Semantic Web Interest Group IRC Chat Logs for 2003-08-11

00:34:49 <GNUWookie> GNUWookie is now known as Wookie-ZZZ

04:17:32 <AaronSw> Anyone know if ARP/RDF Validator does DTDs?

06:08:25 <Wookie-ZZZ> Wookie-ZZZ is now known as GNUWookie

07:38:03 <GNUWookie> GNUWookie is now known as md-work

08:14:06 <jql> jql is now known as jql-zzz

08:33:07 <mea_culpa> mea_culpa is now known as EdTivrsky

10:16:48 <gromgull> Hi all, does anyone know of any efforts in making a RDF-stylesheet language? Something to navigate the graph and make up some xhtml?

10:17:11 <darobin> yes there are efforts...

10:17:21 <darobin> but... I don't remember where :(

10:17:37 <darobin> not to make xhtml though, to render the graph nicely

10:17:58 <darobin> IIRC it was IsaViz

10:18:14 <darobin> there http://www.w3.org/2001/11/IsaViz/

10:18:26 <darobin> "Since version 2.0, IsaViz can render RDF graphs using GSS (Graph Stylesheets), a stylesheet language derived from CSS and SVG for styling RDF models represented as node-link diagrams"

10:18:56 <libby> hm, interesting

10:19:08 <gromgull> hmm...

10:19:13 <gromgull> sounds fun.

10:19:23 <jeen> not exactly what you want but maybe the cuypers project is interesting to you as well.

10:19:30 <libby> gromgull, I've been wondering about creatinbg an 'rdfpath' adjunct to xslt to do this

10:19:36 <darobin> I like the idea very much, haven't had time to look into it closer though

10:19:49 <jeen> they do multimedia presentation and generation based on RDF description, combining XSL with rdf qls.

10:19:50 <libby> I probbaly wont get around to it, but shellac sounds quite keen

10:20:10 <darobin> libby: it seems from the discussion on xml-dev that someone at Extreme Markup generalised XPath to work on graphs

10:20:15 <libby> is that the sort of thing?

10:20:33 <libby> ndw (norm walsh) has rdftwig, that what you mean?

10:20:44 <jeen> here's a link http://homepages.cwi.nl/~media/cuypers/

10:20:44 <dajobe> I thought Uche did that for 4suite?

10:20:50 <libby> I think there was something else too...

10:20:50 <gromgull> libby: I've had this idea as well... but time has not permitted me to do it.. I was hoping someone else had done the dirty work :)

10:20:58 <libby> heh

10:21:12 <libby> that's the sort of thing you want though...?

10:21:21 <gromgull> yes

10:21:26 <libby> ah cool jeen

10:22:16 <darobin> graphs and trees: http://lists.xml.org/archives/xml-dev/200308/msg00159.html

10:22:17 <gromgull> omg, it uses prolog, eclipse, sesame, cocoon and tomcat. I was hoping more for a single jar file :)

10:22:30 <libby> I went to an xml summer school recently ad was struck by the possibilities of slotting in 'rdfpath' to xslt, e.g. in saxon, which seems easy to extend

10:22:58 <libby> but I'm not sure if a path-like interface to a rdf ql would be intuitive or not

10:23:11 <libby> cheers darobin

10:24:07 <jeen> yeah, it's rather big but it has some very cool ideas imo

10:24:49 <gromgull> Surely (a subset) of N3 syntax could be used for such a rdfpath

10:25:45 <gromgull> Liek the wonderful examples from "Short cuts and long cuts" : ":me.rel:parent^rel:parent"

10:25:54 <libby> I think theer's a case for making the syntax as close to xoath as possible

10:26:13 <libby> but I think you'd need to experiment a bit, see if it made intuitive sense to xm programmers

10:26:14 <jeen> serql uses a N3-like syntax for path expressions. works very nicely.

10:26:47 * jeen is away: lunch

10:30:53 <libby> hey shellac

10:31:12 <shellac> hello

10:33:21 <libby> so shellac, it's all gone quiet now, but gromgull was looking for templating languages for rdf...

10:33:32 <libby> looks like there's a lot of activity in this area

10:34:41 <shellac> like what I'm about to do

10:34:52 <gromgull> Tell me!

10:34:54 <gromgull> about it.

10:34:59 <shellac> ok...

10:35:25 <shellac> I using saxon, and I'm trying to do things like this:

10:35:30 <libby> and hurry and write it :)

10:35:51 <shellac> foaf:person/knows/@name

10:36:38 <shellac> i.e. try to trick xml people by mapping (dynamically) an xpath to a path in an rdf graph

10:36:38 <gromgull> good...

10:36:51 <gromgull> Have you implemented any of this :) ?

10:37:07 <shellac> I have a name for it: tree hugger

10:37:10 <libby> aw

10:37:31 <shellac> I going to try to get soemthing working this week

10:37:35 <libby> I really want him to do a QL-backed version rather than an api-backed version

10:37:39 <gromgull> I can see this O'reilly book cover already :)

10:37:50 <shellac> always important

10:38:46 <shellac> it's a proof of concept really, so will perform terribly on SQL-backed models

10:39:13 <libby> nooo!

10:39:31 <shellac> essentially it will map XML-Node to and RDF node or property

10:39:45 <shellac> the RDF nodes have attributes

10:39:56 <libby> why not just translate the rdfpath syntax to squish or rdql rdf ql and passit off to one of those at the backend?

10:40:11 <shellac> that's harder to do in saxon

10:40:20 <libby> really?

10:40:32 <libby> but it'd be worth it....

