Semantic Web Interest Group IRC Chat Logs for 2004-08-09

00:00:04 <mattmcc> As a fan of content negotiation, I'd prefer to see the RDF provided by the same URI, but a different content type.

00:00:26 <mattmcc> Allowing clients that don't care about the markup to retrieve just the RDF, and vis versa.

00:00:33 <mterry> mattmcc: Well, that works only if the user agent is interested in only one of those two things.

00:00:40 <mterry> What about a browser than understands both.

00:00:53 <mattmcc> mterry: Not necessarily. A UA can re-request the URI with a different Accept header.

00:01:00 <mterry> But does it know to?

00:01:09 <mterry> If you link it?

00:01:16 <mattmcc> If you write the UA to do so, sure. :)

00:01:36 <mterry> Or use a type="" argument for your <link>, but I don't think the UA is required to follow that hint.

00:01:59 <mterry> I'm interested in getting RDF/XHTML bundled today, not on tomorrow's browsers. Does it make semantic sense?

00:02:36 <mattmcc> Well, today's browsers aren't going to care about the RDF any more than they'd care about a <link>.

00:02:41 <mterry> i.e. is the XHTML really just "spilling out" or is the XML being parsed intelligently when mozilla displays it?

00:03:00 <mattmcc> That depends on the content type.

00:03:16 <eaon> (and no browser aims to implement rdf like that for now, right?)

00:03:18 <mterry> mattmcc: OK, so modern browsers don't care, but special case spiders might, browser plugins might.

00:03:25 <mterry> application/xml

00:03:33 <eaon> browser plugins already care about <link>s

00:03:47 <eaon> see the foaf plugin and the doap plugin for firefox

00:03:51 <mterry> eaon: And I care about bundling, not RDF in a different document.

00:03:57 <crschmidt> (hey, i wrote one of those!)

00:04:05 <mterry> Yeah, <link> solves my problem. But I thought this did too.

00:04:11 <jsled> crschmidt++

00:04:31 <mterry> I just don't like <link>. It offends my RDF side as requiring any thing that wants to know my RDF to first understand XHTML.

00:06:05 <danbri> <link/> was there first...

00:06:20 <danbri> and its worth looking in <link>s, since they often have pointers to more RDF elsewhere

00:06:49 <eaon> link has a broader (much) support than rdf

00:06:56 <mattmcc> I'd also wonder about how much RDF you'll be repeating with each RDF-XHTML document. Author information, for example.

00:06:59 <mterry> OK, I understand that <link>'s have their place and they exist today. But that doesn't solve the problem for an RDF UA or the desire to bundle.

00:07:25 <eaon> there's no rdf ua that handles data like that - at least i don#t know one

00:07:47 <mterry> mattmcc: Sure, some data is replicated as makes sense. But that's more a problem for the author than the UA, and PHP et al solve that anyway.

00:08:11 <mterry> Well, why not? It's just an RDF document, all RDF UA's handle it.

00:09:02 <mterry> RDF-only UAs and HTML-only UAs handle my approach fine. I'm concerned about a UA that understands both. Would what I give it make semantic or syntactic sense?

00:09:44 <eaon> it won't make anything more meaningful just because you wrap it

00:09:52 <CloCkWeRX> doap plugin for firefox?

00:09:54 <CloCkWeRX> oooh

00:09:59 <mattmcc> mterry: Eh, if you bring PHP into the picture, then I'd submit that makes bundling irrelevant, since you can generate content at two locations just as easilly as you can at one.

00:10:03 <eaon> if you integrate rdf in xhtml 2.0 gives it more meaning

00:10:08 <crschmidt> CloCkWeRX: yeah, http://crschmidt.net/doap/doaper.html

00:10:16 <CloCkWeRX> YAY

00:10:27 <crschmidt> CloCkWeRX: You missed that a few weeks ago? :)

00:10:39 <mterry> eaon: Again, I don't want to wait for XHTML2, and I'm not interested in making it more meaningful, I just want to not lose meaning.

00:11:14 <mterry> mattmcc: That depends and is pretty orthogonal to my question.

00:11:19 <eaon> mterry: you're not losing any meaning with a link

00:11:43 <mterry> Except you require the UA to know XHTML.

00:11:47 <Arnia> mterry: What do you mean by 'RDF-UA' btw. What UAs exist for RDF?

00:12:07 <mterry> Arnia: Any computer program that understands RDF.

00:12:37 <CloCkWeRX> i've been having a social life

00:12:39 * danbri -> zzz # night all...

00:12:45 <Arnia> Bye danbri

00:12:54 <eaon> sleep well danbri

00:12:54 <mterry> Bye, thanks for the help.

00:13:18 <CloCkWeRX> or rather girls who come, ransack, and leave; depriving me of geekery

00:13:35 <eaon> mterry: so, some "rdf-uas" will go like that: &lt;html&gt;

00:13:52 <Arnia> mterry: Well, dead text is dead text.

00:13:54 <mterry> eaon: I don't understand what you mean?

00:13:57 <Arnia> Its meaningless

00:14:04 <mattmcc> Not all RDF tools deal in RDF/XML.

00:14:10 <Arnia> Meaning comes through manipulation of structure

00:14:25 <mterry> mattmcc: Fine, I restrict my meaning of RDF-UA to RDF/XML-UA. Can we please move on?

00:15:00 <eaon> mterry: essentially... what about the "rdf-uas" that don't understand xhtml?

00:15:31 <mterry> eaon: That's fine. I don't expect them to. They will skip the XHTML, but they will understnad the metadata about it as well as any otherr RDF I slip in.

00:15:40 <CloCkWeRX> heh

00:15:46 <mterry> Just as XHTML-UA's will understand the XHTML and ignore the RDF.

00:15:55 <CloCkWeRX> you'd actually have to link to a stylesheet for XHTML

00:16:32 <mterry> CloCkWeRX: Hmm?

00:16:34 <eaon> mterry: so ... how will rdf uas access this kind of data?

00:16:43 <eaon> mterry: there is no "browse" function in them

00:16:53 <mterry> eaon: What kind of data? The RDF? It's an RDF document!

00:16:57 <eaon> mterry: it will be the same when you use a link

00:17:01 <mterry> If they get pointed to it.

00:17:13 <mterry> I'm talking like, you know, those FOAF explorer kind of things.

00:17:13 <eaon> the also get pointed to your document

00:17:15 <CloCkWeRX> if the rdf/xml-ua is doing any XML + CSS; you could easily make the XHTML *presentable* to a visual user agent with stylesheets

00:17:16 <mterry> Imagine one of those.

00:17:42 <eaon> mterry: foaf explorer spits out everything, even if it doesn't know what it is - so you'll end up having dead xhtml in it ;)

00:17:55 <mterry> That's true, but I'm required to make the XHTML inside a parseType="Literal" part. Would it be parsed correctly for that?

00:18:45 <mterry> eaon: Fine, that's FOAF explorer's problem. But you understand the kind of RDF-UA I mean?

00:18:55 <eaon> mterry: it will be parsed as literal and shown like one - but it will be totally useless to foaf explorer as it just displays the whole document

00:19:02 <eaon> the thing is

00:19:07 <eaon> the ua you think about

00:19:10 <eaon> DOESNT EXIST :)

00:19:44 <mterry> I know. But I am curious if my approach is clean. I know it works on all current UA's (well, gecko, IE, and some RDF stuff).

00:20:00 <mterry> I want to know if a UA was written to all the specs, if it would understand my document.

00:20:11 <mterry> i.e. is what I'm doing actually spec-valid and semanticly sound?

00:20:18 <mterry> The same question I asked at the beginning.

00:20:32 <eaon> hehe, again, the xhtml you put in there, is *meaningless*

00:20:38 <mterry> eaon: Ok.

00:20:46 <mterry> Because of the parseType="Literal"?

00:21:13 <eaon> because you're not defining anything meaningful (rdf) in it.

