Semantic Web Interest Group IRC Chat Logs for 2003-10-08

00:58:43 * uberfunk is away: chow

02:25:33 <tav> tav is now known as tav|offline

02:43:10 * uberfunk is back (gone 01:44:25)

02:50:57 <mdupont-ZZZ> mdupont-ZZZ is now known as mdupont

05:08:40 <karl_> karl_ is now known as karlcow

05:09:52 <mdupont> mdupont is now known as md-away

08:13:45 <tav|offline> tav|offline is now known as tav

10:14:51 <dajobe>http://www.tbray.org/ongoing/When/200x/2003/10/07/Bristol

10:14:52 <dc_rdfig> A: http://www.tbray.org/ongoing/When/200x/2003/10/07/Bristol from dajobe

10:15:02 <dajobe> A:|West England Web Architecture - Tim Bray

10:15:02 <dc_rdfig> Titled item A.

10:15:59 <dajobe> A:*For some reason, Bristol seems to be a nexus of Semantic Web energy: ...*

10:15:59 <dc_rdfig> Added comment A1.

10:16:09 <dajobe> A:... * I had a deranged vision of the RDFistas staging a coup-de-main and taking the TAG hostage*

10:16:09 <dc_rdfig> Added comment A2.

10:25:31 <jeen> That comment about the beer is rather awkwardly placed. (right above that picture with the tub...)

10:29:51 <dajobe> I should get one of our friends at HP to wear an RDF t-shirt and stalk him

10:29:59 <dajobe> in a nice way :)

11:45:19 <KimmoA> Hmm. This was the closest to a W3C IRC chat I could find ;-)

11:45:30 <KimmoA> Their own server seemed down :-S

11:46:44 <kao> hm, what do you mean by "W3C IRC" chat? covers quite a number of topics...

11:47:36 <KimmoA> Well... Most smart people are away on #HTML at EFnet. I'd like to know if MS IE actaully doesn't support :before

11:48:11 <KimmoA> I was so happy when I found about about the content property together with :before pseudo-element.

11:48:23 <KimmoA> Opera and Mozilla renders it correctly (how suprising)

11:48:29 <kao> oh, thats *very* far off topic here :-)

11:48:32 <D[a]vey> KimmoA: #web on here is what you need :)

11:48:33 <KimmoA> Bah.

11:48:35 <D[a]vey> D[a]vey is now known as Davey

11:48:40 <JibberJim> You'd want a CSS channel, not an HTML one. Inserting content with presentation is generally a bad idea in any case.

11:49:17 <KimmoA> Uh... #HTML generally is for anything web-related, just as #Windows98 on DalNET can help me with anything PC-related.

11:50:39 <KimmoA> Alright, then... See you ;)

11:55:15 <Davey> JibberJim: do you have any Annotea experience?

11:56:52 <JibberJim> I wrote a couple of test clients Davey

11:56:58 <Davey> aah, cool :D

11:57:03 <Davey> I'm writing a client/server in PHP atm

12:28:19 <Davey> Hey tim-mit

12:30:51 * tim-mit is in a teleconf

12:46:21 <Davey> hmm, this part of Annotea is confusing

12:46:31 <Davey> <a:created>1999-10-14T12:10Z</a:created>

12:46:31 <Davey> <d:date>1999-10-14T12:10Z</d:date>

12:46:36 <Davey> whats the difference? :/

12:47:10 <dajobe> look up the namespaces

12:47:33 <Davey> sure, I understand that, but I don't see why you need two.

12:47:55 <dajobe> there are plenty of times you give a doc a date and publish it a different date

12:48:06 <Davey> right, ok, thats what I was thinking.

12:48:15 <Davey> So, the first is creation, and second is publishing.

12:48:21 <dajobe> it could be

12:48:46 <Davey> I don't understand *why* annotations would need that...

12:50:32 <JibberJim> Because you can add an annotation that you created elsewhere, for example a blog entry about an article could be cited by someone else as an annotea comment - the entry was created at a different time to the annotation act itself.

12:52:07 <dajobe> I sometimes edit entries I created earlier

12:52:11 <dajobe> this is all very common.

12:53:50 <Davey> yup, ok

13:24:18 <Davey> Davey is now known as D[a]vey

13:52:23 <D[a]vey> dajobe: should <d:date> be the *current* date, as in, at the time of the message.

13:53:04 <dajobe> you never said what d: meant

13:55:44 <D[a]vey> its dublincore :/

13:56:15 <JibberJim> In my implementations of annotea, the d:date is the date of the submission, in fact I don't think it'e required at all, and you can leave the server to create the time-stamp.

13:56:35 <JibberJim> EricP will of course know more about this than me, it's been some time since I've done any annotea work.

13:57:14 <D[a]vey> JibberJim: ok, I'll just make it default to the client local time

14:01:49 <josek> DaJobe, are you around?

14:02:00 <dajobe> busy, rdf core pubbing today

14:02:11 <josek> ok! I'll mail you and be patient :) thanks

14:26:02 <roGer^work> anyone that has been working with jena 1.6 and knows how to remove the default of an Alt ?

15:03:15 <tim-lex> tim-lex is now known as timbl

15:05:25 <D[a]vey> D[a]vey is now known as Davey

15:10:30 <dajobe> very odd, w3c link validator is taking 5 (min)-10+ seconds per-URL to validate www.w3.org URLs

15:10:58 <Davey> wow.

15:11:19 <dajobe> please don't try it; i'm 15 minutes into validating a big doc

15:11:23 <JibberJim> The validator box was being (unintentionally I believe) DOS'd a couple of days ago, maybe it's started up again.

15:13:37 <dajobe> make that 30 mins

15:14:01 <josek> dajobe, I have a very slow connection to mit too

15:14:16 <dajobe> it's not my connection to mit that's the problem

15:14:19 <dajobe> it's mit's connection to mit

15:14:43 <josek> some lag builds up before it

15:17:03 <dajobe> maybe you can ask the question while I'm waiting josek

15:17:08 <ear1grey> hmm, i hope you're not relying on the graph output - it was a bit broken for PNG and GIF earlier - empty ellipses and rectangles.