10:40:34 <gromgull> libby's idea is better ;0

10:40:36 <gromgull> ;)

10:40:36 <shellac> but if the principle works that will be easy

10:40:49 <libby> yep, a proof of concept would be good

10:41:02 <libby> so uche's done something like this for his ql

10:41:19 <shellac> I agree (re libby) but then I have to do more work

10:41:25 <libby> gromgull, would you be using sql-backed data or files, predominantly

10:42:03 * libby is sure that shellac can knock the whole caboodle up in a week :)

10:42:50 <shellac> the nice thing about doing it this way (with saxon) is that you'll get all of xslt working

10:43:31 <gromgull> Actually, what I have to do at the moment is just to present RDF passed to me by another agent... So you can probably say files...

10:43:32 <libby> and that wouldnt work with the query thing?

10:44:03 <gromgull> but what I really need is somethign that will get instances of something in through and interface, then look up things in the ontologies, which will be kept in a database I imagine to get labels, subclasses etc...

10:44:07 <shellac> whereas the other idea - compiling to squish (or whatever) will not do functions etc

10:44:09 <gromgull> So in a way, you can say both.

10:44:17 <libby> so what's interesting is that there are apparantly no implementations of xslt which are database backed

10:44:47 <shellac> really? that seems odd

10:44:53 <libby> so gromgull, youi will need to switch between data sources?

10:45:20 <gromgull> Yes... sort of. I could always combine them into one big model in memory, it's cheap these days anyway :)

10:45:30 <libby> true

10:45:36 <libby> well xquery is the db one

10:45:58 <libby> I gues you'd need some sort of protocol for changing data sources etc.

10:46:06 <libby> I wonder if xquery does that?

10:46:27 <shellac> you can use document() IIRC

10:46:38 <shellac> something like that, anyway

10:46:39 <libby> hm

10:46:42 <libby> yep

10:46:50 <libby> overlaod it maybe...

10:47:10 <shellac> I'll have to something like that, as well. rdftwig does that as well

10:47:28 <gromgull> rdftwig?

10:47:46 <shellac> nochump http://rdftwig.sourceforge.net

10:47:49 <libby> yep it's pretty cool api-based intereface to rdf via saxon

10:48:14 <shellac> it constructs a tree from a graph using a number of alogrithms

10:48:26 <libby> I find api-based stuff a bit cumbersome to use, but it does actually work :))

10:48:29 <gromgull> Cool...

10:49:21 <shellac> rdftwig is different to my idea in that it sort of 'unrolls' the graph

10:50:03 <shellac> I want to try and make it invisible - trick xml types :-)

10:50:18 <libby> I think that's key, if you can pull it off

10:50:49 <libby> in rdfworld we tend to underestimate the difficulty people have with the syntax oin XML for RDF

10:51:16 <libby> (even though we're all well aware of the issue) - it's worse than we think, because people acanb;t use any of their xml tools to play with rdf

10:51:41 <dajobe> I disagree with that

10:51:51 <libby> ...or even any of the metaphors used in xmlworld

10:52:12 <libby> well, they definitely can;t use heir xml tools

10:52:20 <gromgull> Speaking to some people here, we discussed the possibility of making a canonical XML format of RDF, so that you could canonicalize some RDF which would make it usable with XML Tools like XSLT>

10:52:21 <gromgull> .

10:52:44 <libby> that is an attractive idea

10:52:53 <shellac> but very hard

10:53:12 <swh> it was disscussed at the RDF query BOF session at the web conf. IIRC

10:53:20 <dajobe> it's impossible for certain graphs with bnodes

10:53:22 <libby> I think though that at least at the moment, you'd lose much of the benefit of using RDF, because the interesting ting about RDF is the network effect

10:53:34 <shellac> eg: if there is a canonical format then graph comparison is trivial

10:53:35 <gromgull> hmm....

10:53:46 <libby> and to be standards compliant you'd need to use the standard syntax

10:54:00 <shellac> but that's know to be a hard problem, so the canonicalisation must be hard

10:54:13 <dajobe> see jeremy carroll's canonicalization paper

10:54:16 <gromgull> dajobe: I though the introduction of nodeID solved that?

10:54:18 <libby> max froumentin's xslt rdf parser uses that trick, cannonicalizing and then reparsing

10:55:14 <dajobe> gromgull: no. the bnode names are local and would have to be canonicalised themselves. Graphs with many (all) bnodes will be very hard to canonicalise

10:56:38 <dajobe> matching rdf graphs, carroll: http://www.hpl.hp.com/techreports/2001/HPL-2001-293.pdf

10:56:44 <gromgull> shellac: I did some python hacking for making a canonical form of RDF, which doesn't have anything to do with XML, and doesn't convert back to anything sensible, but it does allow you to do graph isomorphism without getting too heiry.

10:56:46 <gromgull> hairy.

10:56:48 <libby> so one interesting thing is (I think) a conanonical syntax wouldnt be any easier for xml people to read, even if it were easier to process with tools.

10:57:03 <libby> ohh, sounds handy gromgull

10:57:07 <gromgull> True, if anything it would probably be harder to read...

10:57:17 <shellac> as dajobe says: try it with bnode rich graphs

10:57:25 <libby> i.e. it would look like triples, not like xml, maybe something like ntriples in xml

10:57:42 <gromgull> I also played with the thought of doing subgrahp isomorphism by looking at the canonical string form... but it was too messy.

10:57:52 <dajobe> nope, won't work either

10:58:00 <dajobe> we've been down this road already

10:59:03 <shellac> in fairness it's the general case which is the problem

10:59:22 <libby> so...this is why I think we need to abstract right away from the syntax; access to RDF data should be through query languages that make processing quick and intuitive, and not through processing the syntax or via an api

11:01:44 <shellac> always with teh query languages :-)

11:01:47 <libby> heheh

11:02:02 <libby> anyway, you're a convert right? ;)

11:02:33 <libby> you did what I did: tried to use this rdf stuff, got frustrated, and wrote a ql for it!

11:03:14 <shellac> I'd say - if you love iterating, use apis, if not, try a query language

11:03:41 <libby> this is why everyone uses xslt/xpath...