00:21:42 <mterry> eaon: ?? Maybe I don't understand the RDF spec well, but I'm pretty sure that RDF can point to data. Or the current document by saying about="".

00:21:55 <eaon> mterry: the problem is you can't

00:21:55 <mterry> RDF doesn't have to be inside of the thing it describes.

00:22:07 <danbri> B:Quick [http://lists.w3.org/Archives/Public/www-archive/2004Aug/att-0003/servlet_20267.png|graphed example] to show structure.

00:22:08 <dc_rdfig> Added comment B6.

00:22:09 <eaon> mterry: when you have an rdf document the # in a uri points to an rdf:ID

00:22:12 <eaon> not to a html id

00:22:14 * danbri really -> zzz now ;)

00:22:17 <eaon> you have dead text

00:22:22 <eaon> and you can't make it meaningful

00:22:28 <eaon> not even by trying to point to it

00:22:36 <mterry> Well, sure. OK. So I can't point to fragments of the XHTML. But I can talk about the whole thing.

00:22:55 <eaon> same for the <link>

00:23:12 <mterry> eaon: I UNDERSTAND THAT LINK WORKS.

00:23:33 <eaon> that's fine :) i just don't understand why you don't like it

00:23:58 <eaon> the thing is, your solution breaks on various things

00:24:00 <mterry> eaon: Again, because I like the idea of bundling and <link> requires UAs to understand both XHTML and RDF.

00:24:07 <mterry> eaon: OK, that's news.

00:24:07 <Arnia> <link /> IMO is the 'correct' way to do things

00:24:10 <CloCkWeRX> (ab)use the rel attribute of hyperlinks too :D

00:24:10 <mterry> eaon: What things/

00:24:20 <eaon> mterry: again, you can't make things more meaningful

00:24:27 <eaon> mterry: the html content that is

00:24:34 <mterry> eaon: I don't want to make things "more meaningful" I just want to not lose meaning.

00:24:47 <mterry> I want to both talk about the document and display the document in one package.

00:24:49 <eaon> mterry: accessibility breaks, and some rdf interpreting tools won't get your html and just spit out useless html

00:24:56 <eaon> something like "skip" isn't there in rdf

00:25:14 <eaon> what we are trying to say

00:25:17 <eaon> at least i am

00:25:20 <mterry> eaon: Well, sure. But that's why I have parseType="Literal". The RDF doesn't evaluate it.

00:25:21 <eaon> this won't work

00:25:22 <eaon> :)

00:25:44 <mterry> It does work. Just because RDF UAs see some dead text doesn't mean it doesn't work.

00:25:57 <Arnia> Define 'works'

00:26:02 <Arnia> I'm confused on this point

00:26:02 <eaon> hehe

00:26:04 <mterry> Is my usage of parseType="Literal" wrong? I don't really understand the semantics of it.

00:26:45 <mterry> Arnia: It works if RDF-only UAs understand the RDF part, and XHTML-only UAs understnad the XHTML part, and dual UAs understand both.

00:26:59 <mterry> So far, I know that both RDf-only and XHTML-only seem to work.

00:27:06 <eaon> a dual one will also get the link

00:27:09 <mattmcc> And dual UAs don't exist.. :)

00:27:10 <eaon> because thats whats used now

00:27:12 <mterry> eaon: OMG.

00:27:14 <eaon> and that

00:27:14 <Arnia> parseType="Literal" says that whatever is inside this tag should be treated as a literal, XML or otherwise

00:27:33 <mattmcc> I'm not even sure what such a UA would behave like.

00:27:37 <Arnia> (afaik)

00:27:40 <eaon> mterry: i don#t see where the problem is by pointing the "rdf ua" to the rdf part?

00:28:03 <mterry> eaon: You mean like pointing FOAF explorer to it?

00:28:07 <mterry> Sure, I could manually do that.

00:28:18 <mterry> But for autodiscovery, it doesn't work so well.

00:28:20 <eaon> mterry: there is no rdf browser

00:28:34 <mterry> eaon: I know. I'm not talking about browsers, necessarily.

00:28:42 <mattmcc> mterry: Autodiscovery is already here for <link>s.

00:28:44 <mterry> eaon: But there could be an RDF browser, eh?

00:28:48 <eaon> mterry: so? thats what link does

00:28:51 <mterry> mattmcc: I KNOW. I KNOW LINKS WORK.

00:28:51 <mattmcc> See the aforementioned Moz extensions for FOAF & DOAP.

00:28:55 <eaon> mterry: very bad idea

00:28:58 <eaon> mterry: imo of course

00:28:58 <mattmcc> It's there, it's easy.

00:29:11 <eaon> rdf itself isn't meant to be reviewed by humans

00:29:14 <mterry> eaon: OK, so it's a bad idea for what reasons? What actually sucks about it?

00:29:31 <mterry> eaon: I disagree about that, but OK.

00:29:46 <mattmcc> Well, it's true. RDFs primary benefit is that it's machine readable.

00:29:47 <eaon> mterry: erm ...

00:29:48 <mterry> eaon: I mean, I think they were careful to let it be human-readable.

00:29:58 <eaon> mterry: rdf -> tools -> human

00:29:59 <mattmcc> You're thinking of XML.

00:30:02 <mterry> eaon: Not that humans were supposed to do something meaningful with it.

00:30:20 <mterry> mattmcc: OK. Like I said, when I say "RDF" I mean "RDF/XML"

00:30:43 <mterry> Look, I'm not here to be convinced of the wonders of <link>

00:30:53 <mterry> I just want to know what is actually, physically wrong with my approach.

00:31:59 <eaon> mterry: you can't make things more meaningful in the html part - some html browsers *will* be confused

00:32:10 <eaon> rdf is meant to make things more meaningful, so that breaks your thing imo

00:32:29 * Arnia grins at the idea of 'human-readable' together with 'RDF/XML'

00:32:43 <eaon> it may be valid, but it breaks in real life

00:32:53 <eaon> as does some xhtml+css stuff.

00:33:01 <eaon> because it's just not supported

00:33:05 <mterry> eaon: You keep saying "more meaningful". The RDF stuff can talk about the current document. The XHTML is part of the current document. So, I get to talk about the XHTML. Not pieces of it, but the whole thing.

00:33:20 <mterry> eaon: Again, I think my approach does work in real life.

00:33:27 <eaon> more meaningful pieces

00:33:34 <mterry> What doesn't? Please! Tell me what actually is wrong.

00:33:45 <Arnia> mterry: Ok, so where is the document you have stored in the literal

00:33:56 <Arnia> mterry: Where can legacy tools get to it?

00:34:04 <eaon> mterry: you can't make <a href="#something"> for dual agents

00:34:27 <mterry> Arnia: Legacy tools get to it by reading the document and parsing the literal.

00:34:58 <mterry> Is that a problem? An XHTML/RDF UA would be required to skip the literal?

00:35:12 <eaon> no

00:35:14 <eaon> can't be

00:35:28 <eaon> rdf has it's own interpretation of id, as does xhtml

00:35:31 <mterry> OK. Then how best to say, there is some literal XHTML data here?

00:35:39 <mterry> eaon: I understand there are ID problems.

00:35:40 <DanC_mob> Ugh. Flight delayed.

00:35:41 <eaon> there is no way

00:35:55 <mterry> eaon: We don't need to use ID references. Let's move on from the ID problem.

00:36:06 <Arnia> mterry: Make the document a resource, and use <link> to reference an external graph

00:36:12 <eaon> yes! lets whipe all problems away! no problems then ;)

00:36:26 <Arnia> We do need to use ID references for accessibility reasons, and many practical reasons too

00:36:31 <eaon> this is getting silly :)

00:36:36 <mterry> eaon: Well, I don't think I'm wiping it away. I'm willing to live with no XHTML ID references.

00:36:50 <eaon> Arnia: he doesn't care about accessibility obviously, because that would break many agents anyway

00:36:55 <mterry> Arnia: OK, that's a good point.

