Semantic Web Interest Group IRC Chat Logs for 2003-04-10

NOTICE: #rdfig logs are being turned off 2004-12-03. Please switch to the new and shiny #swig channel for Semantic Web Interest Group chat. Change your client to #swig and enjoy the new experience. Or read the latest #swig logs to see what you've been missing :)

01:53:48 <lilo> [Global Notice] Hi all. Freenode staff hope you're enjoying your morning. Lately reliability has been fairly high, but we have an extremely serious server availability problem. We're going to have big problems in the next few days. Please visit http://freenode.net/news.shtml for more information.

06:41:43 * danbri_fr gets all the supporting libraries needed for rubyrdf installed on macosx

06:42:16 <danbri_fr> ...then has ruby 1.6.7 segfault when running his tests

08:16:04 <Jeen> hi libby, dan

08:21:20 <libby> morning!

08:33:19 <danbri_fr> hi there

08:56:02 <arnarl> hi

13:29:15 Topic now RDF/SemWeb wish factory. http://rdfig.xmlhack.com/

13:29:15 Users on #rdfig: logger_2 Morbus danbri_fr alberto AndyS larsbot xower datum grault dc_rdfig deltab sbp` jang taaz fknow em sandro xover mea_culpa eikeon idoru aml ericP mwf TrustBot grove Jeen mariyo nmg arnarl caz dajobe-lap golbeck Wack gcc-O3 libby

13:29:15 <ChanServ> [#rdfig] This channel is logged and blogged: http://logicerror.com/rdfIRCWelcome

13:30:01 Topic now RDF/SemWeb wish factory. http://rdfig.xmlhack.com/

13:30:01 Users on #rdfig: logger_2 Morbus danbri_fr alberto AndyS larsbot xower datum grault dc_rdfig deltab sbp` jang taaz fknow em sandro xover mea_culpa eikeon idoru aml ericP mwf TrustBot grove Jeen mariyo nmg arnarl caz dajobe-lap golbeck Wack gcc-O3 libby

13:30:02 <ChanServ> [#rdfig] This channel is logged and blogged: http://logicerror.com/rdfIRCWelcome

13:30:15 Topic now RDF/SemWeb wish factory. http://rdfig.xmlhack.com/

13:30:15 Users on #rdfig: logger_1 Morbus danbri_fr alberto AndyS larsbot xower datum grault dc_rdfig deltab sbp` jang taaz fknow em sandro xover mea_culpa eikeon idoru aml ericP mwf TrustBot grove Jeen mariyo nmg arnarl caz dajobe-lap golbeck Wack gcc-O3 libby

13:30:16 <ChanServ> [#rdfig] This channel is logged and blogged: http://logicerror.com/rdfIRCWelcome

13:30:20 <dajobe-lap> right, that's it

13:31:00 <arnarl> arnarl is now known as arnarl|away

13:31:08 <dajobe-lap> note to self - fix the hand-edited logger script vs the CVS version

13:33:36 <danbri_fr> can we go back to plain old 'logger' someday? ie not _1... (no big deal tho)

13:40:03 <AndyS> DanBri? - re checking query results

13:40:42 <AndyS> I have changed to using Jeremy graph comparision code now but I used to do something like what you outline

13:41:28 <AndyS> the code is in Jena 1.6.1 (It does bNodes between the two result sets by mapping internal labels IIRC).

13:42:13 <AndyS> Its slow - it was written to check correctly, so simple and verifiable was more important than fast

13:43:13 <danbri_fr> AndyS, thats interesting. Do you treat each row as a separate to graph to map to its corresponding row? or entire resultset as a graph compared?

13:43:29 <danbri_fr> 'mapping internal labels'?

13:45:42 <AndyS> The comparision I used was non-RDF but using an graph with table structure would be the same.

13:46:34 <AndyS> Now I encode the result as per the result-set vocabulary and call actualsResultModel.isomorphicWith(expectedResultModel).

13:46:50 <AndyS> In my old code, I maintained label mappings across rows.

13:47:13 <AndyS> A separate graph per row does not work, me thinks.

13:47:32 <libby> time utc

13:47:35 <libby> .time utc

13:47:35 <datum> 2003-04-10T13:47:35

13:57:37 <DanC_jam>http://www.wiwiss.fu-berlin.de/suhl/bizer/SWTSGuide/index.htm

13:57:38 <dc_rdfig> D: http://www.wiwiss.fu-berlin.de/suhl/bizer/SWTSGuide/index.htm from DanC_jam

13:57:52 <DanC_jam> D:|Semantic Web Trust and Security Resource Guide

13:57:53 <dc_rdfig> Titled item D.

13:58:15 <gcc-O3> does the linkbot remember links from previous days? it only lets me see the last 5

13:58:22 <DanC_jam> D:sent to [daml-security|http://www.daml.org/listarchive/daml-security/0004.html] 30Jan by Chris Bizer

13:58:22 <dc_rdfig> Added comment D1.

13:58:41 <DanC_jam> the bot loses its marbles daily. but http://rdfig.xmlhack.com/ remembers

14:11:31 <libby> whose is the cool datum bot btw?

14:11:52 <Morbus> i think aaron has something to do with it.

14:12:34 <Morbus> anytime he showed up in #disobey, datum magically followed.

14:12:50 <libby> I wonder if it listens for everything with a . in front at start of a line

14:12:50 <libby> I'm wondering because I have a bot that knows about meetings, an I'm not sure what syntax to give it

14:13:05 <libby> heh

14:13:07 <libby> .hello

14:13:12 <Morbus> libby: yeah, it's . for the prefix, you can get a list of commands by msg'ing the bot.

14:13:18 <libby> thanks

14:13:22 <Morbus> i'm pretty sure its a logger too.

14:13:50 <libby> nice

14:28:45 <libby> C4:""

14:28:45 <dc_rdfig> Deleted comment C4.

14:29:04 <libby> .time utc

14:29:04 <datum> 2003-04-10T14:29:04

14:29:49 <libby> BLURB:query test meet agenda item 1: [E] implementing the manifest and resultset formats: how far have people

14:29:49 <dc_rdfig> E: query test meet agenda item 1: [E] implementing the manifest and resultset formats: how far have people from libby

14:30:08 <libby> E:got and what issues have they run into

14:30:08 <dc_rdfig> Added comment E1.

14:30:32 <libby> BLURB:query test meet agnda item 2: [F] query events at www2003

14:30:33 <dc_rdfig> F: query test meet agnda item 2: [F] query events at www2003 from libby

14:30:49 <libby> BLURB:query test meet agnda item 3: [G] Describing triple-pattern queries in RDF/XML, N-Triples, N3.

14:30:50 <dc_rdfig> G: query test meet agnda item 3: [G] Describing triple-pattern queries in RDF/XML, N-Triples, N3. from libby

14:30:55 <ericP> brb (2 mins)

14:31:29 <libby> ok, it's 1430 UTC, so time for RDF query tests meeting...

14:31:29 <libby> ------RDF query tests meet ------

14:31:52 <libby> thanks for coming everyone who is here for query, and apologies to others- we're scheduled to last an hour

14:32:13 <danbri_fr> danbri_fr is now known as danbri

14:32:28 <libby> those who are here, would you add your names to the attendee list?

14:32:37 <libby> BLURB: attendee list for query test meet

14:32:38 <dc_rdfig> H: attendee list for query test meet from libby

14:32:59 <libby> H:[Libby Miller|http://ilrt.org/people/libby/]

14:32:59 <dc_rdfig> Added comment H1.

14:33:27 <libby> the genda for the meeting has been chumped and so is at http://rdfig.xmlhack.com

14:33:48 <libby> darn, there goes alberto....

14:33:49 <danbri> danbri has changed the topic to: #rdfig RDF query testcases collaboration: 14:30UTC 1hr (now!), see http://rdfig.xmlhack.com/

14:33:53 <Jeen> H:[Jeen Broekstra|http://sesame.aidministrator.nl/]

14:33:53 <dc_rdfig> Added comment H2.

14:33:56 <libby> thanks danbri

14:34:10 <tim|mit> H:[Tim Berners-Lee|http://www.w3org/People/Berners-Lee]

14:34:10 <dc_rdfig> Added comment H3.

14:34:35 <danbri> H:[Dan Brickley|http://www.w3.org/People/DanBri/], W3C.

14:34:35 <dc_rdfig> Added comment H4.

14:34:38 <libby> item 1: [E] implementing the manifest and resultset formats: how far have people got

14:34:48 <AndyS> [Andy Seaborne|http://www.hpl.hp.com/semweb/]

14:35:03 <AndyS> H:[Andy Seaborne|http://www.hpl.hp.com/semweb/]

14:35:03 <dc_rdfig> Added comment H5.

14:35:14 <libby> last time seberal people said they would try to implement alberto's manifest format and andy's resultset forma, including mew

14:35:20 <libby> s/me/mew/

14:35:29 <arjohn> H:[Arjohn Kampman|http://sesame.aidministrator.nl/]

14:35:33 <dc_rdfig> Added comment H6.

14:36:04 <AndyS> s!s/me/mew/!s/mew/new/! ?

14:36:14 <arjohn> I have implemented Andy's resultset format in Sesame for SeRQL

14:36:36 <libby> hey jeen, arjohn, tim|mit

14:36:36 <arjohn> a demo is available from http://sesame.aidministrator.nl/sesame/

14:36:36 <libby> nice!

14:36:52 <alberto> H:[Alberto Reggiori|http://reggiori.webweaving.org/]

14:36:58 <libby> are you planning on making any tests avilable as a download some time arjohn?

14:36:58 <dc_rdfig> Added comment H7.

14:37:19 <danbri> arjohn, that's great :)

14:37:26 <arjohn> what exactly do you mean with 'tests"

14:37:28 <AndyS> I looked at Alberto's manifest format and think it is a semantic superset of what I use.

14:37:38 <AndyS> If so - I'll change to use it sometime.

14:37:58 <libby> E:arjohn has implemented andy's resultset format in serql [demo|http://sesame.aidministrator.nl/sesame/]

14:37:59 <dc_rdfig> Added comment E2.

14:38:07 <tim|mit> I have worked a bit on proof but not on converting N3 to Andy's format. The proof work led to a design of a vocabulary for desbribing N3, including RDF of course and also includeing formule, and so including rules.

14:38:14 <ericP> i've made references to andy's resultSet taxonomy from the query survey (http://www.w3.org/2001/11/13-RDF-Query-Rules/#taxonomyInterop)

14:38:22 <Jeen> arjohn: I think the idea is to use the manifests, which specify a query, a source dataset and an expected result set, and use those to test the query engine.

14:38:58 <libby> arjohn, yep, what jeen said - I've had a go for a squish variant: http://swordfish.rdfweb.org/rdfquery/tests/

14:39:00 <Jeen> I am working on implementing such a tool for sesame, and doubtless we'll also come up with some test cases of our own... can't say when yet though.

14:39:11 <libby> this is great :))

14:39:18 <danbri> I spent a bunch of time on creeping professionalisation of rubyrdf, ie. debian packaging, tests etc. Didn't yet switch to new manifest format but most of the infrastructure is in place to do so now.

14:39:27 <libby> E:[squish tests available|http://swordfish.rdfweb.org/rdfquery/tests/]

14:39:27 <dc_rdfig> Added comment E3.

14:39:47 <alberto> my connection today is a bit fuzzy - sorry

14:39:53 <AndyS> The namespace URI for the resul set now does something if you GET it.

14:39:53 <danbri> I had some trouble with deciding whether to do code generation of ruby test scripts (my final design) versus loop through each test file within a single test (original design; annoying).

14:40:01 <danbri> cool, thanks andy

14:40:28 <alberto> how sesame people solved the results set "equality" problem DanBri mentioned in email on www-rdf-rules?

14:40:39 <danbri> I also just sent mail (sorry not <24hr before now) on methods for actually _using_ the resultset format...

14:40:45 <libby> E:see[tests repository |http://www.w3.org/2003/03/rdfqr-tests/] for manifest and resultset formats

14:40:45 <dc_rdfig> Added comment E4.