15:17:21 <dajobe> no images in this doc

15:18:40 <josek> dajobe, thanks. I tested a bit more and found the answer

15:19:00 <josek> I was trying to open multiple bookmarks. I thought I needed a world + model + storage per rdf file I parsed

15:19:19 <josek> but then realized they all share the same world and I just need a model + storage per parsed file

15:19:50 <josek> btw, I tried the LIBRDF_DEBUG option and it doesn't compile anymore

15:20:00 <josek> and it crashes in some cases

15:20:08 <dajobe> I use that all the time

15:20:23 <josek> this time I was running with 0.9.14

15:21:06 <josek> e.g., in rdf_node.c:librdf_free_node, the DEBUG statemens access some members that are not in the structure def.

15:21:10 <dajobe> you don't necessarily need a model&storage per file, you can use one model & storage and use contexts for example

15:21:24 <josek> node->string, node->max_usage

15:21:33 <dajobe> ah, when LIBRDF_DEBUG > 1

15:21:47 <dajobe> yeah, I took that out

15:22:28 <josek> ok, I'll look at the context then. I programmed this quite modularly, so for me it's a localized change

15:22:41 <dajobe> I'll try fixing it

15:22:51 <josek> not urgent (for me now) :)

15:23:13 <josek> this new version of amaya will let you store bookmarks anywhere on the web, open multiple bookmarks at the same time, edit them, save them...

15:23:34 <josek> not sure if i'll finish the changes before the next amaya release, though

15:25:24 <josek> I also added some support so that it can detect bookmark properties inside rdf files and show them as bookmarks. I want to add a daml-oil same-as processing, so that I can say "open this rss file and show it as bookmarks"

15:25:51 <dajobe> ah

15:26:02 <dajobe> I've not looked, but the gnome epiphany browser uses rdf/xml for it's bookmarks

15:26:09 <dajobe> looked=at the schema/properties

15:26:25 <JibberJim> Storing bookmarks remotely is something lots of people seem to want in their browser.

15:26:44 <josek> I'd look at it too to see if our schemas are equivalent

15:26:52 <dajobe> I use my weblog for bookmarks

15:27:12 <dajobe> but I should make an easier simple link storing api for it

15:27:20 <josek> it would be nice reading your weblog and if we have some same-as properties viewing it as bookmarks

15:27:41 <josek> Dom made an XSLT already so that you can view the bookmarks ala RSS.. with XML + css :)

15:28:02 <nmg> very nice - now all I need is the same functionality in mozilla

15:28:29 <josek> you just need to tell us who are the mozilla bookmark guys and put us in contact with them

15:30:42 <dajobe> hurrah, 45 minute link check

15:30:58 <dajobe> timeouts on some docs, I'm not repeating it

17:22:28 <Davey> Davey is now known as D[a]vey

17:28:22 <zoyd> i just tried makeTriples.py from here http://www.xml.com/pub/a/2003/02/12/rdflib.html ...

17:29:07 <zoyd> none of the triples get added to the rdf file, it only contains the namespace declerations.

17:36:31 <eikeon> zoyd: There is a bug in the serializer where triples do not get serialized after the first call to save.

17:37:39 <eikeon> It's been fixed in CVS and will be in a soon to be released bug fix release.

17:37:59 * zoyd goes to grab the cvs

17:40:06 * eikeon looks forward to getting back into the swing of regular rdflib releases... been a few bug fixes that should have been release some time ago.

17:41:29 * zoyd finds rdflib.net to be slow

17:45:31 * eikeon probably should stop all my X11 traffic... am displaying windows remotely from the box that runs rdflib.net

17:46:14 <eikeon> ... while I wait for a new laptop.

17:46:43 <zoyd> still can't get to rdflib.net

17:47:47 * eikeon killed X11 traffic... checking site now.

17:48:09 <eikeon> hum

17:48:12 <eikeon> Waitting

17:49:06 <zoyd> still on sending request to rdflib.net ..

17:49:37 <zoyd> aah .. there it is

17:49:56 <eikeon> Not sure what all was slowing it down.

17:51:03 <zoyd> is the bug fixed in 1.3.0?

17:51:06 * eikeon may need to do some site tweaks soon to :)

17:51:24 <eikeon> No.

17:51:33 <eikeon> Need to make a 1.3.1

17:53:49 * eikeon is away from primary computer... else I would promise one today. Will promise a new version in a few days instead.

17:56:15 <zoyd> eikeon: "anonymous@rdflib.net"?

17:57:47 <zoyd> cvs: "Connection refused"

18:01:50 <bijan> zoyd it's an easy fix to hack in

18:10:29 <zoyd> hmm

18:13:40 <ericm> dajobe, you around?

18:17:27 <libby> ilrt is pretty much offline for some reason. dajobe is about though somewhere

18:17:43 <dajobe> hello

18:18:22 <ericm> hey dajobe! :)

18:18:24 <eikeon> zoyd: Sorry, I don't have anonymous CVS access set up.

18:18:34 <ericm> using your shadow perl script

18:18:49 <eikeon> It is an easy fix though.... and I can point you to the cvsview URL for it.

18:18:49 <dajobe> caveat emptor

18:18:58 <ericm> running into issue but i'm thinking they're more doc related than script at this point...

18:19:08 <ericm> you going to be around however for a bit?

18:19:12 <ericm> (say next hour)?

18:19:14 <dajobe> yes

18:19:20 <ericm> excellent, thanks

18:20:12 <zoyd> eikeon: ok, where's it?

18:22:08 <eikeon> zoyd: http://rdflib.net/cvs/rdflib/rdflib/syntax/serializer.py.diff?r1=1.12&r2=1.13

18:22:31 <ericm> dajobe, docs (minus lbase and rdf-mt) all seem to pass; references on these 2 docs are inconsistent with rest in shadow TR space

18:22:44 <ericm> working on these now

18:22:53 <dajobe> uhoh !

