here; the idea came up in DAML+OIL design discussions too. 16:02:56 <reagleMIT> reagleMIT has joined #swarch 16:03:21 <sandro> em: Mostly I was going to say why this is important, but lets get on with how to solve it. 16:03:35 <reagleMIT> what's the content model of xhtml:meta ? 16:03:42 <sandro> Ralph: Let's stipulate there is a need to put RDF in HTML documents. Any debate? 16:03:43 <reagleMIT> can it take element content? 16:03:46 <reagleMIT> if so, why not? 16:03:55 <DaveB> reagleMIT: currently an empty eleemnt IIRC 16:04:16 <sandro> Mike Dean: It's still valuable, but it's easy for outsiders/newcomers to over estimate the value. 16:04:18 <chaalsBOS> reagleMIT: Was asked and stephen said that would break DTD validity, which is a requirement of the HTML group 16:04:39 <sandro> Steven: I think you'll get far greater use of RDF if it's exposed to the public in HTML. 16:04:47 <danbri> (danbri claps) 16:04:53 <reagleMIT> if Steven has to make a change, is there a DTD featuring corresponding to xsd:any ? 16:05:05 <sandro> em: That's my experience as well. It's required by law in 6 countries! 16:05:07 <reagleMIT> ah, thanks charles 16:05:31 <mdubinko> reagleMIT, no, no such thing in DTD-land 16:05:35 <sandro> TimBL: If it's in HTML meta tag, you wont be able to cut and paste this stuff. 16:05:58 <DanC> yes, people *ARE* getting RDF/XML from XHTML via XSLT. 16:06:24 <sandro> Ralph: As soon as your open up the question of radical changes to syntax, ... you're outside the scope of any WG. 16:06:45 <sandro> JJC: If there's a good RDF serialization for HTML Heads, it shiould be the primary syntax. 16:06:54 <sandro> DanBri: WHat about N-triples? 16:07:04 <Steven> Steven has joined #swarch 16:07:21 <DanC> yes, people *ARE* getting RDF/XML from XHTML via XSLT. http://www.w3.org/2000/06/dc-extract/form 16:07:24 <danbri> ...ntriples as groundwork for potentially doing an n-triples in angle brackets 16:07:49 <sandro> Brian: Is there any support from eg Adobe. (HTML authoriing tools vendors) 16:07:57 <Steven> My example is available at http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2003Feb/0103.html 16:08:14 <DaveB> my (+Arts) ntriples syntax (as a subset of n3 + stuff) is essentially the rdf graph concretized 16:08:18 <sandro> DanC: Back when I was chair of HTML WG, they surel;y wanted to know how to put the document author in the HTML metadata. 16:08:34 <sandro> Thanks, Steven. 16:08:46 <DanC> author name 16:09:17 <sandro> heh 16:09:17 <danbri> (side talk: danbri asks Steven if DTD validation important for HTML because of entities; Steven replies: 'several issues') 16:09:48 <sandro> Ralph: How is DTD for MathML solved....? THeir slide had no doctype? html+math 16:09:56 <sandro> Steven: html+math 16:10:12 <sandro> ...: We have a profile for html+math+svg 16:10:26 <sandro> ...: We had to rename entities before we could combine them. 16:10:34 <sandro> DanC: No doctype? 16:10:46 <GuusS> GuusS has joined #swarch 16:11:06 <sandro> Steven: There is a doctype coming. People will have to use the HTML+Math+SVG doctype. 16:11:53 <sandro> DanC: I do not agree there a requirement for entities. 16:12:37 <sandro> Steven: MathML is geared around entities. 16:12:50 <sandro> DanC: But MathML would be fine without entities. 16:13:01 <sandro> ..: It would be a smaller change. Just use UTF-8. 16:13:08 <DanC> or <mchar> 16:13:20 <sandro> TimBL: Could we put a wrapper in? SOmething with "any" as the content model. 16:13:33 <Stuart> Stuart has joined #swarch 16:14:10 <sandro> Steven: You can say "any" and it means "anything defined in the DTD". In Schema there is a more broad "any" that means "any well-formed XML". 16:14:39 <sandro> ...: It's a load of work (amazing we did it) to do namespaces in DTDs. 16:15:00 <sandro> ...: If we had entity-functionaly elsewhere, we would no longer need DTDs. 16:15:10 <sandro> ...: (IE we could switch to Schema.) 16:15:50 <sandro> Daniel Austin: one approach is to use Schema + DOCType *onyl* for entities, not for validation. 16:16:10 <sandro> danbri: How responsive is XML Schema WG on this? 16:16:17 <sandro> Steven: I'm not optimistic. 16:16:31 <sandro> DanC: It has to be fixed in the XML, not XML Schema. 16:16:37 <JosD___> JosD___ has joined #swarch 16:17:02 <sandro> DA: My impression in XML Core WG is planning to take up entities. 16:17:33 <sandro> TimBL: WHat would HTML WG say about that half-way route? (DA's above) 16:17:45 <sandro> Steven: I don't know. It might be possible to be accepted. 16:17:58 <sandro> Ralph: Can you please explore that question more? 16:18:00 <sandro> Steven: Yes. 16:18:30 <sandro> =======Mimasa takes podium 16:18:31 <DaveB> what was the half-way route? 16:19:04 <sandro> /DTD XHTML 1.1 plus MAthML 2.0 plus SVCS 1.1//EN 16:19:20 <sandro> M: I don't like this, but it's what we have to do at the moment. 16:20:16 <Zakim> Zakim has left #swarch 16:21:05 <sandro> jjc: Do you have to use the default namespace? 16:21:24 <sandro> [ there's some magic to getting namespaces to work in DTDs ] 16:22:12 <sandro> Steven: In DTD you have to say which prefixes you're going to use. 16:22:52 <sandro> em: This is a good example of embedding SVG in HTML, but SVG has a metadata tag, ... but that's lost here? 16:23:10 <sandro> M,SP: Yes. It's lost. You can't do wide-open "any" in DTDs. 16:23:18 <DanC> mimasa sorta answered jjc's questions about namespaces+DTD by showing parts of http://www.w3.org/TR/XHTMLplusMathMLplusSVG/ 16:23:28 <sandro> Ralph: So this is not the full SVG. 16:23:33 <sandro> Steven: RIght. 16:23:56 <sandro> Dan Austin: Schema has several levels of "any". 16:23:57 <danbri> steven: SVG DTD's notion of ANY isn't that of SVG XML Schema, so SVG DTD is partial representation 16:24:43 <danbri> (meeting process questions) 16:24:48 <danbri> steven: this is brainstorming really 16:25:36 <DanC> sandro: how about [basically, treating RDF as foreign string data, ala script] 16:25:48 <DanC> rrs: not just now... 16:26:43 <danbri> jjc: process point; steven's msg was Member-only, this is a public visible meeting 16:26:47 <sandro> Steven: So my proposal is something HTML authors will easily understand and be able to use. 16:26:49 <danbri> steven: I have no objection 16:26:54 <danbri> ...to it being public 16:27:16 <danbri> danc: re steven's proposal, we need to distinguish between uris, and things that look like uris that are strings 16:27:19 <sandro> DanC: How will I know if the homepage was meant as a string or a URI? 16:27:29 <sandro> Steven: uh, yeah. 16:27:48 <sandro> Dan Austin: Why not just put the RDF in a string? 16:28:05 <DaveB> ah, the <script> approach? 16:28:07 <sandro> DanC: yes -- like <script> 16:28:33 <sandro> DanC: Then the tools -- eg SAX processors -- wont get it. 16:29:40 <danbri> jjc: nodeID, datatype literals, uri-vs-string 16:29:47 <danbri> ...missing, would need adding 16:30:44 <sandro> DanBri: How about we make a regular syntax for RDF? 16:31:27 <sandro> DanC: this is implicit parsetype=Resource.......... 16:32:08 <sandro> jjc: this syntax would support a QNAMEs in string 16:32:50 <sandro> jjc: nameQname, nameURI, etc. needed. 16:32:58 <sandro> ...: 6 o r7 16:33:05 <sandro> TImBL: probably NOT a prime number. :-) 16:33:38 <sandro> Ralph's amaya drops into the debugger. 16:33:45 <DanC> [rrs's Amaya is running from a debugger... nifty ;-] 16:34:20 <sandro> DanBri: THere are lots of possible RDF sytnax. This is not far from Query/RUles syntaxes, which need to flag their variables. 16:34:50 <sandro> TimBL: Conversely, a lot of the R.*L languages are not XML and wouold embed in HTML very easily. 16:34:52 <eikeon> eikeon has joined #swarch 16:34:56 <DanC> (it seems pretty far from a query/rules/formula syntax, to me) 16:35:03 <em-lap> RDF/XHTML 16:35:18 <sandro> ...: Or if you go XML, you could sort of reify everything, like in RUleML;'s RDF syntax. 16:35:41 <sandro> JJC: XML Literals are also a problem with these syntaxes. 16:35:59 <JosD___> (quite far froof proof as well) 16:36:01 <sandro> ...: I use them a lot of puting HTML inside RDF 16:36:15 <JosD___> s/froof/proof 16:36:20 <sandro> ...: That wont fit in DTD world. 16:36:22 <Steven> Steven has joined #swarch 16:37:31 <danbri> (thinking about this more, steven's proposal is uncannily close to recent rulemarkup ideas... --> fwd reference to Harold's presentation in a later session) 16:38:08 <sandro> sandro: We either need: get rid of DTDs, use literal, make new RDF syntax. 16:38:55 <sandro> TimBL: we're turning this all backwords, breaking our document format, because our tools (DTDs) are broken. 16:39:11 <sandro> jjc: long URIs 16:39:17 <sandro> mike dean: use entities! :-) 16:39:43 <danbri> I think two paths: migrate from DTDs so we can use XML Schema 'ANY'; and also do an ntriples-in-xml (done w/ rules in mind). 16:39:59 <sandro> DanC: Surpreme Irony. XML folks keep wanting this kind of thing. 16:40:14 <sandro> TimBL: I have an element called element and and attribute called attribute. 16:40:22 <danbri> hmm 16:40:23 <danbri> [[ 16:40:24 <danbri> <resource href="mydoc"> 16:40:24 <danbri> <link rel="content-language-specific" href="mydoc.eng"> 16:40:24 <danbri> <link rel="content-language-specific" href="mydoc.fr"> 16:40:24 <danbri> </resource> 16:40:26 <danbri> ]] 16:40:30 <sandro> Dan Austin: What about embedding the HTML inside the RDF. 16:40:31 <danbri> -- http://www.w3.org/MarkUp/Resource/Specification 16:40:42 <sandro> TimBL: It would work quite well, but might be insentivive. 16:41:15 <sandro> DanC: nealry all browsers would work fine. 16:41:22 <mdubinko> would the documents be delivered with a media type of HTML or RDF? 16:41:22 <sandro> TImBL: It's valid RDF! 16:41:48 <sandro> DanC: but what media type...... 16:41:55 <sandro> Stickler: rdf+html ? 16:42:28 <sandro> Chaals: Sidesteps DTD validations by not doing it. 16:43:15 <sandro> DanC: you could put entities in the RDF document's internal subset. 16:43:35 <sandro> Brian: People want to be able to syn-validate RDF, as like DTDs. 16:43:42 <DanC> (bwm, do you tell them you *can* do XML schema validation of RDF?) 16:44:10 <sandro> JJC: (misssed) 16:44:14 <DaveB> you can XSD-validate, to a certain degree, fixed rdf/xml profiles 16:44:41 <bwm_> bwm agrees with DaveB 16:45:32 <danbri> yup, to a modest agree 16:45:37 <DanC> JJC: name="dc:title" might be DTD-ok, but no good cuz dc: isn't bound. 16:45:50 <danbri> I want to hear more about why folk want DTD validation, apart from entities. 16:46:48 <DanC> I hear some folks rely on DTD-happy systems, and alternatives aren't sufficienty mature. 16:46:59 <Steven> dc: would be bound in the document 16:47:33 <Steven> (See remark dc: isn't bound abouve; not = Dan Connolly :-) 16:48:08 <sandro> Ralph: DTDs and Open Vocabularies seem to clash -- we haven't figured out how to put them together. 16:48:43 <sandro> TimBL: Would Schema solve this? 16:48:58 <sandro> DanC: It's not possible to write one-last-true-schema for RDF/XML. 16:49:27 <sandro> ...: For each property you have to say whether you're going to use it as an element or an attribute. Or at least it's very very hard. 16:49:57 <DanC> pf of "not possible to write one-last-true-schema for RDF/XML": consider rdf:_1, rdf:_2.... QED. 16:49:59 <sandro> Mimasa: If we forget about DTD and schema. Modular Namespaces.... 16:50:11 <sandro> ....: In RelaxNG 16:50:39 <em-lap> http://www.taiopensource.com/relaxng/mns.html#example 16:50:52 <em-lap> http://www.thaiopensource.com/relaxng/mns.html#example 16:51:20 <sandro> Steven: Relax took modularization to heart. 16:52:42 <sandro> Joseph: I consider Schema's "any" feature dangerous, because of its need for deterministic parsing, when you extend (eg merging in another namespace) the schema, it breaks. 16:52:50 <sandro> ...: Relax does much better here. 16:53:04 <bwm_> bwm claims prize for embedding html in RDF and displaying in browser 16:53:29 <Steven> demo! 16:53:40 <sandro> Ralph/DanC: "equivalence class" in XML Schema. No "any element can appear here", but "any element whose type is X can appear here." 16:54:00 <DaveB> surely some proposed rddl solutions are html in rdf 16:54:04 <DanC> where X is, e.g., rdf property element 16:54:20 <sandro> scribe sees bwm's document. 16:55:06 <sandro> TimBL: first do a pass, for every namespace, of gathering data and making a DTD. 16:55:20 <sandro> Dan Austin: that would only validate local chunks, not the document as a whole. 16:55:32 <em-lap> bwm, do you want to show this? 16:55:43 <sandro> DanBri: I want to hear more about why people want to stick with DTDs. 16:55:53 <sandro> Steven: Existing tools..... 16:56:11 <sandro> em-lap, bwm is trying to send it to an archive. 16:57:26 <sandro> Steven: time check.....? 16:57:35 <sandro> em: Next steps? 16:58:07 <shinichi_> shinichi_ has left #swarch 16:58:35 <sandro> NOT-REAL ACTION: Steven push back on alternatives to DTDs 16:58:56 <sandro> Steven: I'm not optimisitics 16:59:15 <sandro> TimBL: Until then, we'll just put our HTML inside RDF. :-) 16:59:33 <sandro> Ralph: What about this meta/meta proposal? 16:59:38 <DaveB> can't say I'm happy with that, as rdf/xml spec editor 16:59:47 <DaveB> - the html in RDF/XML that is 16:59:53 <DaveB> DanC: yes 17:00:08 <sandro> DaveB -- it's as XML literlas, I think. 17:00:27 <DaveB> sandro: ok 17:00:46 <DanC> DanC: DaveB is willing to help the HTML WG with the <meta name="dc:title"> idea 17:00:51 <sandro> Several people express interest in helping HTML WG make a better meta syntax. 17:01:20 <DaveB> groan 17:01:30 <DanC> groan? did I misunderstand? 17:01:31 <DaveB> helping? :) 17:01:48 <DaveB> I'd prefer a few people helping ;) 17:02:00 <sandro> em makes a strong pitch for solving this problem together 17:02:13 <sandro> TimBL: If you do meta/meta, you have to show at least how it maps to RDF. 17:02:55 <sandro> Steven: Even now you see dc.title...... Thjere';s presure here. 17:03:01 <danbri> for record, i volunteer to review proposals, and have a specific interest in commonalities between new rdf dump syntax and work on rule/query markup 17:03:01 <sandro> em: that's mine and I hate it! 17:03:10 <DaveB> ha ha 17:03:27 <sandro> Jos: If you can't do it by value, due it by reference 17:03:28 <danbri> timbl: if you don't say it is rdf, you create a whole new space of html meta property uris 17:03:28 <DaveB> +danbri? 