14:40:56 <arjohn> andy: the 'rs' namespace in the example doesn't seem to be correct.

14:41:22 <Jeen> alberto: we haven't solved it _yet_, but I plan to use the sesame client lib internal table representation of result sets and compare those...

14:41:27 <danbri> ie. need a way to compare actual vs expected values in table of results. I would love to know more about how people here are comparing resultsets

14:41:41 <AndyS> Bother - what is it in the example?

14:41:59 <alberto> the problem is all about those bNodes :-( thanks Jeen

14:42:27 <AndyS> OK - I see I left the placeholder in Will fix.

14:42:27 <danbri> yes, all those bnodes!

14:42:47 <libby> I may have copied the placeholder ns

14:43:12 <danbri> jeen, doesn't that approach rely on your test data (and expected query result doc) knowing which bnode ids will be assigned?

14:43:16 <alberto> I have not been there yet danbri, but I would use some graph equality alg (RDF test cases one in C++ or something)

14:43:32 <libby> Andy mentioneed an algoritm earlier?

14:43:32 <arjohn> First, I copied the placeholder too, but it didn't matcht the resource for 'null' values ;-)

14:43:43 <AndyS> It's right earlier on and in the other files - just the example is messed up

14:43:51 <danbri> it isn't _quite_ same problem as graph compare.

14:44:01 <libby> alberto, it need not be a complete graph matcher though right?

14:44:05 <ericP> danbri, comparing result sets, if the results are reported in a tabular form, you can sort and diff

14:44:08 <danbri> though that gives you most of what you want, and the bnode labelling strategies are similar

14:44:12 <AndyS> I now use graph compare.

14:44:23 <libby> jena graph compare very useful

14:44:24 <tim|mit> Why is it not graph compare?

14:44:26 <danbri> ericp, you can't sort easily, due to different bnode names

14:44:45 * danbri tries to explain his (poss. muddled) thinking:

14:44:45 <tim|mit> Jena graph compare doesn't do datatypes.

14:44:48 <alberto> erip, sort and diff is what SQL people would do, of course! but we have bNodes which you can "not* compare :(

14:44:54 <tim|mit> I made a graph canonicalizer to cann'ze NTriples.

14:45:00 <Jeen> danbri, I haven't considered that yet, but I don't think so. why would there be different bnode names?

14:45:00 <alberto> s/erp/ericp/

14:45:01 <danbri> Two ways we could try to solve this with 'graph compare".

14:45:15 <danbri> (i) take each binding table, and rebind to query to get a graph. and compare those.

14:45:21 <ericP> is presupposing tabular, i assumed a report subset that had only bindable variables or a common string like "-unknown-"

14:45:53 <danbri> ... if different, yes we found a differnce. But could be same and still have different tabular structure, so long as when bound back into graph got the desire triples.

14:46:37 <ericP> i don't think this "hack" sacrifices much and is quite eady to do

14:46:37 <AndyS> Jena2 should do graph compare on the syntactic graph inc datatypes but not sameValueAs on literals

14:46:37 <alberto> ericp, that make sense, like 'undef' in most languages

14:46:37 <danbri> (ii) treat each row as an rdf graph, and compare it to the corresponding row from the other table.

14:46:37 <AndyS> Have we missed something?

14:46:39 <ericP> you just have to convince your reporting system to assign a common label for bnodes

14:46:51 <danbri> ...probs: Firstly, pairing up the rows. But also, we have the constraint not only that the graph the row makes must match its partner, but also the bits must be in the right colum.

14:47:07 <tim|mit> AndyS, I was using Jena 1.6.1 Should I switch to 2?

14:47:44 * danbri stops thinking out loud, catches up reading what you folks said.

14:47:47 <AndyS> tim - yes! (If you don't need persistence)

14:48:16 <AndyS> 1.6.1 has no datatype support expect for reading them (and forgetting them)

14:48:30 <tim|mit> Ok, will do.

14:48:32 <alberto> danbri, (i) is about substituting bound vars into the query patterns and see if we get the original doc ?

14:48:50 <danbri> ericP, re 'common string like -unknown-', if you rewrote each bnode id to be 'unknown', yes you could compare, but there would be cases that test would miss.

14:48:52 <AndyS> danbri (iii) record whole result set in a graph - or is that (i)

14:49:04 <danbri> that was whati meant by (i), yup.

14:49:09 <ericP> danbri, there would be false negatives

14:49:14 <danbri> yup

14:49:41 <ericP> tends not to come up in tabular form as each solution has a a godd number of distinguishing result strings

14:49:42 * Jeen just realizes his question was bit dumb. grokked the problem now :)

14:49:45 <alberto> andy, is Jena model.compare() implementing Jan Grant NTC algorithm ?

14:49:47 <danbri> i think something close to what eric said, but instead of 'unknown' on each bnode, write 'notknownmuchabout-1', 'notknownmuchabout-2' etc on them, based on what we do know.

14:49:47 <tim|mit> I am surprised that eth result set os not alreday descrtibed as an RDF graph.

14:49:48 <AndyS> if (i) is substituting need to be careful about collapsing due to multiple ways to get same subgraph (see report)

14:50:19 <AndyS> Result set is an RDF graph in the result set vocabulary.

14:50:35 <ericP> danbri, uniqifying the bnodes gives you the same problem again

14:50:39 <danbri> eg. one bnode appears to have 3 dc:titles. Another is pointed to by both foo:owns and bar:likes. So we can describe them by their charcteristics a bit. perhaps enough to improve sorting and compare...

14:50:53 <alberto> [NTC: A Simple N-Triples isomorphism utility|http://www.w3.org/2000/10/rdf-tests/rdfcore/utils/ntc/]

14:51:06 <AndyS> alberto - check with Jeremy (who wrote graph compare a while ago).

14:51:19 * tim|mit doesn't understand the bnode problem.

14:51:24 <AndyS> Jeremy looked at the graph isomorphism literature

14:51:30 <tim|mit> Maybe there is a problem with the vocab

14:51:32 <danbri> result set is indeed *described* (almost 'reified' but lets not use that term) in RDF. So hmm could you just diff that? no, because of bnodes...

14:51:40 <danbri> yes jeremys paper was nice...

14:52:04 <alberto> andy, ah ok the ISWC paper one

14:52:06 <libby> E:[A Simple N-Triples isomorphism utility|http://www.w3.org/2000/10/rdf-tests/rdfcore/utils/ntc/]

14:52:10 <dc_rdfig> Added comment E5.

14:52:10 <tim|mit> What is the bnode problem? Just canonicalizing?

14:52:15 <AndyS> I think the bNode problem is for Dan's matching scheme

14:52:46 <arjohn> How does one know that a returned bNode matches a certain bNode from the source data?

14:52:50 <ericP> tim, i believe so

14:52:55 <arjohn> It can have a totally different ID.

14:53:01 <tim|mit> E: [A simple buggy but useful canonnicalizer fro NTriples|http://www.w3.org/2000/10/swap/cant.py]

14:53:01 <dc_rdfig> Added comment E6.

14:53:05 <arjohn> That's the bNode problem, right?

14:53:16 <ericP> it is comupationally hard to know if two graphs "match" if they have many bnodes

14:53:21 <tim|mit> It is soluble in practice

14:53:44 <tim|mit> For the actual graphs in the rdfcore parser tests for example, cant.py will work

14:53:55 <tim|mit> There are wird graphs which it won't work for

14:53:56 <AndyS> Graph compare tries to find a consistent mapping from bNodes in one graph to the bNodes in the other (and match URIs elsewhere)

14:54:02 <tim|mit> s/wird/weird/

14:54:31 <AndyS> JJC's worse case is an all bNode graph of 64+ bnodes, all properties the same. Yuk!

14:55:35 <danbri> am i seeing a problem where this is none? (re bnodes etc). I haven't spent a lot of time on this bit of the puzzle, so don't let me derail things. I'm just looking for some code that'll take a table parsed out of the RDF resultset format, and a table of bindings from talking to my query engine...

14:55:44 <tim|mit> Dr. Jobb's Journal of May 2003 has an article and python code for c'n of graphs.

14:55:55 <ericP> if tere is a large set of tests for which the cant.py algorythm will work, we can sequester those and make it easy to test conformance on those cases.

14:55:55 <danbri> ..and tell me if the answer from the query system is obviously goofy.

14:56:05 <danbri> ...regardless of which parser etc i used, what bnode ids it assigns etc.

14:56:41 <tim|mit> danbri, could you try running them both through cant.py and let me know what happens?

14:56:41 <AndyS> Why not use the general purpose utility? No false results

14:56:55 <danbri> I agree that bnodes are (it seems) used infrequently enough for this not to be a big issue.

14:57:16 <Jeen> unless you query DAML sources, that is :)

14:57:29 <danbri> AndyS, use it by comparing the entire resultsets (turned back into RDF, ie the matched subgraph), you mean?

14:57:29 <AndyS> FOAF?

14:57:48 <DanC_jam> .esw FOAF?

14:57:51 <ericP> exact graph match is not as difficult as graph diff problem which should notice that ugendi718 is a Person and should really be a BluePerson

14:57:51 <libby> perhaps we can try out some algorithms and test them against tims and jjeremy's graph matchers for next time?

14:57:59 <tim|mit> So long as the bNodes are in an environemnt with URIs on arcs and URUIs or lits on nodes, then the graphs c'nize easily.

14:57:59 <DanC_jam> .google FOAF

14:57:59 <datum> FOAF: http://xmlns.com/foaf/0.1/

14:58:00 <dc_rdfig> Label FOAF not found.

14:58:01 <danbri> tim|mit, i'll take a look at cant.py, yep.

14:58:10 <AndyS> No - result set is a graph but not the matched subgraph - it encodes the result set itself.

14:58:28 <libby> that an action danbri?

14:58:31 <AndyS> FOAF has lots of bnodes does it not?

14:58:31 <danbri> Oh so you just compare the more 'meta' graph?

14:58:34 <danbri> yeah, action me.

14:58:52 <danbri> foaf does indeed use a fair number of bnodes (for people, typically). but plenty of uris too.

14:58:54 <libby> if people are agreeable, I'd like to set this aside for now, and spend a few minutes on the manifest

14:58:59 <libby> format

14:59:12 <AndyS> Yes compare the "output" graph

14:59:13 * DanC_jam noodles on topic names for meeting support RDF vocabularies

14:59:15 <libby> has anyone tried implementing that? any issues?

14:59:19 <DanC_jam> MeetingAgents

14:59:43 * DanC_jam realies a meeting is in progress. oops.

15:00:02 <libby> E:ACTION danbri take a look at cant.py for comparing graphs

15:00:04 <dc_rdfig> Added comment E7.

15:00:42 <ericP> tim, i think you can canonicalize bnodes if you have arc labels and no repeated props with bnode objects

15:00:47 <libby> alberto, did you implement the manifest format?

15:01:00 <danbri> I redid my own manifest stuff, as http://www.w3.org/2001/12/rubyrdf/pack/tests/_genq.rb so it should be easy this week to migrate to alberto's.

15:01:15 <alberto> no yet libby, just converted Andy's RDQL jena tests to the new format

15:01:40 <libby> mine were very strightformward to convert. my only issue was the number: some of mine have 5a, 5b etc

15:01:51 <libby> cool

15:01:55 <alberto> I will probably by next meeting, trying out one of those graph-isomorphism algs for matching results

15:02:04 <libby> are you up for making them avilable somewhere alberto, andys?

15:02:11 <AndyS> I look at Albertos manifest - has information I need so converting to it is straight forward

15:02:31 <Jeen> I am only just starting on implementing, but hopefully will have something to show for it in a couple of weeks

15:02:31 <alberto> libby, I dropped and email to andy (cc-ed to you and danbri) - the perl script to convert the jena ones is also attached

15:02:40 <libby> because I think we're almost at the stage where we can try out each other's tests!

15:03:06 <libby> great alberto. I'd like to see them online somewhere whne you and andy are happy with them

15:03:09 <libby> hey shellac

15:03:15 <alberto> andy, can not we give to the test manifest.rdf a URL (temporary) one ? CVS ?