18:23:22 <ericm> if I check the rest in, can you eye-ball a couple to make sure your scripts didn't get a bit carried away :)

18:23:39 <dajobe> if you set debug=1 (or 2?) it'll show you before/after changes

18:23:54 <ericm> debug=2 is how i found out about the problem(s)

18:24:40 <ericm> debug=2 'looks' fine as a dump, but i haven't tested taken a look and/or re-tested on pubrules

18:24:55 <dajobe> ok

18:25:58 <zoyd> eikeon: worked, thanks a lot :)

18:26:24 <ericm> err.. what's the cvs command for recursively checking in a whole set of directories?

18:26:43 <dajobe> cvs diff

18:26:47 <dajobe> oh, checking in

18:26:57 <deltab> I suspect you want import

18:27:10 * dajobe checks out stuff from his backups

18:28:29 <eikeon> zoyd: Your welcome. And I'll get a release out soon.

18:59:10 <dajobe> ericm: the script seems to have done too much; it's edited some body text refering to the older versions, especially in the SOTD - this is for syntax, for example

19:00:04 <dajobe> & TOC

19:00:25 <dajobe> & changes

19:00:28 <dajobe> so much for automation

19:01:01 <dajobe> next time, give the links an html:class and use xslt

19:02:37 <ericm> crap

19:03:26 <ericm> sigh... i dont think i can get all of the references converted in time

19:03:47 <dajobe> the references seem fine

19:03:56 <dajobe> so what to do now?

19:04:11 <ericm> the references all point back into shadow space

19:04:27 <ericm> 2 options

19:04:31 <ericm> 1) fix code

19:04:47 <ericm> 2) hand edit all references to point to soon to be TR space

19:06:11 * DanC has trouble getting to http://ilrt.org/discovery/chatlogs/rdfig/2003-09-17#T17-08-18

19:06:18 <dajobe> ilrt's offline

19:06:20 <dajobe> partially

19:06:23 <DanC> :_{

19:06:29 <dajobe> tcp drops after about 0.5 secs with a RST

19:06:50 <ericm> dajobe, do you have a sense how much work #1 is?

19:07:00 <dajobe> I'll look at it; it might be dooable

19:07:28 * DanC wonders if the white blood cells are winning the battle today in ericm's system

19:07:36 <dajobe> hmm, It really prefers unix line feeds at my end; rdf-mt's in mac

19:07:41 <ericm> http://www.w3.org/2001/sw/RDFCore/TR/shadowtr.pl

19:07:41 <dc_rdfig> B: http://www.w3.org/2001/sw/RDFCore/TR/shadowtr.pl from ericm

19:08:04 * ericm donks head on table.. thinks he's just answered Danc's question

19:08:21 <ericm> B:|Don't click me

19:08:21 <dc_rdfig> Titled item B.

19:08:30 <DanC> LOL

19:08:32 <dajobe> you can zap the url if you want

19:08:50 <DanC> B:http://deleteme.example.org/

19:08:51 <dc_rdfig> Added comment B1.

19:08:55 <DanC> B:=http://deleteme.example.org/

19:08:55 <dc_rdfig> Replaced URL of B.

19:11:27 <ericm> dajobe, please let me know in 10 min if you think #1 is an option... if not, i'm going to suggest we low-brow / brut-force this

19:11:40 <dajobe> I'm trying #1 this sec

19:11:53 * ericm bows deeply to the east

19:13:53 <dajobe> can you convert the mt to have unix line ends?

19:14:13 <ericm> err... not sure; trying

19:15:53 <DanC> perl -i.bak -pe 's/\r/\n/g' foo.html

19:16:10 <DanC> ^converts mac newlines to unix newlines

19:16:42 <dajobe> some success

19:16:51 * DanC supposes ericm knows the analagous emacs prayer for converting newlines

19:18:09 * ericm thought he did... my mac/emacs version however is not accepting this... trying X

19:18:56 <ericm> i thought it was 'raw-text-unix'

19:19:00 <dajobe> this is working better

19:20:16 * ericm notes first time i've encountered a mac / linux emacs experience that was different

19:20:45 * ericm notes... err, other than the nicer interface :)

19:24:43 <dajobe> amusingly, I'm having trouble with my doc

19:25:11 <dajobe> looking better

19:25:35 * ericm crossing fingers

19:25:42 <ericm> brb

19:25:42 <dajobe> did you change & commit mt?

19:27:07 <ericm> yes

19:27:25 <ericm> check out new version ... should be there by now

19:27:28 <dajobe> this is nearly there, minor errors now

19:27:29 <dajobe> yes

19:32:27 <dajobe> I'll commit the changes to shadowtr.pl

19:33:09 <ericm> excellent thanks

19:33:21 <ericm> let me know and i'll run these again

19:33:32 <dajobe> committed

19:33:39 <dajobe> now the minor fixes are more visible

19:34:00 <dajobe> in rdf-mt, it over-edits a para 'Detailed changes from the previous ...'

19:34:14 <dajobe> concepts seems ok

19:34:53 <dajobe> syntax seems ok

19:34:53 <ericm> err.. so its not working now?

19:34:59 <dajobe> it is

19:35:35 <ericm> did you check new files into staging? (/me think you have permissions) or do i need to re-run this?

19:35:44 <dajobe> I didn't check them in

19:35:47 <ericm> ok

19:35:47 <dajobe> if you want me to?

19:36:31 <ericm> no, i can do this... just checking to make sure where not colliding

19:36:38 <dajobe> I won't commit before acking

19:36:48 <dajobe> I was just doing a diff -ur old new for each doc

19:36:59 <dajobe> so anyway line 104 rdf-mt is over edited

19:37:11 <dajobe> syntax, concepts convert ok

19:37:32 <ericm> so this needs to be hand edited back to?