00:36:57 <mattmcc> Well, if all you're looking for is "Does this work for my needs", then of course we can't tell you no.

00:37:05 <Arnia> Plus RDF graphs can contain multiple literals, how would you know which literal was the correct one (for an XML-only agent)

00:37:05 <mattmcc> If you're asking if it works as a general purpose solution, that's another matter.

00:37:12 <mterry> eaon: No, I do care about accessibility. What breaks accessibility now besides IDs.

00:37:24 <mterry> ?

00:37:25 <eaon> mterry: the fact that you use application/xml

00:37:30 <mterry> eaon: Really?

00:37:34 <mterry> eaon: Tell me more.

00:37:37 <eaon> lynx will try to download that

00:37:48 <DanC_mob> The rdf-on-the-outside is perfectly good xml and perfectly good rdf (I assume; haven't looked). It's not ok for the text/html media type, and it's not validator-happy. If that's what you want, then jam on.

00:37:59 <mterry> eaon: OK. Well, fair enough. But that's because lynx is dumb.

00:38:06 <eaon> haha

00:38:10 <mterry> eaon: Not because of a problem with my approach?

00:38:24 <mterry> DanC_mob: I'm sending application/xml

00:38:31 <eaon> mterry: no - application/xml != application/xhtml+xml

00:38:35 <eaon> nor is it text/html

00:38:41 <mterry> DanC_mob: I'm able to validate via xml schema and RDF schema.

00:38:52 <eaon> okay

00:38:54 <eaon> lets make a list

00:38:56 <mterry> eaon: I know, but it is an acceptable mime type for XHTML and is just more generic.

00:38:57 <eaon> - breaks ids

00:39:02 <eaon> - breaks accessibility

00:39:09 <mterry> eaon: How does it break accessibility?

00:39:11 <eaon> - breaks ie rendering (quircks mode)

00:39:23 <eaon> mterry: the id is a part of accessibility

00:39:27 <Arnia> DanC_mob: my personal question is how does a pure XML agent know where to find the appropriate literal. Surely that will require an understanding of RDF?

00:39:30 <eaon> so is browser support

00:39:41 <mterry> eaon: Well, it's namespaced.

00:39:42 <eaon> - doesn't make you able to make chunks of content more meaningful

00:39:57 <eaon> mterry: so what?

00:40:07 <mterry> Oh, I meant to address that to Arnia.

00:40:30 <mterry> eaon: OK, IDs break. But that's a general problem with XML and mixing namespaces. Is that solved by xml:id?

00:40:38 <DanC_mob> Arnia: "generic xml clients" are often xslt-happy; xslt could be used to pick out the literal

00:40:43 <eaon> mterry: not at all

00:40:48 <eaon> an xml id is not an rdf id

00:40:52 <mterry> eaon: OK, IE renders quirks and lynx doens't work (both because of UA deficiencies, not my approach)

00:40:57 <eaon> thats why you DON'T MIX SUCH THINGS :)

00:41:20 <Arnia> DanC_mob: Assuming some properties of the graph? It still feels awkward to me...

00:41:41 <mterry> eaon: OK. But the document as a whole is RDF. Or, the root is RDF. So, wouldn't rdf:ID stuff work for RDF stuff. I dont' know about XHTML ids...

00:41:45 * DanC_mob wonders what motivated mterry to explore this approach. Any particular use case?

00:42:02 <eaon> mterry: example

00:42:11 <eaon>http://niij.org/mterry.rdf#michael

00:42:11 <dc_rdfig> A: http://niij.org/mterry.rdf#michael from eaon

00:42:19 <eaon> give that to a browser

00:42:23 <eaon> xhtml browser

00:42:28 <eaon> and it will look for an xhtml id

00:42:36 <eaon> give it to an rdf app

00:42:41 <mterry> DanC_mob: Honestly, I just wanted something that was totally spec-friendly, would embed together, and would work in most UAs

00:42:41 <eaon> and it will look for the rdf id

00:42:55 <mterry> eaon: well, isn't that based on mime too?

00:42:57 <sbp`> note that eikeon tried out mterry's approach some time ago, if I remember correctly. let me search for a reference

00:43:07 <mterry> eaon: Admittadly, I'm not helping with my application/xml mime.

00:43:20 <eaon> mterry: your mime type doesn't matter

00:43:26 <eaon> mterry: ie doesn't understand rdf

00:43:35 <DanC_mob> Really, mterry? This is just an intellectual exercise? No particular rdf vocab in mind, for example?

00:43:35 <eaon> mterry: it will look for the xhtml stuff

00:43:36 <mterry> eaon: I know.

00:43:54 <mterry> eaon: yes, and that's what I want. I don't expect to make RDF-dumb applications suddenly understand RDF.

00:43:58 <eaon> mterry: so basically, you can't have any ids in the document because that will confuse agents

00:44:05 <DanC_mob> Of course the mime type matters, when it comes to fragids

00:44:09 <mterry> DanC_mob: I wanted to mix dc, foaf, etc.

00:44:15 <sbp`> can't find it

00:44:22 <mterry> DanC_mob: I wanted to mix and match all sorts of RDF namespaces.

00:44:38 <mterry> DanC_mob: All the current methods of embedding were restrictive in what vocabularies you used.

00:44:41 <sbp`> mterry: have you tried GRDDL by any chance?

00:44:44 <sbp`> .g GRDDL

00:44:47 <phenny> GRDDL: http://www.w3.org/2004/01/rdxh/spec

00:44:47 <dc_rdfig> Label GRDDL not found.

00:44:56 <mterry> sbp`: Um, I looked at it.

00:45:02 <mterry> Let me look again.

00:45:15 <eaon> mterry: i find it funny how many people are telling you more or less that your approach is unclean and doesn't work incertain situations

00:45:15 <DanC_mob> I donlt think xhtml specs say what uas do with application/xml eaon

00:45:42 <eaon> DanC_mob: what do you refer to?

00:46:09 <DanC_mob> Agents getting confused by ids

00:46:10 <mterry> eaon: Well, I'm glad I'm serving you some amusement.

00:46:33 <sbp`> there are plenty of other embedding approaches, but none which validate across HTML and XHTML to the W3C MarkUp validator's satisfaction

00:46:45 <mterry> eaon: I haven't actually seen one situation described where I fail. Some ID errors, ok. A UA that doesn't like xml, OK.

00:46:56 <eaon> DanC_mob: you mean it doesn't tell agents to use #id on application/xml? well thats fine but it still will do it probably ;)

00:47:05 <DanC_mob> eaon, pls don't use vage arguments like "how many people..."

00:47:23 <eaon> how many people?

00:47:25 <eaon> where?

00:47:40 <eaon> oh that

00:47:45 <DanC_mob> 10ish lines up

00:48:12 <eaon> then replace it with some

00:48:35 <Arnia> mterry: I take it you're guaranteeing that only one of these literals will exist in any single RDF doc?

00:48:40 <eaon> well, anyway, it's not made to get mixed at the moment

00:48:53 <eaon> if you want to, fine, there will be some problems in some cases

00:48:59 <eaon> anyway

00:49:09 <mterry> Arnia: I'm not sure. That's part of my question. What is the semantic meaning of my approach?

00:49:18 <eaon> this is more or less a discussion that starts and ends at the same point at the moment ;)

00:50:13 <Arnia> mterry: You've just created a triple that has an XML fragment as the object

00:50:19 <eaon> so i'm out of here - it's already quite late here

00:50:25 <DanC_mob> I find it counter-intuitive to meet a goal of working with html consumers by using rdf on the outside, but I cannot dispute that it works in many popular agents.

00:50:27 <eaon> good night guys

00:50:52 <mterry> Arnia: OK. Good... So, how could I make that make sense in RDF? How could I say, here is some data, that may be interepreted as XHTML?

00:50:58 <Arnia> mterry: What the subject and predicate are is up to you

00:51:05 <CloCkWeRX> XSLT :P

00:51:25 <mterry> Arnia: yeah. So, what subject and predicate are best?

