Semantic Web Interest Group IRC Chat Logs for 2002-10-31

00:50:02 * sandro reads over heflin/timbl conversation. excellent start.

01:07:36 * JimStillAtWork wonders "excellent start" of what? i.e. seems like they're still somwhat talking past each other.

01:09:39 <JimStillAtWork> I do like Tim's "<> a ucc:bindingContract." as a way of doing an imports-like thing, think I may like it a lot...gotta think about it though

01:10:03 <JimStillAtWork> JimStillAtWork is now known as jhendler

01:22:10 <ChanServ> [#rdfig] This channel is logged and blogged: http://logicerror.com/rdfIRCWelcome

01:26:54 * sbp notes that if you want to implement daml:imports you have to throw away the subject anyway

01:27:00 <sbp> or else provide a specific one

01:27:07 <sbp> (implement in N3, that is)

01:28:23 <sbp> I had something like { ?x daml:imports [ log:semantics { ?s ?p ?o } ] } => { ?s ?p ?o } .

01:29:18 <sbp> (needs a log:includes in there, actually)

01:30:03 <sbp> but the proposed { [ a :TrueDocument; log:semantics [ log:includes { ?s ?p ?o } ] ] }{ => { ?s ?p ?o } . sits much nicer

01:30:58 * sbp remembers going through a heap of daml files that used daml:imports actually getting CWM to get all the triples from the imported files

01:31:06 <sbp> it only worked on one of the whole lot, IIRC

01:32:56 <sbp> the problem with having the unary function is that you can't say what other people believe, though. no more <someOtherPerson'sFile.rdf> daml:imports <someBeliefs.rdf> . perhaps that's not much of a drawback...

01:35:59 <sbp> Hmm... by searching for { ?x daml:imports <myOntology> } you can find out which people use (same as "believe to be true"?) your ontology, I suppose. you can't really do that with :TrueDocument

01:36:35 <sbp> (unless you had loads of formulae and then searched for { ?F log:includes { <myOntology> a :TrueDocument } })

01:36:58 <deltab> how would an ontology be "true"?

01:38:19 <sbp> I'm not an expert, but I assume that some ontologies are taken as inconsistent. subClass(Human, Tree) for example. so for "true", perhaps substitute "consistent within a chosen application"

01:39:03 <sbp> :ConsistentDocument might fare better

01:39:05 <deltab> ah

01:39:28 <deltab> you'd want to say what it was consistent with, though

01:40:13 <sbp> well, if a formula has a statement in it that says that a certain document is consistent, you'd probably consider it consistent with the other statements in that formula, I suppose

01:46:30 <sandro> I think you're not so much interested in it being internally consistant as it being consistent with all the other data your application is ever going to handle.

01:46:56 <sandro> "True" seems like a nice word for that, but it's not perfect.

01:47:08 <sandro> "Ruling", "Guiding", "Foundational", ....

01:47:24 <sbp> careful: you're heading towards "Axiomatic"

01:47:30 <sbp> BTW, congrats on TAG going to RFC

01:47:36 <sbp> "tag:", rather. heh

01:47:57 <sandro> How about <foo> a owl:Inclusion

01:48:15 <sbp> that I like

01:48:50 <sandro> I'm totally unclear that tag: is going to RFC. What's your knowledge of the status? paf's e-mail was like "it's going to RFC, but does anyone have any reason it shouldn't?" which is *very* confusing.

01:49:38 <sandro> (Axiomatic is really rather different, I think.)

01:50:07 <sandro> (esp when Dan Connolly and Larry Masinter jump in with reasons why it shouldn't!)

01:51:00 * sandro better get back to housecleaning.

01:51:05 <sbp> I think that it's going to be published, but that the review comments might lead to an updated version. actually, I don't know; I hadn't thought about that much. it seems very odd indeed now that you mention it

01:51:41 * sbp had better let you get back to housecleaning :-)

01:52:09 <sandro> And perhaps I misunderstood it. Perhaps paf was looking for "just cause" as in some basic misrepresentation, as opposed to technical argument.

01:52:56 <sbp> sounds fair

01:53:38 <jhendler> hmm, guess imports is unary - we represent it binary, but always in the context of <> -- typically

01:53:46 <jhendler> <> a owl:ontology;

01:53:53 <jhendler> owl:imports foo:.