19:37:37 <dajobe> yeah

19:37:41 <dajobe> maybe I can look into why

19:38:20 <dajobe> schema also has over editing 154; seems to be in similar place (SOTD)

19:39:10 <libby> hello ol :)

19:39:17 * ericm not sure how long dajobe plans on being here

19:39:25 <Ol> hello all :)

19:39:27 <ericm> can you look into code while i hand-edit these?

19:39:30 <dajobe> well I might get a cup of tea

19:39:32 <dajobe> sure

19:40:00 <dajobe> testcases looks ok after -> staging

19:40:09 <libby> I was just trying out the new ical with my p800 ol...still syncing...

19:40:21 <Ol> huhu

19:40:25 <dajobe> primer also

19:40:32 <Ol> you can only blame me for the ical part of the conduit, not the p800 part :)

19:40:40 <libby> heh

19:40:49 <libby> excellent

19:41:05 <libby> I hear timezone support too? not looked yet

19:41:13 <dajobe> lbase also ok afterwards

19:41:20 <dajobe> so basically you need to edit the 2 lines above

19:41:32 <Ol> yes; I will probably post a short mail to rdf-cal today summarizing the new things that might be interesting for rdf-cal experiments

19:41:48 <libby> ooh, thanks yeah

19:42:07 <Ol> so yes, timezone, although still somewhat basic, but we do export rfc-compliant timezone block now.

19:42:18 <libby> oh, very cool

19:43:21 <ericm> dajobe, you're perl script seems to have a problems with '(' or ')'

19:44:15 <Ol> libby: another thing that might interest you is that we now handle the URL field in both backend and UI.

19:44:34 <dajobe> ericm: where?

19:44:43 <libby> oh, that's very interesting

19:45:28 <DanC> http://www.w3.org/2003/08/owl-systems/test-results-out

19:45:28 <dc_rdfig> C: http://www.w3.org/2003/08/owl-systems/test-results-out from DanC

19:45:43 <DanC> C:|OWL Test Results

19:45:43 <dc_rdfig> Titled item C.

19:46:04 <DanC> C:"Approved Tests Passed 0 times: None."

19:46:05 <dc_rdfig> Added comment C1.

19:46:30 <DanC> C:this is one of the coolest collaborations ever

19:46:30 <dc_rdfig> Added comment C2.

19:46:31 <ericm> where again are changes needed? i'm not seeing them in concepts

19:46:34 <dajobe> ericm: so the reason it gets it wrong for MT is that the <hr /> is not after the toc

19:46:47 <ericm> ineresting

19:47:07 <dajobe> I ran shadowtr.pl then diff -ur WD-rdf-concepts-20030117/ staging/rdf-concepts/|less

19:47:30 <dajobe> concepts worked ok

19:47:47 <ericm> which ones had the problem?

19:47:48 <dajobe> it was mt & schema

19:48:16 <dajobe> if you move the <hr /> to after the toc in mt, it'll probably work ok

19:48:28 <dajobe> ditto schema

19:48:39 <dajobe> then it'll likely work automatically

19:49:21 <ericm> adding <hr /> and testing

19:49:41 <libby> it worked ol! very neat :)) and the url thing looks very useful

19:50:35 <Ol> libyy: I'm writing a short mail for the list right now

19:50:53 <Ol> And sorry for being silent for some busy months :)

19:51:31 <libby> I'm not surprised! it's ace

19:51:39 <dajobe> ericm: there are some old refs in mt that need fixing, the shadowtr.pl output gives you them (lines 512, 1397, 1625)

19:53:12 <dajobe> similarly primer is refering to 20030123 lbase in 7389, 7392, 7393 lines that need fixing

19:53:30 <dajobe> I guess I Should make the script handle >2 from_dates

19:54:01 <ericm> you want to try this now?

19:54:40 <dajobe> sorry? I'm just commenting. You should make the fixes I mention above, since the script won't.

19:55:03 * ericm not seeing problem in schema... still trying to resolve this issue

19:55:22 <dajobe> well the diff says:

19:55:28 <dajobe> @@ -154,7 +154,7 @@

19:55:32 <dajobe> (<a href="http://www.w3.org/2001/sw/Activity">Activity Statement</a>).</p>

19:55:32 <dajobe>

19:55:32 <dajobe> <p> Detailed changes from the previous <a

19:55:32 <dajobe> -href="http://www.w3.org/TR/2003/WD-rdf-schema-20030905/">05 September 2003

19:55:32 <dajobe> +href="http://www.w3.org/TR/2003/WD-rdf-schema-20031010/">10 October 2003

19:55:34 <dajobe> -----------

19:55:43 <dajobe> for schema

19:56:31 <dajobe> that's in the SOTD which the script can't fix since there is an <hr /> before it

19:57:37 <dajobe> w3c cvs isn't working for me now

19:59:39 <D[a]vey> Hmm, dajobe, do you know if the non-HTML elements within the <a:body> in the example at http://www.w3.org/2001/Annotea/User/Protocol.html#PostABody are part of the body, or part of the Annotea message?

19:59:54 <dajobe> sorry, I don't have time to look at that right now

19:59:58 <D[a]vey> ah :/

20:00:17 <DanC> http://www.ietf.org/html.charters/crisp-charter.html

20:00:17 <dc_rdfig> D: http://www.ietf.org/html.charters/crisp-charter.html from DanC

20:00:17 <D[a]vey> its the <r:description>, perhaps you know off the top of your head?

20:00:25 <DanC> D:|Cross Registry Information Service Protocol (crisp)

20:00:25 <dc_rdfig> Titled item D.

20:00:33 <dajobe> I've never used annotea

20:00:39 <D[a]vey> dajobe: ah, ok

20:00:53 <DanC> D:I offer a 1000 point bounty for a demonstration HTTP+RDF system that meets their requirements

20:00:53 <dc_rdfig> Added comment D1.

