for the property (P981). Will the new template that I have added now be used in future scrapes? IOW: how to update the instances? --- Cheers!, Wouter. On Wed, May 17, 2017 at 5:52 PM, Andy Mabbett &lt;andy(a)pigsonthewing.org.uk&gt; wrote:

On Wed, May 17, 2017 at 6:14 PM, Andy Mabbett &lt;andy(a)pigsonthewing.org.uk&gt; wrote: Let me indeed be a bit more specific... For instance the Dutch municipality of Apeldoorn (https://www.wikidata.org/wiki/Q3018561) should refer to the Kadaster resource IRI and not to a document IRI (as is currently the case). Specifically, the value of property "BAG-code for Dutch locations" (P981) should be changed for each Dutch municipality. I've edited P981's property "RDF URI template" (P1921) to the correct IRI template, but this template is not yet used for the Dutch municipality instances (also not after purging the cache). There is also a "formatter URL" property (P1630). Maybe that conflicts with the IRI template that I have edited? Or maybe it takes precedence over the RDF URI? --- Cheers, Wouter.

On Wed, May 17, 2017 at 5:52 PM, Andy Mabbett &lt;andy(a)pigsonthewing.org.uk&gt; wrote: The ID usually points to human readable website, whereas in RDF the items use IRI to distinguish between concepts. An IRI points to the related node in the linked data cloud. In the context of being able to link with external (RDF) datasets, using the ID from a specific property does not work.

Using P1921 will not always work. Especially in the case where resources use opaque identifiers either as internal identifiers and/or IRI's, making the transformation between the two impossible. Also having to rely on P1921 to run for example federated queries leads to complex query patterns. You always have to include an additional step where the IRIs are calculated on the fly. On Thu, May 18, 2017 at 12:51 AM, Andy Mabbett &lt;andy(a)pigsonthewing.org.uk&gt; wrote:

On Thu, May 18, 2017 at 8:45 PM, Andy Mabbett &lt;andy(a)pigsonthewing.org.uk&gt; wrote: I agree, but we're in a mixed world right now... semantic web is being picked up, and I think we must take advantage of this (IRIs are identifiers too), but many databases are not semantic and have old fashion IDs. (As maintainer of an ID mapping system called BridgeDb, this is my daily work: dealing with two types of identifiers from two related but different worlds...) I absolutely agree ideally we would have one solution, and while on Wikidata we might try that community consensus, I am not sure I see how you want to achieve that beyond Wikidata at this moment? Egon -- E.L. Willighagen Department of Bioinformatics - BiGCaT Maastricht University (http://www.bigcat.unimaas.nl/) Homepage: http://egonw.github.com/ LinkedIn: http://se.linkedin.com/in/egonw Blog: http://chem-bla-ics.blogspot.com/ PubList: http://www.citeulike.org/user/egonw/tag/papers ORCID: 0000-0001-7542-0286 ImpactStory: https://impactstory.org/u/egonwillighagen

On Sat, May 20, 2017 at 6:47 PM, Andy Mabbett &lt;andy(a)pigsonthewing.org.uk&gt; wrote: Then I think that we were not talking about the same things... what I meant is, restricting myself to chemical databases, ChEBI and ChemSpider have old fashion identifiers and IRIs (based on those identifiers), but HMDB, for example, does not. That's outside Wikidata's control, so Wikidata has to accept both types of identifiers. Right? Egon -- E.L. Willighagen Department of Bioinformatics - BiGCaT Maastricht University (http://www.bigcat.unimaas.nl/) Homepage: http://egonw.github.com/ LinkedIn: http://se.linkedin.com/in/egonw Blog: http://chem-bla-ics.blogspot.com/ PubList: http://www.citeulike.org/user/egonw/tag/papers ORCID: 0000-0001-7542-0286 ImpactStory: https://impactstory.org/u/egonwillighagen

Hi Andy, Thank you for the useful feedback. I'm trying to understand what's the best way for us to link our LOD activities to Wikidata's. Your comments (and also Egon's BTW) are thus much appreciated. On Wed, May 17, 2017 at 5:55 PM, Andy Mabbett &lt;andy(a)pigsonthewing.org.uk&gt; wrote: That's a good idea. That way a LOD agent can traverse from Kadaster to Wikidata (in addition to the other way round). What's the best practice for doing this on our end? Should we create a linkset with statements of the form [1] or are there other/better approaches? [1] <KADASTER-URI> owl:sameAs <WIKIDATA-URI> --- Cheers, Wouter.