11:04:24 <shellac> gormgull: I'll let you know how Tree Hugger gets on

11:04:38 <shellac> I may have a logo by next week :-)

11:04:45 <libby> I can nver think of good names

11:04:49 <libby> that's a good one

11:05:05 <gromgull> Cheers!

11:05:12 <shellac> I'm going to push i as a derogatory term for xml people :-)

11:05:19 <dajobe> lol

11:05:24 <shellac> s/i/it/ - no, wait

11:05:36 <libby> aw, I think it's a sweet name

11:05:45 <libby> it'll just make them look cuter

11:05:51 <gromgull> For now I think I'll stick with writing a java class for each "thing" I want presented and use RDQL...

11:05:56 <shellac> tree surgeons?

11:06:21 <libby> nah, not as good

11:06:23 <shellac> I think that's a good plan, gromgull

11:06:43 <libby> makes sense, though sounds tedious

11:07:18 <shellac> you could automate the process, of course

11:07:25 <libby> annoying to have to hardcode stuff when RDF is supposed to be so flexible

11:07:57 * jeen is back (gone 00:41:09)

11:07:58 <shellac> the main annoyance is the absence of optionals

11:08:09 <libby> that is pretty annoying

11:08:33 <shellac> which I've just finished some hacking on (for squish in ruby)

11:10:02 <shellac> another suggestion for xml people before I go: tree beards

11:10:39 <darobin> makes us sound very wise :)

11:11:17 <shellac> I was thinking beardy tolkein types with a relaxed attitude to hygiene :-)

11:11:34 <libby> hehe

11:11:36 <libby> how rude

11:11:51 <darobin> ah

11:11:55 <darobin> hehehe

11:12:25 * darobin starts picking a few nits from his own beard -- yummy!

11:12:38 * shellac ponders possible shave and change of clothes

11:14:01 <darobin> but what would we call you back then if we were tree huggers? the OIL lobby?

11:14:36 <jeen> greasy... something?

11:16:58 <DanC-AIM> byebye

11:21:50 * shellac returns from changing

11:22:04 <shellac> darobin: why, you...

11:23:58 * shellac --> shoppin'

11:54:08 <libby> hello larsbot! were you at extreme?

12:28:29 <larsbot> hi libby

12:28:31 <larsbot> I certainly was

12:29:16 <darobin> was it good?

12:29:34 <libby> sounded fun

12:31:08 <larsbot> it was very good

12:31:22 <larsbot> lots of topic map stuff, some RDF stuff, and lots of good XML stuff

12:31:36 <larsbot> enjoyed myself a lot

12:46:39 <ericP> ddd

12:47:56 <ericP_> ignore wierd utterances from my irc doppleganger

12:48:01 <ericP_> some client probs...

12:48:29 <danbri> did anyone hear something? me neither...

12:57:42 * ericP feels people are doppleganging up on him

13:13:03 <wkearney99|away> wkearney99|away is now known as wkearney99

13:20:18 <darobin> CC/PP is expressed in RDF right?

13:20:30 <shellac> yep

13:20:53 <darobin> ta. Do you know where I can find interesting (ie more or less real life) examples?

13:21:01 <shellac> um

13:21:35 <shellac> youv've tried the project page?

13:21:49 <darobin> ah I've found some in the wild

13:22:00 <darobin> hmmm

13:22:29 <darobin> ooh, yes indeed, I'd skipped the "Profiles" sections as "profiles of the language", instead "CC/PP Profiles" !

13:22:38 * darobin tweaks his brain optimiser

13:27:32 <danbri>http://www.w3.org/2003/07/25-rdfcore-irc.html

13:27:34 <dc_rdfig> A: http://www.w3.org/2003/07/25-rdfcore-irc.html from danbri

13:27:41 <danbri> A:|Friday's RDFCore IRC logs

13:27:42 <dc_rdfig> Titled item A.

13:28:45 <danbri> A:See also actions in RDF, [http://www.w3.org/2003/07/25-rdfcore-actions.rdf|25-rdfcore-actions.rdf].

13:28:46 <dc_rdfig> Added comment A1.

13:29:25 <danbri> A:oops wrong urls, that was week before.

13:29:25 <dc_rdfig> Added comment A2.

13:36:17 <danbri> A:Ah, last week's was logged at ILRT, see [http://ilrt.org/discovery/chatlogs/rdfcore/2003-08-08.html|2003-08-08]. And week before last's: [http://www.w3.org/2003/08/01-rdfcore-irc.html|irc-transcript], [http://www.w3.org/2003/08/01-rdfcore-actions.rdf|actions]

13:36:17 <dc_rdfig> Added comment A3.

13:41:26 <dajobe> wassup?

13:54:05 <_joshua> Hello

13:54:14 <Schuyler> hey Josh

13:54:20 <_joshua> how's it going?

13:54:45 <Schuyler> Going pretty good. I'm in Boston. (Cambridge, actually.)

13:55:10 <_joshua> Conference?

13:58:11 <Schuyler> Just plain work. Sent to get braindump from the manuscript conversion staff.

13:59:31 <_joshua> oh

13:59:41 <_joshua> clearly you should open an NYC office

14:16:27 <Schuyler> clearly :-)

14:25:11 <eamonn__> eamonn__ is now known as eamonn

14:26:32 <eamonn>http://www2.elsevier.co.uk/~tony/spec/rss/mod_context.html

14:26:33 <dc_rdfig> B: http://www2.elsevier.co.uk/~tony/spec/rss/mod_context.html from eamonn

14:27:07 <eamonn> B: |RSS 1.0 Module: Context

14:27:07 <dc_rdfig> Added comment B1.

14:28:17 <eamonn> B:|RSS 1.0 Module: Context

14:28:18 <dc_rdfig> Titled item B.