20:01:09 <JibberJim> They're part of the Annotea message, used to tell it what mime-type etc. is used to return that body when the client asks.

20:01:16 <JibberJim> it==server

20:02:17 <D[a]vey> aaah, ok

20:02:36 <D[a]vey> JibberJim: so, whats the best way to decide what the content type is? heh

20:03:20 <Cardinal> Get the object of 'subject: a:body predicate: h:ContentType'

20:03:48 <D[a]vey> Cardinal: can you break that down into english?

20:05:13 <D[a]vey> D[a]vey is now known as Davey

20:06:52 <Cardinal> Actually, the subject would be that blank node.

20:07:08 <Davey> *totally* lost now

20:07:12 <Cardinal> Davey: Well, RDF is a collection of statements that have three parts.

20:07:19 <Davey> I'm a newbie remember :(

20:07:29 <Cardinal> Read the primer yet? :)

20:07:46 <Davey> no, heh, whats the link again?

20:07:55 <Cardinal> http://www.w3.org/TR/rdf-primer/

20:07:55 <dc_rdfig> E: http://www.w3.org/TR/rdf-primer/ from Cardinal

20:08:31 <Davey> E:|RDF Primer

20:08:31 <dc_rdfig> Titled item E.

20:09:07 <Cardinal> Right. Must remember to prepend URI's. :P

20:10:06 <Davey> hmm?

20:10:36 <Cardinal> Well, that shouldn't really have been chumped.

20:10:43 <deltab> what's this? a #html invasion?

20:10:51 <deltab> #web even

20:11:37 <Davey> deltab: heh

20:11:46 <Davey> its the natural evolution deltab :)

20:12:03 <Davey> one day I hope to be as l33t as Cardinal :D

20:21:27 * DanC thinks it's healthy to chump the RDF primer here every few days/weeks/whatever

20:21:51 <DanC> E:good for fun and health

20:21:51 <dc_rdfig> Added comment E1.

20:23:06 * DanC also thinks it's fine to get into the occasional HTML/CSS minutiae here, contrary to a remark that I can't attribute cuz it's scrolled off the end of my buffer

20:31:49 <Davey> ok, in the client I've gotten as far as check for and construct Annotea XML for Annotations with an existing body (I check if the body is *just* a valid URI (using the regex on the URI RFC)) and almost finished the part with new bodies (just doing the content type and stuff part), and on the serverside, I've written enough to check the type of action being requested and to check if the POSTed data contains an existing body

20:35:18 <bijan> timbl?

20:35:21 <Ol> libby: sent

20:35:34 <bijan> Have you been following the sw-meaning stuff recently?

20:35:47 * bijan trying to anticpate not dying in friday's telcon

20:35:56 <bijan> Stupid pi meeting, swslf2f, and ISWC

20:36:20 <DanC> swslf2f?

20:36:25 <bijan> swsl f2f

20:36:28 <DanC> is that one of the Oct FL thingies?

20:36:33 <bijan> Yes

20:36:36 <bijan> In between

20:36:45 <bijan> Oh, and ontoweb OWL implementors meeting

20:36:55 * bijan is sorta presenting in ALL of these things

20:37:36 <timbl> Hi Bijan.

20:37:42 <bijan> hello.

20:37:43 <timbl> No, I am very derelict ;-)

20:38:12 <timbl> I am not up to speed on the sw-meaning.

20:38:26 <bijan> Well, I ask becasue of the current agenda

20:38:33 <bijan> Which is me picking on/at/about you

20:38:35 <bijan> :)

20:38:36 <timbl> I do have a few minutes now if you would like to discuss it .

20:39:16 <timbl> We could map out a bit of the argument here and tackle th remianing on Friday.

20:39:25 <DanC> as sw-meaning co-chair, I consider timbl and bijan up to date almost by defintion. i.e. if you guys haven't read a piece of mail, I don't consider that the forum has consumed it.

20:39:36 <bijan> I've read up :)

20:41:06 <timbl> Got antying in mind?

20:41:31 <bijan> well, I replied to your reply

20:41:45 <bijan> Sandro posted a "running code" version of what he thinks is your view

20:41:57 <bijan> There seems to be some consensus against the "strong imports" view.

20:42:01 <bijan> Including from you

20:42:44 <bijan> I.e., use of a uri requires (soemthing equivalent to) importing that uri's "namespace" part

20:43:11 <timbl> Wherethe strong imports view is ... let me guess... that the document which contains the triple s p o is deemed to contain the triples from the dereferent of p?

20:43:38 <timbl> Right. I think there is a consensus that the document doesn't import it in the sense of #include.

20:43:39 <DanC> .time

20:43:39 <datum> Wed, 08 Oct 2003 20:43:39 GMT

20:43:44 <DanC> .time CDT

20:43:44 <datum> Wed, 08 Oct 2003 15:43:44 CDT

20:43:49 <DanC> .time EDT

20:43:49 <datum> Wed, 08 Oct 2003 16:43:49 EDT

20:44:40 <bijan> So, a related question is, "is my document inconsistent if *if* it were to do the imports closure of the used uris, the result would be inconsistent"

20:44:55 <timbl> I saw that question.

20:45:09 <timbl> In systems like this, there are many forms of "inconsistent".

20:45:16 <timbl> Just like there are many forms of "valid".

20:45:41 <timbl> A working group can dfeine one, and if it is useful, then people will use the term and the concept.

20:45:41 <bijan> I wouldn't go there

20:45:47 <bijan> Just coing a fresh term :)

20:47:01 <bijan> Perhaps a similarly related question: "Is it in any sense 'ok' (i.e., endorsable or merely not frowned upon) to use an alternative, conflicting ontology for a term"

20:47:09 <timbl> we could coin "OWL closure conistent" to mean what you said -- that a document's ontological closure was consistent with the OWL axioms.