15:03:15 <shellac> hi libs

15:04:06 <alberto> libby, it would be cool if you could try (at some moment) your syntax-translator/mapping tool on the manifests :-)

15:04:06 <libby> so one issue I had was that I was tryint to separate out the tests where you need to use constraints like ~ or > or eq from ones taht were just a stright conjucntive query

15:04:08 <AndyS> alberto - you mean upload it to W3C WWW/CVS?

15:04:13 <danbri> My tests have some more stuff to do with testing the syntax of a particular query language, or maybe all these samey query languages. Number statements in the query, etc.

15:04:37 <libby> ...so I put the word 'constraint' in all the relevant summaries, but then those which only do conjuctive queries wouldnt be able to fuind out which they could not use!

15:04:41 <danbri> ... i think i should just try to mix in bits of my namespace into general framework from alberto's doc.

15:04:49 <alberto> andys, yes or somehwere else - just to refere to it

15:04:52 <danbri> I hereby promise to do just that!

15:05:02 <libby> excellent

15:05:15 * DanC_jam likes actions chumped under agenda items

15:05:18 <AndyS> Can do but I'm away from this PM! Might not get done unless someone else does it (to be realistic)

15:05:29 <libby> E:ACTION danbri mix in his namespace for syntax of query with alberto's format

15:05:29 <dc_rdfig> Added comment E8.

15:05:39 * DanC_jam thx

15:05:43 <AndyS> You could send to www-archive@w3.org

15:05:57 <libby> good itea andys

15:06:10 <danbri> E:...in rubyrdf test cases, and circulate pointer

15:06:18 <libby> or to rdf-rules?

15:06:18 <dc_rdfig> Added comment E9.

15:06:37 <danbri> yes, lets use www-rdf-rules...

15:06:37 <Graco> hi

15:06:44 <libby> ericp, do you have plans to try out these test frameworks with algae?

15:06:45 <danbri> hi Graco

15:07:18 <Graco> what about this channel ?

15:07:18 <ericP> try out these test frameworks, yup

15:07:35 <tim|mit> danbri, I'd like to sync with you on your action item too

15:08:13 <libby> cool ericp. do you ahve a test suite?

15:08:28 <libby> tim|mit is that an action?

15:09:12 <libby> graco, the channel is for discussing RDF http://www.w3.org/RDF. we're mid meeting (it ends in 20 minurtes), discussing RDF query testcases

15:09:37 <libby> ok, how about next agenda item:

15:09:51 <alberto> H:[Jena RDQL tests using new manifest|http://rdfstore.sourceforge.net/2003/04/10/rdql-tests/Manifest.rdf]

15:09:55 <libby> 2. RDF query events at www2003

15:09:55 <dc_rdfig> Added comment H8.

15:09:55 <tim|mit> I'd like Danbri's action on possibly refactoring namespaces (IIRC) to include checking with me as I have $SWAP/reify# and $SWAP/reason# which are related

15:09:59 <ericP>http://dev.w3.org/cvsweb/perl/modules/W3C/Rdf/test/

15:10:00 <dc_rdfig> I: http://dev.w3.org/cvsweb/perl/modules/W3C/Rdf/test/ from ericP

15:10:13 <ericP> I: | algae test cases

15:10:13 <dc_rdfig> Added comment I1.

15:10:13 <danbri> tim|mit, ACK'd. I'll check with you...

15:10:18 <ericP> (reply to libby's query)

15:10:25 <libby> thansk ericp :)

15:10:52 <Graco> libby: ok tnx

15:10:59 <AndyS> Tim - are those compatible with http://www.w3.org/2003/03/rdfqr-tests/recording-query-results.html

15:11:14 <alberto> H: Alberto and Andy's work to map Jena RDQL tests to new manifest [see here|http://rdfstore.sourceforge.net/2003/04/10/rdql-tests/Manifest.rdf]

15:11:17 <dc_rdfig> Added comment H9.

15:11:25 <libby> E:tim|mit I'd like Danbri's action on possibly refactoring namespaces (IIRC) to include checking with me as I have $SWAP/reify# and $SWAP/reason# which are related

15:11:25 <dc_rdfig> Added comment E10.

15:11:49 <libby> em: Alberto and Andy's work to map Jena RDQL tests to new manifest [see here|http://rdfstore.sourceforge.net/2003/04/10/rdql-tests/Manifest.rdf]

15:11:52 <libby> arse

15:11:56 <libby> E:Alberto and Andy's work to map Jena RDQL tests to new manifest [see here|http://rdfstore.sourceforge.net/2003/04/10/rdql-tests/Manifest.rdf]

15:11:56 <dc_rdfig> Added comment E11.

15:12:02 <libby> thanks alberto

15:12:19 <libby> danbri, Iwondred if you'd say somethoign about the bof at www2003?

15:12:32 * danbri strikes 'arse' from the arsing record. this is a W3C-related forum, sweary mary ;)

15:12:39 <danbri> OK, this can be quick.

15:12:47 <alberto> libby, danbri, myself and andy would wait for www-rdf-rules to get some code running those tests

15:12:49 <libby> sorry danbri

15:13:23 <danbri> There is a conference coming up that some of your will have heard of, some of you may even be attending. WWW2003. I've signed up for a BOF (birds of a feather) lunchtime session on RDF Query testcases.

15:13:30 <alberto> SF URL is enough for the moment but I will email to www-archive too later

15:13:42 <AndyS> I already have the tests running expect I use a reduced manifest format (will fix that "soon")

15:14:01 <danbri> See url in the agenda for details. I am still waiting for final 'ack' from the organisers, i expect they're overloaded, but I think the meeting is a 'go'.

15:14:17 <libby> F:[see Wiki entry for WWW2003 Budapest Birds of a feather meet|http://esw.w3.org/topic/RdfQueryTestingBudapestMeeting]

15:14:17 <dc_rdfig> Added comment F1.

15:14:22 <libby> F:<danbri> See url in the agenda for details. I am still waiting for final 'ack' from the organisers, i expect they're overloaded, but I think the meeting is a 'go'.

15:14:23 <dc_rdfig> Added comment F2.

15:14:29 <danbri> I'm aware that not everyone can be there, but for those of us who are, its a chance to have a higher-bandwidth connection...

15:14:43 <alberto> danbri, I will be there :)

15:15:05 <libby> great alberto

15:15:08 <danbri> As with the 'rdf query futures' discussion at the W3C Tech Plenary last month, I'll try to make sure there's a decent record for non-attendees. Hopefully realtime / irc.

15:15:10 <Jeen> it looks like I may join in after all in Budapest...

15:15:13 <alberto> just registered yesterday - shall we add our names to that WiKi page

15:15:17 <danbri> alberto, excellent. Jeen too, cool :)

15:15:38 <libby> excellent jeen! this'll be great :)

15:15:46 <libby> go for it alberto

15:15:49 <danbri> anyone else? AndyS, do you know if you'll be able to attend?

15:15:56 <danbri> ericp, you'll be there, right?

15:15:57 <AndyS> I should be there

15:16:01 * danbri will, semi-obviously.

15:16:06 <libby> :)

15:16:22 <libby> :) andys

15:16:22 * libby too

15:16:22 <ericP> danbri, there is WWW2003 query BOF? yes.

15:16:26 <danbri> cool

15:16:31 <danbri> yes! see url above.

15:16:43 <AndyS> Is it about query or just query tests?

15:16:49 <danbri> sorry i'm learning a new laptop, am having trouble typing/thinking/navigating as fast as poss.

15:17:06 <AndyS> High-bandwidth might suggest "query" which keeps on coming up here

15:17:15 <libby> I guess nearer the time we can decide on some specific ideas for what to do, based on 3 in url above.

15:17:22 <danbri> we only currently have a lunch slot so i kept it focussed to the query testing thing, but am open to suggestions.

15:17:45 <libby> I think we need very specific goals fo rsuch a short meet

15:17:45 <ericP> are folks likely to want to get together outside of the query BOF?

15:17:50 <danbri> As I started thinking of things we could cover, i realised a day would be easy to fill.

15:17:59 * Jeen would

15:18:00 <libby> for another bof?

15:18:04 * danbri will be around all week, evening/bar BOF possible too

15:18:12 <AndyS> Good to rules folks, DQL folks, involved

15:18:35 <AndyS> s/to/to get/

15:19:08 <libby> maybe we coudl ahve an evening bof on one toic and keep the lunch one testing

15:19:08 <danbri> Yes, i noticed some new stuff from daml folks lately

15:19:08 <danbri> are we done with this agenda item? Libby could you mention the BOF in the meeeting summary? (sorry i need to step out the meeting briefly...)

15:19:08 <ericP> i'm at a DAML meeting now, will look for DQL folks who will be in budapest

15:19:14 <libby> sure

15:19:14 <libby> thanks danbri

15:19:22 <libby> thanks ericp, good plan

15:19:32 <tim|mit> Andy, you asked about compat of my vocabs with your result vocab. I reify the term which something is bound to as a thing which is explcitly eitehr a symbol with URI or a literal with string. You seem to just ahve a literal

15:19:36 <danbri> ericp, thanks!

15:19:36 <libby> ok, last agenda item 10-ish minutes

15:19:49 <libby> query test meet agnda item 3: [G] Describing triple-pattern queries in RDF/XML, N-Triples, N3.

15:19:56 <alberto> waves to danbri and ericp

15:20:25 <libby> re queries in ntriples,and in rdf generally, we started to talk about this last time, and it was interesting but we had to stop

15:20:48 <libby> I've been playing around with this a lot becauseI wanted a very small implementation -> no xml parsing

15:21:29 <AndyS> tim - can that capture something result-set vocab can't? e.g. unnamed vars (bNodes in paths??)

15:21:29 <tim|mit> This is mostly the same as describing RDF, except that it incldues variables.

15:21:29 <alberto> libby, is this also related to the "constraints" story? i.e. how to expressed complex queries

15:21:42 <libby> the characteristics I've found I need are some way of identifying the variables you actually want to return

15:21:51 <AndyS> libby - got an example?

15:21:52 <alberto> yes with (no standard) bArcs

15:21:57 <ericP> mike dean will be at WWW2003 query BOF. he is involved in DQL.

15:22:19 * ericP waves to alberto

15:22:27 * tim|mit wonders whether this IRC chat should somehow be merged with the DAML query bof

15:22:33 <libby> if people qwant to ad themselve to http://esw.w3.org/topic/RdfQueryTestingBudapestMeeting re BOF as alberto suggested, that's be great

15:22:44 <alberto> libby, I think if you have the manifest they are in Andy's result sets

15:22:56 <libby> there's a daml query BOF tim|mit?

15:23:17 <tim|mit> There is/was a breakout, no, ericP?

15:23:19 <tim|mit> was ...

15:23:23 * AndyS wonders if there is a ref to DAML query BOF

15:23:36 <ericP> there was a query breakout at the DAML PI meeting

15:23:43 * libby vaguely remmeber someone talkign about it - maybe there were notes?

15:23:49 <ericP> 1300-1600Z yesterday

15:24:03 * libby remembers ericp talking about it :)

15:24:13 <tim|mit> Any takeways to summarize in one line?

15:24:34 <ericP> benjamin grossof is doing a summary right now. he will make hi slides public.

15:24:42 * tim|mit apologizes for distracting from agenda

15:25:49 <danbri> 'This is mostly the same as describing RDF, except that it incldues variables.'

15:25:57 <danbri> eek, netsplit.

15:26:06 <danbri> lets wait a minute see if it rejoins.

15:26:25 <alberto> auch....

15:27:15 <ChanServ> [#rdfig] This channel is logged and blogged: http://logicerror.com/rdfIRCWelcome

15:27:19 <Jeen> I'm actually very interested in that, it's on our todo list to come up with an n3/rdf notation for serql queries. but no work done on it yet...

15:27:39 <danbri> Are we all back?

15:28:07 <libby> I've been using Ntriples and you end up with this hacked together version of ntriples with preds as variables and extra syntax for the varaibles you want, and maybe special predicates for constraints like string matching and so on

15:28:07 * tim|mit notes netsplit over

15:28:07 <danbri> looks like it.