14:28:48 <libby> hey eamonn

14:28:57 <eamonn> hi Libby

14:29:36 <eamonn> we put out B on Friday to rss_dev but it seems pretty quiet there

14:30:39 <libby> yep, I saw it

14:30:49 <libby> i giess rss is pretty quiet

14:31:41 <eamonn> maybe they are all firefighting echo/atom or whatever it is today

14:31:53 <libby> yeah, probably

14:32:42 <eamonn> do you know if there is still a process for publishing modules through rss_dev, or has it broken down?

14:33:17 <libby> I think the whole rss 1.0 wg thing is sorta wound down, could be wrong

14:33:28 <libby> not been any process stuff on there for ages

14:34:44 <eamonn> guess I'm still catching up with the present then ...

14:59:42 <eamonn> any questions about mod_context?

15:03:03 <libby> "In these examples we assume a uniform means of querying an RSS aggregator for an RSS 1.0 feed - using an OpenURL as a query API. "

15:03:17 <libby> intersting. though I don;t find it very intuitive

15:04:01 <bijan_> bijan_ is now known as bijan

15:05:13 <wkearney99> where is RDF's URI concatenation handling documented?

15:07:48 <jeen> syntax spec I think. checking...

15:07:48 <eamonn> that's a general complaint that we get about OpenURL - partly because context needs more definition

15:12:53 <AaronSw>http://www.intertwingly.net/blog/1557.html#c1060611075

15:12:54 <dc_rdfig> C: http://www.intertwingly.net/blog/1557.html#c1060611075 from AaronSw

15:13:00 <AaronSw> oops

15:13:02 <AaronSw> C:|ignore me

15:13:02 <dc_rdfig> Titled item C.

15:13:04 <AaronSw>http://www.intertwingly.net/blog/1557.html

15:13:04 <dc_rdfig> D: http://www.intertwingly.net/blog/1557.html from AaronSw

15:13:11 <AaronSw> D:|RDF can be readable?

15:13:12 <dc_rdfig> Titled item D.

15:13:25 <AaronSw> D:Sam Ruby tests my RDF integration with Pie/Echo/Atom

15:13:25 <dc_rdfig> Added comment D1.

15:13:45 <AaronSw> D:My evil scheme is all going according to plan. ;)

15:13:45 <dc_rdfig> Added comment D2.

15:15:11 <MikeM> eamonn, are you involved with the open url stuff?

15:15:54 <MikeM> oh wait... ok, I remember you from the conference call....duh

15:16:33 <eamonn> yep

15:19:49 <danbri> D:Those files don't look like rdf to me, parseType=Resource missing maybe?

15:19:50 <dc_rdfig> Added comment D3.

15:19:50 <dajobe> D:close but not quite. In Sam's maxAtom.rdf property element <generator> has two 'nodes' inside

15:19:51 <dc_rdfig> Added comment D4.

15:21:58 <sbp> D:The Validator passes them. Check out [http://intertwingly.net/stories/2003/08/10/atom.dtd|atom.dtd] for the missing attributes

15:21:59 <dc_rdfig> Added comment D5.

15:23:08 <AaronSw> D:Read the DTD, folks. It adds the parseTypes in

15:23:09 <dc_rdfig> Added comment D6.

15:23:17 <AaronSw> D6:""

15:23:17 <dc_rdfig> Deleted comment D6.

15:23:20 <AaronSw> oops, what sbp said

15:24:32 <Schuyler> AaronSw: word up!

15:25:19 <sbp> D:Clever. Very clever. In fact, a little more than clever--pretty much Swartzesque. But <id> gets hung out to dry as a literal when it *really* wants to be an rdf:about on the enclosing typed node

15:25:19 <dc_rdfig> Added comment D6.

15:25:53 <danbri> D:Oh, interesting. I was wondering about that idea the other day, but concluded (forget why) that it was infeasible. Clever! I was thinking about xmlns: attributes more than parseType...

15:25:54 <dc_rdfig> Added comment D7.

15:25:57 <dajobe> D:RDF/XML does not use Infoset DTD infoitems

15:25:58 <dc_rdfig> Added comment D8.

15:26:30 <danbri> D:Yeah, most RDF parsers won't grok it. Does ARP and the validator?

15:26:31 <dc_rdfig> Added comment D9.

15:26:49 <sbp> D:No PSVI at all? Well, the validator validates the PSVI, so that ought to be changed. It's pretty misleading.

15:26:50 <dc_rdfig> Added comment D10.

15:26:57 <dajobe> D:XML parsers have a Big Switch set to validate/WF - most people set it to WF check only

15:26:58 <dc_rdfig> Added comment D11.

15:27:37 <dajobe> D:don't confuse PSVI with Infoset. Former is W3C XML schemas only.

15:27:38 <dc_rdfig> Added comment D12.

15:27:53 <danbri> D:See [http://intertwingly.net/stories/2003/08/10/atom.dtd|atom.dtd]

15:27:54 <dc_rdfig> Added comment D13.

15:28:12 <AaronSw> D:More a PDVI than a PSVI. ARP and the Validator grok it fine.

15:28:12 <dc_rdfig> Added comment D14.

15:28:24 <dajobe> D:that's never been proof it's right :)

15:28:25 <dc_rdfig> Added comment D15.

15:28:35 <dajobe> D:it accepted rss2 for a long while

15:28:36 <dc_rdfig> Added comment D16.

15:28:58 <AaronSw> D:The rdf-syntax-grammar spec doesn't say it excludes DTDs and the Infoset includes them.

15:28:58 <dc_rdfig> Added comment D17.

15:29:16 <dajobe> D:it lists the infoitems it accepts and doesn't use explicitly

15:29:17 <dc_rdfig> Added comment D18.

