Hey Paolo I believe latest SPARQL 1.1 drafts add a REPLACE() function which is equivalent to the XPath fn:replace() function (bar some differences in handling literal forms as per the SPARQL 1.1 string functions spec) that should be able to do what you want. AFAIK this should be in the latest ARQ trunk unless I am mistaken.
For general "How do I do X with SPARQL?" type questions answers.semanticweb.com has plenty of those kinds of questions and people are pretty good at coming up with answers. Cheers, Rob -----Original Message----- From: Paolo Castagna [mailto:[email protected]] Sent: Wednesday, November 30, 2011 9:26 AM To: [email protected] Subject: How to construct new URIs via SPARQL CONSTRUCT? Hi, I am using SPARQL CONSTRUCT queries to derive a dataset (from Geonames, DBPedia and, eventually, other data sources) of all Italian regions, provinces and 'comuni'. Italy is divided in 20 regions. A region can be divided in its provinces. There are 110 provinces in Italy. Finally, a province can have 'comuni' (or other smaller administrative entities). There are 8094 of these in Italy (according to Geonames). This is a fragment of my SPARQL CONSTRUCT query: ---- PREFIX italy: <http://example.com/italy/> ... CONSTRUCT { ?region_uri rdf:type italy:Region . ?region_uri rdfs:label ?region . ?region_uri italy:contains ?p . ?province_uri rdf:type italy:Province . ?province_uri rdfs:label ?province . ?province_uri italy:contains ?comune_uri . ?province_uri italy:isContained ?region_uri . ?comune_uri rdf:type italy:Comune . ?comune_uri rdfs:label ?comune . ?comune_uri italy:isContained ?province_uri . ... } WHERE { ... BIND ( iri(concat("http://example.com/italy/";, lcase(?region))) AS ?region_uri ) BIND ( iri(concat("http://example.com/italy/";, lcase(?region), '/', lcase(?province))) AS ?province_uri ) BIND ( iri(concat("http://example.com/italy/";, lcase(?region), '/', lcase(?province), '/', lcase(?comune))) AS ?comune_uri ) } ---- This is a fragment of what the query above generates, for 'Roncone' which is a 'comune' in the province of Trento which is a province in the 'Trentino-Alto Adige' region: ---- <http://example.com/italy/regione autonoma trentino-alto adige/provincia di trento/roncone> a italy:Comune ; rdfs:label "Roncone" ; italy:isContained <http://example.com/italy/regione autonoma trentino-alto adige/provincia di trento> . ---- All this works well, but I have not been able to strip out "regione autonoma" and replace ' ' with '_' when I generate my new URIs. This, instead, is what I'd like to achieve: ---- <http://example.com/italy/trentino-alto_adige/trento/roncone> a italy:Comune ; rdfs:label "Roncone" ; italy:isContained <http://example.com/italy/trentino-alto_adige/trento> . ---- Or, if possible (even better) this: ---- italy:trentino-alto_adige/trento/roncone a italy:Comune ; rdfs:label "Roncone" ; italy:isContained italy:trentino-alto_adige/trento . ---- Is this possible to do this with a SPARQL CONSTRUCT query without writing a new SPARQL/ARQ function? I searched in ARQ, but I did not find a 'replace' function. Is there one? Would it be an useful addition? Also, I know that URIs should be opaque, but I though it would help humans to follow a pattern {ns} + {region}/{province}/{comune}. What do you think about this? Do you have other suggestions (in addition to Geonames, DBPedia and Freebase) of good datasets for geographical informations about Europe and/or Italy? Last but not least, this message is probably not completely on-topic on the jena-users mailing list. Are these type of messages (on SPARQL queries and/or data modelling using RDF) fine on jena-users? If not, what would be a better mailing list for this sort of questions (I have a few on 'cheese' :-))? Thanks, Paolo
