- From: Jerven Bolleman <me@jerven.eu>
- Date: Fri, 2 Oct 2015 16:47:49 +0200
- To: Mikel Egaña Aranguren <mikel.egana.aranguren@gmail.com>
- Cc: Richard Boyce <rdb20@pitt.edu>, HCLS <public-semweb-lifesci@w3.org>
- Message-ID: <CAHM_hUMJThi3ckz2rnNSxL_u8KYJNgrvWLQWdYOcYY2Y666iuw@mail.gmail.com>
Hi, In Oracle, I think you can create a view on a select on sem_match. e.g. CREATE VIEW proteinOrganism AS SELECT p AS PROTEIN, o AS ORGANISM FROM TABLE(SEM_MATCH( 'PREFIX c:<http://purl.uniprot.org/core/> SELECT ?p ?o WHERE {?p a c:Protein . ?p c:organism ?o}', SEM_MODELS('UNIPROT_2015_08'), null, null, null)); Something similar is possible in Virtuoso, and surely DB2 In PostgreSQL, you could use https://github.com/cyga/www_fdw/wiki/Documentation to map sparql results into views. However, all will requires certain maintenance workloads. For the experimental stuff you could look at MonetDB/RDF or Virtuoso-CS (not yet public) http://homepages.cwi.nl/~duc/papers/emergentschema_www15.pdf Regards, Jerven On Fri, Oct 2, 2015 at 3:04 PM, Mikel Egaña Aranguren < mikel.egana.aranguren@gmail.com> wrote: > Hi; > > My understanding is that Andrea wants the opposite: rewrite SQL queries to > SPARQL queries, not access to SQL DBs with SPARQL. Anyway if the latter is > the case, apart from D2R, the new Stardog release includes virtual graphs > to access tabular (SQL, CSV files, ... ) data: > http://docs.stardog.com/#_virtual_graphs. > > Mikel > > 2015-10-02 14:34 GMT+02:00 Richard Boyce <rdb20@pitt.edu>: > >> Hi Andrea, I think that D2R Server is very helpful here: >> http://d2rq.org/d2r-server >> >> All can be done with a mapping file that you configure. The server >> provides a SPARQL web query interface (SNORQL) but also can dump to an RDF >> file that you load in a separate store. >> >> hope it helps, >> -R >> >> >> On 10/02/2015 07:37 AM, Andrea Splendiani wrote: >> >> Hi, >> >> I am wondering if some of you knows of some tool that can translate (a >> subset) of SQL to SPARQL (perhaps requiring some constraints on the RDF >> representation). >> In principle it should be simple: classes can appear as tables, URIs as >> IDs, datatype properties as columns and object properties as Fkeys. >> Is there something implementing this translation available, that some of >> you know ? One current option (I think) is via Oracle, but I am wondering >> if there is something like-weight. >> >> As of why I am interested in it... it's curious: I may have an RDF graph >> representing a unified set of sources (some of which native in RDF, some of >> which virtualized from SQL). >> Still people like to query the sql sources in sql, just because they know >> it better. But like this, they miss the whole integration (and a more >> flexible data model). >> >> best, >> Andrea >> >> >> >> -- >> Richard D Boyce, PhD >> Assistant Professor of Biomedical Informatics >> Faculty, Center for Pharmaceutical Policy and Prescribing >> Faculty, Geriatric Pharmaceutical Outcomes and Gero-Informatics Research and Training Program >> University of Pittsburghrdb20@pitt.edu >> Office: 412-648-9219 >> Twitter: @bhaapgh >> >> > > > -- > Mikel Egaña Aranguren, Ph.D. > > http://mikeleganaaranguren.com > > > -- Jerven Bolleman me@jerven.eu
Received on Friday, 2 October 2015 14:48:19 UTC