00:51:26 * CloCkWeRX be's annoying and pretends to miss the point

00:51:27 <CloCkWeRX> :)

00:51:29 <mterry> Ok, about XSLT.

00:51:38 <mterry> And GRDDL?

00:51:58 <DanC_mob> Use subject of "", I.e. This doc, and make up a property for this purpose; e.g. representationForHumans

00:52:07 <mterry> Does that require UA to know about XHTML?

00:52:35 <DanC_mob> Yes and no...

00:52:40 <mterry> DanC_mob: OK, is a UA that only understands XHTML going to be parsing this in spec mode and understand it?

00:53:02 <mterry> DanC_mob: Like, am I relying on non-spec behavior to have XHTML agents parse the literal as XHTML?

00:53:15 <DanC_mob> Grddl works with xhtml documents, which are expected to be grokked by xhtml uas. But it also works with any other sort of xml

00:53:35 <DanC_mob> Non-spec: yes, completely

00:54:09 <mterry> DanC_mob: poops.

00:54:12 <sbp`> not only non spec, it's *counter* spec

00:54:24 <mterry> sbp: Really? It's just a literal.

00:54:25 <sbp`> user agents should dispatch off of the root element's namespace

00:54:31 <sbp`> which in this case is RDF

00:54:41 <DanC_mob> Xhtml agents aren't specified to do *anything* with application/xml nor docs whose root is not <html>

00:54:45 <MarkB> sbp`: bzzt!

00:54:49 <sbp`> so your document is an RDF document, and XHTML UAs should say "I do not understand this"

00:54:57 <sbp`> after XML functions

00:55:29 <DanC_mob> Xhtml ua might also be an xml us, sbp. Should is too strong. May.

00:55:43 <DanC_mob> s/us/ua/

00:55:53 <sbp`> but an XML UA wouldn't render it as XHTML

00:56:04 <DanC_mob> Why not?

00:56:12 <sbp`> because it's not XHTML

00:56:21 <DanC_mob> A stylesheet pi would be more spec-happy

00:56:34 <DanC_mob> The embedded stuff us xhtml

00:57:04 <sbp`> if I hand you an unidentified XML dialect with a piece of XHTML embedded in it, you shouldn't display it as XHTML: you don't know what the surrounding context is

00:57:06 <mterry> OK. So, I link to the standard XHTML stylesheet.

00:57:29 <mterry> Does that make it better?

00:57:30 <DanC_mob> Afaik, there's no prohibition on xml uas doing anything they like with xml, including hunting inside for xhtml. I don't recommend it, but I'm not aware of any spec that prohibits it

00:57:33 <sbp`> "standard XHTML stylesheet"? the one that you created?

00:57:38 <mterry> sbp: Really? I thought they were supposed to ignore it.

00:57:56 <mterry> sbp: No, the one that comes with the spec. The one that describes default styles.

00:58:20 <sbp`> isn't that a CSS stylesheet?

00:58:26 <mterry> sbp: sure.

00:58:34 <sbp`> that doesn't define behaviours for elements, only styles

00:58:42 <sbp`> welcome to inactive links

00:58:51 <DanC_mob> Easy enough to cook up peekInsideForXhtml.xsl

00:59:15 <mterry> sbp: oh, true.

00:59:18 <sbp`> as for ignoring content, XHTML UAs should ignore elements they don't understand inside *XHTML* documents, sure

00:59:33 <mterry> sbp: OK, but that's not a general XML thing?

00:59:42 <mterry> sbp: Or RDF/XLL thing?

00:59:50 <sbp`> nope: other XML applications can mandate other behaviours

01:00:13 <DanC_mob> Sbp, there are no (strictly conforming) xhtml docs with unknown elements.

01:00:22 <mterry> sbp: OK, does RDF talk about it?

01:00:29 <sbp`> DanC: if I created an XML application that said "you must ignore all XHTML in instance documents", and then you used your peekInsideForXhtml.xsl on an instance of that application, it would not be conforming

01:00:39 <mterry> DanC_mob: That's a point. I'm not strictly conforming. But I am conforming, I think.

01:00:52 <DanC_mob> Right, sbp. So?

01:01:01 <sbp`> DanC: right. that's just a notice about user agent fallback in HTML 4.01

01:01:21 <sbp`> so that's answering what mterry is asking: does it conform to specification

01:01:51 <DanC_mob> You're conforming to xml and to rdf specs, but not to xhtml spec, mterry. Ironic, since your target is html browsers

01:02:11 <mterry> DanC_mob: OK. I could live with that.

01:02:24 <DanC_mob> Sbp, thatls why I said I don't recommend peeking inside.

01:02:50 <mterry> DanC_mob: Am I really breaking XHTML? Would an XML UA that knows XHTML make sense of what I'm doing, or does the parseType="Literal" mess that up?

01:02:59 <sbp`> I agree and second your unrecommendation

01:03:24 <DanC_mob> mterry, just realize you're depending on stuff that standards don't guarantee and (imo) are not likely to ever guarantee.

01:04:10 <mterry> DanC_mob: OK. But surely XHTML inside of XML makes sense even if it isn't strictly conforming.

01:04:18 <DanC_mob> An xhtml us is only required to deal with strictly conforming xhtml docs. As soon as it seed <rdf>, all its obligations are gone.

01:05:00 <mterry> DanC_mob: Yeah...

01:05:13 <DanC_mob> "Makes sense" is an aesthetic judgement, in this case (I.e. Lacking documented community consensus). It does not appeal to me.

01:05:15 <MarkB> don't think so, DanC_mob ....

01:06:31 <sbp`> aren't we looping a bit here? XHTML UA shouldn't do anything with RDF/XML -> UA might be a generic XML UA -> generic XML UA has undefined behaviour -> behaviour is defined by RDF/XML -> rendering XHTML behaviour is fine for XML UA then -> but wait, the targets UA are XHTML UAs...

01:06:44 <MarkB> xhtml 1.0 permits non-validating processors which only test for well-formedness

01:06:52 <sbp`> s/targets UA/target UAs/

01:07:02 <MarkB> http://www.w3.org/TR/2002/REC-xhtml1-20020801/#uaconf

01:07:24 <DanC_mob> Excerpt, pls?

01:07:27 <sbp`> it does not permit them at all

01:07:30 <DanC_mob> Low bandwidth here

01:07:43 <sbp`> #uaconf simply states some conformance constraints

01:08:17 <mterry> Actually, my target could be an XML UA that knows XHTML.

01:08:24 <mterry> In that case, is it right?

01:08:31 <DanC_mob> I'm pretty sure xhtml ua behaviour is only specified in the case of strictly conforming docs. (Sigh)

01:08:34 <sbp`> the only normatively conformant XHTML document are those that conform to all of the normative conformance requirements in XHTML 1.0, which includes the dreaded Appendix C, and DTD validation

01:08:44 <sbp`> DanC_mob: right

01:08:45 <mterry> DanC_mob: hmm...

01:09:11 <sbp`> which is a PITA, but it's in the spec.

01:09:19 <MarkB> In order to be consistent with the XML 1.0 Recommendation [XML], the user agent must parse and evaluate an XHTML document for well-formedness. If the user agent claims to be a validating user agent, it must also validate documents against their referenced DTDs according to [XML].

01:09:21 <mterry> DanC_mob: OK, so I'm relying on slightly out-of-spec behavior? Or maybe not-so-slight.

01:09:31 <MarkB> (from #uaconf)

01:10:30 <DanC_mob> Mark, but whether the ua dtd-validates or not, it's only obliged to do anything in particular with strictly conforming docs

01:10:32 <DanC_mob> No?

01:10:44 <MarkB> don't think so ... more ...

01:10:46 <sbp`> non-validated XHTML documents are not Strictly Conforming XHTML Documents in the sense of section 3.1.1 of the same recommendation

01:11:07 <MarkB> If a user agent encounters an element it does not recognize, it must process the element's content.