15:29:20 <sbp> interesting to see an attempt to bring people over to the RDF camp--by trying to avoid all its negative connotations--being poured down the drain. people detest RDF/XML, and whether their reasons are sound or not, it's becoming the flow of things

15:30:04 <dajobe> yawn

15:30:32 <dajobe> D:the dtd idea is interesting, it's just not something I'd expect to work

15:30:33 <dc_rdfig> Added comment D19.

15:30:50 <AaronSw> D:The Infoset seems pretty clear to me when it says "[attributes] An unordered set of attribute information items, one for each of the attributes (specified or defaulted from the DTD) of this element."

15:30:50 <dc_rdfig> Added comment D20.

15:31:35 <danbri> D:But you can't mandate DTD processing just by citing a DTD.

15:31:35 <dc_rdfig> Added comment D21.

15:31:46 <AaronSw> I agree.

15:31:48 * dajobe hunts for the big switch

15:32:16 <dajobe> oh, I can't do it. validation requries a DOM for libxml, I recall now

15:32:21 <sbp> Pie/Echo/Atom is yet another format that should clearly--clearly--be an application of RDF. I haven't seen many people from the RDF camp asking why it is that they haven't used RDF, that's all... I'd've thought that a few questions would be being raised, seeing as how it's the current BigThing

15:32:56 <danbri> I had a crack at it, wasn't v pretty in rdf as they wanted properties of literals so extra level indirection was needed

15:33:17 <Schuyler> sbp: We Already Have RSS 1.0.

15:33:19 <danbri> If i get time I'll try it in soap encoded format

15:33:21 <danbri> true

15:33:23 <sbp> subversive attempts to pull people over, such as Aaron's thingy, are all well and good, but if there's a strong political aversion to RDF, then that's a problem for anyone working with RDF

15:33:45 <Schuyler> sbp: the aversion is based on FUD. and the rdf:Sequence bit of RSS 1.0.

15:33:47 <danbri> +1

15:34:12 <danbri> also that we haven't really explained / shown RDF's advantages well enough yet. imho.

15:34:19 <Schuyler> d'accord

15:34:31 <Schuyler> I didn't get it at all until a couple months ago either

15:34:40 <Schuyler> it's the cool apps that sell people

15:34:43 <Schuyler> (mudlondon in my case)

15:35:05 <sbp> RSS has its problems, hence Pie/Echo/Atom. some are political, some are technical... I guess it's usable, but that's not the point--the point is that there's a format being developed which will be used by thousands of people that should clearly be an RDF application and isn't for political reasons alone

15:35:13 <danbri> :)

15:35:13 <Schuyler> but the effort to drag Atom back towards RDF is a good one

15:35:26 <sbp> RDF advantages: yeah... and those advantages have to be greater than the RDF/XML serialization defecit

15:35:29 <Schuyler> sbp: You're right! Semantic Avengers! To the RDF Cave!

15:35:47 <sbp> no need to patronize me

15:35:57 <danbri> The way to get Atom to use RDF is to sell people on the merits of RDF, not scold them for being unbelievers or try to persuade them that rdf:about= attributes are not toooo ugly.

15:36:11 <Schuyler> sbp: I was just being silly, I'm sorry. I totally agree with you -- we need a call to action.

15:36:23 <Schuyler> +1 to that also

15:36:37 <AaronSw> What are the advantages again?

15:36:59 <AaronSw> Easy merging and mixing,

15:37:02 <danbri> data mixing. loosly coupled extension.

15:37:17 <sbp> danbri: agreed. but people also need to measure shifting the serialization around into RDF form in parsing not aesthetic terms

15:37:19 <danbri> cf discussion w/ jon udell, linked from http://weblog.infoworld.com/udell/

15:37:34 <sbp> Schuyler: my apologies. I misread, and thank you

15:37:48 <ktest> danbri: and others comments e.g. Marc Canter

15:38:14 <sbp> it's too hot here... :-)

15:38:39 <ktest> danbri: http://blogs.it/0100198/2003/08/08.html

15:39:12 <ktest>http://www.rolandTanglao.com/2003/08/08.html#a5122

15:39:14 <dc_rdfig> E: http://www.rolandTanglao.com/2003/08/08.html#a5122 from ktest

15:39:54 <sbp> extensibility seems to be the biggest plus for me, I think. without a formal model, I think that extensions of the RSS 2.0 type could have a greater tendency towards being underspecified. with RDF, you know exactly how the added information relates to what's currently there, in terms of the RDF model

15:40:14 <Schuyler> yeah, I would call it "upward compatibility"

15:40:19 <danbri> I can't read Marc's blog easily... it is almost impossible to tell which is "Marc's voice" from those he is quoting. Just html formatting issue I guess... but I always come away confused.

15:40:25 <dajobe> +1

15:40:29 <ktest> +1

15:47:03 <sbp> hi Jon

15:48:49 <Talliesin> hi sean

15:49:25 <Talliesin> 'sup?

15:50:08 <sbp> just reading though atom-syntax archives at the moment...

15:50:16 <sbp> you?

15:50:29 <Talliesin> Trying to find gainful employment.

15:50:39 <Talliesin> Anyone built a semantic job-finder yet?

15:51:58 <Talliesin> More accurately I was trying to find gainful employment, but then I decided to give up for a while and chat on IRC instead.

15:52:38 <sbp> heh, heh. not yet. there was some RSSjobs thing about lately, but I didn't take any notice of it. but I think that IRC is a far better waste of time, anyway

15:52:59 <Talliesin> RSSjobs?

15:53:24 <sbp> it was on daypop for quite a while. http://rssjobs.com/rssjobs/index.jsp

15:57:36 <Talliesin> Something FOAF-based would be more interesting (it might not actually work or anything, but it would be more interesting).

15:59:11 <libby> e.g. lookig for a persion, don;t know their name; do know their interests and skills

15:59:12 <sbp> heh, yeah. foaf:employmentStatus

15:59:14 <libby> - a query!