15:28:07 <alberto> danbri, do you have any test of n-triples for triple-patterns with constraints in?

15:28:07 <tim|mit> In N3, you write a query as the LHS of a rule

15:28:28 <libby> in a way it's a bit of a red herring, but it's nice to ahve a very simple syntax

15:28:49 <danbri> I'm sorta torn. I would like (before www2003) to show multiple systems using a basic (ntriples) representing of triples-to-match. But ultimately i think rdf/xml/n3/etc is the way to go

15:28:55 <AndyS> I had an experimental query/LHS=> result set term rule - worked nicely.

15:29:01 <libby> yes, ericp's document very useful for looking at ways of doing it with N3, ntriples etc

15:29:11 <tim|mit> There are a bunch of cwm tests which have queries and returned results

15:29:11 <danbri> ....as it is more annotational, gives you a place to put extra info without altering the base syntax.

15:29:11 <Jeen> well, someone i know is working on embedding serql queries in XSLT. for him an xml/rdf representation would be dead-useful.

15:29:48 <alberto> danbri, I agree - we first need the "filtering" part working i.e. grep

15:29:48 <danbri> alberto, i don't. I have no implemention of the constraints stuff. I wasn't keen on the way you andy and libby did it in squishql/rdql as a special case...

15:29:50 <alberto> without constraints - thanks

15:30:07 <tim|mit> { ?x :born ?b. ?b time:greaterThan "1967-12-031" } => { (?x ?y ) a :result }.

15:30:33 <alberto> at the moment I used RDQL-Document to tell to the machine that the URI (URL) points to a query specific textual file

15:30:38 <danbri> I think the 'constraints' stuff could be represented as rdf properties, esp now rdf core seem to admit that literals are (kinda) resources.

15:30:39 <tim|mit> I am very much infavor of the constrints being specified as RDF preciates too.

15:30:51 <libby> alberto, I've got an ntriples with a made-up predicate test somewhere

15:30:55 <alberto> i.e. I can put whatever I want in that, also constraints :)

15:30:59 <danbri> In fact I'd like to urge folks here to have a look at the xslt/xml query functions and operators spec, if you've not already.

15:31:08 <danbri> linked from www2003 bof page, hang on...

15:31:31 <danbri> ...idea is that some of their functions, the binary ones at least, we cuold claim were rdf predicates

15:31:35 <libby> hm, we're out of time

15:31:41 <alberto> url danbri of that xslt/xml query func?

15:31:53 <Jeen> that sounds useful

15:32:02 <libby> ----query testcases meeting adjourned-----

15:32:09 <tim|mit>http://www.w3.org/2000/10/swap/doc/Functions.html

15:32:09 <dc_rdfig> J: http://www.w3.org/2000/10/swap/doc/Functions.html from tim|mit

15:32:11 <libby> - but don;t let that stop you :)

15:32:16 <danbri>http://www.w3.org/TR/xquery-operators/

15:32:18 <dc_rdfig> K: http://www.w3.org/TR/xquery-operators/ from danbri

15:32:30 <danbri> K:|Xquery/xslt functions and operators

15:32:31 <dc_rdfig> Titled item K.

15:32:54 <alberto> danbri, very interesting :-)

15:32:56 <tim|mit> J: |List of XMLQ/XPath query functions and operators, with some comments about RDF.

15:32:56 <dc_rdfig> Added comment J1.

15:32:59 <danbri> K:Could these be used as RDF properties? The binary ones anyway! A non-existent swad-europe tshirt to anyone who implements such a thing...

15:33:00 <dc_rdfig> Added comment K1.

15:33:03 <tim|mit> J:In progress

15:33:03 <dc_rdfig> Added comment J2.

15:33:25 <Jeen> cool, a shirt!

15:33:29 <danbri> K:(I expect the tshirts to exist at some point later this year...)

15:33:30 <libby> heheheh

15:33:30 <dc_rdfig> Added comment K2.

15:33:39 <AndyS> Tim - The link http://www.w3.org/2000/10/swap/doc/Rules.html does not work for IE (problem with xmath DTD)

15:33:39 <libby> you'll have to make them now danbri

15:33:52 <libby> does IST SWAP hacve shirts jeen?

15:33:54 * danbri nods

15:34:00 <DanC_jam> .time

15:34:01 <datum> Thu, 10 Apr 2003 15:34:00 GMT

15:34:07 <tim|mit> K:See [Cwm Built-in properties|http://www.w3.org/2000/10/swap/doc/CwmBuiltIns]

15:34:07 <dc_rdfig> Added comment K3.

15:34:15 <ericP> i have too many tshirts now. how about socks and/or underwear?

15:34:15 <Jeen> libby, I'll have to check, I don't think so. only silly flyers I'm afraid.

15:34:32 <libby> us too at the moment...but soon...

15:34:33 <Jeen> I can pester some people into making shirts though, if you're really keen on one :)

15:34:46 * libby always like smore shirts

15:35:08 <libby> wow thre are tons of xquery operators

15:35:14 <danbri> ericP, underwear is a possibility but i expect it'll be shirt-based.

15:35:20 * ericP withdraws his minority, though sensible) request

15:35:20 <libby> dates ones woudl be useful

15:35:23 <danbri> (ie wear it how you like)

15:35:36 <danbri> libby, are you volunteering to look at the date ones?

15:35:37 <libby> socks would be good. everyone needs more socks

15:35:58 <libby> maybe

15:36:02 * libby noncommital

15:36:03 <Jeen> heh

15:36:15 <libby> I've tried to implement this sort of thing before and got in a terrible mess

15:36:18 * DanC_jam wonders if the ThinkGeek folks would/could help with SemanticWebGear

15:36:38 <tim|mit> K3:See [Cwm Built-in properties|http://www.w3.org/2000/10/swap/doc/CwmBuiltins]

15:36:38 <dc_rdfig> Replaced comment K3.

15:36:52 <danbri> okay, don't wanna overload you with it. I was thinking more along lines of 'does xquery pull one way; rdf-cal pull another' w.r.t. rep of time stuff in practical rdf use?

15:37:08 <Jeen> yeah, I was wondering about those. they look like a sensible starting point to me.

15:37:16 <libby> I'll have a skim, see how it goes.

15:37:29 <Jeen> (the cwm builtins, that is)

15:37:37 <danbri> Jeen, would you have any time to take a closer look at the xquery operators?

15:37:59 <tim|mit> (Danbri, http://www.w3.org/2000/10/swap/doc/Rules.html works for me on IE on OSX)

15:38:23 <danbri> I have an action from Semantic Web Coordination Group to seek an rdf representation. Hmm I should ask libby for a proper agenda slot on that next time...

15:38:36 <danbri> tim, it wasn't me w.r.t. Rules.html

15:38:57 <libby> ok, danbri

15:39:18 <tim|mit> Sorry - (AndyS, http://www.w3.org/2000/10/swap/doc/Rules.html works for me on IE on OSX)

15:39:22 <AndyS> Tim - not for IE6 for several people I have asked. Mozilla works :-) I'll email you the error

15:39:36 <tim|mit> tx

15:39:50 <libby> btw, I shoudl have said that we have query meetings evry other week at 1430 UTC, so the nest one is 2003-04-24

15:40:08 <Jeen> danbri, can't promise anything, will see what I can do though.

15:40:11 <Jeen> sorry to be vague

15:40:51 <tim|mit> Anyone looking for that shirt: it is quite easy to add built-ins to cwm.

15:40:55 <tim|mit> All contributions braetfully accepted.

15:41:04 <tim|mit> All contributions gratefully accepted.

15:41:28 <Jeen> I have to leave now, cheers everyone. same time in two weeks I reckon?

15:41:31 <libby> C:next meeting [http://www.timeanddate.com/worldclock/fixedtime.html?day=24&month=4&year=2003&hour=14&min=30&sec=0&p1=0|2003-04-24 UTC]

15:41:32 <dc_rdfig> Added comment C5.

15:41:32 <tim|mit> You tke something like http://www.w3.org/2000/10/swap/cwm_math.py and clone it

15:41:45 <libby> actually next meet is rather close to easter. ah well....

15:41:54 <tim|mit> Whatever happend to time URis anyway

15:41:59 <libby> thanks jeen, yep, 2 weeks

15:42:03 <Jeen> ok, seeya!

15:42:04 <tim|mit> Thank you, Libby!

15:42:40 <tim|mit> Regrets for next one - I will be in Wales.

15:43:20 <ericP> tim, would a meeting instance uri be a combination of a general meeting uri and a time uri?

15:43:42 <ericP> re whales, bring galoshes

15:43:44 <tim|mit> I don't think URIs should be used for coupound things.

15:43:55 <tim|mit> But for poinst in geospace and times they are interesting.

15:44:02 <ericP> i guess that works for wales, too

15:44:18 * DanC_jam wonders what a compound thing is

15:44:30 <ericP> tim mentioned time uirs

15:44:55 <tim|mit> compiund thing: a VEvent.

15:44:55 <tim|mit> eg

15:45:02 <ericP> i considered the case where one want to use that time uri to identiify a repeated event

15:45:13 <tim|mit> i don't think that compound datatypes are very useful as URIs

15:45:28 <tim|mit> However, simple datatypes as URIs often are.

15:45:32 <ericP> i don't either, just wanted to think about it for a moment

15:46:14 <tim|mit> Libgbry could have posted time:2004-04-24T15:00Z and your software could have been set yup to dot the right thing with it

15:46:41 <ericP> http://...rdfig-query-biweekly + time:20030410-1014:30

15:46:41 <dc_rdfig> L: http://...rdfig-query-biweekly from ericP

15:46:45 <Graco> excuse me, is there a defined standard for biological experiment and biological data

15:46:48 <Graco> ?

15:47:18 <tim|mit> I guess saying "2004-04-24T15:00Z"^^<xsd:time> is equiavelnt.

15:47:20 <ericP> no standard, and don't know of any schemas

15:47:36 <tim|mit> Graco, biochem?

15:47:40 <shellac> any particular part of biology?

15:47:56 <alberto> see you, leaving now...

15:48:07 <tim|mit> There is a lot of work in biochemistry and genetics

15:48:09 <gcc-O3> Graco: OASIS might have some schemas

15:48:14 <gcc-O3> or maybe working groups

15:48:20 <alberto> thanks libby for organising

15:48:38 <libby> you're welcome :)

15:48:43 <libby> bye alberto

15:48:43 <ericP> yes, tx

15:48:52 <shellac> JibberJim: if the logs are off why did you [off]? :-)

15:49:16 <tim|mit> Graco: A good person to ask would be Eric Neumann /Beyond Genomics

15:49:29 <Graco> tnx gcc-O3

15:49:56 <Graco> tim|mit: and where can i looking for him ?

15:50:06 <libby> logger 2 is here, but the logs webpages aren;t up yet

15:51:14 * danbri had to step out, sorry.

15:51:25 <danbri> thanks and bye and everything!

15:53:47 <DanC_jam> pls put the topic back if the meeting is over

15:54:07 <Graco> bye bye boys

15:54:39 <libby> boys!

16:04:47 * AndyS waves goodbye

16:04:59 <AndyS> Thanks to Libby for organising the meeting

16:05:02 <libby> bye ands! thanks for coming

16:05:05 <libby> :)

16:05:17 <libby> glad you can probably mae it to www2003

17:22:07 Topic now #rdfig RDF query testcases collaboration: 14:30UTC 1hr (now!), see http://rdfig.xmlhack.com/

17:22:07 Users on #rdfig: logger_1 DanC_jam deltab Morbus Oblomov larsbot xower datum grault dc_rdfig sbp` jang taaz fknow em sandro xover mea_culpa eikeon idoru aml ericP mwf TrustBot grove mariyo arnarl|away golbeck Wack gcc-O3 tim|mit dajobe

17:22:07 <ChanServ> [#rdfig] This channel is logged and blogged: http://logicerror.com/rdfIRCWelcome

17:36:48 <dajobe> the logger logs are back now

17:53:37 <tim|mit>http://www.ralphm.net/world

17:53:37 <dc_rdfig> M: http://www.ralphm.net/world from tim|mit

17:54:03 <tim|mit> M:|Fun world map of geographical presence

17:54:03 <dc_rdfig> Titled item M.