01:11:13 <MarkB> (also from #uaconf)

01:11:34 <DanC_mob> Mterry, if you used an <?xml-stylesheet?> thingy with peekInside.xsl, you'd be in line with 3 w3c recs and ie.

01:11:36 <mterry> Great. So, XML stuff that understands XHTML but not RDF should be fine, according to spec?

01:11:41 <sbp`> what does "process" mean in that context?

01:11:44 <mterry> But if it understands both? Maybe not.

01:11:49 <mterry> DanC_mob: Ok.

01:11:49 <DanC_mob> But I still wouldn't like it :)

01:12:09 <mterry> DanC_mob: But how does peekInside.xsl know if it is a RDF bot or an XHTML browser?

01:12:22 <CloCkWeRX> mterry, it doesn't give a damn

01:12:41 <CloCkWeRX> the browser requests it and transforms it

01:12:44 <CloCkWeRX> the bot doesn't

01:13:10 <mterry> CloCkWeRx: So, what does peekinside do? Does it just take out the XHTML? In that case, how is the RDF ever referenced?

01:13:15 <DanC_mob> PeekInside assumes its reader is rendering for humaan consumption (dunno if the stylesheet PI spec makes that explicit or if this would be greymarket)

01:13:33 <MarkB> sbp; can't find a ref, but I can only see it meaning one thing; "don't ignore"

01:13:55 <mterry> DanC_mob: grey market, I'm pretty sure.

01:14:03 <mterry> DanC_mob: Any UA may follow the PI.

01:14:21 <DanC_mob> Peekinside could look for a specific rdf property, forHumanConsumption, applied to ""

01:15:37 * DanC_mob kinda likes that idea; wonders if sbp understands it well enough to explore in detail

01:15:59 <DanC_mob> Or asHTML

01:16:15 <mterry> Where is that property set?

01:16:17 <DanC_mob> Or htmlAlt

01:16:31 <DanC_mob> In the enclosing rdf

01:17:30 <mterry> OK. But how do we stop the peekinside from going off if an RDF bot comes?

01:17:35 <DanC_mob> <> mterry:htmlRepresentation """<html>...."""^^XML

01:18:40 <DanC_mob> I expect rdf bots to ignore stylesheet pis.

01:18:59 <MarkB> DanC_mob, not in Ottawa yet?

01:19:19 <DanC_mob> I thiby way of precedent, I'm pretty sure xml schema consumers skip stylesheet pis

01:19:19 <mterry> DanC_mob: I don't think we can assume that.

01:19:52 <DanC_mob> In ORD. Flight to YOW delayed a bit. Eta was 10:33pET. Dunno now

01:20:04 <MarkB> ick

01:20:37 <DanC_mob> Not too bad

01:21:58 <MarkB> oh, you mean eta @ yow for 10:33? thought you meant departure

01:22:04 <DanC_mob> Boarding...

01:22:20 <MarkB> have a good trip

01:22:28 <MarkB> p.s. big volleyball tourney at moonie's bay this week

01:22:41 <Arnia> Have a good journey

01:22:46 <mterry> I think if this doesn't fly, I might as well use <link>. :-/ Thank you everyone for the discussion. I appreciate your time and effort.

01:26:15 <DanC_mob> Moonie's bay? Wherezat?

01:26:27 <MarkB> beech just south of downtown

01:26:37 <DanC_mob> In ottawa?

01:26:41 <MarkB> yup

01:27:00 <MarkB> 10 min cab ride down the canal

01:27:03 <MarkB> from downtown

01:27:24 <DanC_mob> Ooh! Ooh! Details?

01:27:50 <MarkB> i'll see what I can dig up ...

01:28:06 <MarkB> just a local tourney

01:28:35 * DanC_mob wonders about walking on... To heck with TAG mtg :)

01:28:50 <MarkB> p.s. "mooney" - always get that wrong

01:28:52 <MarkB> heh

01:28:55 <DanC_mob> 2s? 6s?

01:29:26 <MarkB> DanC_mob not sure. matches are usually 6 on 6 on weeknights

01:30:00 * DanC_mob prefers 2s

01:30:26 <MarkB> there were some 2s

01:30:43 <MarkB> probably 20 courts(?) or so

01:31:37 <DanC_mob> Gerald talks about some huge vball tourney... I wonder if that's in Ottawa... I thought it already happened this year

01:32:38 <DanC_mob> Hope...

01:32:48 <MarkB> hope beech, yah, happened last month

01:32:55 <MarkB> that's at mooney's bay

01:33:02 <MarkB> http://www.geocities.com/dw2345/Beach.htm

01:33:10 <MarkB> hmm, doesn't mention it

01:33:37 <DanC_mob> . http://impressive.net/people/gerald/2001/07/07/hope.html

01:34:35 <DanC_mob> Sbp? Thoughts on "" :htmlAlt idea?

01:36:05 * DanC_mob invites mark to read about my blogworthy friday http://dm93.org/z2001/ShareLifeThruArt

01:38:32 <DanC_mob> They just made the exit row blurb... I wonder if that's a residue from a union debate. Flight attendants wanted the roomy seats, not the little jump seats. ;-)

01:38:36 <MarkB> sweet night. my friday was take out swiss chalet on the beech at mooney's bay. my 3 yr old pretty much requires retiring 8ish 8-)

01:39:26 <DanC_mob> Beech... Sounds nice too.

01:39:42 <DanC_mob> 8ish bed times can be good too!

01:40:15 <MarkB> for mom & dad 8-)

01:40:34 * DanC_mob wonder what software markb uses to share his thoughts with the world...

01:41:11 <DanC_mob> Wordpress is tempting

01:41:13 * MarkB doesn't share much in the way of thoughts, though wondered if setting up a personal Wiki DanC-style would help; http://www.markbaker.ca/mbwiki/

01:42:12 <DanC_mob> No? I see a 7aug blog entry

01:42:41 <MarkB> those aren't of the personal type, which is what I thought you meant

01:42:47 <sh1mmer> DanC is flashmobbing?

01:43:08 <DanC_mob> Flashmwhat?

01:43:19 <sh1mmer> .g flash mob

01:43:22 <phenny> flash mob: http://www.flashmob.com/

01:43:25 <DanC_mob> Hi tom

01:43:30 <sh1mmer> Hiya :)

01:44:20 <sh1mmer> Flashmobs are where you get a bunch of random people to congregate and do something random by using phone/sms/mms/email etc

01:44:32 <DanC_mob> "Gizmos must be put awayk msg here.

01:44:35 <sh1mmer> but all really at the drop of a hat

01:45:16 <sh1mmer> DanC_mob hows it going anyway? How is the clan?

02:22:20 <eaon> eaon is now known as eaon|zZz

04:08:14 <DanC_mob> .time

04:08:59 <DanC_mob> .time

04:09:03 <DanC_mob> .t

04:09:05 <phenny> Mon, 09 Aug 2004 04:09:03 GMT

04:09:17 <DanC_mob> DanC_mob is now known as DanC_blog

04:09:37 <DanC_blog> DanC_blog is now known as DanC_mob

04:10:50 <DanC_mob> Well... That was... interesting. Got the 3rd degree from canadian immigration. The had access to some file that let them know to check for a scar on my left wrist!

04:14:28 <mmealling> fairly spooky

04:34:11 <karlcow>http://www.la-grange.net/2004/08/05#rdf-poete

04:34:13 <dc_rdfig> B: http://www.la-grange.net/2004/08/05#rdf-poete from karlcow

04:35:18 <karlcow> B:|RDF For Poets

04:35:18 <dc_rdfig> Titled item B.

04:35:24 <karlcow> B: (in French) a very basic introduction to the notion of graph and RDF for everyone without technical knowledge.

04:35:24 <dc_rdfig> Added comment B1.

06:22:55 * jdougan is away: Later, eh.