20:47:13 <d2m_> d2m_ is now known as d2m

20:47:42 <bijan> That's a bit ambiguosu

20:48:03 <bijan> Since there's already an explicit owl:imports consistency notionj

20:48:07 <timbl> Say it better.

20:48:44 <bijan> I would just call the document "consistent". If it conflicts with soem other docuemnt, that document set is inconsistent

20:48:44 <timbl> OWL-consistent under ontological closure?

20:49:17 <bijan> ontological closure isn't crips

20:49:21 <bijan> crisp

20:49:21 <timbl> so the document set we are talking about here is the ontological closure and some representation of the OWL model theory.

20:49:26 <timbl> crips?

20:49:28 <bijan> Because of the "ontological" not because of the "closure"

20:49:34 <bijan> crisp :)

20:49:53 <timbl> I defined it i thought quite crisply.

20:50:03 <bijan> Not all the documents "behind" a uri will be "ontologies" (however you partion that out)

20:50:13 <bijan> Oh, is this th =pt thing?

20:50:22 <timbl> right.

20:50:32 <bijan> Ah.

20:50:51 <bijan> Yes, that operational definition seems as robust as owl:imports

20:50:55 <timbl> for any s x o or s rdf:type x look up x and load it recursively

20:51:02 <bijan> Take that as praise or blame, as your inclination :)

20:51:26 <bijan> I'm just saying that that may not map onto an intuitive notion very naturally

20:51:59 <timbl> Actually, it maps pretty darn well. It is i think what the daml validator does by defualt. It was what my validator separately did by default.

20:52:23 <bijan> Perhaps.

20:52:30 <bijan> Hmm.

20:52:35 <bijan> Why not o rdf:type x?

20:52:50 <timbl> >

20:53:02 <bijan> And is it on the entailment closure of the document, or just the explicit triples?

20:53:18 <timbl> ? s was a variable use o if you like?

20:53:25 <bijan> no

20:53:28 <bijan> I mean you said

20:53:38 <bijan> For any s x *o*, but only looked up the type of *s*

20:53:52 <DanC> timbl, please don't rebind "consistent" and "valid". I spent 18 months learning what the logicians mean by those words, and they spent 50 years writing it down.

20:54:06 <bijan> (70, at least)

20:54:11 <timbl> Ok.

20:54:15 <bijan> (2500, by some measures)

20:54:27 <bijan> (And undergrads *STILL* die hard on it :))

20:54:39 <timbl> I am just saying Bijan can't bind them to these particular intersting peoperties, we have tomake up new names.

20:54:55 <bijan> I can't bind, what?

20:55:00 <timbl> for any s x o or s rdf:type x look up x and load it recursively

20:55:10 * DanC is noodling with terms like web-justifyable conclusion

20:55:14 * bijan didn't do any binding

20:55:22 <bijan> Ah, s is in differnt scopes

20:55:24 <bijan> I see.

20:55:31 <timbl> (you can't use "consisetnt" for "consistentwithetheowlspecsanditsontologicalclosue")

20:55:42 <bijan> Did I?

20:56:00 <bijan> I mean "consistent" in the sense of "possible all statmetn co-true"

20:56:14 <bijan> That's all

20:56:40 <bijan> In an owl document, the questions is "what is the set of statements 'in" that docuemnt"

20:56:54 <bijan> It's the explicit statements + the imports closure

20:56:57 <bijan> No rebinding needed

20:57:48 <timbl> ((never mind. [bijan So, a related question is, "is my document inconsistent if *if* it were to do the imports closure of the used uris, the result would be inconsistent"] )

20:58:22 <bijan> There's no redefinition there.

20:58:32 <timbl> setfo statemenst whcih is in the documents plus imports -- yes, i think taht;s what the spec says.

20:58:41 <bijan> Exactly

20:58:53 <bijan> And that set is consistent if there's a model

20:58:55 <timbl> Of cousre the set of statements in the RDF document is the set of triples period.

20:59:16 <bijan> And an owl document isn't quite an RDF document

20:59:29 <bijan> Given that owl:imports really is a bit of extra syntax

20:59:39 <bijan> Whichis part of the mess of it.

21:00:04 <timbl> Or rather, the relationship between an RDF document and an RDF graph is not the same as the OWL parser relationship between an RDF document and an RDF graph.

21:00:12 <timbl> The document is still the same document.

21:00:15 <bijan> Ues

21:00:20 <bijan> er.

21:00:23 <bijan> I agree with the former

21:00:29 <bijan> Unsure of the latter

21:01:03 <bijan> Certianly in some sense

21:01:08 <timbl> I agree that owl:imports pretends to be a bit of syntax when it isn't to RDF.

21:01:14 <timbl> Iagree that is messy.

21:02:17 <bijan> Hmm. I was going to say that consistency is defined over the graph, not the document per se, but that's messy too

21:02:25 <timbl> But OWL so long as it is clear on the set of triples it is defining it cna define what it likes. It was just nice when the RDFtriples and OWL tripl;es were the same.

21:02:41 <bijan> It would be slighlty neater if it were rdf:imports ;)

21:02:51 <bijan> Oh eek, convergence

21:02:56 <bijan> We should stop now

21:03:01 <bijan> Declare victory

21:04:01 <bijan> Let me note that we've gotten *way* far away from "ontological committment"

21:04:04 <timbl> slightly neater. But we managed to get aboutEachPrefix out of rdf after several years, and i am not very happy to see new stuff go in whcih could be similarly viewed in future years.

21:04:15 <bijan> Or "commitement to an ontology"

21:04:16 <timbl> Ok., Ontological commitment.

21:04:24 <timbl> So what does it mean?

21:04:32 <bijan> Well, I don't know what *you* meant by that

21:04:43 <bijan> But to me, committmetn is some sort of "must"

21:04:47 <timbl> I think it was a phrase I got form someone else.

21:04:52 <bijan> You weakened to "should"

21:05:07 <bijan> And now, it's ok to "maybe haven't something else that conflicts" :)

21:05:17 <bijan> We're not all that far from "ontological promiscuity"!