17:54:09 <tim|mit> M:Should be RDF ...

17:54:10 <dc_rdfig> Added comment M1.

18:47:55 <lilo> [Global Notice] Hi all. Apologies for the interruption, but we'd better cap test the new server. Users affected, about 1500.... results on WALLOPS if you're interested (/mode yournick +w)....

18:48:37 <lilo> [Server Notice] Hi all. You're on the server to be affected. It'll be pretty floody for a bit, but we should be right back if all goes well.

18:50:28 <ChanServ> [#rdfig] This channel is logged and blogged: http://logicerror.com/rdfIRCWelcome

19:02:49 <libby> sked, all

19:02:50 <skedulike> time now is 2003-04-10 19:02 UTC. Upcoming meetings concerning #rdfig:

19:02:51 <skedulike> 2003-04-16 14:00UTC Geo meeting #rdfig

19:02:51 <skedulike> 2003-04-23 15:00UTC Calendar meeting #rdfig

19:03:03 <libby> sked, help

19:03:03 <skedulike> I'm an interface to an RDF database of meetings

19:03:03 <skedulike> by Libby Miller, based on perl BasicBot.

19:03:03 <skedulike> type 'sked, all' to find all future meetings on #rdfig

19:03:03 <skedulike> you can add a search term after 'sked, all' e.g. 'calendar' 'query' 'geo'

19:03:04 <skedulike> type 'sked, 2003-04-10' to find events on that date.

19:03:04 <skedulike> to add meetings, see http://esw.w3.org/topic/ScheduledTopicChat

19:03:06 <skedulike> you can '/msg skedulike all' and so on too.

19:03:23 <libby> sked, 2003-04-16

19:03:23 <skedulike> time now is 2003-04-10 19:03 UTC. 2003-04-16 meetings concerning #rdfig:

19:03:23 <skedulike> 2003-04-16 14:00UTC Geo meeting #rdfig

19:03:34 <libby> sked, all query

19:03:35 <skedulike> time now is 2003-04-10 19:03 UTC. Upcoming meetings concerning query:

19:03:35 <skedulike> no results

19:03:40 <libby> sked, all geo

19:03:41 <skedulike> time now is 2003-04-10 19:03 UTC. Upcoming meetings concerning geo:

19:03:41 <skedulike> 2003-04-16 14:00UTC Geo meeting #rdfig

19:04:48 <libby> so to make this work well, I need to find a good way of harvesting the dates of meetings off the wiki. at the moment it has hardcoded pages that it knows about, e.g. RdfCalendar, GeoInfo

19:05:10 <DanC_jam> BLURB:Semantic Web Services Initiative Meeting

19:05:10 <dc_rdfig> N: Semantic Web Services Initiative Meeting from DanC_jam

19:05:16 <DanC_jam> N:in Miami, FL

19:05:17 <dc_rdfig> Added comment N1.

19:05:24 <DanC_jam> logger, pointer?

19:05:34 <dajobe> it's logger_1

19:05:35 <DanC_jam> logger_1, pointer?

19:05:35 <DanC_jam> See http://ilrt.org/discovery/chatlogs/rdfig/2003-04-10#T19-05-35

19:05:43 <libby> maybe if there was a way to signal that pages have events on them from the ScheduledTopicChats page?

19:06:02 <DanC_jam> N:Katia convened it just now. [some notes in #rdfig|http://ilrt.org/discovery/chatlogs/rdfig/2003-04-10#T19-05-35]

19:06:03 <dc_rdfig> Added comment N2.

19:06:38 <DanC_jam> N:previous meetings in Sardenia last June, in Portland (prev DAML PI meeting)

19:06:38 <dc_rdfig> Added comment N3.

19:07:26 <DanC_jam> N:Tim Finin has been asked to take some notes. (not sure where he's going to put them; probably www-ws, among other places)

19:07:26 <dc_rdfig> Added comment N4.

19:07:55 <DanC_jam> ooh... 802.11 service thru Sat

19:08:21 <DanC_jam> N:April 10-12, 2003

19:08:22 <dc_rdfig> Added comment N5.

19:09:09 <DanC_jam> N:Tim Finin has accepted scribe duties. (not sure where he's going to put them; probably www-ws, among other places)

19:09:10 <dc_rdfig> Added comment N6.

19:10:07 <DanC_jam> Katia: mailing list details to appera on a slide at the break.

19:10:36 <DanC_jam> N: agenda + overview of current state

19:10:37 <dc_rdfig> Added comment N7.

19:10:46 <DanC_jam> N: agenda + relation with W3C activities

19:10:46 <dc_rdfig> Added comment N8.

19:10:59 <DanC_jam> N: agenda + Business Case for Semantic Web Services

19:10:59 <dc_rdfig> Added comment N9.

19:11:12 <DanC_jam> N: agenda + Presentation (Language Committee)

19:11:12 <dc_rdfig> Added comment N10.

19:11:23 <DanC_jam> N: agenda + Presentation (Architecture Committee)

19:11:23 <dc_rdfig> Added comment N11.

19:11:46 <DanC_jam> N: agenda + Discussion (meeting objectives and Products, Procedures of operation, Action Items)

19:11:47 <dc_rdfig> Added comment N12.

19:12:03 <DanC_jam> N:also present: Dieter Fensel, Murray Burke, ...

19:12:04 <dc_rdfig> Added comment N13.

19:15:45 <dajobe> dajobe has changed the topic to: Semantic Web wish factory http://rdfig.xmlhack.com/

19:16:52 <libby> dave could you link http://www.w3.org/2002/12/cal/ on rdfig.xmlhack.com as the calendar taskforce please? the other page is not used much...

19:17:11 <dajobe> ack

19:17:19 <libby> ta

19:17:45 <libby> or the calendar mailing list, to make it consistent with the others maybe?

19:17:50 <libby> either

19:18:41 <dajobe> the others haven't got anything but the list?

19:19:02 <libby> well there's a testcases page for rdf query (and rules)

19:19:03 <dajobe> i'd add a geo wiki page link sinec that's a community in some sense

19:19:18 <DanC_jam> N6:Tim Finin has accepted scribe duties. (oops; mailing list of this group isn't publicly archived; not clear that DanC is licensed to report on this meeting in public)

19:19:19 <dc_rdfig> Replaced comment N6.

19:19:22 <libby> I think the interest group is the mailing list really

19:19:46 <libby> so it's a secret initiative?

19:19:59 <DanC_jam> libby, your calendar page is still the top google hit for "rdf calendar"; have you made a link to /2002/12/cal/ ?

19:20:06 <libby> oops, no

19:20:14 <libby> didnt realize it was top hit

19:20:17 <libby> will fix

19:20:33 <dajobe> that's why I linked to it

19:20:41 <libby> ah...

19:20:57 <libby> you shouldnt believe evrything google tells you ;)

19:21:12 <DanC_jam> well, I believe lots of people use google to find things.

19:21:19 <DanC_jam> ;)

19:21:19 <dajobe> you should fix the <title> on http://www.w3.org/2002/12/cal/

19:21:24 <dajobe> "Index of /..."

19:21:32 <DanC_jam> the <title> is hard to fix

19:21:58 <DanC_jam> If I change from HEADER.html to Overview.html , I'll have to manually index the contents of the directory

19:22:21 <dajobe> anyway, i've linked to it

19:22:53 <DanC_jam> I agree the <title> sucks. Perhaps the HEADER.html should at least apologize for that.

19:24:29 <Morbus> you can change the header with HEADER.html.

19:24:33 <Morbus> or you should be able to, rather.

19:24:36 <Morbus>http://disobey.com/detergent/

19:24:37 <dc_rdfig> O: http://disobey.com/detergent/ from Morbus

19:25:07 <dajobe> what's in the HEADER.html that does that?

19:25:28 <Morbus> http://disobey.com/detergent/.design/.HEADER.shtml

19:25:39 <Morbus> you've gotta mod your thing.

19:25:45 <Morbus> SuppressHTMLPreamble.

19:25:57 <dajobe> ah

19:26:29 * dajobe realises he already knows this: http://ilrt.org/discovery/chatlogs/rdfig/

19:26:33 <Morbus> http://httpd.apache.org/docs/mod/mod_autoindex.html#indexoptions

19:26:40 <Morbus> heh, heh.

19:28:34 <Morbus> dajobe, you may be interested in: http://disobey.com/detergent/code/request_uri_linker.txt

19:28:46 <Morbus> it just takes the "Index of path/path/path" and turns it into a breadcrumb trail.

19:28:57 <Morbus> its in use on /detergent/

19:29:12 <dajobe> it's not SSI

19:29:16 <dajobe> ... in that area

19:29:42 <Morbus> fair enough.

19:29:54 <dajobe> i've got some indexing stuff already prototyped

19:31:16 <dajobe> I prefer "fried" web pages ("fryed"-en-US ?)

19:31:25 <dajobe> i.e. dynamic pages that have been made static at an appropriate time

19:37:06 <Morbus> ah. ok. first time i've heard that phrase.

19:38:53 <dajobe> lol, I got it backwards

19:38:58 <dajobe>http://www.aaronsw.com/weblog/000404

19:38:58 <dc_rdfig> P: http://www.aaronsw.com/weblog/000404 from dajobe

19:39:07 <dajobe> P:|Bake, Don't Fry - Aaron

19:39:08 <dc_rdfig> Titled item P.

19:41:32 <Morbus> note to self: don't listen to dajobe ;)

19:42:16 <Morbus> my original cms, cowpoke-cms, did what MT did before MT did. i eventually stopped development on that when MT "caught up" as it were, right around 1.4.

19:42:56 <Morbus> thanks for the link.

19:44:30 <dajobe> i'm considering moving to blosxom

19:44:35 <dajobe> more hackable

19:44:41 <dajobe> (or java, py, whatever equiv)

19:44:52 <Morbus> i don't like blosxom due to the one file per post mnemonic.

19:44:57 <Morbus> that just feels massively ooky to me.

19:45:05 <Morbus> i certainly like the ease of grepping, unix tooling, etc.

19:45:26 <Morbus> but, i dunno. the thought of 2000+ files in an uncategorized directory, all named non-standard like, just makes me shiver.

19:45:38 <dajobe> mmm

19:45:46 <dajobe> yeah, lots of stat()s opendir()s etc.

19:45:54 <dajobe> going ot be slow

19:46:01 <Morbus> i think i've yet to find a proper means of using blosxom, so i've not actually implemented the damn thing. i've certainly helped test it. but just not used it production wise.

19:46:03 <dajobe> but blosxom can make a static site too

19:46:29 <Morbus> i was planning on using it for thehorrorsection.com, but it doesn't have user accounts, so it didn't work well with my attempt.

19:46:52 <dajobe> rael said something about using the underlying unix accounts for multi-user

19:46:56 <Morbus> likewise, there's no "remote editing" feature, i don't think.

19:47:07 <dajobe> you can use 'scp'

19:47:12 <dajobe> or anything else that will deliver a file

19:47:16 <Morbus> yeah, but see, that's assuming i'm gonna be at a machine that has that stuff.

19:47:25 <Morbus> if i'm at a public library dummy, i can't post, edit, or do much of anything.

19:47:27 <Morbus> which suxors.

19:47:27 <dajobe> use email & a procmail filter

19:47:43 <Morbus> and editing would be...

19:47:45 <dajobe> you want a web interface?

19:47:52 <Morbus> ultimately, yeah.

19:47:52 <dajobe> How Hard Could That Be

19:48:03 <Morbus> user accounts via unix? bah. can't be installed on a win32 machine now.

19:48:05 <Morbus> ;)

19:48:20 <Morbus> also assumes that the webhost will create x blog users per 1 web host account.

19:48:21 <Morbus> silly.

19:48:40 <dajobe> I wouldn't say blosxom is for anyone bar the tinkering hacker

19:48:43 <dajobe> for the rest, MT

19:48:58 <Morbus> fair enough. if that's the audience, then sure, i gotta change my expectation.s

19:49:23 <Morbus> but the "magic" of blosxom was that its "small", "easily installed", etc.