01:55:21 <jhendler> the question is actually easily asked in an ontology setting - suppose you create an ontology with lots of facts about dogs and cats. I want some way to say "I agree with everything you say about dogs and cats"

01:55:28 <jhendler> if I just say

01:55:36 <jhendler> :foo a sbp:cat.

01:55:49 <jhendler> then how do you know whether I agree with any of the stuff you say about dogs?

01:56:28 <jhendler> Seems to me imports is really a "social" contract (Jeff often calls it "commitment") and we're just looking for the technical content to deliver it.

01:57:18 <jhendler> Note that if there is no explicit imports some people (me for one) feel strongly that it must NOT mean I commit to the ones I don't point at (i.e. I like what Seansays abotu cats, but refuse to see what he says about dogs)

01:57:36 * jhendler ends soliloquey - but at least I'm now on the record :->

01:58:04 <sbp> that seems fair, but I still wonder if saying :foo a sbp:cat even means that you believe anything else I say about cats

01:58:39 <jhendler> sorry, I didn't mean to imply that at all - I was just using cats v. dogs to avoid getting into that issue.

01:58:42 <sbp> I suppose, however, that if you use someone else's identifier, you're also implicitly using their definitions of it

01:59:01 <sbp> ah, I see

01:59:03 <jhendler> I agree if you point to something but don't import the policy is unclear (or should be explicitley defined somewhere)

01:59:53 <jhendler> but Jeff wants some way to say "I agree with everything" without having to either explicitely point at each URI (and even that wouldn't be enough because some things in the graph dn't have URIs)

02:00:15 <jhendler> or having to cut/paste the document.

02:00:23 <sbp> there were a lot of conversations on DC-architecture (I think) a few months ago about the nature of a vocabulary. there seems to be some confusion over the exact meanings of "document", "vocabulary", "ontology", and "formula", and also how they relate

02:00:48 <sbp> (and "namespace", for that matter)

02:00:49 <jhendler> Basically, we need an RDF equivalent of XInclude (but cannot use XInclude since it creates illegal RDF)

02:01:20 <jhendler> i.e. a document with an <rdf:RDF> cannot include another document with an <rdf:RDF> and still be a legal RDF/XML focument

02:02:25 <jhendler> sbp, for what it's worth, there's been argument over those terms a lot longer than that -- I think the first proposal for what an ontology was is dated BC

02:02:52 <sbp> XInclude: why not use XPointer to get to whatever bits you need?

02:03:53 <sbp> B.C.: ah, but there weren't any mailing list back then, so... :-)

02:06:33 <jhendler> Sean, you're right - of course, even if we just go back to my early AI days of discussing this stuff - you weren't born yet...

02:06:43 <jhendler> and there were mailing lists, just not web archives of them :->

02:07:11 <jhendler> Xpointer: how? WOuld mean writing complex crap to do something very simple...

02:07:36 <sbp> well, it's not too bad:-

02:07:37 <sbp> <rdf:RDF [...]><xi:include href="somedoc.rdf#xmlns(rdf=[...])

02:07:37 <sbp> xpointer(rdf:RDF/rdf:Description)" parse="xml"/></rdf:RDF>

02:08:26 <sbp> (it's a shame that all that old information was lost. the Google Usenet archives are just the tip of the iceberg, I suppose)

02:08:35 <jhendler> not quite so simple, because there's other stuff in there you also need to "not include" (for example if that document has an ontology statement, then you cannot have one also)

02:09:02 <sbp> an ontology statement?

02:09:14 <jhendler> sbp - wasn't totally lost, we had things called books (and papers) back then...some of this stuff was written down by guys like Quine and exist in places called libraries ...

02:09:30 <jhendler> <> a owl:ontology;

02:09:31 <sbp> ugh. don't get me started on libraries :-)

02:09:40 <jhendler> owl:imports foo: ;

02:09:51 <jhendler> dc:creator "Hendler";

02:09:52 <sbp> oh, right, you need to be able to parse it as RDF... argh

02:09:59 <jhendler> (other junk);

02:10:05 <jhendler> .