15:59:33 <Talliesin> exactly.

16:00:07 <Talliesin> And the foaf:knows linkage means that now nepotism can work for everyone.

16:00:34 <libby> excellent :)

16:00:35 <Talliesin> (assuming < 6 degrees truism pans out)

16:00:55 <Talliesin> In fact, if I could get someone to pay me to build that then I wouldn't have to find a job either.

16:01:20 <libby> perfect!

16:01:48 <Talliesin> Now I have to explain the concept of semantic web to VCs though :(

16:03:30 <libby> that's where it all gets a bit tricky

16:04:07 * Talliesin starts his plans for a FOAF-based investment-locator but realises the obvious flaw in that approach.

16:04:36 <danbri> I did a jobs'n'cv site prior to foaf, very centralised... was one of the motivations for doing a harvesting-based system...

16:05:35 <Talliesin> oh, that's the worse thing. I've done jobs'n'cv sites too, hence my getting all depressed about the chances of finding work :)

16:05:47 <danbri> heheh

16:05:51 <danbri> good luck w/ the job hunt...

16:06:22 <Talliesin> Cheers. I'm not really depressed (not yet anyway), just bored.

16:10:03 <collord> job search could be a "transaction" handled by agents e.g. in a FIPA environment?

16:10:48 <libby> you poor thing Talliesin, how lame.

16:14:01 <Talliesin> Nah. If I'm still unemployed in a few weeks I'll qualify for "poor thing". As it is there's even the chance I might even get a job that doesn't pay a ridiculously low salary.

16:14:43 <libby> :)

16:22:53 <Wack> flipping burgers

16:23:08 <libby> quite

16:37:17 <Schuyler> collord: FIPA?

16:37:41 <libby>http://ilrt.org/discovery/chatlogs/rdfig/2003-08-11.html#T10-16-48

16:37:42 <dc_rdfig> F: http://ilrt.org/discovery/chatlogs/rdfig/2003-08-11.html#T10-16-48 from libby

16:38:05 <libby> F:|chatting about using RDF with XSLT today

16:38:05 <dc_rdfig> Titled item F.

16:47:14 Disconnected from irc.freenode.net (Connection reset by peer)

16:48:02 Topic now Semantic Web hack-n-chat - http://rdfig.xmlhack.com/ wiki at http://esw.w3.org/topic/ irc logs http://ilrt.org/discovery/chatlogs/rdfig/latest

16:48:02 Users on #rdfig: logger jql-zzz ndw bitsko karlcow swh kota merriam collord grove

17:15:26 <dje> dje is now known as notStable

17:15:36 <notStable> notStable is now known as dje

17:16:15 <ChanServ> [#rdfig] This channel is logged and blogged: http://logicerror.com/rdfIRCWelcome

17:20:46 <ChanServ> [#rdfig] This channel is logged and blogged: http://logicerror.com/rdfIRCWelcome

17:25:43 <lilo> [Global Notice] Hi all. As you're aware, we've had some difficulties with the new code. We're looking at them now. Please bear with us. Thank you for your patience, and thank you for using freenode.

17:40:01 <Md2> sorry, we will relink these servers to the rest of the network as soon as it will be safe to do ti

17:55:09 <Md2> [Global Notice] Hi all. Status report on the website (http://freenode.net/news.shtml), please point it out to whoever you can. Thanks!

18:00:37 <lilo> [Global Notice] Hi all. For updates to status please see http://freenode.net/news.shtml .... we're making progress and may have something for you in the next hour. Thanks!

18:05:14 <collord> heh, spellcheck-o on page 2 of the ora RDF book

18:05:27 <Schuyler> word

18:16:24 <Md2> [Global Notice] Hi all. Status report on the website (http://freenode.net/news.shtml), please point it out to whoever you can. Thanks!

18:25:17 <ChanServ> [#rdfig] This channel is logged and blogged: http://logicerror.com/rdfIRCWelcome

18:28:01 <lilo_> [Global Notice] Hi all. We are substantially up for the moment, though we will have to kill and restart some more servers, obviously, some of them very large

18:28:49 <lilo_> [Global Notice] For status, please check http://freenode.net/news.shtml and we'll try to update it in a timely way if there are further outages while we fix the problems. Thank you for your patience, and thank you for using freenode!

18:42:54 <Schuyler_> Schuyler_ is now known as Schuyler

20:14:49 <lilo_> [Global Notice] Good evening to you all. There's a status update on http://freenode.net/news.shtml and we will be finishing our crit upgrade in a few moments. Expect some disruption which should resolve itself fairly quickly for most users. Thank you for your patience, and thank you for using freenode!

20:23:17 <CaptSolo> hi all

20:23:25 <CaptSolo> what's hot?

20:24:15 <lilo_> [Global Notice] Hi all. "Big re-up" in process. See you on the other side.

20:25:24 <libby> europe

20:25:43 Disconnected from irc.freenode.net (Connection reset by peer)

20:26:05 Topic now Semantic Web hack-n-chat - http://rdfig.xmlhack.com/ wiki at http://esw.w3.org/topic/ irc logs http://ilrt.org/discovery/chatlogs/rdfig/latest

20:26:05 Users on #rdfig: logger merriam _joshua eikeon danbri EdTivrsky golbeck aharth DanC CaptSolo shellac JibberJim

20:26:06 <ChanServ> [#rdfig] This channel is logged and blogged: http://logicerror.com/rdfIRCWelcome

20:36:00 <lilo_> [Global Notice] Hi all. I should mention our next step is to take care of the remaining 10% of the users; you'll see a number of small splits. Any ending status will be displayed on wallops (/mode <yournick> +w)....thanks again and have an excellent night!

20:48:22 <fsparv> Hi, I've been googling around and can't seem to find any examples of code that checks a chunk of RDF to see if it contains assertions that violate a given ontology... anyone know where I could find an example?