19:49:39 <Morbus> i usually associate those with usability and "this is easier than the other product".

19:49:51 <dajobe> those parts are

19:49:59 <Morbus> nothing against rael, of course. i wrote a book with him, he rocks ;)

19:50:02 <dajobe> but not editing, online help etc.

19:50:05 <dajobe> sure

19:50:15 * dajobe sure he had something to ask about that book

19:51:10 <dajobe> i've just finished reading Google Hacks - a great book & format

19:51:28 <dajobe> Mac OS X Hacks is on the pile

19:51:28 <Morbus> yeah. i consulted on the perl scripts for that book.

19:51:30 <Morbus> tis good :)

19:51:42 <Morbus> linux server hacks is pretty good too.

19:51:50 <dajobe> of course, I don't do OSX except on an ibook

19:51:51 <Morbus> its been dogeared by my boss already.

19:52:16 <dajobe>http://www.oreilly.com/catalog/mcosxhks/

19:52:17 <dc_rdfig> Q: http://www.oreilly.com/catalog/mcosxhks/ from dajobe

19:52:23 <dajobe> Q:|Mac OSX Hacks

19:52:24 <dc_rdfig> Titled item Q.

19:54:20 <dajobe>http://www.oreilly.com/catalog/googlehks/

19:54:21 <dc_rdfig> R: http://www.oreilly.com/catalog/googlehks/ from dajobe

19:54:24 <dajobe> R:|Google Hacks

19:54:25 <dc_rdfig> Titled item R.

19:54:31 <dajobe> R:great format, book - I recommend it

19:54:31 <dc_rdfig> Added comment R1.

19:54:52 <dajobe> Q:on my pile, skimming it looks interesting (I'm not primarily an OSX user)

19:54:53 <dc_rdfig> Added comment Q1.

19:56:22 <DanC_jam> N6:Tim Finin has accepted scribe duties. to appear in [www-ws|http://lists.w3.org/Archives/Public/www-ws/]

19:56:22 <dc_rdfig> Replaced comment N6.

19:59:32 <larsbot> em, re review: first point was my reaction as well :-)

20:00:26 <DanC_jam> N:home page is [http://swsc.semanticweb.org/|The international joint Sematic Web Services ad hoc Consortium (SWSC)]

20:00:26 <dc_rdfig> Added comment N14.

20:00:51 * em sees DanC_jam taking notes from break-out meeting... excellent

20:00:55 thanks DanC_jam

20:01:26 <DanC_jam> it's not a break-out. it's its own 3-day deelie. just co-located with DAML PI

20:01:27 steve, larsbot... let me know when you've digested http://lists.w3.org/Archives/Public/www-archive/2003Apr/0020.html and / or if you have a questions

20:01:42 3-day deelie? didn't realize that

20:02:24 * pepper is still digesting...

20:02:40 are there links to other presentations , agenda for this thing? appreciate seeing notes from the parts your attending but if there is more info even better

20:02:56 danbri?

20:03:34 any foafers here by chance? i'm curious if any best practices for foaf have been writtent yet?

20:04:12 <DanC_jam> my impression of the foaf world is that it operates by monkey-see-monkey-do

20:04:49 yep... just wondering if any of the silverbacks have been recording their journey

20:06:24 <DanC_jam> N:seems there was a kick-off meeting in Innsbruk [sp?].

20:06:24 <dc_rdfig> Added comment N15.

20:06:36 <larsbot> it's Innsbruck,

20:07:09 * larsbot has read through the review now

20:07:30 * pepper is also ready

20:07:39 * em wait for pepper... then summarize key points and ask questions to start dialog

20:08:20 <pepper> you want us to summarize the key points, em, or will you? or shall we go through your comments?

20:08:32 quick review of http://www.ontopia.net/topicmaps/materials/identitycrisis.html

20:08:36 1) good paper

20:08:40 2) please change title

20:08:47 <pepper> 1) thx

20:08:50 <pepper> 2) sorry

20:08:56 <pepper> apologies for the title of the paper. not meant to be confrontational. intended to attract attention (and get paper accepted for XML Europe :)

20:09:10 <pepper> as you know - it was taken from an article on XML.com

20:09:15 3) best practices... what to put into specs vs what to leave to communities to resolve

20:09:39 4) possible next step suggestions for supporting #3

20:10:12 <pepper> tell us when we get to </ol>

20:10:14 re #3 .. does the notion of best practices make sense? and not sticking this directly into spec?

20:10:29 * em thinks chumping what were talking about makes sense

20:10:43 <pepper> best practices are always useful, but i don't think they are the key issue here

20:10:57 <DanC_jam> I http://esw.w3.org/topic/SubjectIndicator

20:10:59 <DanC_jam>http://esw.w3.org/topic/SubjectIndicator

20:11:00 <dc_rdfig> S: http://esw.w3.org/topic/SubjectIndicator from DanC_jam

20:11:15 <larsbot> em: #3: I think you need a combination

20:11:27 <pepper> it's possible to have a model which mitigates against "bad sense"

20:11:29 <DanC_jam> I don't mind "crisis" in the title. "Curing" I could perhaps live without.

20:11:45 "best practices are always useful, but i don't think they are the key issue here" ok... then lets explore on this for a bit...

20:11:49 <larsbot> best practices a la foaf can do most of it, but it's useful to have the concept of subject indicator

20:11:59 got urls handy?

20:12:06 <larsbot> for what?

20:12:29 http://www.ontopia.net/topicmaps/materials/identitycrisis.html

20:12:29 <dc_rdfig> T: http://www.ontopia.net/topicmaps/materials/identitycrisis.html from em

20:12:31 <larsbot> subject indicators? I'd point to Steve's paper :-)

20:12:40 <pepper> also: http://www.ontopia.net/tmp/pubsubj-gentle-intro.htm#s.2.4

20:12:47 T:|Review of Steve's paper

20:12:47 <dc_rdfig> Titled item T.

20:12:51 <jhendler> and ontologies might get some mention...

20:13:15 T:quick [review | http://lists.w3.org/Archives/Public/www-archive/2003Apr/0020.html] from em

20:13:15 <dc_rdfig> Added comment T1.

20:13:46 ok... figure if we agree to some points, we've got a handle to hang these off from and then link from http://esw.w3.org/topic/SubjectIndicator

20:13:54 <jhendler> (i.e. TM to OWL looks promising, but we haven't had a place of the "conversation" to occur)

20:14:35 thinks we should have that conversation soon (along with Eric Freese, etc. who've been working on tthis) but suggest not doing both now

20:14:40 <larsbot> jhendler: did you read my paper about TMs and OWL?

20:14:49 <larsbot> em: eh, ok, good thinking

20:14:55 <pepper> larsbot's paper: http://www.ontopia.net/topicmaps/materials/tmrdf.html

20:15:21 <pepper> let's stick to subject indicators...

20:15:29 thanks :)

20:15:57 does the point about IDing the metadata make sense to you?

20:16:17 <larsbot> no, not really

20:16:27 <larsbot> not sure whether the problem is the point or my understanding of it :)

20:16:37 * pepper has the same problem

20:17:05 ok... will try something else...

20:17:12 if i understood you to say...

20:17:35 part of the problem is using the same URI for descibing both the resource and the description of the resource

20:17:44 <pepper> exactly: basic premise is that URIs are being used in two fundamentally different ways. why not formalize this in the way TMs do?

20:17:51 yes|no|kindof?

20:18:16 <pepper> yes

20:18:18 <larsbot> yes

20:18:22 <pepper> kindof :-)

20:18:28 want to hold of a sec on specific solution of 'why not formalize this in the way TMs do?'

20:18:43 ok yes, yes and kindof :)

20:18:44 <pepper> kindof because of the problem inherent in the term "resource"

20:19:08 <tim|mit> I don't believe that there is a well defined function there

20:19:09 fair concern

20:19:37 <pepper> we believe there is a fundamental difference - on the web - between things in general and information resources

20:19:59 <pepper> "resource" blurs this

20:20:09 <tim|mit> If A is craeted to be a description of B (like a library card) fair enough, but for most documents A there is nota well-defined thing which A id descrtibing, and for most things C there is not a well-defined document which describes it.

20:20:36 <larsbot> entirely true, but doesn't really matter

20:20:46 <pepper> tim|mit: true - but you're jumping the gun

20:21:35 <pepper> shall we talk about "the fundamental difference"?

20:22:07 <DanC_jam> N:Austin Tate asks about connections to grid; mentions upcoming grid/web-services ftf thingy. [where when? Jun?]

20:22:08 <dc_rdfig> Added comment N16.

20:22:15 so one way in your example of adding metadata to the description of the resource was to associate dc:create and dc:date to http://www.w3.org/People/EM/contact not http://www.w3.org/People/EM/contact#me

20:22:35 * jhendler asks Lars to chump his paper again - I lost my pointer...

20:22:47 <pepper> maybe. but do we want to discuss examples or fundamentals first?

20:22:47 <larsbot>http://www.ontopia.net/topicmaps/materials/tmrdf.html

20:22:48 <dc_rdfig> U: http://www.ontopia.net/topicmaps/materials/tmrdf.html from larsbot

20:23:49 my preference is on examples actually... as a means to tease out fundamentals ... starting from fundamentals via irc as only means of communication i think has a less likely chance of success

20:24:00 <DanC_jam> N16:[Austin Tate|http://www.aiai.ed.ac.uk/~bat/] asks about connections to grid; mentions upcoming grid/web-services ftf thingy. [where when? Jun?]

20:24:01 <dc_rdfig> Replaced comment N16.

20:24:47 <pepper> why not go back to your point: URIs used for two different purposes?

20:24:51 <DanC_jam> N4:""

20:24:51 <dc_rdfig> Deleted comment N4.

20:25:14 <jhendler> DanC - there's a UK grid/services f2f, I think it was chumped a couple weeks back. also GGF has a group out of the ontology world in the loop

20:25:39 <pepper> 1) to identify a thing directly

20:25:39 <pepper> 2) to identify a thing indirectly via some "description"

20:26:01 my point about URIs used for two different purposes... hypothesis here is that its just bad descriptive practices

20:26:39 * em have tempted to jump to end of review to begin dialog

20:27:03 <pepper> ok, go ahead...

20:27:18 The suggestion the authors (and others @@ ref @@) provide are to

20:27:18 differentiate URIs via different relationships with the object.

20:27:31 <pepper> yes

20:27:32 e.g.

20:27:33 <contact:Document rdf:subject="http://www.w3.org/People/EM/contact#me">

20:27:34 <dc:date>2002/06/04</dc:date>

20:27:34 </contact:Document>

20:27:34 <contact:Person rdf:indicator="http://www.w3.org/People/EM/contact#me">

20:27:34 <contact:fullName>Eric Miller</contact:fullName>

20:27:36 </contact:Person>

20:27:44 note: a point of future discussion along the pedagogical / deployment

20:27:44 axis might be to s/rdf:indicator/rdf:about to reflect current (even if

20:27:44 potentially confused) practice.

20:27:45 <mariyo> hi, there. just got up. following the discussion.

20:27:45 <tim|mit> T:Apparently ignores/unaware of difference bteween contact and contact#me

20:27:46 <dc_rdfig> Added comment T2.

20:27:54 note: a point regarding interop with topicmap community might be to

20:27:54 discuss s/rdf:subject/tm:subject in the above example

20:28:27 <larsbot> let's get this straight: tm:subject == rdf:resource

20:28:41 re note 1 - i think your notion of rdf:indicator is along the lines of somes interpretation on rdf:about

20:28:48 larsbot, agreed

20:28:56 <pepper> tim|mit: yes, i was unaware of some of the nuances of the contact/contact#me distinction. it is rather subtle, you must admit :)

20:29:12 <tim|mit> No, it is not subtle.

20:29:14 pepper, then this is working :)

20:29:35 <pepper> em: absolutely

20:29:51 <pepper> tim|mit: for most users, it's pretty subtle, believe me

20:29:54 <tim|mit> pepper, no - The object identified by something with a "#" is COMPLETELY different from the obejct identified by the one without.

20:30:02 <tim|mit> Not subtle at all.