21:05:18 <bijan> :)

21:05:27 <timbl> maybe haven;t something else which conflicts?

21:05:56 * DanC has not reached enlightenment on what "commitment to an ontology" means... i.e. I haven't written it down in larch

21:06:08 <timbl> yes, commitment is a must. But I think where we have in the past come off the rails is in ditinguish tbetrween the naive protocols and the robust ones.

21:06:20 <bijan> I.e., I can in my docuemnt, or in a docuemtn i import (or merge with or) "ontological" statements which conflict with the uri's owner's ontology

21:06:24 <DanC> larch for "consistent" and "valid" fyi: http://www.w3.org/XML/9711theory/FormalSystem

21:07:11 <timbl> Let's talk about the naive protocol.

21:07:25 <bijan> Well, it was in a fairly late message that you introduced this distinction

21:07:44 <bijan> "When naive, be committent. As you grow more experiences, be cynical, but robust"

21:07:46 <bijan> :)

21:07:49 <bijan> committed

21:07:58 <timbl> Example: Naive protocol: everyone writes who they are in the from feoidl,m and everyone else believes it.

21:08:16 <bijan> From feoidl,m?

21:08:19 <DanC> ETOOMANYTYPOS here too

21:08:21 <bijan> Form field?

21:08:25 <bijan> From Field?

21:08:29 <timbl> Less naive prtocol: everyone nice writes who they are in the from feidl, and no one believes it is the message says anything aboyt enlarging the size of anyting.

21:08:39 <timbl> "From" fields

21:08:53 <timbl> of email

21:08:56 <timbl> sorry

21:08:58 <DanC> roger

21:09:09 <timbl> So, lets ry to make a naive protocol for this.

21:09:26 <timbl> Everyone who makes an ontolog yexpolains what they can about it in owl and publishes that at the namespace.

21:09:42 <bijan> Eek!

21:09:44 <bijan> No logger?

21:09:48 <timbl> Everyone who uses an ontology has read the specs and understands ethm and uses the terms correctly.

21:09:59 <nernst> "ontological commitments are agreements to use the shared vocabulary in a coherent and consistent manner"

21:10:10 * bijan was about to send notification to the sw-meaning list :(

21:10:25 <nernst> gruber, IJHCS 43, 1995

21:10:36 * DanC was counting on logger to capture this magical moment too

21:11:02 <DanC> but don't let that stop you; there's plenty of redundancy.

21:11:04 <timbl> Evryone who reads any RDF message can take the ontological closure of it and hold the person to not only the triplesin the messgae but also the closure(pt).

21:11:32 <DanC> ^name for that, pls?

21:11:38 <timbl> protocol0

21:11:42 <DanC> thx

21:11:45 <timbl> <#protocol0>

21:11:53 <dajobe> i'll turn this xchatlog into a logger log, iyswim

21:11:57 <timbl> <> log:uri ?x

21:12:09 <bijan> dajobe: thanks!

21:12:24 <timbl> So this protocol is open to a number of smags due to bugs and meanness.

21:12:36 <timbl> - net goes down

21:12:49 <timbl> - someonbody writes garmage in the obntology file

21:13:05 <timbl> - someone writes "you owe me $1e6" in an ontologu file etc.

21:13:14 <timbl> So we have many many less naive prtocols.

21:13:21 <DanC> BLURB: sw-meaning foo

21:13:22 <dc_rdfig> F: sw-meaning foo from DanC

21:14:04 <bijan> I have a naiver protocol

21:14:06 <bijan> I think.

21:14:12 <timbl> Mostly these are different forms of caution on the reader's part. (Caveat lector?)

21:14:18 <timbl> ok, you have a p-1?

21:14:46 <bijan> Hang on

21:15:17 * timbl watches while Bijan rummages in his trunk for a very naive protocol

21:15:32 <bijan> I'm still not quite grasping yorus...but

21:16:04 <bijan> Hmm. It may not be a protocol

21:16:35 <DanC> F:[FYI to sw-meaning list|http://lists.w3.org/Archives/Public/public-sw-meaning/2003Oct/0028.html]

21:16:36 <dc_rdfig> Added comment F1.

21:17:03 <bijan> imbl Everyone who uses an ontology has read the specs and understands ethm and uses the terms (no "correctly")

21:17:39 <bijan> Everyone who makes an ontolog yexpolains what they can about it in owl and publishes that at the namespace.

21:17:52 <bijan> s/they can/they want/

21:18:11 <bijan> I'm not sure if this is naive or robust :)

21:18:17 <bijan> Or nuts

21:19:03 <timbl> Well, does it allow me to write a message and maintin that I am using owl:TransitiveProperty to mean th eclass of carrots?

21:19:20 <bijan> "correctly" really sticks; it's hard to characterize

21:19:24 <bijan> Even naively

21:19:36 <bijan> Not in owl dl, yes in owl full

21:20:00 <bijan> I'd wadger

21:20:15 <bijan> Actually,t her'es a bunch of sublties

21:20:32 <bijan> You could definitely say that owl:TransitiveProperty sameClassAs :Carrots

21:20:49 <Davey> [off] wonders if dajobe can correct typos when he converts to logs...

21:20:49 <bijan> What that *means* is a bit trickier to get out

21:21:08 <timbl> "correctly" is what Pat I guess meant when he said, "Yes, of ocurse evryone has to use a URI to mean the same thing" OWTTE.

21:21:12 * DanC wonders if mdean_'s appearance is causally connected to the FYI I sent to sw-meaning

21:21:24 <mdean_> yes

21:21:31 <bijan> Thank you for sending that DanC.

21:21:50 <bijan> I'd need to check Pat's context

21:22:19 <bijan> In any context, every occruance is same denoting

21:22:59 <bijan> I'm pretty sure that's all he claimed

21:23:57 <bijan> In any single graph, any node labeled with string= uris, is the same node

21:24:23 <bijan> On graph merge, same uried nodes are merged

21:24:34 <DanC> no, PatH concurred with more than that in a TAG telcon...