08:10:07 <eaon|zZz> eaon|zZz is now known as eaon

12:03:39 <danbri>http://danbri.org/words/skos.phps

12:03:41 <dc_rdfig> C: http://danbri.org/words/skos.phps from danbri

12:03:54 <danbri> C:|Latest skos.phps source (for Wordpress blog add-on)

12:03:54 <dc_rdfig> Titled item C.

12:04:14 <danbri> C:See [http://rdfig.xmlhack.com/2004/08/08/2004-08-08.html#1092002409.958721|yesterday's notes]. Now with TODO list.

12:04:14 <dc_rdfig> Added comment C1.

12:27:25 <MacIntire> danbri - have you chatted with [GNU-] and mortenf about all their fOAF Wordpress work?

12:28:05 <danbri> w/ mortenf, yeah. I test-drove an early version, and am running it.

12:28:52 <MacIntire> ah cool

12:29:18 <danbri> C:I'd like to be able to cross-link between Wordpress/SKOS category scheme and Wordpress/FOAF person descriptions, so a crawler could scoop up all your friends, contacts etc categories. And then hack the RSS1 generator to allow clients to request items only under certain categories based on this.

12:29:18 <dc_rdfig> Added comment C2.

12:29:19 <MacIntire> You might want to create a page for your work on the Wordpress wiki

12:29:51 <danbri> good idea. am only just getting a feel for what shape this could take. probably be good bundled w/ the foaf thing rather than standalone, if mortenf sympathetic.

12:29:57 <danbri> just a proof of concept for now

12:30:16 <MacIntire> right....but even if it's just a seeAlso....so we know it's under development ;)

12:30:21 <danbri> C:If anyone has a Wordpress blog and wants to test this, might be a good way to get some experimental data...

12:30:21 <dc_rdfig> Added comment C3.

12:30:43 <danbri> C:TODO: link from Wordpress wiki somehow

12:30:43 <dc_rdfig> Added comment C4.

12:30:47 <MacIntire> I'm still waiting for an easy way to assign relationships and add all my friends. ;)

12:30:53 <MacIntire> and add their trust levels

12:31:48 <crschmidt> Danbri: just put that file in my root wp directory, i guess?

12:31:56 <danbri> yep

12:32:11 <danbri> It takes no args, so should be reasonablyish secure

12:32:19 <danbri> It is my first PHP script though ;)

12:33:38 <crschmidt>http://crschmidt.net/wordpress/skos.php

12:33:38 <dc_rdfig> D: http://crschmidt.net/wordpress/skos.php from crschmidt

12:34:01 <crschmidt> D:| Example output for Danbri's skos plugin

12:34:01 <dc_rdfig> Titled item D.

12:34:06 <crschmidt> logger, chump d

12:34:06 <crschmidt> D:See [http://www.ilrt.bris.ac.uk/discovery/chatlogs/rdfig/2004-08-09#T12-34-06|discussion]

12:34:07 <dc_rdfig> Added comment D1.

12:34:25 <danbri> C:More sample data: [http://crschmidt.net/wordpress/skos.php|crschmidt's data].

12:34:26 <dc_rdfig> Added comment C5.

12:34:33 <danbri> thanks :) just added it to C as well for easy finding

12:34:42 * crschmidt nods

12:34:51 <crschmidt> I always forget that I'm going to be chumping things :)

12:35:06 <danbri> C:Now we have two sample data category schemes, how can we map between them?

12:35:06 <dc_rdfig> Added comment C6.

12:35:13 <MacIntire> C: More sample data: [http://www.brain-stream.org/wordpress/skos.php|bkdelons's data].

12:35:13 <dc_rdfig> Added comment C7.

12:35:26 <MacIntire> damn. Can't even spell my own nick

12:35:57 <crschmidt> C7: More sample data: [http://www.brain-stream.org/wordpress/skos.php|bkdelongs's data].

12:35:58 <dc_rdfig> Replaced comment C7.

12:36:10 <MacIntire> thx

12:36:17 <MacIntire> So...what's skos?

12:36:30 <danbri> C:OK danbri:c8 is 'technology', Chris's c5 is 'Technical', bkdelong's c11 is 'Technology'. Heh, who'da thunk it? ;)

12:36:30 <dc_rdfig> Added comment C8.

12:37:38 <crschmidt> .g SKOS RDF

12:37:41 <phenny> SKOS RDF: http://staff.oclc.org/~vizine/GSAFD/SKOS/

12:37:44 <crschmidt> hm.

12:38:15 <danbri> C:[http://www.w3.org/2004/02/skos/|SKOS] is the latest in a long line of RDF vocabularies for representing thesaurus (etc) data., ie. webs of linked concepts. It is a scruffier form of KR than formal ontologists, and maps well both to digital library systems and to simple home-grown hierarchical classifications such as used in bookmarks/weblogs etc.

12:38:16 <dc_rdfig> Added comment C9.

12:38:57 <MacIntire> ah. nifty

12:39:12 <crschmidt> hm. *goes to look at the del.icio.us api*

12:39:21 <MacIntire> Wonder if we could do something with delicious data. I have such a messt bit o categories

12:39:49 <danbri> C:I've been [http://www.w3.org/TandS/QL/QL98/pp/queryservice.html|goofing around] with such structures since I got into RDF, and SKOS is definitely the most polished. It started in SWAD-Europe project and is finding a home in W3C Semantic Web Best Practices Working Group lately.

12:39:50 <dc_rdfig> Added comment C10.

12:40:50 <crschmidt> Does localID need to be a number?

12:41:01 <danbri> C10:I've been [http://www.w3.org/TandS/QL/QL98/pp/queryservice.html|goofing around] with such structures since I got into RDF, as have various other folk. SKOS is definitely the most polished piece of work in this tradition. It started in SWAD-Europe project and is finding a home in W3C Semantic Web Best Practices Working Group lately.

12:41:01 <dc_rdfig> Replaced comment C10.

12:41:13 <MacIntire> eww.....633 categories in delicious.

12:41:24 * danbri tweaked as looked like i was taking credit; Alistair did most of the work on SKOS + folks at ILRT

12:41:41 <danbri> a delicious category dump would be great

12:42:00 <crschmidt> Example delicious data:

12:42:00 <crschmidt> <tags user="crschmidt">

12:42:01 <crschmidt> <tag count="1" tag="3650" />

12:42:01 <crschmidt> <tag count="3" tag="amusing" />

12:42:01 <crschmidt> <tag count="2" tag="apple" />

12:42:13 <DaveP> danbri: Note, links from http://www.w3.org/2004/02/skos/ are 404? Is it your page?

12:42:50 <crschmidt> so, just a skos:Concept; skos:localId "3650" ; skos:prefLabel "3650" .

12:43:05 <danbri> a textual label would be more helpful

12:43:45 <crschmidt> Like wordpress's "description"?

12:44:18 <danbri> I think http://www.w3.org/2004/02/skos/ is still being worked on (by Alistair), he may be moving the guide/ and spec/ into place. I'll ask.

12:44:39 <danbri> skos allows a prefLabel, a definition and a scopeNote

12:44:54 <danbri> my wordpress blurbs were scruffy so I made them into scopeNotes; didn't record definitions...

12:45:06 <crschmidt> with the example data I just offered, what other data/tags should/can we record from that?

12:45:33 <danbri> C:Sample data: [http://danbri.org/words/skos.php|danbri's categories].

12:45:34 <dc_rdfig> Added comment C11.

12:46:08 <danbri> what is '3650'? ah a kind of phone. Ok makes more sense now.

12:46:15 <crschmidt> oh, heh

12:46:16 <crschmidt> sorry :)

12:46:27 <danbri> just make the tags be prefLabel and omit scopeNote unless there's a longer text field in there somewhere

12:46:34 <danbri> I misunderstood the int :)

12:46:36 <crschmidt> No longer text field

12:46:38 <crschmidt> yep

12:46:47 <crschmidt> is having a localId being the same as the name okay?