20:30:19 tim|mit, i think it is bit subtle but ths is the point we want to make clearer to a larger community

20:30:28 in big bold blink letters

20:30:29 <tim|mit> Most users may regard www.example.com and http://www.example.com as the same thing.

20:30:40 <pepper> undoubtedly

20:31:00 <tim|mit> One is a domain name, anothe ris an HTTP URI.

20:31:16 <tim|mit> They are related.

20:31:26 <tim|mit> They are not the ssame thing architecturally.

20:31:33 <bijan> Actually, it is subtle, I think.

20:31:43 <bijan> At least in part because we chain identifications.

20:31:56 <pepper> i know that - but for most users it doesn't mean a thing. look at the <url></url> elements in any set of XML conference proceedings papers.

20:32:23 <pepper> pepper is now known as trotsky

20:32:35 * em pokes trotsky

20:32:40 * trotsky pepper

20:32:41 come back pepper

20:32:45 <trotsky> trotsky is now known as pepper

20:32:52 <tim|mit> Its a bit like the distinction between p and *p in C. subtle.

20:33:10 ok, subtle aside for a sec... pepper, does the distinction make sense?

20:33:12 <bijan> Well, the sematnics are clear

20:33:21 <bijan> But the pragmatics trip people up.

20:33:39 <pepper> em: distinction is clear to me

20:33:47 larsbot?

20:33:56 <larsbot> I understand it, but I don't think it makes any sense

20:34:20 * em interesting test here...

20:34:23 <larsbot> I have two problems with it:

20:34:29 <larsbot> a) URIs should be opaque

20:34:36 pepper, can you explain to larsbot? (you can skip if youd like)

20:34:44 <larsbot> b) it's about what it means to reference a URI

20:34:56 <larsbot> c) http://....#foo can also mean two things

20:35:00 * larsbot can't count

20:35:07 :)

20:35:13 better than a, b, 3

20:36:21 * em suspect that we have agreement here on the 2 differences and the difference is in suggested ways of supporting this...

20:36:31 not sure how to keep testing this suspicion

20:36:37 <tim|mit> a) When used as terms in a semantic web expression, URIs are opaque - its only when you try to look inside them using web specs you get more info.

20:36:47 <eikeon> One can not ask a server for http://...#foo -- which has been the main thing I have been running into in practice.

20:36:49 <tim|mit> b) refernce a uri? <-- ??

20:37:13 <larsbot> b) when you say (http://www.ontopia.net, dc.

20:37:21 <larsbot> :creator, Person X)

20:37:33 <tim|mit> c) In my architecture, http://...#foo can't mean two things. You sacrifice teh ability to refernce a part of te document for the ability to refernce an abstract thing.

20:37:56 <larsbot> c) well, in that case it is at least unambiguous

20:38:10 i think (c) is identified in paper

20:38:20 <larsbot> b) in the triple I gave the URI of the subject is interpreted one way

20:38:34 <pepper> paper objects to having to make that sacrifice...

20:39:06 <larsbot> b) however, the URI rdf:type is interpreted in a different way

20:39:14 <tim|mit> pepper doesn't have that luxury in any programming languge

20:39:21 <tim|mit> ... that i can think of

20:39:46 <tim|mit> how do you refer to the bit of file which talks about the integer foo in main.c?

20:39:50 * mariyo thinks that you are always doing both referencing part of doc and abstract thing. can't get away from that fact.

20:40:27 <pepper> tim|mit: if you have a sufficiently powerful addressing language you can do that

20:41:09 <pepper> XML lets you do it

20:41:28 XML?

20:41:34 <mariyo> ?

20:41:34 <tim|mit> xml isn't a langauge by itself

20:41:36 a bit more specific?

20:41:56 <pepper> we're talking about the ability (and the requirement) to be able to address parts of documents

20:42:07 <pepper> that's easy with xml

20:43:06 <tim|mit> That is XML as a markup language, which only has bits of documents, but no meaning.

20:43:07 <pepper> i don't think it's at all unreasonable to want to be able to use the fragment identifier to address... a fragment

20:43:36 * em thinks irc is sometimes a crappy medium... not sure how best to get back to tradeoffs of different approaches

20:43:46 * eikeon wants to be able to use an abstract URI in a request to a server and perhaps get 303ed to a concrete document.

20:44:14 * pepper suggests everyone lets em steer the discussion

20:44:49 so if i understand correctly...

20:44:59 1) URI/URIref

20:45:00 <mariyo> what eikeon is saying is very practical. it would be great.

20:45:20 2) differnet relations

20:45:49 1) elaborating... contact / contact#me

20:46:21 2) elaborating ... :foo rdf:subject :contact#me. :foo rdf:indicator :contact#me

20:46:47 is this a correct characterization of the 2 approaches discussed so far that differentiate the problem?

20:46:53 yes|no|kindof

20:47:45 <pepper> yes, if i understood your question correctly: 1) is tim|mit's proposal, 2) is the TM approach

20:47:45 * em waits for feedback on summary

20:48:33 <pepper> may i summarize, em?

20:48:37 ok; so we agree that (subtle or not) we're bothing trying to do the same thing ... just have different ideas on the 'best' approach

20:48:48 pepper, you have the floor

20:49:09 <pepper> 1) we agree that URIs are being used for 2 different purposes and this is a problem

20:49:53 <pepper> 2) tim|mit wants to ban the use of one URI for two different purposes by tying the purpose to the form of the URI/URIref

20:50:18 <pepper> 3) TM approach is to accept that one and the same URI/URIref may be used in two ways and provide syntax to distinguish them

20:50:51 <pepper> </ol>

20:51:36 <tim|mit> Let's call the thing with the # a Universal Term (UT) "URUref" is overloaded with relative/absolute distinction.

20:51:48 thats a better way to summarize than i did (but I think the term 'ban' is perhaps a bit harse) .. URI/URIref to distinguish use vs additinoal syntax to distinguish use

20:52:22 <pepper> bit harsh: point taken

20:52:30 <pepper> prohibit would be better

20:52:30 larsbot? does peppers summary make sense?

20:52:36 <larsbot> em: yes, it does

20:52:40 <tim|mit> 2) tim wants to point out that th difference between URI and UT is just the distinction one needs

20:52:57 tim|mit, can you elaborate a bit more on this

20:52:58 <pepper> UT?

20:53:09 <tim|mit> re 3) The TM approach works but makes the assumption that there iare well-defined functions which there are not.

20:53:28 <tim|mit> elaborating on 2:

20:53:36 <pepper> (sorry - missed the UT def above)

20:54:20 <tim|mit> When you make a new application, you invent a new langauge, and it has terms. Most langauges use identifiers local to the document. Some langauges can be webized by extending them to use universal terms.

20:54:40 <tim|mit> A universal term plays the same role in the langauge as the term

20:54:57 <DanC_jam>http://www.eclipse.org/

20:54:57 <tim|mit> Eg in a python document, it identifies a module or a class etc

20:54:57 <dc_rdfig> V: http://www.eclipse.org/ from DanC_jam

20:55:06 <DanC_jam> V:Eclipse -- a Java IDE

20:55:07 <dc_rdfig> Added comment V1.

20:55:32 <tim|mit> The power of the web is that you can make new terms in an unrestricted way, by taking concpets out of any new langauge.

20:55:43 <DanC_jam> V:also: [new eclipse 2.1 deb|http://lists.debian.org/debian-java/2003/debian-java-200304/msg00018.html] Apr 2003 to debian-java

20:55:44 <dc_rdfig> Added comment V2.

20:55:44 <tim|mit> The semantic web -- rdf and layers above -- is one such langauge.

20:55:52 <dajobe> V:and a Perl IDE, and a C IDE,...

20:55:52 <dc_rdfig> Added comment V3.

20:55:54 <pepper> tim: don't understand your elaboration of 2). could you elaborate on the 'but' in 3)?

20:56:01 <DanC_jam> V:Stefan D. uses this (on Windows XP)

20:56:02 <dc_rdfig> Added comment V4.

20:56:27 <tim|mit> The cool thingabout a UT is taht you can form one in such a way that it can be used to get you a document written in that langauge and the local term within that document.

20:57:22 <tim|mit> There's a standard for this on eth web that you take the URI of the document, a "#" , and he local term and concatenate them (munging the locla term if necessary)

20:58:03 pepper, does tim|mit

20:58:10 's explanation help?

20:58:17 <DanC_jam> V:cvs support

20:58:18 <dc_rdfig> Added comment V5.

20:58:18 <pepper> no, see above

20:58:21 <pepper> tim: don't understand your elaboration of 2). could you elaborate on the 'but' in 3)?

20:59:04 <tim|mit> Repharsing 2: The fact that you jump into a whole new world when you add the "#" is an important flexibility point in weg architecture.

20:59:09 <tim|mit> web

20:59:09 "<tim|mit> re 3) The TM approach works but makes the assumption that there iare well-defined functions which there are not."

21:00:03 <tim|mit> The TM approach gives you two ways of using an identifier, so that one way refers to the bit of the document, and the other way refers to the thing.

21:00:11 <tim|mit> The problem is with "the".

21:00:20 <pepper> which "the"?

21:00:28 <niq> tim|mit - did you just say Humpty Dumpty invented the sem web?

21:00:38 <tim|mit> In an RDF document, foo.rdf#bar refers to one thing, but as parts of dopcument, variosu bits scattered all over it.

21:00:43 <pepper> there are four "the"s in your sentence...

21:00:53 <tim|mit> Humpty Dumpty?

21:01:00 * jhendler had a slide once for the web -- "the" with a kill ring over it...

21:01:14 <tim|mit> The TM approach gives you two ways of using an identifier, so that one way refers to *the* bit of the document, and the other way refers to *the* thing.

21:01:33 <niq> tim|mit: " When you make a new application, you invent a new langauge, and it has terms."

21:01:44 <pepper> and *the* problem was?

21:01:49 <tim|mit> What part of the document corresponds to foo.rdf#bar?

21:02:02 <niq> Humpty Dumpty: "words mean what I say they mean" (or somesuch)

21:02:12 <pepper> presumably the element with the ID "bar"

21:02:22 <jhendler> the organization of these UTs in the SW also a challenge -- OWL first step in that direction

21:02:24 <larsbot> that depends on the MIME-type of the document

21:02:34 * pepper was assuming RDF

21:02:37 * tim|mit niq, Winnie-the-Pooh: "Don't you understand what *the* means?"

21:02:53 <bijan> It's Winnie-*ther*-Pooh!

21:03:09 <pepper> quiet in the 1 and 9s, please

21:03:23 <tim|mit> pepper, no - RDF deos not use xml:ID -- - any part of the document which refers to #bar is referting to the same node, no part of the document uniquely is a special "defining" place.

21:03:30 * jhendler notes that the Pooh book spelled "OWL" as "OWL" which is where the acronym for our langauge came from :->

21:03:47 * em folks... we're at top of hour and i've got to run to next meeting unfortunatly... i'd like to recap a bit here if possible.

21:04:00 <pepper> go ahead, em

21:04:31 <tim|mit> pepper, the corollary of the problem is : what does http;//www.w3.org/2000/10/whatever refer to as a thing?

21:04:48 1) i think we agree to peppers summary 123 ... we're arguing on which approach is 'best' (architecture, deployment, pedegogical, etc.)

21:05:14 if this is the case, i think this is helpful as this was not reflected in the paper

21:05:40 <pepper> really? i thought it was quite explicit in the paper...

21:06:53 <pepper> 1) "The root of the problem is that one URI is being used in two quite distinct ways"

21:06:56 i thought the problems with URI/URIref we captured but not the fact that in short this is the same functionality that distinguishing these syntactically provides

21:06:56 <tim|mit> I feel I have a consistent archietcture and the paper doesn't address the problems with it. Eg I don't agree with (1) for cwm.

21:07:08 <pepper> em: it's *

21:07:14 <pepper> em: it's *not* the same functionality

21:07:32 <pepper> approach 2) prevents you from using one URI/URIref in two different ways

21:07:53 good point...

21:07:57 <pepper> why *shouldn't* people be able to do that?