17:03:47 <sandro> TimBL: href="data:application/rdf+xml; ...." (sandro's solution) 17:04:18 <DanC> DanC: I may have misrepresented DaveB's offer 17:04:30 <sandro> but today's talk is about embedding. 17:04:46 <sandro> Dan Austin: Can you put it at the end? 17:04:49 <DanC> DanC: ... so I'm not quite sure what he's offering. 17:04:57 <DaveB> DanC: I'm offering to work with html wg 17:05:09 <sandro> TimBL: You may need to keep it in the <head> to get nice behavior from old browsers. 17:05:31 <sandro> em: HTTP HEAD. 17:05:44 <sandro> [ do that ever read the HTTP-EQUIV meta...?????????? wonders sandro ] 17:06:14 <sandro> DanBri: Markup in body, use case: [didnt get it] 17:06:31 <DaveB> DanC: I was just cautious about having lots of helpers. I claim I'm an expert on RDF and XML, I'd want helpers who were too (+HTML) 17:06:44 <sandro> ADJOUNRED . RETURN AT 1:25. 17:08:01 <sandro> ========================= 17:08:01 <sandro> (for presentations on semantic layering) 17:08:18 <PStickler> PStickler has joined #swarch 17:08:46 <mdean> mdean has joined #swarch 17:08:57 <danbri> danbri has joined #swarch 17:11:06 <micah> micah has joined #swarch 17:12:21 <bwm> bwm has joined #swarch 17:12:26 <bwm> http://www.w3.org/2001/sw/RDFCore/20030305-swarchmeeting/experiment/World%20Wide%20Web%20Consortium.html 17:12:33 <sandro> EDIT ME 17:18:37 <mdubinko> mdubinko has joined #swarch 17:19:07 <danbri> sandro.edit() 18:04:18 <DaveB> DaveB has joined #swarch 18:22:50 <bwm> bwm has joined #swarch 18:25:17 <libby> libby has joined #swarch 18:26:32 <bwm> bwm has joined #swarch 18:27:27 <libby> "This is a little fancy that lets other users (on the same subnet) find out about you" 18:27:34 <libby> (tiny download) 18:27:53 <libby> assuming you want to be found out about of course 18:29:38 <RalphS> RalphS has joined #swarch 18:29:42 <mdean> mdean has joined #swarch 18:30:25 <simonSNST> simonSNST has joined #swarch 18:32:05 <lasf2f> lasf2f has joined #swarch 18:33:10 <ircleuser> ircleuser has joined #swarch 18:34:03 <bwm> pat begins 18:34:12 <DanC> -------------------------- 18:34:13 <bwm> appologies for unreadability of slides 18:34:22 <DanC> === 13:30 - 15:30 First afternoon session 18:34:22 <DanC> Semantic Layering. presentation by Peter F. Patel-Schneider, Pat Hayes 18:34:24 <danbri> (just found http://www.xml.com/pub/a/2003/03/05/social.html -- kendall clark on social meaning) 18:34:51 <bwmscribe> pat adjusts magnification of the slides 18:34:54 <DanC> note to self: get pat's jpg file into http space 18:35:05 <bwmscribe> no url to refer to 18:35:18 <bwmscribe> a picture of an rdf universe 18:35:30 <JosD___> JosD___ has joined #swarch 18:35:35 <bwmscribe> no constraints on anything - classes can contain themselves etc 18:35:43 <bwmscribe> owl has a very different picture 18:36:02 <bwmscribe> owl is neater 18:36:40 <bwmscribe> an owl dl universe is neat; segregated; separation between classes and individuals etc 18:36:52 <bwmscribe> classifcal fol way of building an interpretation 18:37:11 <bwmscribe> pat has fun zooming slide 18:37:30 <bwmscribe> owl full says scruffy is the way to go 18:37:49 <bwmscribe> owl classes as rdf classes etc 18:37:58 <bwmscribe> just use the owl vocabulary freely in rdf 18:38:16 <las> las has joined #swarch 18:38:27 <Tantek> Tantek has joined #swarch 18:38:29 <bwmscribe> a different way to do this is to retain the neatness of the owl picture 18:38:59 <bwmscribe> s/owl/owl dl/ 18:39:17 <bwmscribe> some classes are owl dl classes, similarly for properties etc 18:40:26 <bwmscribe> owl lets you reason about the number of things in a set 18:41:20 <bwmscribe> the process of embedding the more complex owl syntax into triples introduces things into the owl universe, e.g. lists, which don't bother RDF but do bother owl dl. 18:41:52 <bwmscribe> jos asks what the compliment of owl:thing would be in the diagram 18:42:15 <bwmscribe> there is a class called owl:nothing which cannot contain owl:things but can contain other things. 18:42:30 <bwmscribe> pfps: owl:nothing is the empty set 18:43:22 <chaalsBOS> chaalsBOS has joined #swarch 18:43:57 <bwmscribe> if you stick to the appropriate subset of rdf you can be neat and work in owl dl 18:44:17 <bwmscribe> within that subset the semantics of owl dl and owl full are equivalent 18:44:36 <bwmscribe> problems is its hard to determine if you are in this subset. 18:45:05 <bwmscribe> folks think you just check the syntax, but that isn't good enough because entailments can take you out of the subset. 18:45:53 <bwmscribe> why bother? 18:46:13 <bwmscribe> reasoning about owl dl is a more comfortable place to be 18:46:34 <bwmscribe> neatness guaratees powerful efficient reasoning engines 18:47:07 <bwmscribe> pat didn't expect the equivalence to work - and its not too outragously complicated 18:47:30 <bwmscribe> pat shows rdf, rdfs and rdf-dt in lbase 18:47:53 <bwmscribe> and owl as well, I think - not sure 18:47:59 <DanC> ACTION: PatH send 'OWL, RDFS, RDF in lbase' to connolly, em, www-archive 18:48:23 <Nobu> Nobu has joined #swarch 18:48:40 <bwmscribe> danbri: would this be easier in owl abstract syntax rather than rdf representation 18:48:55 <bwmscribe> patH: yes it would be a standard translation of a DL into fols 18:48:58 <bwmscribe> s/fols/fol/ 18:49:32 <bwmscribe> PatH hands over pfps 18:49:48 <DaveB> but OWL's abstract syntax is triples 18:50:45 <GuusS> GuusS has joined #swarch 18:50:57 <JosD___> Brian, it's called LBASE-new.html 18:51:36 <sanScribe> And now PFPS with the glass half-empty view of layering...... 18:52:04 <sanScribe> pfps: (there's ONLY ONE DROP of beer in the glass.) 18:52:39 <sanScribe> ...: We did the impossible. Why did we have to do research here? 18:52:57 <danbri> dancon: because they wouldn't let me close the WG 18:53:23 <jhendler_> we had to do research because the neats couldn't live with the scruffiness of Owl Full 18:53:24 <sanScribe> ...: Logic layering (proposition, DL, Prop-modal, FOL, HOL) is old. 18:53:36 <sanScribe> ...: (many branches, some dont layer) 18:53:50 <sanScribe> ...: (bump around modal) 18:53:56 <danbri> p: "modal sorta sits off to the side" 18:54:45 <sanScribe> ...: Different syntax for logics, compatible semantics. 18:55:00 <sanScribe> ...: eg FOL does *not* allow p ^ q 18:55:02 <bwmscribe> pats lbase doc: http://lists.w3.org/Archives/Public/www-archive/2003Mar/att-0018/LBASE-new.html 18:55:13 <sanScribe> ...: I can write a parse for any of these (in a day or so). 18:55:37 <sanScribe> ...: But in SemWeb, we need to use the same syntax?! 18:56:30 <em-lap> q? 18:56:37 <Zakim> Zakim has joined #swarch 18:56:46 <sanScribe> ...: Traditionally, compatible is just fine. Same-syntax is not needed. 18:57:15 <sanScribe> ...: Of course there are lots of other langs, like Montague, that are the result of a philosopher getting too much rope and hanging himself. 18:58:09 <sanScribe> ...: Montague wanted to write the ultimate logic. It's the spruce goose. :-) 18:58:44 <sanScribe> ...: SemWeb side is like working in a straightjacket. 18:58:51 <sanScribe> [ This is all slide 2 ] 18:59:13 <bwmscribe> bwm wonders why these syntactic constraints were imposed 18:59:44 <sanScribe> http://www-db-out.bell-labs.com/user/pfps/talks/semantic-layering/slide2-0.html 19:00:37 <sanScribe> DanC: WebOnt decided on the straightjacket, biased by the charter. 19:01:00 <DanC> JJC actually said that, but I agree. 19:01:07 <sanScribe> pfps: Allowing extensions to the syntax would have gotten us out of straightjacket. 19:01:21 <sanScribe> timbl: all the rules languages extend syntax. 19:01:46 <sanScribe> pfpfs: We were able to do this because OWL doesnt need variables. Ohhh, can we do FOL without variables? 19:01:52 <sanScribe> ph: don't go there! 19:02:29 <danbri> I really don't think 'every layer (above XML) must use RDF triples' is accepted even amongst RDF-enthusiasts. 19:02:37 <sanScribe> jjc: I challenge that semantic compatibile is rosy. They took a 100 years, we took one. 19:03:28 <DanC> I expect RDF syntax to grow forAll/variables. i.e. not just triples-as-are... but I didn't think the ontology layer was the layer in which to do it. 19:03:54 <sanScribe> Ian: RDF is not such a simple, well-understood layer as Proposition Logic. 19:05:05 <sanScribe> pfps: 1. syntax of triples is so impoverished (a,b,c) and each term has to denote, all in a set. 19:05:22 <sanScribe> .... (4 problems with RDF) 19:05:51 <sanScribe> ...: * every triple denotes, it has its own little say. 19:06:16 <sanScribe> ...: (see Dark Triples.) 19:06:36 <RalphS> PatH: we got along fine with this restricted syntax in lisp 19:07:06 <sanScribe> ...: Rephrase: 1a Every Triple has it say, 1b Every term has to denote. 19:07:39 <em-lap> q? 19:07:52 <sanScribe> ...: Literals denote kind of like a built-in functional term. 19:08:07 <em-lap> ack danc 19:08:08 <Zakim> DanC, you wanted to say that syntactic interoperability is a constraint derived from the principle of partial understanding 19:08:22 <sanScribe> ...: 1. Syntax is confiding, 2 semantics of terms in confing, 3. semantics of triples-asserted is confining. 19:08:32 <sanScribe> DanC: Lack of n-ary in RDF hurts a lot. 19:08:54 <sanScribe> DanC: Shared Syntax comes from a need for partial understanding. 19:09:22 <sanScribe> pfps: I don't believe that fixing the syntax helps partial understanding. 19:10:18 <sanScribe> ..: Because then you have to encoded other things (eg disjunctions) in a way which is inscrutible. 19:10:47 <sanScribe> DanC: We didnt need universals for OWL. When we need them, I expect we'll extend the syntax. 19:11:09 <sanScribe> jjc: own has disjunctions 19:11:25 <sanScribe> pfps: it has a certain kind of disjunctive expression 19:12:13 <timbl> timbl has joined #swarch 19:12:14 <sanScribe> slide 5. http://www-db-out.bell-labs.com/user/pfps/talks/semantic-layering/slide5-0.html 19:12:59 <sanScribe> http://www-db-out.bell-labs.com/user/pfps/talks/semantic-layering/slide6-0.html 19:14:12 <RalphS> pfps: the "myFavoriteMartian" property [might] contribute something to the meaning of the restriction but you don't know what 19:14:14 <sanScribe> skipping up to slide 11 19:14:27 <RalphS> ... but you can't ignore it 19:14:30 <sanScribe> http://www-db-out.bell-labs.com/user/pfps/talks/semantic-layering/slide11-0.html 19:16:16 <las> las has joined #swarch 19:16:27 <sanScribe> pfps: You need comprehension principles; from nothing you can infer the existence of every list.. 19:16:27 <sanScribe> DanC: section 2 of kiff says all lists exist. This is not odd. 19:16:34 <RalphS> RalphS has joined #swarch 19:16:51 <sanScribe> ph: I don't mind comprehension priniciples! 19:17:14 <sanScribe> pfps: people often hang themselves with these. 19:17:51 <simon> simon has joined #swarch 19:17:53 <bwmscribe> bwmscribe has joined #swarch 19:18:05 <sanScribe> : Because everything is at the top level, ... you can have loops, liar paradox 19:18:40 <sanScribe> ...: comprehension principles might encompass liar paradox; trivial theory 19:18:47 <sanScribe> ...: bad 19:18:59 <sanScribe> http://www-db-out.bell-labs.com/user/pfps/talks/semantic-layering/slide12-0.html 19:19:19 <sanScribe> ..: So we'll violate RDF and say *ONLY* the tree-structured statements exist. 19:19:43 <sanScribe> ...: So some inferences you might think exist do not. 19:20:04 <Nobu> Nobu has joined #swarch 19:21:07 <Nobu> Nobu has joined #swarch 19:21:11 <sanScribe> jjc: you can wait until someone asks.... 19:21:19 <sanScribe> pfps: thank you Socrates 19:21:40 <sanScribe> jjc: infinite stuff makes some implementation approaches impossible. 19:22:02 <sanScribe> jjc: you can't list all the conclusions of an owl KB 19:22:06 <sanScribe> ian: no, of course not. 19:22:21 <sanScribe> pfps: concluding with slide 8. my proposal..... 19:22:31 <sanScribe> ...: weakest version 19:22:42 <sanScribe> ...: You can change the semantics as long as it works out right in the end. 19:22:55 <sanScribe> ...: That's wrong for the SemWeb. 19:23:03 <sanScribe> (slide 9) 19:23:18 <sanScribe> ...: We want the language to be in the superset relation, as in FOL and HOL. 19:24:18 <sanScribe> ...: Weak semantic corresopnence. [Unclear]. As in modal and DL, which act the same even when they look the same inside. 19:24:30 <sanScribe> PFPS: we should break straightjacket by 19:24:36 <sanScribe> ...: 1) allow syntactic extensions 19:24:46 <sanScribe> ...: 2) allow expansions of semantics. 19:24:50 <ryanlee> ryanlee has joined #swarch 19:25:43 <sanScribe> Brian: RDF is for stating simple facts. We also need a language for describing types and classes. WHy use the same? 19:26:07 <sanScribe> pfps: There is a strange attractiveness to eating your own dogfood. 19:26:17 <sanScribe> Brian: I make frames for pictures, but I don't eat them. 19:26:41 <sanScribe> DanC: why is less-than-is-transitive anything other than a simple fact. 19:26:54 <sanScribe> Ian: For 100 years people figured out it was a bad idea! 19:27:02 <sanScribe> PH: No no, it's fine. We got it to work! 19:27:15 <sanScribe> PH: Even God has had model theories! 19:27:30 <DanC> "... even God has to have model theories ..." -- PatH 19:27:37 <sanScribe> PFPS: Given you had to do the stuff on the side, what's the benefit of having the syntaxes the same. 19:28:23 <sanScribe> PH wearing JH's had: Pragmatic Utility to passing complex RDF through simple RDF parsers. It's not a miracle, but you can get down a pipe. 19:28:35 <sanScribe> pfps: Let's make the pipe bigger. 19:29:09 <sanScribe> Lynn: You don';t get to make the pipe -- you need to pick one existing pipe. 19:29:15 <sanScribe> pfps: My pipe is a string of bits. 19:29:31 <sanScribe> Lynn: there are some pipes out there which we want to use. 19:29:45 <sanScribe> PFPS: I'd be relatively happy picking XML as my pipe. 19:30:38 <em-lap> q+ ian 19:30:42 <em-lap> q+ path 19:30:44 <sanScribe> EricP: WHen you encode you HOL's in a simple graph, you need to hide certain bits from the simple graph. Need reification or something. [ discomfort from PH and PFPS ] What subset have you had to hide from naive processors? 19:30:44 <em-lap> q? 19:31:09 <sanScribe> pfps: Hiding wasn't allowable, so we hid it in a blizzard. 