12:48:07 <danbri> sure

12:48:25 <danbri> its the identifier within a system, bit like the local bit in a qname i guess

12:51:38 <crschmidt>http://crschmidt.net/delicious.skos

12:51:39 <dc_rdfig> E: http://crschmidt.net/delicious.skos from crschmidt

12:51:55 <crschmidt> E:| Del.Icio.Us Dump

12:51:55 <dc_rdfig> Titled item E.

12:52:02 <crschmidt> E: Working towards creating a script for this

12:52:03 <dc_rdfig> Added comment E1.

12:52:14 <crschmidt> What are the _:c$foo again?

12:52:20 <crschmidt> I think those are qname things?

12:52:51 <crschmidt> Okay, so it's a bnode

12:53:05 <crschmidt> ( n3 tutorial in 20 lines, by sbp, http://miscoranda.com/109 )

12:54:09 <crschmidt> Okay, what's a bnode? :)

12:55:21 <danbri> its a node in an RDF graph that doesn't have a URI, and so is only understood by its relatinoships to other things

12:55:28 <crschmidt> Hm.

12:55:40 <crschmidt> Delicious tags actually have URIs though, right?

12:55:48 <crschmidt> So I don't need to treat them as bnodes?

12:55:55 <MacIntire> crschmidt -did you autogenerate your delicious categories?

12:56:13 <crschmidt> MacIntire: The api lets you dump them, http://del.icio.us/api/tags/get

12:56:17 <danbri> chris, can you add the N3 header stuff to delicious.skos too?

12:56:18 <MacIntire> cause I have 633 and hand munging it would suck.

12:56:31 <crschmidt> E: Delicious tag dump at http://del.icio.us/api/tags/get

12:56:32 <dc_rdfig> Added comment E2.

12:56:50 <MacIntire> ooh. nice

12:56:56 <crschmidt> danbri: done

12:57:09 <danbri> ta :)

12:57:31 <MacIntire> ok...next question crschmidt - got something I can use to transform the output to skos ? ;)

12:57:37 <crschmidt> cat example | sed -e 's/<tag.*tag="\(.*\)".*$/a skos:Concept; skos:localId "\1" ; skos:prefLabel "\1"; ./' | grep -v "/tags"| less | tail -n 40 > /home/httpd/htdocs/delicious.skos

12:57:42 <danbri> also something like "_:c6 a "

12:58:03 <crschmidt> danbri: i've got the a in there, what i was wondering is if I can use a URI rather than a bnode

12:58:12 <danbri> ah right

12:58:26 <danbri> yeah would be good. Prolly also for skos.php too

12:58:27 <[GNU]> crschmidt: are you aware that the delicious url is only avail to delicious developers?

12:59:04 <crschmidt> [GNU]: It's in the delicious api docs at http://del.icio.us/doc/api

12:59:20 <crschmidt> [GNU]: it's "logged in users only", but that doesn't seem too difficult?

12:59:42 <[GNU]> ne, not to diffucult... but may confuse ppl

12:59:55 <crschmidt> [GNU]: well, it asks you to login when you go there ;)

13:00:16 <MacIntire>http://www.brain-stream.org/delicious.skos

13:00:17 <dc_rdfig> F: http://www.brain-stream.org/delicious.skos from MacIntire

13:00:21 <crschmidt> MacIntire: if you do use that, know that the "40" is just to get rid of the top two lines

13:00:23 <[GNU]> and if i cancel no link to the singup :(

13:00:47 <crschmidt> so you should change it to like, 634, probably

13:00:51 <MacIntire> hehehe

13:01:05 <MacIntire> should have it do a wc of the file to determine how to nix the top two lines

13:01:33 <crschmidt> MacIntire: no, i should have it read into a perl script and nix them in a better way, you wanted it now though, not in a couple hours when I get enough time to play with it ;)

13:02:39 <MacIntire> hehe

13:03:01 <crschmidt> danbri: have added URIs at the beginning

13:04:21 <crschmidt> E: Okay, this should really use URIs - I've added them, using the general delicious tags so they aren't user specific

13:04:22 <dc_rdfig> Added comment E3.

13:04:40 <crschmidt> E: I'll write a script for this later, right now I'm just doing a quick command line hack to generate it

13:04:40 <dc_rdfig> Added comment E4.

13:05:25 <crschmidt> E: However, by using the general del.icio.us tag, we ensure that the categories will be smushable, so all the delicious skos we see can be aggregated to generate a giant list of the tags used on delicious in rdf form

13:05:25 <dc_rdfig> Added comment E5.

13:06:02 <crschmidt> E: By using the general del.icio.us tag, we ensure that the categories will be smushable, so all the delicious skos we see can be aggregated to generate a giant list of the tags used on delicious in rdf form

13:06:02 <dc_rdfig> Added comment E6.

13:06:11 <crschmidt> E6: ""

13:06:11 <dc_rdfig> Deleted comment E6.

13:06:22 <crschmidt> E5: By using the general del.icio.us tag, we ensure that the categories will be smushable, so all the delicious skos we see can be aggregated to generate a giant list of the tags used on delicious in rdf form

13:06:22 <dc_rdfig> Replaced comment E5.

13:08:38 <crschmidt> E: Thoughts appreciated

13:08:38 <dc_rdfig> Added comment E6.

13:09:11 <crschmidt> MacIntire: You should add the namespace URIs to the top of your delicious dump

13:14:33 <MacIntire> oh right....what should I put at the top/bottom of the file, crschmidt?

13:14:52 <crschmidt> bottom is fine as it is

13:15:07 <crschmidt> Top should be:

13:15:08 <crschmidt> @prefix rdfs: <http://www.w3.org/2000/01/rdf-schema#> .

13:15:08 <crschmidt> @prefix skos: <http://www.w3.org/2004/02/skos/core#> .

13:15:20 <crschmidt> I"m not sure if the rdfs schema is even needed, but danbri has it in there

13:16:12 <MacIntire> gotcha...done

13:17:28 <crschmidt> you might also want to add <http:\/\/del.icio.us\/tag\/\1> before "a skos:Concept" in the sed thingy

13:17:38 <crschmidt> that way youy can attach the skos info to URIs

13:17:54 <crschmidt> Although, to be honest, I'm not sure if that's the right way to do it

13:18:03 <crschmidt> so you might want to wait until an n3 wizard or something comes back :)

13:18:10 <MacIntire> I'll hold off until you do a little more hacking ;)

13:18:32 <MacIntire> Did you setup a blog entry or wiki thing with this anywhere I can delicious for future reference? ;)

13:18:56 <crschmidt> logger, pointer

13:18:56 <crschmidt> See http://www.ilrt.bris.ac.uk/discovery/chatlogs/rdfig/2004-08-09#T13-18-56

13:19:00 * crschmidt points :)

13:19:24 <crschmidt> hm. that seems to be the text version, with no anchors

13:19:32 <crschmidt>http://www.ilrt.bris.ac.uk/discovery/chatlogs/rdfig/2004-08-09.html#T12-41-41

13:19:33 <dc_rdfig> G: http://www.ilrt.bris.ac.uk/discovery/chatlogs/rdfig/2004-08-09.html#T12-41-41 from crschmidt

13:19:40 <crschmidt> Bah.

13:19:47 <crschmidt> I didn't mean to do that :P

13:20:04 <libby> just title it, all will be well

13:20:22 <crschmidt> G:| Discussion of SKOS del.icio.us data

13:20:23 <dc_rdfig> Titled item G.

13:23:41 <libby> ta

13:35:36 <danbri> <crschmidt> I"m not sure if the rdfs schema is even needed, but danbri has it in there

13:35:47 <danbri> rdfs bit not needed; i did use rdfs:label previously...

13:36:05 <crschmidt> danbri: none of the skos output is valid ntriples, according to raptor/rapper

13:37:01 <lyle> lyle is now known as lyle|afk

13:37:51 <danbri> danbri@fireball:~/Desktop$ rapper -i turtle http://danbri.org/words/skos.php -o ntriples

13:37:57 <danbri> works...