21:08:37 rather i was thinking more along the lines of merging infromation (metadata about data, metadata about metadata, etc.)

21:08:52 <tim|mit> There is a *lot* to be gained in the architecture from a single idnetifier identifying a single thing.

21:08:54 why *shouldn't* people be able to do that?.. tim|mit was talking about costs of this

21:09:15 <pepper> really? didn't realise that

21:09:19 <tim|mit> There is a *lot* to be gained in the architecture from a single idnetifier identifying a single thing -- across ALL applications which use UTs.

21:09:21 <pepper> what are the costs?

21:09:35 <larsbot> tim|mit: you don't have to give that up

21:09:44 <tim|mit> How not?

21:09:56 * em trying to negotiate delay for next meeting... failing

21:10:00 <larsbot> the way FOAF does

21:10:11 * tim|mit wondering what this meeting was

21:10:14 <larsbot> use blank nodes for resources which are not network-retrievable

21:10:40 <larsbot> unfortunately, that means RDF properties have to be represented by blank nodes :-(

21:11:02 pepper, larsbot... not sure on next steps, suggestions?

21:11:16 <pepper> f2f in london?

21:11:23 tim|mit, meeting to review pepper's paper http://www.ontopia.net/topicmaps/materials/identitycrisis.html

21:11:27 <pepper> when are you coming to europe, em?

21:11:34 <mariyo> good idea.

21:11:38 www2003 certainly

21:11:45 <tim|mit> Requirement: To be able to give in one application, an identifier which when passed anywhere in the universe will, without further context apart from the core web specs (URI etc) , identify exactly the same thing.

21:11:54 <pepper> oh... an excuse for /me to go to budapest. goody.

21:12:05 <pepper> coming, larsbot?

21:12:06 <dajobe> (rdf properties can't be blank nodes, not allowed at the moment)

21:12:49 <pepper> em?

21:12:53 weve been thinking of arraging rdf/tm bof (owl as contraint, talk about query requirements, etc...) make sense to add this to list?

21:13:03 i think we have a different view on deployment costs

21:13:03 <bijan> tim|mit: er..uris don't do that

21:13:13 <bijan> As they can change denotation over time

21:13:19 <tim|mit> If you have to use an rdf path expression to identify something, then you can make a new URI scheme out of that, and that then becaoems your identifer space.

21:13:23 <pepper> em: certainly - if we can make it to the conf

21:13:25 <bijan> (at least on some views)

21:13:32 <tim|mit> bijan, how so?

21:13:44 budapest lovely that time of year :)

21:13:47 <tim|mit> Aah, time

21:14:05 * bijan notes that Fieldlings view holds that they *don't* change over time

21:14:16 <jhendler> me and larsbot have spent some time trying to figure out wheree we can get a join SW / TM activitiy together

21:14:28 <jhendler> WWW 2003 has not replied to our request for a BOF

21:14:32 <bijan> Which is interesting given his differences about what uris can denote

21:14:37 <jhendler> Looking at possibility of XML or ISWC

21:14:42 <bijan> (what which uris...)

21:14:47 <bijan> And how they do so.

21:14:47 * pepper v. tempted by budapest. get back to em by email?

21:14:49 <jhendler> Clear would be nice to get some people together for a f2f

21:14:51 <tim|mit> BIjan, most systems are built to wither ignore time, or to take it into account explicitly, or to change with it. But in each case, the concepts such as color are still useful even their definition ignores time.

21:15:00 doesnt see SW/TM connection rather RDF/TM ... both seem to be working towards Semantic Web

21:15:13 'pepper v. tempted by budapest. get back to em by email?' deal

21:15:13 <bijan> Hmm. Uris aren't concepts

21:15:14 <pepper> can i quote you on that? :)

21:15:25 <bijan> They can denote concepts.

21:15:27 <mariyo> :)

21:15:30 <bijan> So :color can change

21:15:31 got to run folk.... thanks all

21:15:36 <bijan> Even if color doesn't :)

21:15:38 <mariyo> bye

21:15:40 <pepper> thx. see you later

21:15:56 <larsbot> thanks, em

21:15:58 <jhendler> yes em - agree RDF/OWL/TM

21:16:09 <jhendler> or RDF = RDF/RDFS/OWL

21:16:10 <pepper> pepper is now known as koshoo

21:16:24 <tim|mit> Time-varying concepts are actually useful sometimes.

21:16:41 <bijan> Tim, I'm not clear that ignoring, takign into acount, and changing are the only options.

21:16:47 * koshoo needs to get to bed. long day tomorrow.

21:16:50 <bijan> Or than any of them is quite what happens on the web.

21:17:17 <bijan> Hmm. There's two senses (at least) to "time varying concept". One is that you have two *different concepts* at two differnt times

21:17:36 <bijan> THe other is that the cocnept itself...

21:17:38 <bijan> Hmm.

21:17:40 <dajobe> I assume larsbot, grove_ & steve (oops missed) will be at xml europe?

21:17:44 <bijan> Not sure what I want to say there.

21:17:55 <tim|mit> Bijan,.... theer are many ways to model time.

21:17:59 <larsbot> dajobe: yes, we will

21:18:05 <bijan> There's the trivial, e.g., that the extension of a concept can be different in different circumstances.

21:18:07 <DanC_jam> N:danc has to leave presently (during Language Committee item). above agenda covers today. tomorrow and Sat are for technical committees, I gather.

21:18:08 <dc_rdfig> Added comment N16.

21:18:14 <tim|mit> But to bring it up in every discussion tends to be distracting.

21:18:15 <bijan> At least contingent concepts

21:18:33 <bijan> I'm not talking about modling time.

21:18:49 <bijan> If we depend on sharing denotions of uri, then time matters.

21:19:01 <bijan> Since concurrent execution is fairly rare :)

21:19:05 <bijan> Er...

21:19:11 <bijan> relatively speaking.

21:19:31 <tim|mit> In practice, the time-varying aspects for HTTP URIs are complex social ones.

21:19:33 <bijan> Or more tot he point, that you might be able to apply the same technqiues you apply to temporal slipage to otehr sorts of context slippage

21:20:18 <bijan> Actually, i suspect you ahve a choice: let time slip or let context/use slip

21:20:33 <tim|mit> in practice, one tends to set up a subsystem fo teh whole semantic wen in which there are certian assumptions made which allow time to be factored out.

21:20:53 <tim|mit> s/wen/web/g

21:21:14 <tim|mit> slip?

21:21:19 <jhendler> bijan, concurrent execution rare, but URI change on SW can be relatively slow - so not sure that isn't a red herring - context slippage an interesting point

21:22:18 <bijan> jim, sure, except if "slow change" is ok why isn't "fast" change? Ambiguity is ambiguity.

21:22:25 <tim|mit> Example: the web page which lists the working groups in the W3C. most people don't aorry abot exactly which time period it is valid for.

21:22:59 <bijan> tim: sure.

21:23:05 <jhendler> bijan, question is what needs to be modeled how, not necessarily what is theoretically possible.

21:23:11 <tim|mit> For example, if you check whether you are interested in each group, then you check every few weeks, and so if it is a little out of date, you'll catch it next time.

21:23:20 <jhendler> (neat v. scruffy?)

21:23:20 <bijan> jhendler: exactly

21:23:31 <tim|mit> You *could* model it explcitly, but you don't do tyhat in RDF most of the time.

21:23:47 <bijan> So that abiguity doesn't matter?

21:24:03 <jhendler> fast changing stuff would more likely be done by having the URI do an "indirect reference" of some type

21:24:20 <bijan> I mean, if the uri denotes "the currrent members of teh WG", then it changes. I.e., it doesn't refer to one thing accross different applications

21:24:33 <tim|mit> Actually, once I did write a script which cheked out a version as of now, and a version as of 6 months ago, and generated a list of new groups. That system had an explit model of time - crude thoug hit was.

21:24:45 <bijan> THat suggests to me that *that propertY* (not changing blah blah) maybe isn't quite as important as you made it out to be.

21:25:20 <jhendler> but the property is the same (the list of WGs) the content of that property changed

21:25:38 * bijan isn't quite confortable with that, btw, I'm doing a bit of exploratory devil's advocacy here

21:26:02 * jhendler promises not to fire Bijan if he (Bijan) wins the debate :->

21:26:09 <bijan> Heh

21:26:18 * tim|mit bijn is now known as Bijan666

21:26:32 <bijan> Anyone for web services?

21:26:42 <bijan> Can I interest you in a gopher server?

21:26:58 <tim|mit> Go on...

21:27:03 <bijan> Did I mention my forthcoming patent on distributed, scalable,d isorganized KR systems?

21:27:09 <jhendler> I think there is something of the name, content of name issue - probably worth teasing out

21:27:24 <bijan> "content" of the name? Do you mean denotation?

21:27:29 <jhendler> bijan - too late, Amit Sheth already pattened that as soon as you've published...

21:27:35 <bijan> Damn.

21:27:37 <tim|mit> I think you can model the changing and not changing parts.

21:28:05 <jhendler> well, sort of, but -- what Tim said (changing/not changing) more than denotation per se

21:28:07 <bijan> Well, natch. The real question is what do we *have* to do, what *ought* we do, and what do we *want* to do :)

21:28:35 <tim|mit> I think right now in the discussion of URU vs UT and so on, we ought to not get caught in th etime discussion.

21:28:41 * bijan notes that he tends to reduce it to denotation for simplicity;must wat that.

21:28:51 <tim|mit> If you want to get caught in a discussion, then get caught in the "generic resource" discussion

21:28:51 * jhendler notes the Amit Sheth note should probably have been [off]

21:29:42 <tim|mit> Suppose something denotes "The W3c home p[age" and two other things denoet "The W3C home page(in english)" and "The W3C home page (in french".

21:29:52 <jhendler> this is the content slippage Bijan refered to :->

21:29:54 <tim|mit> What is the relationship between these things?

21:30:19 <tim|mit> This the Greeks discusses with respect to substances, I belive.

21:30:37 <bijan> In discussion with eikeon, we realized that we *reallY* wanted convienent ways to move between denoting the object and the various represenations.

21:30:40 <tim|mit> The "W3C home page" is a generic identifier.

21:31:29 <bijan> I.e., have an axis of uris that let you name the various represenations you GET from it

21:31:45 <tim|mit> I have to join a telcon.

21:31:49 <bijan> Have fun :)

21:31:53 <tim|mit> tx

21:31:56 <tim|mit> cu

21:32:00 <bijan> ta

21:32:01 <tim|mit> tim|mit is now known as tim-away

21:33:09 <jhendler> M: the English version is [here | http://www.ralphm.net/world?language=en#]

21:33:09 <dc_rdfig> Added comment M2.

21:33:21 <jhendler> M2: " "

21:33:21 <dc_rdfig> Replaced comment M2.

21:33:29 <jhendler> N: the English version is [here | http://www.ralphm.net/world?language=en#]

21:33:30 <dc_rdfig> Added comment N17.

21:33:53 <jhendler> N17: ""

21:33:53 <dc_rdfig> Deleted comment N17.

21:34:12 <jhendler> M2: the English version is [here | http://www.ralphm.net/world?language=en#]

21:34:12 <dc_rdfig> Replaced comment M2.

21:34:39 * jhendler having trouble typing sigh

21:34:54 <jhendler> bye tim|mit, bye bijan

21:59:38 <ChanServ> [#rdfig] This channel is logged and blogged: http://logicerror.com/rdfIRCWelcome

22:03:49 <ChanServ> [#rdfig] This channel is logged and blogged: http://logicerror.com/rdfIRCWelcome

23:12:52 <DanC-AIM> .time

23:12:53 <datum> Thu, 10 Apr 2003 23:12:53 GMT

23:13:22 <DanC-AIM> Hi fro MIA, flt 1477

23:32:31 <golbeck> IM *on the plane*

23:32:33 <golbeck> how cool

23:35:25 <DanC-AIM> Love this http://dm93.org/z2001/WearableGizmo

23:36:09 <DanC-AIM> /me reads bake don't fry...

The IRC chat here was automatically logged without editing and contains content written by the chat participants identified by their IRC nick. No other identity is recorded.