19:31:26 <sanScribe> Ian: More on pipe analogy. We didnt pick it, it chose us. 19:31:34 <danbri> fair comment! 19:31:37 <em-lap> ack ian 19:31:40 <em-lap> ack path 19:31:56 <sanScribe> PH: EricP's way is misleading. We didnt have to hide anything. 19:33:50 <sanScribe> ericP: would a quote operator have helped? 19:33:58 <Bernard> Bernard has joined #swarch 19:34:10 <sanScribe> pfps: In OWL we never have to write down those primitive triples. 19:35:14 <sanScribe> ph: If you come to this world from logic, it's like falling down the rabbit hole.... 3 of them. Cherry picker holding cherry picker holding holder. 19:35:37 <RalphS> Sandro: I did unification back in August using reification 19:36:11 <RalphS> PatH: the truth predicate that you then need for de-reification is scary 19:36:49 <shellac> shellac has joined #swarch 19:40:17 <sanScribe> DanC: Peter, will you write up what you said in 5-page form? 19:41:25 <sanScribe> Ben Grosof: Problems/opportunities in layer. RDF syntax as encoding as RuleML. 19:41:53 <sanScribe> ...: OWL fits in more naturally.... 19:42:43 <sanScribe> PFPS: sounds like Owl-DL philosophy 19:43:38 <sanScribe> TimBL: You're talking about reification -- the RDF your describting your expression with is different from the instance RDF -- tool reuse is nice, but they don't really connect. 19:44:26 <sanScribe> BG: What about rules which can talk about RDF triples only.....? 19:44:49 <sanScribe> TimBL: (misssed) 19:45:07 <danbri> timbl said that lots of folk have a 'triples plus ...<something>' language on top of rdf 19:45:09 <sanScribe> Ian: Partial fix perhaps: stratified semantics for RDF, instead of recursive semantics. 19:46:32 <sanScribe> BG: second. RDF and RDFS are sufficiently rich ............(lost) 19:46:42 <sanScribe> em: strata == modules? 19:46:54 <sanScribe> em: separate namespaces, specific strata, ....? 19:46:56 <DaveB> I assume Ian's refering to rdfs(fa), 2 years too late IMHO 19:47:21 <sanScribe> Ian: Layer1-class, Layer2-class, .... no looping. 19:47:30 <sanScribe> PH: You could use owl:Class and owl:Property 19:47:46 <sanScribe> Ian: you might be able look at it. 19:47:53 <sanScribe> DanC: like ML type checker. 19:48:07 <sanScribe> TimBL: does that work on unbounded, open, semantic web. 19:48:14 <sanScribe> PH: You're safe until you pull in more data. 19:48:26 <sanScribe> TimBL: when you publish, OTHER PEOPLE combine you. 19:48:46 <sanScribe> ...: Different views. 19:48:53 <sanScribe> Ian: <nods> 19:49:30 <sanScribe> TimBL: Loops can emerge from different parts. 19:49:56 <sanScribe> DanC: Does there have to be a bottom layer. 19:50:15 <timbl> I said pointed out that BenG's use of RDF with RuleML he was describing was to use RDF to describe the otheer langauge, keeping RDF on a completely different level, making no connection netween assertions and object sin RDF and in teh other language. 19:50:42 <sanScribe> Ian: Two things in same strata.... 19:50:58 <sanScribe> ...: If you fixed names, class0, ...... 19:51:08 <sanScribe> PH: there are better things than stratified type theory. 19:51:27 <sanScribe> Ian: What's "works" 19:53:20 <sanScribe> BG: start with simple case -- first stratum -- classes do not have classes or proptiers or instances, ... DL constraint -- Simple==no constraints so Higher Order, vs Simple==clear constaints for tractability. 19:54:55 <sanScribe> ...: It doesnt make sense to treat the whole web as scopeless, as one KB. 19:55:56 <sanScribe> [ bit about lists ] 19:56:21 <sanScribe> TimBL: it's reasonable to have granularity, but the web works by people combining KBs. 19:56:49 <sanScribe> BG: Sure - if you have 3 KBs meeting DL restriction, you know the merge of them also meets it. 19:57:01 <sanScribe> TimBL: you migth pick up KBs out of google. 19:57:24 <sanScribe> q? 19:57:58 <sanScribe> PH: it should be up the READER to check cohernece. 19:58:36 <sanScribe> BG: Inferential machinery constructs a 3rd (perhaps virtual) KB. It has finite scope. 19:58:52 <sanScribe> TimBL: what if 3rd KB barfs? 19:59:12 <sanScribe> BG: an ecology will support people publ;ishing KBs which can be combined well! 19:59:42 <sanScribe> ...: Advertize that you are recursion free, etc, etc, quadtatic, etc. ... 20:00:04 <sanScribe> PH: Or even which heuristics are likely to work. Little hacks, repairs. 20:01:12 <sanScribe> Frank: This is no different from Data Sources on the web. One says Foo, another says Not Foo; you require that the party finding both to find out who to believe, etc. 20:01:51 <sanScribe> TimBL: sure, one of them is wrong. 20:02:50 <sanScribe> ...... 20:02:59 <sanScribe> em: WHat about real systems> 20:03:09 <sanScribe> Ian: My system detects conflicts. 20:03:26 <sanScribe> TimBL: we're talking about constraints on class hierarchy cycles. 20:04:10 <sanScribe> pfps: You beleive either class A is below, above, or we have a cycle. Like Each source is right or wrong. 20:04:47 <sanScribe> ph: some rdf-fcomments requesting stratifivcation. 20:05:10 <sanScribe> DanC: we could have a note which said " Here is a way to do stratified. " 20:05:35 <sanScribe> JJC: stratified RDF is just intersection of RDFS & OWL Lite. 20:05:47 <sanScribe> Ian: THat's PFPS trivial 2-layer strat. 20:05:58 <RalphS> JJC: the stratified RDFS is simply the intersection of RDFS and OWL lite 20:06:01 <sanScribe> DanC: But that's editoriallly expensive. Hard tor read it all. 20:06:13 <RalphS> PatH: agree 20:06:25 <sanScribe> ian; Syntax of owl-lite is another layer. 20:07:15 <sanScribe> PH: just use OWL vocabularies without OWL meanings! :-) 20:08:49 <sanScribe> PH: Who uses OWL terms without importing OWL semantics? 20:09:01 <sanScribe> [ various hands, including TimBLs to everyeone's surprise. ] 20:11:20 <sanScribe> RRSAgent, pointer? 20:11:20 <RRSAgent> See http://www.w3.org/2003/03/06-swarch-irc#T20-11-20 20:11:45 <sanScribe> ============================== 20:11:49 <sanScribe> RECESS for 45 minutes 20:52:01 <sanScribe> test 20:52:56 <libby> libby has joined #swarch 20:53:03 <bwm> bwm has joined #swarch 20:55:30 <JosD___> JosD___ has joined #swarch 20:57:06 <pfps> pfps has joined #swarch 20:57:47 <danbri> ping 20:57:52 <danbri> irc seems ok to me 20:58:06 <bwm> me too 20:58:54 <danbri> http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2002Jun/0180.html -- some early text from pathayes & me that became (prev version) 'social meaning' section of Concepts 20:59:09 <bwmscribe> neg comments about section 4 of rdf concepts doc 20:59:28 <bwmscribe> it would be nice if the protagonists could beat each other up instead of the editors 21:00:04 <bwmscribe> desired output from this meeting is advice for the rdfcore wg 21:00:20 <danbri> http://www.w3.org/2001/sw/meetings/tech-200303/social-meaning.html -- jjc's agenda for 'social meaning' discussion here 21:00:27 <bwmscribe> it is not an objective to make changes to the doc 21:00:41 <bwmscribe> options: 21:00:48 <bwmscribe> accept section 4 21:00:54 <bwmscribe> cut section 4 21:00:58 <bwmscribe> edit section 4 21:00:59 <DaveB> who's speaking? 21:01:02 <bwmscribe> end of list 21:01:06 <bwmscribe> jjc 21:01:15 <bwmscribe> jjc leading section 21:01:47 <bwmscribe> bijan identified issuettes 21:02:37 <bwmscribe> jjc suggests straw polls on issuettes questions 21:03:53 <bwmscribe> danc: is this a live with thing or a preference 21:03:56 <bwmscribe> jjc: preference 21:05:10 <bwmscribe> Section 4 as a whole: 21:05:20 <bwmscribe> accept: 1 21:05:34 <bwmscribe> cut: 7 21:06:25 <bwmscribe> aggressive edit: 6 21:06:54 <bwmscribe> who can't live with: 21:07:18 <bwmscribe> endorsing: 4 21:07:28 <bwmscribe> cut: 2 21:08:13 <bwmscribe> major edit: 2 21:08:19 <bwmscribe> endorsing s/4/6/ 21:10:54 <bwmscribe> authoritive definition of URI's: i.e. who gets to say what a URI denotes 21:12:48 <danbri> something like "RDF graphs have propositional content. Their meaning is fixed by a bunch of hairy stuff only partly understood and documented (eg. implicit theory of reference associated with URIs). Minor health warning. The End." 21:12:53 <danbri> ...would satisfy me. 21:14:03 <bwmscribe> who wants the sentence [[The social conventions surrounding use of RDF assume that any RDF URI reference gains its meaning from some defining individual, organization or context. ]] 21:14:26 <PStickler> Sounds fine to me 21:14:30 <bwmscribe> A: no substanstive change:6 21:14:36 <bwmscribe> C: cut: 4 21:14:40 <bwmscribe> s/4/5/ 21:14:54 <bwmscribe> cant live iwth: 21:14:58 <bwmscribe> as is: 4 21:15:00 <bwmscribe> cut: 2 21:15:16 <danbri> My 'E' vote meant "we should note the importance of reference in the task of fixing meaning" 21:15:19 <bwmscribe> more work: 2 21:16:00 <PStickler> My 'E' vote is wanting the simple clarification that the URI and URIREG conventions indicate *who* assigns meaning, not what the meaning is. 21:16:44 <bwmscribe> defining information: [[An RDF graph may contain "defining information" that is opaque to logical reasoners. This information may be used by human interpreters of RDF information, or programmers writing software to perform specialized forms of deduction in the Semantic Web.]] 21:17:14 <bwmscribe> a: is ok 1 21:17:21 <bwmscribe> c: cut: 8 21:17:27 <bwmscribe> eikeon: more work 2 21:17:37 <bwmscribe> cant live 21:17:45 <bwmscribe> as is: 8 21:17:50 <bwmscribe> cut: 2 21:17:54 <bwmscribe> more work: 2 21:18:25 <bwmscribe> s/eikeio/ edit 21:18:34 <bwmscribe> asserting rdf 21:20:45 <bwmscribe> [[RDF/XML expressions, i.e. encodings of RDF graphs, can be used to make claims or assertions. Not every RDF/XML expression is asserted. ]] about the 'real' world. Such expressions are said to be asserted. 21:21:04 <bwmscribe> preference: 21:21:20 <bwmscribe> a: its ok - 3 21:21:32 <bwmscribe> cut: 4 21:21:34 <bwmscribe> edit: 2 21:21:46 <bwmscribe> can't live with: 21:22:57 <las> las has joined #swarch 21:23:25 <Bernard> Bernard has joined #swarch 21:23:35 <bwmscribe> as is: 1 21:23:49 <em-lap> em-lap has joined #swarch 21:23:56 <bwmscribe> s/1/3/ 21:24:09 <bwmscribe> cant live: 1 21:24:12 <bwmscribe> more work: 0 21:24:50 <ericP> ericP has joined #swarch 21:25:12 <bwmscribe> relationship to http, ignoring the 404 21:25:26 <bwmscribe> keep: 2 21:25:34 <bwmscribe> cut: 6 21:25:38 <chaalsBOS> chaalsBOS has joined #swarch 21:25:44 <bwmscribe> edit: 0 21:25:52 <bwmscribe> s/0/1/ 21:25:55 <bwmscribe> cut live 21:25:58 <bwmscribe> as is: 3 21:26:14 <bwmscribe> cut 0 21:26:20 <bwmscribe> more work 1 21:26:25 <Tantek> Tantek has joined #swarch 21:27:17 <bwmscribe> mechanical inference and social meeting [[Human publishers of RDF content commit themselves to the mechanically-inferred social obligations. 21:27:17 <bwmscribe> The meaning of an RDF document includes the social meaning, the formal meaning, and the social meaning of the formal entailments. ]] 21:27:50 <bwmscribe> prefer: 21:27:57 <bwmscribe> as is: 3 21:28:03 <bwmscribe> get rid: 5 21:28:17 <bwmscribe> edit: 2 21:28:28 <bwmscribe> can't live with: 21:28:37 <bwmscribe> as is: 7 21:28:45 <bwmscribe> cut: 1 21:29:11 <bwmscribe> more work 0 21:29:28 <bwmscribe> slipperyness of intended meaning 21:30:42 <em-lap> RRSAgent, pointer? 21:30:42 <RRSAgent> See http://www.w3.org/2003/03/06-swarch-irc#T21-30-42 21:30:56 <bwmscribe> [[A combination of social (e.g. legal) and technical machinery (protocols, file formats, publication frameworks) provide the contexts that fix the intended meanings of the vocabulary of some piece of RDF, and which distinguish assertions from other uses (e.g. citations, denials or illustrations).]] 21:31:15 <danbri> we didn't have 'intended' in the original, http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2002Jun/0180.html 21:32:31 <simon> simon has joined #swarch 21:33:03 <bwmscribe> [[The social machinery includes the form of publication: publishing some unqualified statements on one's World Wide Web home page would generally be taken as an assertion of those statements. But publishing the same statements with a qualification, such as "here are some common myths", or as part of a rebuttal, would likely not be construed as an assertion of the truth of those statements.]] 21:33:55 <PStickler> The URI specs don't fix the meaning. They fix the *authority* who gets to fix the meaning. A schema provided by that authority can fix the meaning of a given URI. 21:34:30 <sanScribe> timbl: We're defining a system, mostly via natural language ... specs describe how it's supposed to work. 21:35:09 <sanScribe> ...: Notionally, the URI authority is supposed to define the meaning, but if it doesn't then things may or may not break. 21:35:29 <sanScribe> ...: Just people people don't always stick to the spec doesnt mean we shouldn't define it cleanly. 21:36:04 <sanScribe> Bijan: I disbelieve nearly everything you just said. If everything you said was false, nothing would be broken. 21:36:14 <sanScribe> ...: Most of what you said is misleading at least. 21:36:31 <DanC> DanC has joined #swarch 21:37:15 <sanScribe> TimBL: okay to rip out "legal" 21:37:29 <sanScribe> Bijan: I don't beleive you can fix one interpretation of a document. 21:38:00 <sanScribe> Bijan: you said "it means this" and that doesnt fly and is not necessary. 21:38:03 <las> las has joined #swarch 21:38:06 <sanScribe> TimBL: you're modeling too formally. 21:38:22 <sanScribe> ...: "H1 is a top level heading". 21:38:57 <sanScribe> On Slipperyness: 21:39:58 <sanScribe> a: 3 21:40:10 <chaalsBOS> s/3/4/ 21:40:11 <sanScribe> (modulo "intended") 21:40:19 <sanScribe> c: 5 21:40:30 <sanScribe> e: 6 21:40:36 <sanScribe> timbl: put it in primer 21:40:46 <sanScribe> a: lots and lots 21:41:01 <sanScribe> cant live with lots and lots 21:41:09 <sanScribe> cant live with c -2 21:41:13 <sanScribe> cant live with more work 0 21:41:23 <sanScribe> Third Party Publication 21:42:24 <RalphS> RalphS has joined #swarch 21:42:42 <sanScribe> a: 0 21:42:49 <sanScribe> c: 8 21:42:59 <sanScribe> e: 5 21:43:13 <sanScribe> a: -7 21:43:20 <sanScribe> c: -1 21:43:26 <sanScribe> e: 0 21:45:34 <sanScribe> Insult 21:47:34 <bwm_> bwm_ has joined #swarch 21:47:35 <sanScribe> TimBL: If this is all describing the framework, but Bijan is generally correct about how this applies as a matter of law. 21:47:35 <sanScribe> PH: We can have non-legal examples. 