13:38:03 <crschmidt> oh

13:38:07 <crschmidt> it's turtle, hh

13:38:14 <danbri> python cwm.py http://danbri.org/words/skos.php --rdf

13:38:17 <danbri> ok too

13:38:38 <crschmidt> yeah, i was doing -i ntriples

13:39:01 <crschmidt> I misunderstood what I was generating :)

14:06:47 <lyle|afk> lyle|afk is now known as lyle

14:22:37 <crschmidt> is there a way to say "This URL is a user specific version of a general URL?"

14:22:52 <crschmidt> del.icio.us tags have a generic form - /tag/tagname

14:22:56 <crschmidt> as well as a user form:

14:22:58 <crschmidt> /username/tagname

14:23:19 <crschmidt> I want to be able to describe both of those, and to say /crschmidt/3650 is a subset of /tag/3650

14:23:41 <jsled> owl:sameAs?

14:24:10 <jsled> which says neither of those things, fwiw. :)

14:24:45 <crschmidt> eh, sameAs doesn't seem like what i want

14:34:46 <bengee> maybe </crschmidt/3650> <skos:narrowerPartitive> </tag/3650> ?

14:34:54 * bengee not sure about skos semantics, just an idea...

14:40:06 <crschmidt> Sounds right to me!

14:43:27 <crschmidt> E: Now includes "narrowerInstantive" with user-specific info

14:43:27 <dc_rdfig> Added comment E7.

14:52:36 <danbri> hmm not sure re tags

14:53:05 <danbri> in del.icio.us, is /crschmidt/3650 always a specialization of /tag/3650 ?

14:53:15 <crschmidt> yep

14:53:21 <crschmidt> a narrowerinstantive

14:53:24 <danbri> or could the two '3650s' be coincidentally?

14:53:26 <danbri> ok great

15:18:13 <crschmidt> I'm liking the idea of getting a lot of people to dump delicious skos data, then creating a database of it. you could then build a browser that allowed you to browse tags -> people -> tags -> people

15:18:45 <crschmidt> crschmidt -> amusing -> drumm, crschmidt -> drumm -> blogging -> ...

15:21:59 <danbri> and match against wordpress installs

15:22:05 <danbri> anyone done a tags plugin for wp?

15:22:15 <crschmidt> well, aren't categories basically tags?

15:22:25 * crschmidt maybe confused

15:22:30 <danbri> bbiab sorry

17:42:42 * sanctius is away: ?_ absent pour le moment _?

17:53:03 * CLoCKWeRX hrms and considers the usefulness of a story related ontology...

18:17:20 <balbinus> balbinus is now known as stork

19:37:46 <golbeck_> golbeck_ is now known as golbeck

19:58:35 <ubu> wow, look at all the people. <oldandintheway>I remember when rdfig only drew a few hardcores</oldandintheway>

20:00:17 <crschmidt> ah, how things have changed

20:01:30 <ubu> indeed

20:01:47 <dajobe> my favourite subject http://www.ilrt.bris.ac.uk/discovery/chatlogs/rdfig/stats.html

20:03:47 <ubu> so, getting to the point... i'm looking for a grammar to mix in with FOAF that provides more fine-grained relationship definitions

20:04:10 <crschmidt>http://www.perceive.net/schemas/20021119/relationship/

20:04:12 <dc_rdfig> H: http://www.perceive.net/schemas/20021119/relationship/ from crschmidt

20:04:13 <crschmidt> Something along the lines of that?

20:04:24 <crschmidt> H:| Relationship schema

20:04:24 <dc_rdfig> Titled item H.

20:04:29 <crschmidt> H: Eric Vitiello

20:04:30 <dc_rdfig> Added comment H1.

20:04:39 <crschmidt> Hm, I'm thinking that's not the most recent one

20:05:10 <crschmidt> Ah, yes

20:05:17 <ubu> yeah, i've seen that one, but i've also seen one that is far more detailed

20:05:22 <crschmidt> H: Replaced by the [http://purl.org/vocab/relationship/ newer schema]

20:05:22 <dc_rdfig> Added comment H2.

20:05:40 <crschmidt> H: "Added the following properties: lostContactWith, knowsOf, wouldLikeToKnow , knowsInPassing, knowsByReputation, closeFriendOf, hasMet, worksWith, colleagueOf , collaboratesWith, employerOf, employedBy, mentorOf, apprenticeTo, livesWith, neighborOf , grandparentOf, lifePartnerOf, engagedTo, ancestorOf, descendantOf, participantIn, participant"

20:05:40 <dc_rdfig> Added comment H3.

20:06:42 <ubu> does the expanded version have any real traction (i.e is anyone actually *using* it) ?

20:07:32 <ubu> it would suit my app just fine but i don't want to create a hedge

20:07:36 <crschmidt> I believe so, although to be honest, the best person I know to ask that question to would be Julian Bond, as he's the one who talks most commonloy about having to deal with multiple revisions of the schema

20:08:34 <ubu> ah, thanks for the tip

22:15:11 <eaon> eaon is now known as eaon|zZz

22:42:57 <danbri> dc_rdfig:view?

22:42:58 <dc_rdfig> D: Example output for Danbri's skos plugin (http://crschmidt.net/wordpress/skos.php)

22:42:59 <dc_rdfig> E: Del.Icio.Us Dump (http://crschmidt.net/delicious.skos)

22:43:00 <dc_rdfig> F: http://www.brain-stream.org/delicious.skos

22:43:01 <dc_rdfig> G: Discussion of SKOS del.icio.us data (http://www.ilrt.bris.ac.uk/discovery/chatlogs/rdfig/2004-08-09.html#T12-41-41)

22:43:02 <dc_rdfig> H: Relationship schema (http://www.perceive.net/schemas/20021119/relationship/)

22:45:39 <danbri> C:I updated [http://danbri.org/words/skos.phps|the code] a bit, and added a few more notes etc., fixed some bugs. Next stop: RDF/XML?

22:45:40 <dc_rdfig> Added comment C12.

23:06:20 <crschmidt> E: See also [http://crschmidt.net/delicious.skos.txt bash script] for generating skos data from del.icio.us tags

23:06:20 <dc_rdfig> Added comment E8.

23:06:26 <jsled> crschmidt: hey ... it occurs to me that narrowerInstantive might not be what you want.

23:06:29 <jsled> heh

23:06:33 <crschmidt> jsled: heh

23:06:52 <crschmidt> jsled: I'm all ears.

23:08:10 <jsled> well, </user/jsled/gnucash> isn't an narrower instance of </tag/gnucash>.

23:09:38 <jsled> it seems partative.

23:11:15 <crschmidt> I suppose I may not understand what you mean.

23:11:22 <crschmidt> /tag/gnucash has the full set of posts

23:11:26 <crschmidt> oh, hm

23:11:31 <crschmidt> maybe i do understand

23:11:33 <jsled> is

23:11:45 <crschmidt> the concepts themselves are the same

23:12:08 <crschmidt> the content contained may be a smaller subset, but I'm not describing that data, right?

23:12:36 <jsled> I believe so.

23:12:48 <crschmidt> hm hm hm.

23:13:05 <crschmidt> I want to be able to tie to users and to the concepts.

23:13:16 <crschmidt> Becuase I think that could create a valuable use case.

23:13:27 <jsled> but ... you really want something like prOWL ... a practical OWL.

23:13:38 <jsled> that has things like "users" and "sessions" and "shopping carts" and stuff.

23:14:51 <jsled> maybe not; you just want to indicate "subset of".

23:15:19 <jsled> But all the terms in SKOS-Core <http://www.w3.org/2001/sw/Europe/reports/thes/1.0/guide/#3.9.2> seem to have ... more ... semantics than that.

23:16:27 <jsled> hmm. what else would prOWL have?

23:16:42 <jsled> users are the big one.

23:18:01 <crschmidt> I'm still not sure what inScheme should point to.