20:48:45 <bijan> Hmm.

20:49:08 <bijan> Well, what exactly do you want?

20:49:12 <bijan> I.e., to test for consistency?

20:49:18 <bijan> To do simple range/domain validation?

20:49:38 <fsparv> domain/rainge/cardinality datatypes etc

20:49:43 <bijan> What kind of ontology are you thinking about

20:49:57 <bijan> Do you need complete/effective reasoning?

20:50:52 <fsparv> at the moment all I am interested in is the answer to the question, does this chunk of assertions that I have generated (as a result of user input) fit

20:51:12 <bijan> Well, what kind of fittness :)

20:51:22 <bijan> Do you mean, are they *consistent*?

20:51:42 <bijan> (I.e., given a kb made up of the ontolgoy + these assertions, is that kb consistent)

20:52:18 <bijan> You can try ours: http://www.mindswap.org/2003/pellet/

20:52:18 <fsparv> did the user answer "one I think" where the number 1 was required (consistant given the ontology, yes)

20:52:37 <bijan> Well, "one I think" isn't in owl or rdf :)

20:52:44 <bijan> No "I think" :0

20:53:15 <fsparv> a silly example of course but a simple one too...

20:54:14 <bijan> Any SHIF or SHIQ (or similar) reasoner will do this, to varying degrees

20:54:34 <bijan> FaCT has source code avaialbe (but isn't directly RDF aware)

20:55:01 <bijan> Pellet is a SHIN (most of OWL DL) reasoner written in java,b ut not with many services or optimizations (yet)

20:55:32 <fsparv> ok, I have to admit I don't know what you mean by SHIF/SHIQ/SHIN

20:55:36 <bijan> If you work out a suitible translation, you can use most first order automated theorem provers or propositional modal logic provers

20:55:46 <bijan> They are varieties of description logic

20:55:53 <bijan> Characterized by their expressivity

20:56:26 <bijan> SHIF is the boolean connectives (and, not, or, i.e., intersectionOf, complementOf, unionOf)

20:56:56 <bijan> Plus, existential and universal quantifiers (someValuesFrom, allValuesFrom)

20:57:14 <bijan> Plus Transitive roles an role hierarchies (all this == the S in SHIF)

20:57:17 <bijan> er

20:57:26 <fsparv> ok

20:57:26 <bijan> S is everyhting but role heriarchies

20:57:30 <bijan> Role hierachys == H

20:57:35 <bijan> I is inverse roles

20:57:51 <bijan> F is functional roles (min/max/card limited to 0 and 1)

20:58:28 <fsparv> ok so what is Q and N

20:58:43 <bijan> N is generalized cardinalities

20:58:47 <bijan> min 100 max 200

20:58:54 <bijan> Q is qualified carndinalities

20:59:15 <bijan> peole *have* 10 fingers, and *have* 10 toes

20:59:18 <bijan> Q is not in OWL

20:59:28 <bijan> But is in DAML+OIL as the cardinalityQ constructs

20:59:38 <bijan> So OWL is strictly less expressive than DAML+OIL

20:59:46 <fsparv> oh yes I remember reading about those... why isn't it in OWL?

21:00:02 <bijan> O is "nominals", oneOf and hasValue

21:00:11 <bijan> OWL Lite is SHIF(D) (d is for datatypes)

21:00:20 <bijan> OWL DL is SHION(D) (roughly)

21:00:30 <bijan> (the D's aren't quite right,b ut close enough)

21:00:38 <bijan> DAML+OIL is SHIOQ(D)

21:01:05 <CaptSolo> yo bijan :>

21:01:08 <bijan> OWL Full is none of these :) but an undecidable (and slightly odd) fragment of first order logic

21:01:11 <bijan> Hello

21:01:20 <fsparv> hehe ok

21:01:57 <bijan> BUT, for your purposes, Pellet should be fine

21:02:07 <bijan> may, in fact, be the easiest current option to get going with

21:02:27 <bijan> Just take your ontology

21:02:37 <bijan> append your test assertions

21:02:41 <bijan> and feed to the webform

21:03:41 <bijan> If you are java savvy it's pretty easy to use

21:03:54 <bijan> Code not documented yet.

21:04:12 <bijan> The Overview page will have some of what I've been saying, plus pointers to descriptions of the algorithms involved

21:04:35 <fsparv> Ok I will check it out... Source is available?

21:04:37 <sandro> Bijan, do you have an ontology for reporting your results on the test suite?

21:04:48 <bijan> Should be

21:04:57 <bijan> I know we have a subversion repository

21:05:01 <bijan> But i ahven't used it :)

21:05:13 <bijan> I will pester folks to make a tarball of source in the next few days

21:05:18 <fsparv> oh cool. I havn't had a good reason to use subversion yet :)

21:05:20 <bijan> sandro, uhhhhhhh

21:05:25 <bijan> I think ronwalf does

21:05:48 <sandro> if he wanted to drop me a URL for where he keeps the test results, that'd be nice.

21:05:54 * sandro wonders off, family time.

21:06:16 <bijan> Er..

21:06:22 <bijan> logger, pointer

21:06:22 <bijan> See http://ilrt.org/discovery/chatlogs/rdfig/2003-08-11#T21-06-22

21:06:29 <bijan> ronwalf: check archives :)

21:06:34 <bijan> sandro just wandered off :)

21:10:54 <ronwalf> sandro: Current version outputs results in RDF

21:11:46 <ronwalf> Well, my current version. I need to check that change into subversion

21:15:13 * GNUWookie lurks

21:16:16 <ericP_> ericP_ is now known as ericP

21:20:03 <sandro> ronwalf, where are those results on the web...?