21:47:55 <sanScribe> Frank: I don't recall anywhere in HTML anything about libelous statements. But there's nothing peculiar about RDF in this regard. 21:48:35 <RalphS> RalphS has joined #swarch 21:48:40 <sanScribe> Mike Dean: we need a libel markup language...... 21:48:54 <Nobu> Nobu has joined #swarch 21:48:58 <sanScribe> DanC: Proximate Cause, Bernstein, Copyright law, links as speech, .... 21:50:30 <sanScribe> Bijan: I was commenting on the "mechanically inferred" part, NOT the insult bit. 21:51:20 <danbri> danbri has joined #swarch 21:51:44 <ericP> minutes are being written to http://www.w3.org/2003/03/06-SW-Arch 21:52:16 <sanScribe> RRSAgent, pointer? 21:52:16 <RRSAgent> See http://www.w3.org/2003/03/06-swarch-irc#T21-52-16 21:52:30 <sanScribe> Hrm. They are also being written there. 22:07:11 <sanScribe> TimBL: The essential point is that the RDF spec must say that the meaning of each URIRef term is based on the URI spec tree. 22:07:54 <jhendler_> jhendler_ has joined #swarch 22:14:19 <danbri> yeah, useful discussion 22:23:38 <Zakim> Zakim has left #swarch 22:24:06 <RalphS> [em, do you think that Tim has given an answer to Sun's question of "how _do you want me_ to refer to a piece of equipment?"] 22:24:46 <RalphS> [... a _clear_ answer for Sun's purpose, that is ;)] 22:28:31 <em-lap> [Ralph, no i do not] 22:28:34 <jhendler_> is this being scribed somewhere?? 22:28:42 <em-lap> RRSAgent, pointer? 22:28:42 <RRSAgent> See http://www.w3.org/2003/03/06-swarch-irc#T22-28-42 22:28:59 <bwm_> frank: we need to say things about how rdf should be used 22:29:03 <RalphS> [em, that was my impression too -- I hope we don't adjourn before getting him back up to answer] 22:29:35 <RalphS> Frank: it is important that the RDF spec say something about intended meaning 22:29:47 <bwm_> jjc: we need to come to a consusus if possible about what this meeting thinks 22:31:10 <em-lap> [ralph, i absolutely agree] 22:33:47 <bwm_> patrick s returns to point about having this covered in another document 22:34:09 <bwm_> bwm points out that cut doesn't mean cut forever, the issue is when do we do the work 22:34:32 <bwm_> danc calls for who would be interested in contributing to further work in another forum 22:34:51 <bwm_> pfps says that those voting to cut also think there is something important to say here 22:35:28 <GuusS> GuusS has joined #swarch 22:35:31 <bwm_> Propose section 4 should be struck and the CG should ensure this area is addressed in another document. 22:35:47 <bwm_> pfps there are shadows of section 4 elsewhere 22:36:29 <bwm_> DanC: Propose rdfms-assertion should be postponed and rdfcore relieved of responsibility 22:36:59 <bwm_> RalphS: does this mean that we'd have to change the namespace when that new doc was published? 22:37:10 <danbri> http://www.w3.org/TR/rdf-schema/#ch_comment in RDFS is in a similar vein. 22:37:27 <danbri> [[ 22:37:28 <danbri> A variety of documentation forms can be combined to indicate the intended meaning of the classes and properties described in an RDF vocabulary. Since RDF vocabularies are expressed as RDF graphs, vocabularies defined in other namespaces may be used to provide richer documentation. 22:37:29 <danbri> ]] 22:37:29 <bwm_> DanC: who would spend an hour a week 22:37:37 <bwm_> lots of people responded 22:37:52 <bwm_> danc call for folsk to approach him privately about chairing and editing 22:38:02 <bwm_> propose we suggest 22:38:05 <jhendler_> only an hour a week? Sounds liek vacation :-> 22:38:09 <bwm_> - section 4 is struck 22:38:24 <bwm_> - postpone rdfsm-assertion 22:38:39 <bwm_> - CG prioritise highly work on this issue 22:39:39 <bwm_> http://www.w3.org/2000/03/rdf-tracking/#rdfms-assertion 22:40:23 <bwm_> timbl: if its a tag issue, then timbl will ask the tag if they want to give it to us and they'll probably bite our arms off. 22:41:13 <RalphS> PatH: whatever the TAG says does matter here and we need to take it into account 22:41:49 <bwm_> add to proposal: coordinate with the TAG over URI denotation 22:41:53 <bwm_> danc proposes 22:41:55 <RalphS> the issue about adding word regarding URI denotation 22:41:57 <bwm_> timbl seconds 22:42:00 <bwm_> none apposed 22:42:04 <bwm_> no abstentions 22:42:08 <bwm_> carried unanimously 22:42:29 <bwm_> action jeremy to email to rdf comments 22:43:09 <bwm_> danc propose to break 22:43:22 <bwm_> em-lap: meeting is in riverfront room tomorrow 22:43:29 <bwm_> same floor just down the corridor 22:44:05 <bwm_> frankm: if you're driving: road conditions are hazardous - 100 care pile up, lots of accidents 22:44:15 <bwm_> meeting adjourned 22:45:04 <danbri> am wondering whether www-rdf-logic is a good forum to continue this discussion... 22:47:35 <ryanlee> ryanlee has left #swarch 22:53:51 <jhendler_> if anyone sees this - can you ask Bijan to call me or rejoin irc? thanks 22:54:10 <SethR> SethR has joined #swarch 23:15:41 <libby> libby has joined #swarch 23:31:19 <libby> ------query and rules bof notes ------ 23:31:47 <libby> http://www.w3.org/2001/11/13-RDF-Query-Rules/ 23:32:02 <libby> - eric introduces RDF query 23:32:06 <libby> ericP 23:32:47 <GuusS> GuusS has joined #swarch 23:33:51 <JosD___> JosD___ has joined #swarch 23:35:54 <libby> picture which was the result of the first BOF on query: http://swordfish.rdfweb.org/photos/2003/03/05/2003-03-05-Pages/Image4.html 23:36:19 <libby> benjamin asks q about relative expressivity of N3 and algae 23:36:41 <libby> algae doesnt have "for some", N3 does. 23:39:51 <libby> ericp talks about SQL backends for algae, including access to a 'normal' realtional database 23:41:14 <libby> [aside - Jan Grant is doing something similar - mapping RDf queries to arbitrary SQL databases] 23:43:35 <libby> [scribe not catching all the detail of the discussion, sorry] 23:53:54 <Nobu> Nobu has joined #swarch 23:55:58 <libby> the next IRc meeting http://lists.w3.org/Archives/Public/www-rdf-rules/2003Feb/0014.html about tests for RDf queries 23:56:23 <libby> Mike Dean talks about his hierarchy of RDF QLs 23:57:26 <libby> http://www.daml.org/2002/03/tutorial/slide122-0.html 23:57:34 <libby> talks about DQL: 23:57:57 <libby> http://www.daml.org/2002/08/dql/ 23:58:09 <libby> no official syntax yet, but a model 23:58:27 <libby> "must bind" "may bind" "don't bind" 23:58:30 <libby> for variables 23:58:57 <libby> for bnode processing - dont return the answer 23:59:35 <libby> also skolem functions

14:01:24 RRSAgent has joined #swarch 14:01:30 Zakim has joined #swarch 14:01:45 em-lap has changed the topic to: semweb arch tech plen meeting - 2002-03-06 14:01:51 DanC has joined #swarch 14:02:10 RRSAgent, pointer? 14:02:10 See http://www.w3.org/2003/03/06-swarch-irc#T14-02-10 14:03:11 hi 14:05:00 do the bots need/want op? 14:05:33 don't think so. 14:05:51 agenda + zakim how-to 14:06:06 Zakim, who's here? 14:06:07 sorry, DanC, I don't know what conference this is 14:06:09 On IRC I see DanC, Zakim, RRSAgent, em-lap, DaveB 14:06:16 Zakim, this is SWARCH 14:06:17 sorry, DanC, I do not see a conference named 'SWARCH' 14:06:22 Zakim, list 14:06:22 I see WS_XMLP(tp)8:00AM, Team_Ralph's(test) 14:07:37 bh has joined #swarch 14:08:01 pfps has joined #swarch 14:08:12 timbl__ has joined #swarch 14:08:22 Hi everyone 14:08:55 JosD__ has joined #swarch 14:09:05 bwm has joined #swarch 14:09:21 EricM convenes the meeting... 14:09:33 NOTICE: records of the meeting are public. 14:10:02 horrocks has joined #swarch 14:10:23 ericm: welcome 14:10:51 GuusS has joined #swarch 14:10:51 PStickler has joined #swarch 14:11:06 sandro has joined #swarch 14:11:12 mdean has joined #swarch 14:11:19 this is an experiment in bringing people with different backgrounds together 14:11:40 em presents http://www.w3.org/2003/03/06-swarch/ 14:12:08 em-lap: accomplished a lot in last two years 14:12:20 sandro has joined #swarch 14:12:24 conferences dedicated to sw 14:12:25 tools 14:12:40 em can't keep up with number of new tools 14:12:57 areas where less successful - areas that need more work 14:13:11 discussing these over next two days 14:13:20 sandro has joined #swarch 14:13:25 goal: identifying barriers to adoption 14:13:45 input to w3c on where to focus phase two of the activity 14:14:18 meeting nodes are public 14:14:57 em thanks danc for helping put agenda together 14:15:04 RRSAgent, pointer? 14:15:04 See http://www.w3.org/2003/03/06-swarch-irc#T14-15-04 14:15:24 em goes over agenda 14:16:29 http://www.w3.org/2001/sw/meeting/tech-200303/ 14:18:58 ?? announces bof - sorry forgot your name 14:19:22 em hands over to DanC 14:19:38 Benjamin Grosof announced the BOF this evening 14:19:43 (sp?) 14:20:17 danc: presenting on w3c 14:20:32 -> http://www.w3.org/2000/01/ala2349/all.htm 14:20:47 What is the topic of the BOF? SW Arch? 14:20:52 [Note that this logger, unlike #rdfig's logger, does NOT record /me actions.] 14:21:15 Surely the BOF is about RuleML. 14:22:44 dan talks about w3c organisation 14:23:12 danc talks about w3c history html, style sheets, ... 14:24:31 danc wonders how much people want to know about process 14:24:45 danc tells them anyway 14:26:52 http://www.w3.org/2001/02pd/rec54.svg I think 14:28:52 danc shows process diagram generated from rdf model of w3c process 14:28:57 http://www.w3.org/2001/02pd/rec54.png 14:30:17 ArtB has joined #swarch 14:30:34 timbl notes the graph is not to scale, when discussing participation 14:32:52 timbl observes that the process looks like the choreography of a bunch of services 14:33:11 DanC asks what people mean when they talk about sw services 14:33:24 ircleuser has joined #swarch 14:34:15 danc moves on to web architecture 14:34:49 http://www.w3.org/TR/webarch 14:35:46 oh he switched http:///www.w3.org/2001/tag/2002/11ac-tag-tb/slide3-0.html 14:36:26 a lot of what sw architecture is to take db and kr technologies and take names and turn them into uri's and see what happens 14:36:45 I'd call them Naming, Knowledge Representation, and Data Transfer. 14:37:05 sandro refers to the tripod of web architecture, I think 14:37:59 danc complains that user defined xml schema datatypes don't have a uri 14:38:08 Identified by URI or URIref? ;-) 14:39:11 path requests permission to beat on danc 14:39:28 PStickler, DanC, TimBL, and the TAG use the term "URI" the way the RDF specs use the word "URIRef". 14:39:30 jhendler_ has joined #swarch 14:40:04 Should the term URIref be deprecated and just use URI? 14:40:22 path says there is a difference between addressing (location) and naming (denotation) 14:40:48 sandro: notes that we are using the term URI to mean URI REF 14:41:05 pfps notes we don't own the terms URI and URIREF 14:41:28 tag is proposing people should use xml 14:41:42 running xslt on rdf hurts 14:42:18 jjc: we tend to use qnames as an abbreviation for URIREFS's which does not fit with xml usage 14:42:23 DanC: issue 8 on the tag list 14:42:41 now on slide 6 of 9 14:43:06 separate presentation from semantics 14:43:24 slide 8 14:43:58 another way to look at sw architecture is what happens when you take formal langauges and descriptions and connect that to the web 14:44:25 danc closes with demo 14:45:04 RalphS has joined #swarch 14:45:21 danc has 200+lines of python that takes any sql database to produce RDF and then styled client side 14:46:10 uses xslt script to format the rdf 14:46:27 roll call 14:46:39 brastow 14:46:45 boley 14:47:04 brickley 14:47:09 britton 14:47:16 carroll 14:47:17 chen 14:47:21 connolly 14:47:24 jos 14:47:24 dean 14:47:36 greeblaat 14:47:36 danbri has joined #swarch 14:47:39 sando 14:47:42 path 14:48:00 horrocks 14:48:22 jemio 14:48:51 macdougall 14:49:00 mcbride 14:49:05 miller 14:49:08 miller, libby 14:49:15 miller, eric 14:49:23 pfps 14:49:27 pattnaik 14:49:47 risteff 14:49:48 schmidt 14:49:54 not schmidt 14:50:08 schreiber 14:50:19 stickler 14:50:26 swick 14:50:34 volz 14:50:38 willard 14:50:39 williams 14:51:15 s/sando/sandro/ 14:51:35 grossof at bit 14:51:38 s/bit/mit/ 14:51:41 horan is here 14:52:02 benjamin grosof (bgrosof@mit) 14:52:04 http://ebusiness.mit.edu/bgrosof/ 14:52:31 timbl is here but was not on that list 14:52:31 timbl also present 14:52:42 Hendler monitoring IRC, apologies for not being able to be there in person. 14:52:49 TimBL takes stage 14:53:00 TimBL talking about http://www.w3.org/DesignIssues/diagrams/SemWave.svg 14:53:35 I'm surprised at how many people are absent. WIll they be here later, or is this a reflection of the difference between intent and reality. 14:54:22 some discussion of taking a break 14:54:26 Jim, this is really great! 14:54:32 html working planning to join us at 10.45 14:54:39 libby has joined #swarch 14:54:44 break till 1005 14:54:58 i.e. 15 minute break 14:55:34 I cited it; was good for waving arms about "the beach" saying where things are 15:00:43 shellac has joined #swarch 15:06:51 emmanuel has joined #swarch 15:07:25 AndyS has joined #swarch 15:12:01 lol 15:13:02 TimBL talking about http://www.w3.org/DesignIssues/diagrams/SemWave.svg 15:14:05 TimBL: Layer cake on left, from bottom: Markup, Data, Ontology, Rules, Proof (and using crypto, to bootstrap a secure system) 15:14:54 ...: Markup gave you Interop with an App (Application Domain) 15:15:17 ...: RDF gives you cross-app interop, so your bank statement works in your calendar. 15:16:41 ...: across top: Researchy, Common FOrmats, Web Standards, Wide Deployment 15:17:26 ..: (when you start writing lots of converters between the common formats, it's time to join a WG to establish a web-standard format.) 15:18:21 ...: webizing KIF is a pretty cool idea, etc 15:18:49 ...: THe "semantic web wave" is to make the whole stack reach wide deployment. 15:19:22 ...: THe front lines: wide deployment of RDF, Standardization of OWL, common formats for rules, research for proofs. 15:20:00 (that is, the backslash diagonal.) 15:20:46 JosD__ has joined #swarch 15:21:02 ...: It would be nice if there were already published rules in many domains. (Rules == Derivation Rules/Axioms, ...?) 15:21:25 ...: we'll need to index all the rules, pick them up on the fly, etc. 15:21:26 time zone data in RDF, FY: http://www.w3.org/2002/12/cal/#tzd 15:21:35 ...: But I'm probably preaching to the converted. 15:21:59 Jos: How does this connect to the semantic web bus? 15:22:47 TimBL: That's another picture. 