- From: Paul Gearon <gearon@ieee.org>
- Date: Mon, 29 Jun 2009 13:02:57 -0500
- To: "Seaborne, Andy" <andy.seaborne@hp.com>
- Cc: "public-rdf-dawg@w3.org" <public-rdf-dawg@w3.org>
On Mon, Jun 29, 2009 at 9:21 AM, Seaborne, Andy<andy.seaborne@hp.com> wrote: > >> -----Original Message----- >> From: public-rdf-dawg-request@w3.org [mailto:public-rdf-dawg- >> request@w3.org] On Behalf Of Simon Schenk >> Sent: 29 June 2009 10:36 >> To: public-rdf-dawg@w3.org >> Subject: UNSAID vs MINUS >> >> ACTION-32 http://www.w3.org/2009/sparql/track/actions/32 >> >> To revive the discussion on negation, I would like to point you to the >> overview of options for implementation at: >> http://www.w3.org/2009/sparql/wiki/Design:Negation >> >> In particular, binary operators vs. FILTER expressions, and set based >> MINUS in SeRQL vs UNSAID are discussed. Note that MINUS in SeRQL != >> MINUS in Mulgara (?), which has the same semantics as UNSAID. >> >> Cheers, >> Simon > > Good time to revive it. > > (I reorganised the page end of Friday - didn't remove any text intentionally but I put the two > options in separate sections. I generally called it NOT EXISTS.) > > To make progress on this, let's take it in two steps. > > 1/ Decide on UNSAID vs Set-MINUS, that is resolve to explore one design in depth > > 2/ Then take as a sub-issue of NOT EXISTS as to have a graph operator and/or a FILTER > expression. > > I propose we adopt the UNSAID/NOT EXISTS design. The Set-MINUS seems to have no > advantages because it is equivalent to an NOT EXISTS form, but requires each side to be a > set, which would need to be worked through the rest of the language. I agree with this proposal. The UNSAID approach makes it very easy to express a lot of concepts, but Set-MINUS can be very hard to structure some queries. Given that Set-MINUS is a degenerate form of UNSAID then I think this strengthens the argument for UNSAID. > On the second part, I found it to be no more work as the graph operator is the introduction of > algebra-level filter. It was a straighforward implementation in Mulgara. It took longer to attempt the algebraic description than it took to write the code. Regards, Paul Gearon
Received on Monday, 29 June 2009 18:03:44 UTC