mahound: A small example, since not a lot of documentation is available
qp: D-VFS will utilize URIs for referencing files
danbri: Backwards-sparql for annotating tab-separated files...
dajobe: or could you just read in a sparql result set (xml, turtle, ...) as a table of bindings and then run the construct
danbri: Yeah exactly, ... the sparql resultset case is left as an exercise to the reader :) I forgot to mention in the writeup....
danbri: yep re html tables too. It's a fairly pure set of complexity/usefulness tradeoffs. You could map all kinds of input formats, but eventually you end up with a super rich mapping language that can scare people off. So Lqraps is an experiment on the simplistic end of the spectrum, saying 'ok all these text dump files people load into Access ... would a couple extra comment lines in them make them RDFy?
danbri: My implementation is a very surface syntax thing, doesn't operate on the templates with any subtlety. This means it can't skip bad triples per the construct sparql spec. A real impl would build upon a SPARQL parser.
mortenf: semi-related, see also CSV-SPARQLer, which uses the regex "^\w+:\S+$" to match uris
dajobe: or could you just read in a sparql result set (xml, turtle, ...) as a table of bindings and then run the construct
danbri: Yeah exactly, ... the sparql resultset case is left as an exercise to the reader :) I forgot to mention in the writeup....
danbri: yep re html tables too. It's a fairly pure set of complexity/usefulness tradeoffs. You could map all kinds of input formats, but eventually you end up with a super rich mapping language that can scare people off. So Lqraps is an experiment on the simplistic end of the spectrum, saying 'ok all these text dump files people load into Access ... would a couple extra comment lines in them make them RDFy?
danbri: My implementation is a very surface syntax thing, doesn't operate on the templates with any subtlety. This means it can't skip bad triples per the construct sparql spec. A real impl would build upon a SPARQL parser.
mortenf: semi-related, see also CSV-SPARQLer, which uses the regex "^\w+:\S+$" to match uris