21:20:52 <ronwalf> sandro: http://owl.mindswap.org/2003/submit-rdf/253.rdf is a recent test run

21:21:17 <ronwalf> No real ontology has been defined yet

21:21:21 <GNUWookie> hi ericP

21:21:24 <ronwalf> But I have some stubs in place

21:21:40 <ericP> heya, t'sup?

21:22:15 <ronwalf> Anyone have a good ontology to describe test runs?

21:22:28 <GNUWookie> ronwalf: good question!

21:22:42 <sandro> I'm looking around for one. (well, I have one in my head.... but I thought I should look around first.)

21:22:48 <GNUWookie> ericP: jql was looking for some help with Rdf::Core

21:23:09 <GNUWookie> ronwalf: I think that test runs are program invokations

21:23:17 <ericP> i assume the perl module and not the working group

21:23:21 <GNUWookie> so you need a way to describe the software runs

21:23:26 <GNUWookie> yes the perl stuff

21:23:48 * GNUWookie is hit by an idea

21:23:52 <GNUWookie> ouch!

21:23:52 <ericP> i've only used it tangentially, mostly used my own for apps

21:24:10 <GNUWookie> ok

21:24:39 <GNUWookie> i see, so algea is an app that just uses

21:24:40 <sandro> I'm thinking something like [ a test:SuccessfulExecution; test:system <http....surnia>; test:startTime ...; test:endTime ...; test:test <.../Manifest001#test>;]

21:24:40 <ronwalf> I know there are other people trying to pass the OWL tests. Who else is putting the results in RDF?

21:24:40 <GNUWookie> it

21:24:53 <GNUWookie> ronwalf: OWL tests?

21:25:01 <ronwalf> The owl test cases

21:25:02 <GNUWookie> ah

21:25:09 <GNUWookie> then the result is a proof?

21:25:35 <sandro> I'm guessing I'll be putting everyone's results in RDF for the Implementation Report. If they don't provide them in a nice RDF form, I'll just have to convert them.

21:25:38 * GNUWookie makes the connection - proof = invocation= run=test

21:25:51 <bijan> Wrong connection

21:25:52 <ronwalf> sandro: Give me an ontology and I'll convert

21:25:57 <GNUWookie> run = time period

21:26:16 <dajobe_> dajobe_ is now known as dajobe

21:26:18 * GNUWookie looks to bijan for guidance

21:26:47 <GNUWookie> maybe a proof implies a run

21:26:50 <sandro> Thank, ronwalf -- I'm not sure I know the right things to say. I think it's basically a tuple of (which test, which installation of some software, pass/fail/unknown/, timing stuff).

21:27:00 <JimJibber> JimJibber is now known as JibberJim

21:27:12 <GNUWookie> so test => proof => run => invokation => time period

21:27:21 <GNUWookie> in terms of implications

21:27:23 <bijan> The tests results are generally "pass" or "fail", or perhaps, "yes, this is consistent, as the manifest claims"

21:28:28 * GNUWookie is attacked by a vision of invokations of functions as micro proofs

21:28:36 <sandro> On for instance an OWL Inconsistency test, I think we should report differently if the tested system says "Consistent" vs "Unknown". They are both kinds of failure, but "Consistent" is a MUST NOT and "Unkown" is a SHOULD NOT.

21:28:50 * GNUWookie gasps

21:29:08 <GNUWookie> what about the contents of the test

21:29:14 * sandro has no idea what GNUWookie is talking about, btw....

21:29:18 <GNUWookie> that is also rdf

21:29:20 * sandro needs to run off gain....

21:29:28 <GNUWookie> sandro: join the club

21:29:33 <bijan> We don't care.

21:29:42 * GNUWookie is lonly in his madness

21:29:55 <bijan> The only thing that matters, re: process, about the tests, is getting the desired answer for each given test

21:30:23 <GNUWookie> ok, i know that euler has a vocab to describe proofs

21:30:27 <bijan> The reasoning tests are testing whether a resoner can correctly determine the consistency of a kb (or the entailment relation between two of them)

21:31:08 <ronwalf> sandro: testSystem is a good idea. The testTime might only be useful in making performance comparisons between implementations...

21:32:19 <ronwalf> sandro: Note that I'm bad and don't label the units in that RDF. The testTime is in milliseconds

21:34:06 <bijan> duh duh duh dATATYPES!

21:42:04 <ericP> sandro, are you basing the test manifests on http://www.w3.org/2003/03/rdfqr-tests/recording-query-results.html ?

21:43:33 <ericP> which defines classes (ResultSet, Solution, Binding) and properties (solution, size, resultVariable, binding, variable, value).

21:44:00 <bijan> Which almost all are pointless for these tests :)

21:44:14 <bijan> "Consistent, Inconsistent, Unknown"

21:44:20 <bijan> "Entails, dosen't entail, unknown"

21:46:12 <ericP> possibly one or two will be relevent

21:46:42 <ronwalf> bijan: er, like "PYMDTHM2.3413S"^^xsd:duration ?

21:46:49 * ronwalf icks

21:46:55 <bijan> YEp :)

21:47:08 <ericP> for instance, soundness and completeness tests could prove their consistency by reporting all the entailments in a ResultSet

21:47:11 <bijan> Dinner

21:47:28 <ericP> "Dinner" is a class?

21:47:34 <ericP> it's capitolized like one

21:48:34 <ericP> libby around?

21:49:30 <ronwalf> [ a meal:Dinner; mail:foodItem [a food:Spinach];]

21:49:35 <ronwalf> Or something like that

21:53:46 <deltab> ronwalf: I've not seen any indication that you can use Y, M, D etc. without corresponding numbers

21:54:44 <ronwalf> deltab: Probably not - I just popped open google and glanced at http://www.w3.org/TR/xmlschema-2/#duration

21:55:17 <ronwalf> It would be PT2.341S I believe

22:50:07 <GNUWookie> GNUWookie is now known as Wookie-ZZZ