02:10:35 <jhendler> yeah - that RDF parsing thing is the rub - the XML mechanisms have a little bit of trouble with it :->

02:11:10 * jhendler looks at time - GACK, promised wife I'd be home 15 minutes ago, gotta drive home faster than the speed of light! See you yesterday, JH

02:11:14 <sbp> I still reckon that we should have done RDF, and then SGML, XML, and whatnot

02:11:17 <sbp> heh. c'ya

02:11:39 <jhendler> sbp - right, but somehow those other folks out there seem to do that other stuff...

02:11:41 <jhendler> ciao

02:58:47 * sandro wonders seriously if a little change in RDF/XML wouldn't solve a lot -- allow an extra rdf:RDF to be present and be ignored!

04:06:25 * chaalsMEL waves

11:30:54 <JibberJim> Not an RDF question, but it's a web-arch one, so I'm going to give it a go....

11:32:03 <JibberJim> If I give a 30* redirect in response to a request, I'm supposed to put a "hyperlink" in the body for http 1.0 clients, what sort of mime-type should I use for that, e.g. if I'm redirecting SVG, or RDF documents?

12:21:04 <PatrickS> testing...

12:21:24 <sandro> Hello, Patrick.

12:21:42 <PatrickS> Hi Sandro

12:22:10 <PatrickS> I'm testing an irc client on my phone, so typing os slooooow...

12:27:46 <sandro> ahhhh! cool!

12:29:37 <PatrickS> Well, it seems to work OK, but I won't be saying alot due to the typing effort

12:30:05 <PatrickS> (which I'm sure some folks are happy to hear ;-)

12:30:16 <sandro> ha!

12:30:36 <sandro> I dont know what you're talking about. :)

12:30:48 <sandro> Can you do voice at the same time?

12:30:57 <PatrickS> You're so kind...

12:31:35 <PatrickS> On this phone at the moment, no, but ... ;-)

12:34:05 <PatrickS> The new GPRS capable phones (7650) can do simultaneous data and voice

12:34:51 <PatrickS> and there is an irc client for the 7650, but the 9210 (my phone) doesnt do GPRS

12:35:21 <PatrickS> and the 7650 also has a built in camera...

12:36:18 <PatrickS> but this is an rdf channel so I'll shut up... :)

12:36:48 <JibberJim> [After the short advertisements we return to regular programs...]

12:37:05 <JibberJim> [Note other phones are available]

12:37:11 <PatrickS> sorry ;+)

12:40:40 <PatrickS> Bye

13:27:42 <shellac_>http://brownsauce.sourceforge.net/

13:27:43 <dc_rdfig> A: http://brownsauce.sourceforge.net/ from shellac_

13:28:04 <shellac_> A|BrownSauce RDF Browser

13:28:13 <shellac_> A:|BrownSauce RDF Browser

13:28:13 <dc_rdfig> Titled item A.

13:28:51 <shellac_> A:Browse the Semantic Web like any other part of the web.

13:28:51 <dc_rdfig> Added comment A1.

13:29:17 <shellac_> A:Instructions ae a little weak, currently.

13:29:17 <dc_rdfig> Added comment A2.

13:29:31 <shellac_> A2:Instructions are a little weak, currently.

13:29:31 <dc_rdfig> Replaced comment A2.

13:30:29 <shellac_> A:Currently FOAF is the best place to browse. Lots of seeAlsos.

13:30:29 <dc_rdfig> Added comment A3.

13:31:07 <shellac_> A:BrownSauce is very customisable in appearance, thanks to CSS.

13:31:08 <dc_rdfig> Added comment A4.

13:42:39 <shellac_> A:Uses rdfs:label to hide URIs (if available). Eg customisation makes foaf:name a subproperty of label

13:42:40 <dc_rdfig> Added comment A5.

13:56:06 <shellac__> A:[http://brownsauce.sourceforge.net/instructions/screenshot.png|Screenshot]

13:56:06 <dc_rdfig> Added comment A6.

14:04:03 <shellac_> A:[http://brownsauce.sourceforge.net/instructions/screenshot.png|Screenshot]

14:04:03 <shellac_> Hey...

14:21:23 <ChanServ> [#rdfig] This channel is logged and blogged: http://logicerror.com/rdfIRCWelcome

14:26:10 <lilo> [Global Notice] Hi all. We've experienced some difficulties relating to network configuration files, and they should not recur anytime soon. Thanks.