15:22:59 ...: Layer cake uses Resistor Color Coding. 15:24:03 what's the URI for the SW bus slide? 15:24:15 ...: lower bus (data/ont) is wide interoperability; higher bus (rules/logic) is less so -- may not expect other engines to infer same things 15:24:46 ...: BUT please don't bring heurisitcs, etc, down to lower DATA bus. 15:24:53 ...: Just index the metadata. 15:25:21 Jeremy: How do you do signatures for RDF? 15:25:39 ...: (or do we need it?) 15:25:53 TimBL: Do we want a canonical RDF? 15:26:00 that's rdf graphs - not the serialization 15:26:04 Nobu has joined #swarch 15:26:08 sweb-bus is at http://www.w3.org/DesignIssues/diagrams/sweb-bus.png 15:26:37 Jeremy: We have a theoretical solution at HP on canonicalization; do we need it? 15:26:49 DanC cuts discussion off here. 15:27:14 =============== 15:27:30 List of RDF Vocabularies, Applications, .... 15:29:26 DanC uses: DC, FOAF, swap/contact, cyc travel, swap log, swap math, .... 15:29:31 adding from other people 15:29:39 ANnotea, EARL, 15:29:53 Test Vocabls for RDF 15:30:07 AgroVoc 15:30:11 RDFS 15:30:15 RSS 15:30:18 Wordnet 15:30:21 OWL 15:30:22 DAML 15:30:24 chaalsBOS has joined #swarch 15:30:31 DAML+OIL 15:30:54 Art and Architecture Thesaurus 15:31:17 MIT Process Handbook 15:31:25 & contracting ontology 15:31:33 (Ben Grosof) 15:31:54 Union List of Artists Names (Guus) 15:32:00 QFX (tim) 15:32:27 MathIntOnt (international mathemantical ontology) Harold 15:32:45 rdf:type (DanBri) 15:32:50 National Cancer Institute Vocabulary to be released soon in OWL Lite 15:32:52 dmoz 15:33:21 Jim these are vocabularies You Actually Use 15:33:30 Sun Product extension to DC 15:33:35 might this be useful? http://swordfish.rdfweb.org/discovery/2003/02/stats/ - stats from my crawler 15:33:39 Technology Area (also sun) 15:34:04 XSD 15:34:56 openCYC 15:35:06 =========== 15:35:08 DAML Services http://www.daml.org/services/ 15:35:21 DanBri -- Mozilla uses some vocabulary internally 15:35:39 DAML Time http://www.cs.rochester.edu/~ferguson/daml/ 15:36:01 DanBri: So how do we scale up this excersise to the web? 15:36:05 various datasets (instances) at http://www.daml.org/data/, particularly countries 15:36:32 bwm uses vcard and some home grown - actions, calendar, filesys, meeting 15:37:19 ...: RDF gets fun when it's all linked together and on the web. Crawling them is great! 15:37:40 ...: (showing Daniel Krech's stuff.) 15:37:44 ...: top of the pops 15:38:11 dunno if that one counts [orthopaedics leg length discrepancy measurement|http://www.agfa.com/w3c/2002/10/medicad/op/lldmP.n3] 15:38:12 ...: My home page uses lots of vocabs. 15:39:01 also rdf & owl testcase vocabulary 15:39:23 ...: All you need is rdfs:seeAlso 15:39:30 we use the computer science ontologies first developed for SHOE and then ported to DAML and now OWL a lot 15:39:37 ...: (ayf.pl) 15:40:47 a new "currency ontology" was just added to Daml library, we've started to use that 15:41:06 ...: So what to do with what you find? List of known ontologies, NOT based on any central registry. 15:41:07 any pointer Jim? 15:41:25 rdfschema.info 15:42:20 ...: ... of course there are plenty of RDF islands that wont be found by this crawler. 15:42:39 http://www.daml.ecs.soton.ac.uk/ont/currency.daml 15:42:45 ...: THIS IS NOT ALL THE RDF OUT THERE. THIS IS ONE ISLAND. 15:43:04 DanC: if you added DAML, this would bog down. 15:43:16 sanScribe: who was talking above, danbri? 15:43:25 Yes, DanBri. 15:43:52 Sun-person: How do you know what the meaning of these vocab terms is??? 15:44:15 DanC: Each of these is linked to a schema, PLUS you can see how it's used and try to figure it out backwards. 15:45:03 TimBL: If you use a vocab in the public, there's an expectation that you'll provide a schema answering people's questions about it. 15:45:32 (I note rss 1.0 breaks that - 404 on all vocab terms) 15:45:36 DanBri: I'm hopinh we can measure the adoption of language features this way. 15:46:17 ============== Waiting for HTML Delegation 15:46:48 10 or so HTML folks come in. 15:47:25 DanC: How many people have tried to put RDF in HTML and found it hurts? [ a few hands ] 15:47:30 I briefed danbri on stuff rdf/html for me 15:48:06 DanC: Some people are paying attention to RDF inside comments inside HTML! 15:48:19 DanC: yes, movable type, CC 15:48:25 Actually only 3 HTML folk come in -- others are getting coffee. 15:48:33 http://www.w3.org/2003/03/06-xhtmlrdf 15:49:41 Ralph: Thanks to Steven and HTML folks for coming over for this. 15:49:45 mdubinko has joined #swarch 15:50:17 Ralph: The situation is hurting us all. It hurts me personally..... 15:50:20 I'm okay, DanC. 15:50:24 thx. 15:50:41 Ralph: At DC meetings year ago this came up, and we still haven't solved it. 15:50:58 hmm, above page links neither to the relevant TAG issue, or rdf core LC issue 15:51:26 ======= Introductory remarlks from folks ---------- 15:51:32 or what I actually wrote in the rdf/xml spec 15:51:36 Steven Pemberton: Our prefered requirements 15:51:38 daveb, can you post those urls pls? 15:51:56 ...: HTML needs to be DTD validatable. (A consensus position in our group.) 15:51:56 http://www.w3.org/2003/03/06-xhtmlrdf.html 15:52:07 I disagree that XHTML needs to be DTD-validatable. 15:52:33 ...: A possible representation of RDF .... 15:52:33 RDF in HTML RDFCore last call issue: http://www.w3.org/2001/sw/RDFCore/20030123-issues/#hodder-01 15:52:41 (for tracking) 15:52:47 ...: RDF/XML Syntax 15:53:54 ...: 15:53:54 ...:etc an RDF syntax which is HTML/Meta friendly. 15:53:56 TAG issue: RDF in XHTML-35 http://www.w3.org/2001/tag/ilist#RDFinXHTML-35 15:54:06 bwm has joined #swarch 15:54:15 Brian: Is it a goal to represent an arbitrary RDF graph, or only some of RDF. 15:55:24 Steven: All of RDF. 15:55:44 Patrick_S has joined #swarch 15:55:51 aside: are there other XML languages that share RDF's unpredictability, ie. use of words from content domain ('Person' etc) as XML element names? 15:55:51 RalphS has joined #swarch 15:56:15 bwm_ has joined #swarch 15:56:25 shinichi_ has joined #swarch 15:56:42 in the TAG, a relevant issue is assigned to me: RDFinXHTML-35 : Syntax and semantics for embedding RDF in XHTML http://www.w3.org/2001/tag/ilist#RDFinXHTML-35 15:57:03 all of rdf would be very tricky - xml literals, datatypes, bnodes ... 15:58:26 a previous similar design (1995, timbl, raggett): http://www.w3.org/MarkUp/Resource/Specification 15:58:36 somebody scribing? 15:58:40 "There are two specs, XSLT and Schema, which do not agree on the meaning of QNames in attribute values. " <---- hmm. 15:58:47 Nobu has joined #swarch 15:58:53 on behalf of Jim Gettys, I note that putting metadata at the top puts it in the critical path re user-perceived latency. consider allowing it at the end too 15:59:03 em asks about embedding svg in html and how that relates 15:59:07 sanScribe has joined #swarch 15:59:17 jjc: two considerations 15:59:23 Tantek has joined #swarch 15:59:33 jjc:whatever is done for xhtml should be reusable elsewhere 15:59:38 se lost our scribe due to flood rules or some such. ouch. 15:59:43 s/se/we/ 15:59:53 brian is scribing for now 16:00:06 thx 16:00:33 links from rdf meta to the html 16:01:01 sp: good approach - use markup and say that it is also rdf information 16:01:04 jeremy offered to cite work by van Harmelen et. al on the sort of linking he mentioned. 16:01:07 sandro has joined #swarch 16:01:26 timbl__: concerned that is fancy and exciting but will increase complexity 16:01:41 jjc: goal is to improve consistency by writing stuff once. 16:02:11 sandro: if we embed rdf in html what are the semantics of frag ids 16:02:20 sp:defined by the mime type 16:02:39 em-lap: many communities are grappling with how to embed meta data into html 16:02:43 hmm... xpath-based "macros" in RDF would be handy for going from dc:title to html here; the idea came up in DAML+OIL design discussions too. 16:02:56 <reagleMIT> reagleMIT has joined #swarch 16:03:21 <sandro> em: Mostly I was going to say why this is important, but lets get on with how to solve it. 16:03:35 <reagleMIT> what's the content model of xhtml:meta ? 16:03:42 <sandro> Ralph: Let's stipulate there is a need to put RDF in HTML documents. Any debate? 16:03:43 <reagleMIT> can it take element content? 16:03:46 <reagleMIT> if so, why not? 16:03:55 <DaveB> reagleMIT: currently an empty eleemnt IIRC 16:04:16 <sandro> Mike Dean: It's still valuable, but it's easy for outsiders/newcomers to over estimate the value. 16:04:18 <chaalsBOS> reagleMIT: Was asked and stephen said that would break DTD validity, which is a requirement of the HTML group 16:04:39 <sandro> Steven: I think you'll get far greater use of RDF if it's exposed to the public in HTML. 16:04:47 <danbri> (danbri claps) 16:04:53 <reagleMIT> if Steven has to make a change, is there a DTD featuring corresponding to xsd:any ? 16:05:05 <sandro> em: That's my experience as well. It's required by law in 6 countries! 16:05:07 <reagleMIT> ah, thanks charles 16:05:31 <mdubinko> reagleMIT, no, no such thing in DTD-land 16:05:35 <sandro> TimBL: If it's in HTML meta tag, you wont be able to cut and paste this stuff. 16:05:58 <DanC> yes, people *ARE* getting RDF/XML from XHTML via XSLT. 16:06:24 <sandro> Ralph: As soon as your open up the question of radical changes to syntax, ... you're outside the scope of any WG. 16:06:45 <sandro> JJC: If there's a good RDF serialization for HTML Heads, it shiould be the primary syntax. 16:06:54 <sandro> DanBri: WHat about N-triples? 16:07:04 <Steven> Steven has joined #swarch 16:07:21 <DanC> yes, people *ARE* getting RDF/XML from XHTML via XSLT. http://www.w3.org/2000/06/dc-extract/form 16:07:24 <danbri> ...ntriples as groundwork for potentially doing an n-triples in angle brackets 16:07:49 <sandro> Brian: Is there any support from eg Adobe. (HTML authoriing tools vendors) 16:07:57 <Steven> My example is available at http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2003Feb/0103.html 16:08:14 <DaveB> my (+Arts) ntriples syntax (as a subset of n3 + stuff) is essentially the rdf graph concretized 16:08:18 <sandro> DanC: Back when I was chair of HTML WG, they surel;y wanted to know how to put the document author in the HTML metadata. 16:08:34 <sandro> Thanks, Steven. 16:08:46 <DanC> author name 16:09:17 <sandro> heh 16:09:17 <danbri> (side talk: danbri asks Steven if DTD validation important for HTML because of entities; Steven replies: 'several issues') 16:09:48 <sandro> Ralph: How is DTD for MathML solved....? THeir slide had no doctype? html+math 16:09:56 <sandro> Steven: html+math 16:10:12 <sandro> ...: We have a profile for html+math+svg 16:10:26 <sandro> ...: We had to rename entities before we could combine them. 16:10:34 <sandro> DanC: No doctype? 16:10:46 <GuusS> GuusS has joined #swarch 16:11:06 <sandro> Steven: There is a doctype coming. People will have to use the HTML+Math+SVG doctype. 16:11:53 <sandro> DanC: I do not agree there a requirement for entities. 16:12:37 <sandro> Steven: MathML is geared around entities. 16:12:50 <sandro> DanC: But MathML would be fine without entities. 16:13:01 <sandro> ..: It would be a smaller change. Just use UTF-8. 16:13:08 <DanC> or <mchar> 16:13:20 <sandro> TimBL: Could we put a wrapper in? SOmething with "any" as the content model. 16:13:33 <Stuart> Stuart has joined #swarch 16:14:10 <sandro> Steven: You can say "any" and it means "anything defined in the DTD". In Schema there is a more broad "any" that means "any well-formed XML". 16:14:39 <sandro> ...: It's a load of work (amazing we did it) to do namespaces in DTDs. 16:15:00 <sandro> ...: If we had entity-functionaly elsewhere, we would no longer need DTDs. 16:15:10 <sandro> ...: (IE we could switch to Schema.) 16:15:50 <sandro> Daniel Austin: one approach is to use Schema + DOCType *onyl* for entities, not for validation. 16:16:10 <sandro> danbri: How responsive is XML Schema WG on this? 16:16:17 <sandro> Steven: I'm not optimistic. 16:16:31 <sandro> DanC: It has to be fixed in the XML, not XML Schema. 16:16:37 <JosD___> JosD___ has joined #swarch 16:17:02 <sandro> DA: My impression in XML Core WG is planning to take up entities. 16:17:33 <sandro> TimBL: WHat would HTML WG say about that half-way route? (DA's above) 16:17:45 <sandro> Steven: I don't know. It might be possible to be accepted. 16:17:58 <sandro> Ralph: Can you please explore that question more? 16:18:00 <sandro> Steven: Yes. 16:18:30 <sandro> =======Mimasa takes podium 16:18:31 <DaveB> what was the half-way route? 16:19:04 <sandro> /DTD XHTML 1.1 plus MAthML 2.0 plus SVCS 1.1//EN 16:19:20 <sandro> M: I don't like this, but it's what we have to do at the moment. 16:20:16 <Zakim> Zakim has left #swarch 16:21:05 <sandro> jjc: Do you have to use the default namespace? 16:21:24 <sandro> [ there's some magic to getting namespaces to work in DTDs ] 16:22:12 <sandro> Steven: In DTD you have to say which prefixes you're going to use. 16:22:52 <sandro> em: This is a good example of embedding SVG in HTML, but SVG has a metadata tag, ... but that's lost here? 16:23:10 <sandro> M,SP: Yes. It's lost. You can't do wide-open "any" in DTDs. 16:23:18 <DanC> mimasa sorta answered jjc's questions about namespaces+DTD by showing parts of http://www.w3.org/TR/XHTMLplusMathMLplusSVG/ 16:23:28 <sandro> Ralph: So this is not the full SVG. 16:23:33 <sandro> Steven: RIght. 16:23:56 <sandro> Dan Austin: Schema has several levels of "any". 16:23:57 <danbri> steven: SVG DTD's notion of ANY isn't that of SVG XML Schema, so SVG DTD is partial representation 16:24:43 <danbri> (meeting process questions) 16:24:48 <danbri> steven: this is brainstorming really 16:25:36 <DanC> sandro: how about [basically, treating RDF as foreign string data, ala script] 16:25:48 <DanC> rrs: not just now... 16:26:43 <danbri> jjc: process point; steven's msg was Member-only, this is a public visible meeting 16:26:47 <sandro> Steven: So my proposal is something HTML authors will easily understand and be able to use. 16:26:49 <danbri> steven: I have no objection 16:26:54 <danbri> ...to it being public 16:27:16 <danbri> danc: re steven's proposal, we need to distinguish between uris, and things that look like uris that are strings 16:27:19 <sandro> DanC: How will I know if the homepage was meant as a string or a URI? 16:27:29 <sandro> Steven: uh, yeah. 16:27:48 <sandro> Dan Austin: Why not just put the RDF in a string? 