- From: Brian McBride <bwm@hplb.hpl.hp.com>
- Date: Mon, 30 Sep 2002 10:00:57 +0100
- To: "Patrick Stickler" <patrick.stickler@nokia.com>, "ext Eric Miller" <em@w3.org>, "w3c-rdfcore-wg" <w3c-rdfcore-wg@w3.org>
Hi Patrick, Thank you for this proposal. Whatever decision we take, tidy, untidy or "we ain't sayin", we will need to write up a rationale for why that is the best decision in the circumstances. I think we agreed Friday that we would write up the rationale and then make a decision. I would like to continue with that plan. In saying that, I don't consider this to be an open ended process. We do need to get done, and soon. I will be proposing a detailed process in the next couple of days. What I currently have in mind is to discuss this issue at the next two teleconferences and then have a vote. I think that gives us one more chance to review the arguments and come to greater consensus. It also gives us one more chance to find a new way of looking at the problem which leads to a new proposal(I know there is one out there) that might attract stronger consensus. So I urge the WG not to think just in terms of tidy versus untidy, but in terms of what it is we really need and whether there might be a third way. I note Pat has trying. With regard to your proposal, my immediate concern is the effect on interoperability. To help us make a decision, can you do an analysis for us on the effect this proposal would have on interoperability. Brian At 10:58 30/09/2002 +0300, Patrick Stickler wrote: >I would like for consideration of the following proposal >to be added to the aggenda for the October 4 telecon, >preferably including a vote for or against its adoption. > > > >----- Original Message ----- >From: "ext Eric Miller" <em@w3.org> >To: <w3c-rdfcore-wg@w3.org> >Sent: 26 September, 2002 21:55 >Subject: Reopening tidy/untidy decision > > > > > Achieving concensus is a difficult task at best, and this particular > > issue of tidy/untidy literals has unfortunately plagued us more than any > > would have liked. > > > > We need to finish the stablization of the core RDF > > specifications as soon as possible. > > > > Our deadline has passed. Please look again at this decision and give > > careful consideration to the impact on the WG schedule. > > > > -- > > eric miller http://www.w3.org/people/em/ > > semantic web activity lead http://www.w3.org/2001/sw/ > > w3c world wide web consortium http://www.w3.org/ > > > >It does not appear that the present RDF Core WG will achieve any >reasonably concensus on this issue, and it is unlikely to be >productive to extend a debate which has remained deadlocked for >over a year without significant new input into the process, which >is unlikely to come, apart from further implementational experience. > >I therefore propose to the RDF Core WG that the issue of string versus >value based interpretation of inlined literals be deferred to a future >working group, and that the WG complete all its specifications as >expediently as possible reflecting the decisions made to date, >and complete its charter. > >The RDF MT should assign no interpretation whatsoever to inlined literals, >and should license no entailments involving triples containing inlined >literals. > >Inlined literals remain "semantic wildcards" which are fully ambiguous >to the RDF MT and have no explicit nor consistent meaning across >applications, unless such a consistent intepretation is agreed to >by applications separately from and in addition to the RDF MT. The >RDF Core WG does not suggest any particular mechanism(s) for such >extra-MT agreements. > >The representation of inline literals in the abstract graph syntax >remains as decided, with each occurrence of an inlined literal represented >by a distinct node in the graph, without any special prefixation or other >modification of the node label relating to its unique occurrence. > >Thus, the following entailment would NOT hold, insofar as the RDF MT >is concerned: > >IF > ?s ?p "LLL" . >THEN > ?s ?p _:x . > >and therefore NEITHER of the following datatyping entailments would hold: > >IF > ?s ?p "LLL" . > ?s ?q "LLL" . >THEN > ?s ?p _:x . > ?s ?q _:x . > >IF > ?s ?p "LLL" . > ?s ?q "LLL" . > ?p rdfs:range ?d . > ?q rdfs:range ?d . > ?d rdf:type rdfs:Datatype . >THEN > ?s ?p _:x . > ?s ?q _:x . > >The interpretation of inlined literals, and the relations between inlined >literals and rdfs:range assertions or any other assertions in the graph >remains undefined by the RDF MT and up to each individual application to >decide if and what interpretations might be assigned to inlined literals. > >The following graph does not constitute a type clash, as no interpretation >is provided for the inline literal by the RDF MT itself: > > ?s ?p "LLL" . > ?p rdfs:range ?d . > ?d rdf:type rdfs:Datatype . > >Individual applications, however, are free to interpret the above >range assertion as affecting the interpretation of the inlined >literal in some manner, if they so choose. Or they are free to ignore >the presence of the range assertion as having no affect whatsoever on >the intepretation of the inlined literal. The RDF MT does not support, >require, presume, suggest, or prefer either choice. > >Member organizations are encouraged to continue research and evaluation >of systems based on various interpretations, as well as to publish >proposed or preferred solutions and experience as W3C Notes or by other >means, and it is hoped that further experience gained from such work will >prove to be valueable input to future working groups addressing this issue. > >Patrick > >[Patrick Stickler, Nokia/Finland, (+358 40) 801 9690, >patrick.stickler@nokia.com] > >
Received on Monday, 30 September 2002 05:03:55 UTC