17:07:42 <Mike> Mike is now known as KillEmAll

17:43:06 <sandro> logger,pointer?

17:43:06 <sandro> See http://ilrt.org/discovery/chatlogs/rdfig/2002-10-31#T17-43-06

17:57:49 <sbp> sandro: part of the XInclude problem is that when you include a bit of RDF with relative links, those relative links are now going to be wrong

17:58:26 * JibberJim doesn't see the need for include anyway, triples are triples surely?

17:58:38 <sbp> (unless that's your intended effect, such as mirroring a schema that uses rdf:ID throughout)

17:59:04 <sbp> Jim: it's for when you want to assert those triples without copying and pasting by hand

17:59:17 <sbp> it could be useful for if you have a bit of RDF that's constantly being updated

17:59:19 <JibberJim> oh right, yeah that makes sense.

17:59:34 * sbp has to go

18:03:32 <sandro> dajobe, do you see a big problem with allowing rdf:RDF elements to be nested (with no semantic effect), just to allow XInclude to work right?

18:03:50 <dajobe> hmm

18:04:00 <dajobe> an rdf:RDF scopes the rdf graph (as triples)

18:04:08 <dajobe> so you are talking about graph merging

18:04:20 <dajobe> therefore, only rdf:RDF at the top-level seems to make most sense

18:04:26 <dajobe> i.e. inside <rdf:RDF> itself

18:04:36 <sandro> Nah, just syntactic inclusion. The inner rdf:RDF would just be ignored.

18:04:47 <dajobe> no, you have to be careful

18:04:56 <dajobe> 'cos you must start the rdf/xml striping with nodes

18:04:58 <sandro> Oh yes -- only at the top leverl, right.

18:05:16 <dajobe> there's another place they could go too

18:05:17 <sandro> I agree -- below the top level you get tangled in striping.

18:05:25 <dajobe> which takes a set of nodes, inside parseType="Collection"

18:05:50 <dajobe> basically anywhere http://ilrt.org/discovery/2001/07/rdf-syntax-grammar/#nodeElementList is used

18:05:54 <sandro> There's lots of fun places to put them, but that's not necessary for what JHendler wants.

18:06:34 <sandro> If this solves WebOnt's "imports" debate, it could be a Good Thing.

18:06:34 <dajobe> but making RDF/XML work with/require? XInclude is another issue

18:06:41 <dajobe> we'd have to consider the impact

18:06:49 <dajobe> does the infoset requirements change

18:07:10 <sandro> (I'm not requesting this btw, just trying to understand the impact.)

18:07:17 <dajobe> sure

18:07:56 <sandro> Raptor still doesnt actually do web-fetching does it? It would need to for xinclude. :-)

18:08:18 <dajobe> it'd have to be optional I expect

18:08:21 <dajobe> raptor - no

18:08:40 <dajobe> I've only just got all the URI handling / resolving working

18:08:52 <dajobe> I think libxml2 would do it automatically though

18:08:59 <dajobe> so wouldn't affect apps of it, such as raptor

18:09:06 <dajobe> dunno if expat does

18:10:36 <sandro> So if you pass the right flags to libxml2, raptor would just work with includes? Cool!

18:10:55 <dajobe> I guess

18:11:01 <dajobe> libxml2 is pretty complete with XML tech

18:11:26 <dajobe> you could even try that yourself sandro ;)

18:11:57 <sandro> indeed. :)

18:12:44 <dajobe> yes, the libxml2 source has xinclude.c

18:17:44 <dajobe> seems to be for DOM api only though, which I don't use

18:19:21 <sandro> I'm looking at python interfaces at the moment.

19:06:04 <ronwalf> danbri__: I heard you were interested in a soap interface to parka

19:06:53 <JibberJim> That's a ghost of danbri ronwalf, he's not really there.

19:07:26 <ronwalf> Perhaps he'll see it later

19:08:04 <KillEmAll> KillEmAll is now known as CruncyTaco

19:08:10 <CruncyTaco> CruncyTaco is now known as CrunchyTaco

19:46:37 <ChanServ> [#rdfig] This channel is logged and blogged: http://logicerror.com/rdfIRCWelcome

19:50:16 <lilo> [Global Notice] Hi all. We're experiencing network-wide problems and will keep you posted. We hope the instabilities are mostly past.