16:28:05 <DaveB> ah, the <script> approach? 16:28:07 <sandro> DanC: yes -- like <script> 16:28:33 <sandro> DanC: Then the tools -- eg SAX processors -- wont get it. 16:29:40 <danbri> jjc: nodeID, datatype literals, uri-vs-string 16:29:47 <danbri> ...missing, would need adding 16:30:44 <sandro> DanBri: How about we make a regular syntax for RDF? 16:31:27 <sandro> DanC: this is implicit parsetype=Resource.......... 16:32:08 <sandro> jjc: this syntax would support a QNAMEs in string 16:32:50 <sandro> jjc: nameQname, nameURI, etc. needed. 16:32:58 <sandro> ...: 6 o r7 16:33:05 <sandro> TImBL: probably NOT a prime number. :-) 16:33:38 <sandro> Ralph's amaya drops into the debugger. 16:33:45 <DanC> [rrs's Amaya is running from a debugger... nifty ;-] 16:34:20 <sandro> DanBri: THere are lots of possible RDF sytnax. This is not far from Query/RUles syntaxes, which need to flag their variables. 16:34:50 <sandro> TimBL: Conversely, a lot of the R.*L languages are not XML and wouold embed in HTML very easily. 16:34:52 <eikeon> eikeon has joined #swarch 16:34:56 <DanC> (it seems pretty far from a query/rules/formula syntax, to me) 16:35:03 <em-lap> RDF/XHTML 16:35:18 <sandro> ...: Or if you go XML, you could sort of reify everything, like in RUleML;'s RDF syntax. 16:35:41 <sandro> JJC: XML Literals are also a problem with these syntaxes. 16:35:59 <JosD___> (quite far froof proof as well) 16:36:01 <sandro> ...: I use them a lot of puting HTML inside RDF 16:36:15 <JosD___> s/froof/proof 16:36:20 <sandro> ...: That wont fit in DTD world. 16:36:22 <Steven> Steven has joined #swarch 16:37:31 <danbri> (thinking about this more, steven's proposal is uncannily close to recent rulemarkup ideas... --> fwd reference to Harold's presentation in a later session) 16:38:08 <sandro> sandro: We either need: get rid of DTDs, use literal, make new RDF syntax. 16:38:55 <sandro> TimBL: we're turning this all backwords, breaking our document format, because our tools (DTDs) are broken. 16:39:11 <sandro> jjc: long URIs 16:39:17 <sandro> mike dean: use entities! :-) 16:39:43 <danbri> I think two paths: migrate from DTDs so we can use XML Schema 'ANY'; and also do an ntriples-in-xml (done w/ rules in mind). 16:39:59 <sandro> DanC: Surpreme Irony. XML folks keep wanting this kind of thing. 16:40:14 <sandro> TimBL: I have an element called element and and attribute called attribute. 16:40:22 <danbri> hmm 16:40:23 <danbri> [[ 16:40:24 <danbri> <resource href="mydoc"> 16:40:24 <danbri> <link rel="content-language-specific" href="mydoc.eng"> 16:40:24 <danbri> <link rel="content-language-specific" href="mydoc.fr"> 16:40:24 <danbri> </resource> 16:40:26 <danbri> ]] 16:40:30 <sandro> Dan Austin: What about embedding the HTML inside the RDF. 16:40:31 <danbri> -- http://www.w3.org/MarkUp/Resource/Specification 16:40:42 <sandro> TimBL: It would work quite well, but might be insentivive. 16:41:15 <sandro> DanC: nealry all browsers would work fine. 16:41:22 <mdubinko> would the documents be delivered with a media type of HTML or RDF? 16:41:22 <sandro> TImBL: It's valid RDF! 16:41:48 <sandro> DanC: but what media type...... 16:41:55 <sandro> Stickler: rdf+html ? 16:42:28 <sandro> Chaals: Sidesteps DTD validations by not doing it. 16:43:15 <sandro> DanC: you could put entities in the RDF document's internal subset. 16:43:35 <sandro> Brian: People want to be able to syn-validate RDF, as like DTDs. 16:43:42 <DanC> (bwm, do you tell them you *can* do XML schema validation of RDF?) 16:44:10 <sandro> JJC: (misssed) 16:44:14 <DaveB> you can XSD-validate, to a certain degree, fixed rdf/xml profiles 16:44:41 <bwm_> bwm agrees with DaveB 16:45:32 <danbri> yup, to a modest agree 16:45:37 <DanC> JJC: name="dc:title" might be DTD-ok, but no good cuz dc: isn't bound. 16:45:50 <danbri> I want to hear more about why folk want DTD validation, apart from entities. 16:46:48 <DanC> I hear some folks rely on DTD-happy systems, and alternatives aren't sufficienty mature. 16:46:59 <Steven> dc: would be bound in the document 16:47:33 <Steven> (See remark dc: isn't bound abouve; not = Dan Connolly :-) 16:48:08 <sandro> Ralph: DTDs and Open Vocabularies seem to clash -- we haven't figured out how to put them together. 16:48:43 <sandro> TimBL: Would Schema solve this? 16:48:58 <sandro> DanC: It's not possible to write one-last-true-schema for RDF/XML. 16:49:27 <sandro> ...: For each property you have to say whether you're going to use it as an element or an attribute. Or at least it's very very hard. 16:49:57 <DanC> pf of "not possible to write one-last-true-schema for RDF/XML": consider rdf:_1, rdf:_2.... QED. 16:49:59 <sandro> Mimasa: If we forget about DTD and schema. Modular Namespaces.... 16:50:11 <sandro> ....: In RelaxNG 16:50:39 <em-lap> http://www.taiopensource.com/relaxng/mns.html#example 16:50:52 <em-lap> http://www.thaiopensource.com/relaxng/mns.html#example 16:51:20 <sandro> Steven: Relax took modularization to heart. 16:52:42 <sandro> Joseph: I consider Schema's "any" feature dangerous, because of its need for deterministic parsing, when you extend (eg merging in another namespace) the schema, it breaks. 16:52:50 <sandro> ...: Relax does much better here. 16:53:04 <bwm_> bwm claims prize for embedding html in RDF and displaying in browser 16:53:29 <Steven> demo! 16:53:40 <sandro> Ralph/DanC: "equivalence class" in XML Schema. No "any element can appear here", but "any element whose type is X can appear here." 16:54:00 <DaveB> surely some proposed rddl solutions are html in rdf 16:54:04 <DanC> where X is, e.g., rdf property element 16:54:20 <sandro> scribe sees bwm's document. 16:55:06 <sandro> TimBL: first do a pass, for every namespace, of gathering data and making a DTD. 16:55:20 <sandro> Dan Austin: that would only validate local chunks, not the document as a whole. 16:55:32 <em-lap> bwm, do you want to show this? 16:55:43 <sandro> DanBri: I want to hear more about why people want to stick with DTDs. 16:55:53 <sandro> Steven: Existing tools..... 16:56:11 <sandro> em-lap, bwm is trying to send it to an archive. 16:57:26 <sandro> Steven: time check.....? 16:57:35 <sandro> em: Next steps? 16:58:07 <shinichi_> shinichi_ has left #swarch 16:58:35 <sandro> NOT-REAL ACTION: Steven push back on alternatives to DTDs 16:58:56 <sandro> Steven: I'm not optimisitics 16:59:15 <sandro> TimBL: Until then, we'll just put our HTML inside RDF. :-) 16:59:33 <sandro> Ralph: What about this meta/meta proposal? 16:59:38 <DaveB> can't say I'm happy with that, as rdf/xml spec editor 16:59:47 <DaveB> - the html in RDF/XML that is 16:59:53 <DaveB> DanC: yes 17:00:08 <sandro> DaveB -- it's as XML literlas, I think. 17:00:27 <DaveB> sandro: ok 17:00:46 <DanC> DanC: DaveB is willing to help the HTML WG with the <meta name="dc:title"> idea 17:00:51 <sandro> Several people express interest in helping HTML WG make a better meta syntax. 17:01:20 <DaveB> groan 17:01:30 <DanC> groan? did I misunderstand? 17:01:31 <DaveB> helping? :) 17:01:48 <DaveB> I'd prefer a few people helping ;) 17:02:00 <sandro> em makes a strong pitch for solving this problem together 17:02:13 <sandro> TimBL: If you do meta/meta, you have to show at least how it maps to RDF. 17:02:55 <sandro> Steven: Even now you see dc.title...... Thjere';s presure here. 17:03:01 <danbri> for record, i volunteer to review proposals, and have a specific interest in commonalities between new rdf dump syntax and work on rule/query markup 17:03:01 <sandro> em: that's mine and I hate it! 17:03:10 <DaveB> ha ha 17:03:27 <sandro> Jos: If you can't do it by value, due it by reference 17:03:28 <danbri> timbl: if you don't say it is rdf, you create a whole new space of html meta property uris 17:03:28 <DaveB> +danbri? 17:03:47 <sandro> TimBL: href="data:application/rdf+xml; ...." (sandro's solution) 17:04:18 <DanC> DanC: I may have misrepresented DaveB's offer 17:04:30 <sandro> but today's talk is about embedding. 17:04:46 <sandro> Dan Austin: Can you put it at the end? 17:04:49 <DanC> DanC: ... so I'm not quite sure what he's offering. 17:04:57 <DaveB> DanC: I'm offering to work with html wg 17:05:09 <sandro> TimBL: You may need to keep it in the <head> to get nice behavior from old browsers. 17:05:31 <sandro> em: HTTP HEAD. 17:05:44 <sandro> [ do that ever read the HTTP-EQUIV meta...?????????? wonders sandro ] 17:06:14 <sandro> DanBri: Markup in body, use case: [didnt get it] 17:06:31 <DaveB> DanC: I was just cautious about having lots of helpers. I claim I'm an expert on RDF and XML, I'd want helpers who were too (+HTML) 17:06:44 <sandro> ADJOUNRED . RETURN AT 1:25. 17:08:01 <sandro> ========================= 17:08:01 <sandro> (for presentations on semantic layering) 17:08:18 <PStickler> PStickler has joined #swarch 17:08:46 <mdean> mdean has joined #swarch 17:08:57 <danbri> danbri has joined #swarch 17:11:06 <micah> micah has joined #swarch 17:12:21 <bwm> bwm has joined #swarch 17:12:26 <bwm> http://www.w3.org/2001/sw/RDFCore/20030305-swarchmeeting/experiment/World%20Wide%20Web%20Consortium.html 17:12:33 <sandro> EDIT ME 17:18:37 <mdubinko> mdubinko has joined #swarch 17:19:07 <danbri> sandro.edit() 18:04:18 <DaveB> DaveB has joined #swarch 18:22:50 <bwm> bwm has joined #swarch 18:25:17 <libby> libby has joined #swarch 18:26:32 <bwm> bwm has joined #swarch 18:27:27 <libby> "This is a little fancy that lets other users (on the same subnet) find out about you" 18:27:34 <libby> (tiny download) 18:27:53 <libby> assuming you want to be found out about of course 18:29:38 <RalphS> RalphS has joined #swarch 18:29:42 <mdean> mdean has joined #swarch 18:30:25 <simonSNST> simonSNST has joined #swarch 18:32:05 <lasf2f> lasf2f has joined #swarch 18:33:10 <ircleuser> ircleuser has joined #swarch 18:34:03 <bwm> pat begins 18:34:12 <DanC> -------------------------- 18:34:13 <bwm> appologies for unreadability of slides 18:34:22 <DanC> === 13:30 - 15:30 First afternoon session 18:34:22 <DanC> Semantic Layering. presentation by Peter F. Patel-Schneider, Pat Hayes 18:34:24 <danbri> (just found http://www.xml.com/pub/a/2003/03/05/social.html -- kendall clark on social meaning) 18:34:51 <bwmscribe> pat adjusts magnification of the slides 18:34:54 <DanC> note to self: get pat's jpg file into http space 18:35:05 <bwmscribe> no url to refer to 18:35:18 <bwmscribe> a picture of an rdf universe 18:35:30 <JosD___> JosD___ has joined #swarch 18:35:35 <bwmscribe> no constraints on anything - classes can contain themselves etc 18:35:43 <bwmscribe> owl has a very different picture 18:36:02 <bwmscribe> owl is neater 18:36:40 <bwmscribe> an owl dl universe is neat; segregated; separation between classes and individuals etc 18:36:52 <bwmscribe> classifcal fol way of building an interpretation 18:37:11 <bwmscribe> pat has fun zooming slide 18:37:30 <bwmscribe> owl full says scruffy is the way to go 18:37:49 <bwmscribe> owl classes as rdf classes etc 18:37:58 <bwmscribe> just use the owl vocabulary freely in rdf 18:38:16 <las> las has joined #swarch 18:38:27 <Tantek> Tantek has joined #swarch 18:38:29 <bwmscribe> a different way to do this is to retain the neatness of the owl picture 18:38:59 <bwmscribe> s/owl/owl dl/ 18:39:17 <bwmscribe> some classes are owl dl classes, similarly for properties etc 18:40:26 <bwmscribe> owl lets you reason about the number of things in a set 18:41:20 <bwmscribe> the process of embedding the more complex owl syntax into triples introduces things into the owl universe, e.g. lists, which don't bother RDF but do bother owl dl. 18:41:52 <bwmscribe> jos asks what the compliment of owl:thing would be in the diagram 18:42:15 <bwmscribe> there is a class called owl:nothing which cannot contain owl:things but can contain other things. 18:42:30 <bwmscribe> pfps: owl:nothing is the empty set 18:43:22 <chaalsBOS> chaalsBOS has joined #swarch 18:43:57 <bwmscribe> if you stick to the appropriate subset of rdf you can be neat and work in owl dl 18:44:17 <bwmscribe> within that subset the semantics of owl dl and owl full are equivalent 18:44:36 <bwmscribe> problems is its hard to determine if you are in this subset. 18:45:05 <bwmscribe> folks think you just check the syntax, but that isn't good enough because entailments can take you out of the subset. 18:45:53 <bwmscribe> why bother? 18:46:13 <bwmscribe> reasoning about owl dl is a more comfortable place to be 18:46:34 <bwmscribe> neatness guaratees powerful efficient reasoning engines 18:47:07 <bwmscribe> pat didn't expect the equivalence to work - and its not too outragously complicated 18:47:30 <bwmscribe> pat shows rdf, rdfs and rdf-dt in lbase 18:47:53 <bwmscribe> and owl as well, I think - not sure 18:47:59 <DanC> ACTION: PatH send 'OWL, RDFS, RDF in lbase' to connolly, em, www-archive 18:48:23 <Nobu> Nobu has joined #swarch 18:48:40 <bwmscribe> danbri: would this be easier in owl abstract syntax rather than rdf representation 18:48:55 <bwmscribe> patH: yes it would be a standard translation of a DL into fols 18:48:58 <bwmscribe> s/fols/fol/ 18:49:32 <bwmscribe> PatH hands over pfps 18:49:48 <DaveB> but OWL's abstract syntax is triples 18:50:45 <GuusS> GuusS has joined #swarch 18:50:57 <JosD___> Brian, it's called LBASE-new.html 18:51:36 <sanScribe> And now PFPS with the glass half-empty view of layering...... 18:52:04 <sanScribe> pfps: (there's ONLY ONE DROP of beer in the glass.) 18:52:39 <sanScribe> ...: We did the impossible. Why did we have to do research here? 18:52:57 <danbri> dancon: because they wouldn't let me close the WG 18:53:23 <jhendler_> we had to do research because the neats couldn't live with the scruffiness of Owl Full 18:53:24 <sanScribe> ...: Logic layering (proposition, DL, Prop-modal, FOL, HOL) is old. 18:53:36 <sanScribe> ...: (many branches, some dont layer) 18:53:50 <sanScribe> ...: (bump around modal) 18:53:56 <danbri> p: "modal sorta sits off to the side" 18:54:45 <sanScribe> ...: Different syntax for logics, compatible semantics. 18:55:00 <sanScribe> ...: eg FOL does *not* allow p ^ q 18:55:02 <bwmscribe> pats lbase doc: http://lists.w3.org/Archives/Public/www-archive/2003Mar/att-0018/LBASE-new.html 18:55:13 <sanScribe> ...: I can write a parse for any of these (in a day or so). 