On Mon, Mar 28, 2016 at 6:02 PM Christopher Johnson <christopher.johnson@wikimedia.de> wrote:
Hi,

I have encountered a problem with federating Wikidata entities with LOD based on the abundant external resource "literal identifiers" that are represented in Wikidata.  An external resource identifier should include the entire IRI to the resource, not just the "id", otherwise the external resource cannot be federated in a SPARQL query without concatenating the IRI to the literal identifier and binding to a new variable.  

I think that Wikidata properties that expect explicit Identifiers in the range should thus be defined as "object properties" rather than "datatype properties".   The javascript id linking gadget in the UI also typically borks the link when a IRI is inputted as an identifier making it less likely for people to do it properly.  I assume that this is a "policy" that was established for bot id imports, that unfortunately has produced a lot of "not so useful" data.

The triple pattern fragment implementation that I am working with does not allow bind grouping in the sparql query, so this unfortunately makes federation of Wikidata entities by Identifiers nearly impossible.  Hence, the only federation that I have been able to do effectively is "to" Wikidata not "from" it....

If anyone is interested, I have setup Wikidata as a TPF datasource here:
http://orbeon-bb.wmflabs.org/wdqs-sparql  The API works by passing an encoded subject, predicate or object parameter like:
http://orbeon-bb.wmflabs.org/wdqs-sparql?object=%22Berlin%22%40de  


That's being worked on with https://phabricator.wikimedia.org/T121274 and https://phabricator.wikimedia.org/T73992 and was one of the main reasons we introduced a new datatype for identifiers recently.

Cheers
Lydia 
--
Lydia Pintscher - http://about.me/lydia.pintscher
Product Manager for Wikidata

Wikimedia Deutschland e.V.
Tempelhofer Ufer 23-24
10963 Berlin
www.wikimedia.de

Wikimedia Deutschland - Gesellschaft zur Förderung Freien Wissens e. V.

Eingetragen im Vereinsregister des Amtsgerichts Berlin-Charlottenburg unter der Nummer 23855 Nz. Als gemeinnützig anerkannt durch das Finanzamt für Körperschaften I Berlin, Steuernummer 27/029/42207.