18:55:37 <sanScribe> ...: But in SemWeb, we need to use the same syntax?! 18:56:30 <em-lap> q? 18:56:37 <Zakim> Zakim has joined #swarch 18:56:46 <sanScribe> ...: Traditionally, compatible is just fine. Same-syntax is not needed. 18:57:15 <sanScribe> ...: Of course there are lots of other langs, like Montague, that are the result of a philosopher getting too much rope and hanging himself. 18:58:09 <sanScribe> ...: Montague wanted to write the ultimate logic. It's the spruce goose. :-) 18:58:44 <sanScribe> ...: SemWeb side is like working in a straightjacket. 18:58:51 <sanScribe> [ This is all slide 2 ] 18:59:13 <bwmscribe> bwm wonders why these syntactic constraints were imposed 18:59:44 <sanScribe> http://www-db-out.bell-labs.com/user/pfps/talks/semantic-layering/slide2-0.html 19:00:37 <sanScribe> DanC: WebOnt decided on the straightjacket, biased by the charter. 19:01:00 <DanC> JJC actually said that, but I agree. 19:01:07 <sanScribe> pfps: Allowing extensions to the syntax would have gotten us out of straightjacket. 19:01:21 <sanScribe> timbl: all the rules languages extend syntax. 19:01:46 <sanScribe> pfpfs: We were able to do this because OWL doesnt need variables. Ohhh, can we do FOL without variables? 19:01:52 <sanScribe> ph: don't go there! 19:02:29 <danbri> I really don't think 'every layer (above XML) must use RDF triples' is accepted even amongst RDF-enthusiasts. 19:02:37 <sanScribe> jjc: I challenge that semantic compatibile is rosy. They took a 100 years, we took one. 19:03:28 <DanC> I expect RDF syntax to grow forAll/variables. i.e. not just triples-as-are... but I didn't think the ontology layer was the layer in which to do it. 19:03:54 <sanScribe> Ian: RDF is not such a simple, well-understood layer as Proposition Logic. 19:05:05 <sanScribe> pfps: 1. syntax of triples is so impoverished (a,b,c) and each term has to denote, all in a set. 19:05:22 <sanScribe> .... (4 problems with RDF) 19:05:51 <sanScribe> ...: * every triple denotes, it has its own little say. 19:06:16 <sanScribe> ...: (see Dark Triples.) 19:06:36 <RalphS> PatH: we got along fine with this restricted syntax in lisp 19:07:06 <sanScribe> ...: Rephrase: 1a Every Triple has it say, 1b Every term has to denote. 19:07:39 <em-lap> q? 19:07:52 <sanScribe> ...: Literals denote kind of like a built-in functional term. 19:08:07 <em-lap> ack danc 19:08:08 <Zakim> DanC, you wanted to say that syntactic interoperability is a constraint derived from the principle of partial understanding 19:08:22 <sanScribe> ...: 1. Syntax is confiding, 2 semantics of terms in confing, 3. semantics of triples-asserted is confining. 19:08:32 <sanScribe> DanC: Lack of n-ary in RDF hurts a lot. 19:08:54 <sanScribe> DanC: Shared Syntax comes from a need for partial understanding. 19:09:22 <sanScribe> pfps: I don't believe that fixing the syntax helps partial understanding. 19:10:18 <sanScribe> ..: Because then you have to encoded other things (eg disjunctions) in a way which is inscrutible. 19:10:47 <sanScribe> DanC: We didnt need universals for OWL. When we need them, I expect we'll extend the syntax. 19:11:09 <sanScribe> jjc: own has disjunctions 19:11:25 <sanScribe> pfps: it has a certain kind of disjunctive expression 19:12:13 <timbl> timbl has joined #swarch 19:12:14 <sanScribe> slide 5. http://www-db-out.bell-labs.com/user/pfps/talks/semantic-layering/slide5-0.html 19:12:59 <sanScribe> http://www-db-out.bell-labs.com/user/pfps/talks/semantic-layering/slide6-0.html 19:14:12 <RalphS> pfps: the "myFavoriteMartian" property [might] contribute something to the meaning of the restriction but you don't know what 19:14:14 <sanScribe> skipping up to slide 11 19:14:27 <RalphS> ... but you can't ignore it 19:14:30 <sanScribe> http://www-db-out.bell-labs.com/user/pfps/talks/semantic-layering/slide11-0.html 19:16:16 <las> las has joined #swarch 19:16:27 <sanScribe> pfps: You need comprehension principles; from nothing you can infer the existence of every list.. 19:16:27 <sanScribe> DanC: section 2 of kiff says all lists exist. This is not odd. 19:16:34 <RalphS> RalphS has joined #swarch 19:16:51 <sanScribe> ph: I don't mind comprehension priniciples! 19:17:14 <sanScribe> pfps: people often hang themselves with these. 19:17:51 <simon> simon has joined #swarch 19:17:53 <bwmscribe> bwmscribe has joined #swarch 19:18:05 <sanScribe> : Because everything is at the top level, ... you can have loops, liar paradox 19:18:40 <sanScribe> ...: comprehension principles might encompass liar paradox; trivial theory 19:18:47 <sanScribe> ...: bad 19:18:59 <sanScribe> http://www-db-out.bell-labs.com/user/pfps/talks/semantic-layering/slide12-0.html 19:19:19 <sanScribe> ..: So we'll violate RDF and say *ONLY* the tree-structured statements exist. 19:19:43 <sanScribe> ...: So some inferences you might think exist do not. 19:20:04 <Nobu> Nobu has joined #swarch 19:21:07 <Nobu> Nobu has joined #swarch 19:21:11 <sanScribe> jjc: you can wait until someone asks.... 19:21:19 <sanScribe> pfps: thank you Socrates 19:21:40 <sanScribe> jjc: infinite stuff makes some implementation approaches impossible. 19:22:02 <sanScribe> jjc: you can't list all the conclusions of an owl KB 19:22:06 <sanScribe> ian: no, of course not. 19:22:21 <sanScribe> pfps: concluding with slide 8. my proposal..... 19:22:31 <sanScribe> ...: weakest version 19:22:42 <sanScribe> ...: You can change the semantics as long as it works out right in the end. 19:22:55 <sanScribe> ...: That's wrong for the SemWeb. 19:23:03 <sanScribe> (slide 9) 19:23:18 <sanScribe> ...: We want the language to be in the superset relation, as in FOL and HOL. 19:24:18 <sanScribe> ...: Weak semantic corresopnence. [Unclear]. As in modal and DL, which act the same even when they look the same inside. 19:24:30 <sanScribe> PFPS: we should break straightjacket by 19:24:36 <sanScribe> ...: 1) allow syntactic extensions 19:24:46 <sanScribe> ...: 2) allow expansions of semantics. 19:24:50 <ryanlee> ryanlee has joined #swarch 19:25:43 <sanScribe> Brian: RDF is for stating simple facts. We also need a language for describing types and classes. WHy use the same? 19:26:07 <sanScribe> pfps: There is a strange attractiveness to eating your own dogfood. 19:26:17 <sanScribe> Brian: I make frames for pictures, but I don't eat them. 19:26:41 <sanScribe> DanC: why is less-than-is-transitive anything other than a simple fact. 19:26:54 <sanScribe> Ian: For 100 years people figured out it was a bad idea! 19:27:02 <sanScribe> PH: No no, it's fine. We got it to work! 19:27:15 <sanScribe> PH: Even God has had model theories! 19:27:30 <DanC> "... even God has to have model theories ..." -- PatH 19:27:37 <sanScribe> PFPS: Given you had to do the stuff on the side, what's the benefit of having the syntaxes the same. 19:28:23 <sanScribe> PH wearing JH's had: Pragmatic Utility to passing complex RDF through simple RDF parsers. It's not a miracle, but you can get down a pipe. 19:28:35 <sanScribe> pfps: Let's make the pipe bigger. 19:29:09 <sanScribe> Lynn: You don';t get to make the pipe -- you need to pick one existing pipe. 19:29:15 <sanScribe> pfps: My pipe is a string of bits. 19:29:31 <sanScribe> Lynn: there are some pipes out there which we want to use. 19:29:45 <sanScribe> PFPS: I'd be relatively happy picking XML as my pipe. 19:30:38 <em-lap> q+ ian 19:30:42 <em-lap> q+ path 19:30:44 <sanScribe> EricP: WHen you encode you HOL's in a simple graph, you need to hide certain bits from the simple graph. Need reification or something. [ discomfort from PH and PFPS ] What subset have you had to hide from naive processors? 19:30:44 <em-lap> q? 19:31:09 <sanScribe> pfps: Hiding wasn't allowable, so we hid it in a blizzard. 19:31:26 <sanScribe> Ian: More on pipe analogy. We didnt pick it, it chose us. 19:31:34 <danbri> fair comment! 19:31:37 <em-lap> ack ian 19:31:40 <em-lap> ack path 19:31:56 <sanScribe> PH: EricP's way is misleading. We didnt have to hide anything. 19:33:50 <sanScribe> ericP: would a quote operator have helped? 19:33:58 <Bernard> Bernard has joined #swarch 19:34:10 <sanScribe> pfps: In OWL we never have to write down those primitive triples. 19:35:14 <sanScribe> ph: If you come to this world from logic, it's like falling down the rabbit hole.... 3 of them. Cherry picker holding cherry picker holding holder. 19:35:37 <RalphS> Sandro: I did unification back in August using reification 19:36:11 <RalphS> PatH: the truth predicate that you then need for de-reification is scary 19:36:49 <shellac> shellac has joined #swarch 19:40:17 <sanScribe> DanC: Peter, will you write up what you said in 5-page form? 19:41:25 <sanScribe> Ben Grosof: Problems/opportunities in layer. RDF syntax as encoding as RuleML. 19:41:53 <sanScribe> ...: OWL fits in more naturally.... 19:42:43 <sanScribe> PFPS: sounds like Owl-DL philosophy 19:43:38 <sanScribe> TimBL: You're talking about reification -- the RDF your describting your expression with is different from the instance RDF -- tool reuse is nice, but they don't really connect. 19:44:26 <sanScribe> BG: What about rules which can talk about RDF triples only.....? 19:44:49 <sanScribe> TimBL: (misssed) 19:45:07 <danbri> timbl said that lots of folk have a 'triples plus ...<something>' language on top of rdf 19:45:09 <sanScribe> Ian: Partial fix perhaps: stratified semantics for RDF, instead of recursive semantics. 19:46:32 <sanScribe> BG: second. RDF and RDFS are sufficiently rich ............(lost) 19:46:42 <sanScribe> em: strata == modules? 19:46:54 <sanScribe> em: separate namespaces, specific strata, ....? 19:46:56 <DaveB> I assume Ian's refering to rdfs(fa), 2 years too late IMHO 19:47:21 <sanScribe> Ian: Layer1-class, Layer2-class, .... no looping. 19:47:30 <sanScribe> PH: You could use owl:Class and owl:Property 19:47:46 <sanScribe> Ian: you might be able look at it. 19:47:53 <sanScribe> DanC: like ML type checker. 19:48:07 <sanScribe> TimBL: does that work on unbounded, open, semantic web. 19:48:14 <sanScribe> PH: You're safe until you pull in more data. 19:48:26 <sanScribe> TimBL: when you publish, OTHER PEOPLE combine you. 19:48:46 <sanScribe> ...: Different views. 19:48:53 <sanScribe> Ian: <nods> 19:49:30 <sanScribe> TimBL: Loops can emerge from different parts. 19:49:56 <sanScribe> DanC: Does there have to be a bottom layer. 19:50:15 <timbl> I said pointed out that BenG's use of RDF with RuleML he was describing was to use RDF to describe the otheer langauge, keeping RDF on a completely different level, making no connection netween assertions and object sin RDF and in teh other language. 19:50:42 <sanScribe> Ian: Two things in same strata.... 19:50:58 <sanScribe> ...: If you fixed names, class0, ...... 19:51:08 <sanScribe> PH: there are better things than stratified type theory. 19:51:27 <sanScribe> Ian: What's "works" 19:53:20 <sanScribe> BG: start with simple case -- first stratum -- classes do not have classes or proptiers or instances, ... DL constraint -- Simple==no constraints so Higher Order, vs Simple==clear constaints for tractability. 19:54:55 <sanScribe> ...: It doesnt make sense to treat the whole web as scopeless, as one KB. 19:55:56 <sanScribe> [ bit about lists ] 19:56:21 <sanScribe> TimBL: it's reasonable to have granularity, but the web works by people combining KBs. 19:56:49 <sanScribe> BG: Sure - if you have 3 KBs meeting DL restriction, you know the merge of them also meets it. 19:57:01 <sanScribe> TimBL: you migth pick up KBs out of google. 19:57:24 <sanScribe> q? 19:57:58 <sanScribe> PH: it should be up the READER to check cohernece. 19:58:36 <sanScribe> BG: Inferential machinery constructs a 3rd (perhaps virtual) KB. It has finite scope. 19:58:52 <sanScribe> TimBL: what if 3rd KB barfs? 19:59:12 <sanScribe> BG: an ecology will support people publ;ishing KBs which can be combined well! 19:59:42 <sanScribe> ...: Advertize that you are recursion free, etc, etc, quadtatic, etc. ... 20:00:04 <sanScribe> PH: Or even which heuristics are likely to work. Little hacks, repairs. 20:01:12 <sanScribe> Frank: This is no different from Data Sources on the web. One says Foo, another says Not Foo; you require that the party finding both to find out who to believe, etc. 20:01:51 <sanScribe> TimBL: sure, one of them is wrong. 20:02:50 <sanScribe> ...... 20:02:59 <sanScribe> em: WHat about real systems> 20:03:09 <sanScribe> Ian: My system detects conflicts. 20:03:26 <sanScribe> TimBL: we're talking about constraints on class hierarchy cycles. 20:04:10 <sanScribe> pfps: You beleive either class A is below, above, or we have a cycle. Like Each source is right or wrong. 20:04:47 <sanScribe> ph: some rdf-fcomments requesting stratifivcation. 20:05:10 <sanScribe> DanC: we could have a note which said " Here is a way to do stratified. " 20:05:35 <sanScribe> JJC: stratified RDF is just intersection of RDFS & OWL Lite. 20:05:47 <sanScribe> Ian: THat's PFPS trivial 2-layer strat. 20:05:58 <RalphS> JJC: the stratified RDFS is simply the intersection of RDFS and OWL lite 20:06:01 <sanScribe> DanC: But that's editoriallly expensive. Hard tor read it all. 20:06:13 <RalphS> PatH: agree 20:06:25 <sanScribe> ian; Syntax of owl-lite is another layer. 20:07:15 <sanScribe> PH: just use OWL vocabularies without OWL meanings! :-) 20:08:49 <sanScribe> PH: Who uses OWL terms without importing OWL semantics? 20:09:01 <sanScribe> [ various hands, including TimBLs to everyeone's surprise. ] 20:11:20 <sanScribe> RRSAgent, pointer? 20:11:20 <RRSAgent> See http://www.w3.org/2003/03/06-swarch-irc#T20-11-20 20:11:45 <sanScribe> ============================== 20:11:49 <sanScribe> RECESS for 45 minutes 20:52:01 <sanScribe> test 20:52:56 <libby> libby has joined #swarch 20:53:03 <bwm> bwm has joined #swarch 20:55:30 <JosD___> JosD___ has joined #swarch 20:57:06 <pfps> pfps has joined #swarch 20:57:47 <danbri> ping 20:57:52 <danbri> irc seems ok to me 20:58:06 <bwm> me too 20:58:54 <danbri> http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2002Jun/0180.html -- some early text from pathayes & me that became (prev version) 'social meaning' section of Concepts 20:59:09 <bwmscribe> neg comments about section 4 of rdf concepts doc 20:59:28 <bwmscribe> it would be nice if the protagonists could beat each other up instead of the editors 21:00:04 <bwmscribe> desired output from this meeting is advice for the rdfcore wg 21:00:20 <danbri> http://www.w3.org/2001/sw/meetings/tech-200303/social-meaning.html -- jjc's agenda for 'social meaning' discussion here 21:00:27 <bwmscribe> it is not an objective to make changes to the doc 21:00:41 <bwmscribe> options: 21:00:48 <bwmscribe> accept section 4 21:00:54 <bwmscribe> cut section 4 21:00:58 <bwmscribe> edit section 4 21:00:59 <DaveB> who's speaking? 21:01:02 <bwmscribe> end of list 21:01:06 <bwmscribe> jjc 21:01:15 <bwmscribe> jjc leading section 21:01:47 <bwmscribe> bijan identified issuettes 21:02:37 <bwmscribe> jjc suggests straw polls on issuettes questions 21:03:53 <bwmscribe> danc: is this a live with thing or a preference 21:03:56 <bwmscribe> jjc: preference 21:05:10 <bwmscribe> Section 4 as a whole: 21:05:20 <bwmscribe> accept: 1 21:05:34 <bwmscribe> cut: 7 21:06:25 <bwmscribe> aggressive edit: 6 21:06:54 <bwmscribe> who can't live with: 21:07:18 <bwmscribe> endorsing: 4 21:07:28 <bwmscribe> cut: 2 21:08:13 <bwmscribe> major edit: 2 21:08:19 <bwmscribe> endorsing s/4/6/ 21:10:54 <bwmscribe> authoritive definition of URI's: i.e. who gets to say what a URI denotes 21:12:48 <danbri> something like "RDF graphs have propositional content. Their meaning is fixed by a bunch of hairy stuff only partly understood and documented (eg. implicit theory of reference associated with URIs). Minor health warning. The End." 21:12:53 <danbri> ...would satisfy me. 21:14:03 <bwmscribe> who wants the sentence [[The social conventions surrounding use of RDF assume that any RDF URI reference gains its meaning from some defining individual, organization or context. ]] 21:14:26 <PStickler> Sounds fine to me 21:14:30 <bwmscribe> A: no substanstive change:6 21:14:36 <bwmscribe> C: cut: 4 21:14:40 <bwmscribe> s/4/5/ 21:14:54 <bwmscribe> cant live iwth: 21:14:58 <bwmscribe> as is: 4 21:15:00 <bwmscribe> cut: 2 21:15:16 <danbri> My 'E' vote meant "we should note the importance of reference in the task of fixing meaning" 21:15:19 <bwmscribe> more work: 2 21:16:00 <PStickler> My 'E' vote is wanting the simple clarification that the URI and URIREG conventions indicate *who* assigns meaning, not what the meaning is. 21:16:44 <bwmscribe> defining information: [[An RDF graph may contain "defining information" that is opaque to logical reasoners. This information may be used by human interpreters of RDF information, or programmers writing software to perform specialized forms of deduction in the Semantic Web.]] 21:17:14 <bwmscribe> a: is ok 1 21:17:21 <bwmscribe> c: cut: 8 21:17:27 <bwmscribe> eikeon: more work 2 21:17:37 <bwmscribe> cant live 21:17:45 <bwmscribe> as is: 8 21:17:50 <bwmscribe> cut: 2 21:17:54 <bwmscribe> more work: 2 21:18:25 <bwmscribe> s/eikeio/ edit 21:18:34 <bwmscribe> asserting rdf 21:20:45 <bwmscribe> [[RDF/XML expressions, i.e. encodings of RDF graphs, can be used to make claims or assertions. Not every RDF/XML expression is asserted. ]] about the 'real' world. Such expressions are said to be asserted. 21:21:04 <bwmscribe> preference: 21:21:20 <bwmscribe> a: its ok - 3 21:21:32 <bwmscribe> cut: 4 21:21:34 <bwmscribe> edit: 2 21:21:46 <bwmscribe> can't live with: 21:22:57 <las> las has joined #swarch 21:23:25 <Bernard> Bernard has joined #swarch 21:23:35 <bwmscribe> as is: 1 21:23:49 <em-lap> em-lap has joined #swarch 21:23:56 <bwmscribe> s/1/3/ 21:24:09 <bwmscribe> cant live: 1 21:24:12 <bwmscribe> more work: 0 21:24:50 <ericP> ericP has joined #swarch 21:25:12 <bwmscribe> relationship to http, ignoring the 404 21:25:26 <bwmscribe> keep: 2 21:25:34 <bwmscribe> cut: 6 21:25:38 <chaalsBOS> chaalsBOS has joined #swarch 21:25:44 <bwmscribe> edit: 0 21:25:52 <bwmscribe> s/0/1/ 21:25:55 <bwmscribe> cut live 21:25:58 <bwmscribe> as is: 3 21:26:14 <bwmscribe> cut 0 21:26:20 <bwmscribe> more work 1 21:26:25 <Tantek> Tantek has joined #swarch 21:27:17 <bwmscribe> mechanical inference and social meeting [[Human publishers of RDF content commit themselves to the mechanically-inferred social obligations. 21:27:17 <bwmscribe> The meaning of an RDF document includes the social meaning, the formal meaning, and the social meaning of the formal entailments. ]] 21:27:50 <bwmscribe> prefer: 21:27:57 <bwmscribe> as is: 3 21:28:03 <bwmscribe> get rid: 5 21:28:17 <bwmscribe> edit: 2 21:28:28 <bwmscribe> can't live with: 21:28:37 <bwmscribe> as is: 7 21:28:45 <bwmscribe> cut: 1 21:29:11 <bwmscribe> more work 0 21:29:28 <bwmscribe> slipperyness of intended meaning 21:30:42 <em-lap> RRSAgent, pointer? 21:30:42 <RRSAgent> See http://www.w3.org/2003/03/06-swarch-irc#T21-30-42 21:30:56 <bwmscribe> [[A combination of social (e.g. legal) and technical machinery (protocols, file formats, publication frameworks) provide the contexts that fix the intended meanings of the vocabulary of some piece of RDF, and which distinguish assertions from other uses (e.g. citations, denials or illustrations).]] 21:31:15 <danbri> we didn't have 'intended' in the original, http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2002Jun/0180.html 21:32:31 <simon> simon has joined #swarch 21:33:03 <bwmscribe> [[The social machinery includes the form of publication: publishing some unqualified statements on one's World Wide Web home page would generally be taken as an assertion of those statements. But publishing the same statements with a qualification, such as "here are some common myths", or as part of a rebuttal, would likely not be construed as an assertion of the truth of those statements.]] 21:33:55 <PStickler> The URI specs don't fix the meaning. They fix the *authority* who gets to fix the meaning. A schema provided by that authority can fix the meaning of a given URI. 21:34:30 <sanScribe> timbl: We're defining a system, mostly via natural language ... specs describe how it's supposed to work. 21:35:09 <sanScribe> ...: Notionally, the URI authority is supposed to define the meaning, but if it doesn't then things may or may not break. 21:35:29 <sanScribe> ...: Just people people don't always stick to the spec doesnt mean we shouldn't define it cleanly. 21:36:04 <sanScribe> Bijan: I disbelieve nearly everything you just said. If everything you said was false, nothing would be broken. 21:36:14 <sanScribe> ...: Most of what you said is misleading at least. 21:36:31 <DanC> DanC has joined #swarch 21:37:15 <sanScribe> TimBL: okay to rip out "legal" 21:37:29 <sanScribe> Bijan: I don't beleive you can fix one interpretation of a document. 21:38:00 <sanScribe> Bijan: you said "it means this" and that doesnt fly and is not necessary. 21:38:03 <las> las has joined #swarch 21:38:06 <sanScribe> TimBL: you're modeling too formally. 21:38:22 <sanScribe> ...: "H1 is a top level heading". 21:38:57 <sanScribe> On Slipperyness: 21:39:58 <sanScribe> a: 3 21:40:10 <chaalsBOS> s/3/4/ 21:40:11 <sanScribe> (modulo "intended") 21:40:19 <sanScribe> c: 5 21:40:30 <sanScribe> e: 6 21:40:36 <sanScribe> timbl: put it in primer 21:40:46 <sanScribe> a: lots and lots 21:41:01 <sanScribe> cant live with lots and lots 21:41:09 <sanScribe> cant live with c -2 21:41:13 <sanScribe> cant live with more work 0 21:41:23 <sanScribe> Third Party Publication 21:42:24 <RalphS> RalphS has joined #swarch 21:42:42 <sanScribe> a: 0 21:42:49 <sanScribe> c: 8 21:42:59 <sanScribe> e: 5 21:43:13 <sanScribe> a: -7 21:43:20 <sanScribe> c: -1 21:43:26 <sanScribe> e: 0 21:45:34 <sanScribe> Insult 21:47:34 <bwm_> bwm_ has joined #swarch 21:47:35 <sanScribe> TimBL: If this is all describing the framework, but Bijan is generally correct about how this applies as a matter of law. 21:47:35 <sanScribe> PH: We can have non-legal examples. 21:47:55 <sanScribe> Frank: I don't recall anywhere in HTML anything about libelous statements. But there's nothing peculiar about RDF in this regard. 21:48:35 <RalphS> RalphS has joined #swarch 21:48:40 <sanScribe> Mike Dean: we need a libel markup language...... 21:48:54 <Nobu> Nobu has joined #swarch 21:48:58 <sanScribe> DanC: Proximate Cause, Bernstein, Copyright law, links as speech, .... 21:50:30 <sanScribe> Bijan: I was commenting on the "mechanically inferred" part, NOT the insult bit. 21:51:20 <danbri> danbri has joined #swarch 21:51:44 <ericP> minutes are being written to http://www.w3.org/2003/03/06-SW-Arch 21:52:16 <sanScribe> RRSAgent, pointer? 21:52:16 <RRSAgent> See http://www.w3.org/2003/03/06-swarch-irc#T21-52-16 21:52:30 <sanScribe> Hrm. They are also being written there. 22:07:11 <sanScribe> TimBL: The essential point is that the RDF spec must say that the meaning of each URIRef term is based on the URI spec tree. 22:07:54 <jhendler_> jhendler_ has joined #swarch 22:14:19 <danbri> yeah, useful discussion 22:23:38 <Zakim> Zakim has left #swarch 22:24:06 <RalphS> [em, do you think that Tim has given an answer to Sun's question of "how _do you want me_ to refer to a piece of equipment?"] 22:24:46 <RalphS> [... a _clear_ answer for Sun's purpose, that is ;)] 22:28:31 <em-lap> [Ralph, no i do not] 22:28:34 <jhendler_> is this being scribed somewhere?? 22:28:42 <em-lap> RRSAgent, pointer? 22:28:42 <RRSAgent> See http://www.w3.org/2003/03/06-swarch-irc#T22-28-42 22:28:59 <bwm_> frank: we need to say things about how rdf should be used 22:29:03 <RalphS> [em, that was my impression too -- I hope we don't adjourn before getting him back up to answer] 22:29:35 <RalphS> Frank: it is important that the RDF spec say something about intended meaning 22:29:47 <bwm_> jjc: we need to come to a consusus if possible about what this meeting thinks 22:31:10 <em-lap> [ralph, i absolutely agree] 22:33:47 <bwm_> patrick s returns to point about having this covered in another document 22:34:09 <bwm_> bwm points out that cut doesn't mean cut forever, the issue is when do we do the work 22:34:32 <bwm_> danc calls for who would be interested in contributing to further work in another forum 22:34:51 <bwm_> pfps says that those voting to cut also think there is something important to say here 22:35:28 <GuusS> GuusS has joined #swarch 22:35:31 <bwm_> Propose section 4 should be struck and the CG should ensure this area is addressed in another document. 22:35:47 <bwm_> pfps there are shadows of section 4 elsewhere 22:36:29 <bwm_> DanC: Propose rdfms-assertion should be postponed and rdfcore relieved of responsibility 22:36:59 <bwm_> RalphS: does this mean that we'd have to change the namespace when that new doc was published? 22:37:10 <danbri> http://www.w3.org/TR/rdf-schema/#ch_comment in RDFS is in a similar vein. 22:37:27 <danbri> [[ 22:37:28 <danbri> A variety of documentation forms can be combined to indicate the intended meaning of the classes and properties described in an RDF vocabulary. Since RDF vocabularies are expressed as RDF graphs, vocabularies defined in other namespaces may be used to provide richer documentation. 22:37:29 <danbri> ]] 22:37:29 <bwm_> DanC: who would spend an hour a week 22:37:37 <bwm_> lots of people responded 22:37:52 <bwm_> danc call for folsk to approach him privately about chairing and editing 22:38:02 <bwm_> propose we suggest 22:38:05 <jhendler_> only an hour a week? Sounds liek vacation :-> 22:38:09 <bwm_> - section 4 is struck 22:38:24 <bwm_> - postpone rdfsm-assertion 22:38:39 <bwm_> - CG prioritise highly work on this issue 22:39:39 <bwm_> http://www.w3.org/2000/03/rdf-tracking/#rdfms-assertion 22:40:23 <bwm_> timbl: if its a tag issue, then timbl will ask the tag if they want to give it to us and they'll probably bite our arms off. 22:41:13 <RalphS> PatH: whatever the TAG says does matter here and we need to take it into account 22:41:49 <bwm_> add to proposal: coordinate with the TAG over URI denotation 22:41:53 <bwm_> danc proposes 22:41:55 <RalphS> the issue about adding word regarding URI denotation 22:41:57 <bwm_> timbl seconds 22:42:00 <bwm_> none apposed 22:42:04 <bwm_> no abstentions 22:42:08 <bwm_> carried unanimously 22:42:29 <bwm_> action jeremy to email to rdf comments 22:43:09 <bwm_> danc propose to break 22:43:22 <bwm_> em-lap: meeting is in riverfront room tomorrow 22:43:29 <bwm_> same floor just down the corridor 22:44:05 <bwm_> frankm: if you're driving: road conditions are hazardous - 100 care pile up, lots of accidents 22:44:15 <bwm_> meeting adjourned 22:45:04 <danbri> am wondering whether www-rdf-logic is a good forum to continue this discussion... 22:47:35 <ryanlee> ryanlee has left #swarch 22:53:51 <jhendler_> if anyone sees this - can you ask Bijan to call me or rejoin irc? thanks 22:54:10 <SethR> SethR has joined #swarch 23:15:41 <libby> libby has joined #swarch 23:31:19 <libby> ------query and rules bof notes ------ 23:31:47 <libby> http://www.w3.org/2001/11/13-RDF-Query-Rules/ 23:32:02 <libby> - eric introduces RDF query 23:32:06 <libby> ericP 23:32:47 <GuusS> GuusS has joined #swarch 23:33:51 <JosD___> JosD___ has joined #swarch 23:35:54 <libby> picture which was the result of the first BOF on query: http://swordfish.rdfweb.org/photos/2003/03/05/2003-03-05-Pages/Image4.html 23:36:19 <libby> benjamin asks q about relative expressivity of N3 and algae 23:36:41 <libby> algae doesnt have "for some", N3 does. 23:39:51 <libby> ericp talks about SQL backends for algae, including access to a 'normal' realtional database 23:41:14 <libby> [aside - Jan Grant is doing something similar - mapping RDf queries to arbitrary SQL databases] 23:43:35 <libby> [scribe not catching all the detail of the discussion, sorry] 23:53:54 <Nobu> Nobu has joined #swarch 23:55:58 <libby> the next IRc meeting http://lists.w3.org/Archives/Public/www-rdf-rules/2003Feb/0014.html about tests for RDf queries 23:56:23 <libby> Mike Dean talks about his hierarchy of RDF QLs 23:57:26 <libby> http://www.daml.org/2002/03/tutorial/slide122-0.html 23:57:34 <libby> talks about DQL: 23:57:57 <libby> http://www.daml.org/2002/08/dql/ 23:58:09 <libby> no official syntax yet, but a model 23:58:27 <libby> "must bind" "may bind" "don't bind" 23:58:30 <libby> for variables 23:58:57 <libby> for bnode processing - dont return the answer 23:59:35 <libby> also skolem functions