21:24:50 <bijan> DanC: i'm only going from the mail I recall

21:25:03 <bijan> If there's an email where he went further I'd love a pointer.

21:25:06 <DanC> to wit "Parties who wish to communicate about something agree upon a shared set of identifiers and on their meanings." -- http://www.w3.org/TR/webarch/

21:25:39 * timbl can't remember when it was

21:26:30 * bijan notes that danc's to wit is a lot more constrained than timbl's quote

21:27:06 <DanC> but it's not as constrained as "In any context, every occruance is same denoting" is it?

21:27:09 <bijan> E.g., I may choose not to communicate with someone (e..g, the uri owner), hence I might be free to disagree on the meaning

21:27:23 <bijan> DanC: unclear

21:27:27 <DanC> ok, unclear

21:27:36 <bijan> Hmm. I didn't want to get as far as same denoting

21:27:44 <bijan> That's why I keep to "same node"

21:27:50 <DanC> to use the web is to agree to communicate with everybody, I think. in a way. hmm.

21:27:59 <bijan> uh...

21:28:04 * bijan stops using the web

21:28:35 <bijan> And surely not

21:29:01 <ChrisDodo> anyone have RDF primers for, well, marketing people?

21:29:22 * bijan does! "RDF Good. Fund Me. Semantic Semantic"

21:29:29 * DanC thinks maybe some of the PRISM materials might suit ChrisDodo's needs

21:29:38 <ChrisDodo> PRISM?

21:29:39 <DanC> hoot!

21:29:45 <DanC> .google prism RDF

21:29:46 <datum> prism RDF: http://www.xml.com/pub/a/2003/02/12/rdflib.html

21:29:53 <DanC> hmm

21:30:12 <bijan> Plus, I imagine pat meant "communicate for some purpose, or period of time, or some topic"

21:30:28 <dajobe> ChrisDodo: this might be handy http://www.semaview.com/c/RDFXML.html but it's more technical and aimed at tech/managers

21:30:36 <DanC> http://prismstandard.org/

21:30:36 <dc_rdfig> G: http://prismstandard.org/ from DanC

21:30:47 <bijan> Not, "if I've ever wanted to communicate with you, on ANY topic, however constraint, I must agree on every possible identifer and their meanings"

21:30:48 <timbl> Pat made the pointtaht you can never say that a term has the same denotation for two difefrent agents. I described the communication rpocess by which tao agents exchange messages until they hve eliminated many many interpretations, remaining only with those in which the term is used consistently with the large set of exchanged messages. Frok then on, whenever one detects in the other a messae which says soemthing

21:30:50 <DanC> G:|PRISM: Publishing Requirements for Industry Standard Metadata

21:30:50 <dc_rdfig> Titled item G.

21:30:54 <bijan> You coudlnt' have disputes over terms!

21:31:40 <DanC> you have to quote terms if you want to dispute them, no?

21:32:04 <ChrisDodo> cool, ta, I'll have a look through, and see if any are not-too-bamboozling.

21:32:16 <bijan> Not always. But even so, if to communicate I have to agree with you on ALL terms

21:32:18 <DanC> G:perhaps a good source of marketing materials about RDF

21:32:18 <dc_rdfig> Added comment G1.

21:32:29 <bijan> The fact that I can quote 'foo' doesn't help

21:32:38 <ChrisDodo> i guess sem web explanations would help too.

21:32:46 <timbl> Well, in fact you are communicating with me (more or less) and we don't agree on ALL terms ... but a few.

21:32:48 <DanC> I consider the web community that group of people/agents who agree what URIs mean.

21:33:00 <bijan> I mean, what can I say, "I think 'foo' means what 'bar' means" DanC: 'But we agreed that 'foo' meant 'baz' otherwise we coudn't communicate!"

21:33:02 * timbl is lonely ;-)

21:33:23 <bijan> Right, we agree *enough* on *enough* terms

21:33:39 <bijan> We certainly don't agree completely on all, or perhaps even *any*, terms

21:33:59 <timbl> Suppose I say "foo rdf:type owl:TransitiveProperty".

21:34:02 <bijan> And if not we too, then certainly not everyone who uses the web :)

21:34:16 <bijan> we two

21:34:18 <bijan> even

21:35:25 * bijan so supposes

21:35:55 * timbl wodners where these ruminations (a) came from and (b) are going to.

21:36:21 * DanC encourages bijan and timbl to ignore me and get back to protocol0 and the like

21:37:06 <bijan> Let me grant protocol0, for the moment, the status of completely worked out and nailed down in clear language, ready for spec inclusion

21:38:08 <bijan> There's still the prior question of whether this needs to be specified now, for example, to make RDF documents "meaningful" (as you have several times opined that they were not (yet)"

21:38:39 <bijan> Whether the naive protocol is a *correct* protocol, or the right one for the web or the semantic web

21:39:16 <bijan> Whether it + some robustification shoudl be the standard protocol

21:39:16 <timbl> Theer is a need tosay not very much, but something.

21:39:28 <bijan> That argument has yet to be made

21:39:37 <timbl> There isno "standard" protocol, i think.

21:39:40 <bijan> I consider protocol0 to say *quite* a bit

21:39:46 <bijan> Ergo, no need to standardize?

21:40:16 <bijan> I really do want to drag some of this back to "what should wgs or the tag *do*"

21:40:18 <timbl> No, the real algorithms a real agent will use to detremine what to trust, when to dereference things, and so on, will be many and varied.

21:40:37 <bijan> So, you don't think that's tied to meaningfulness of RDF?

21:40:42 <bijan> Anymore?

21:40:42 <timbl> But yes, there needs to be something basic whcih defiens (naively) what and RDF document leans.

21:40:54 <bijan> To the left!

21:40:57 <bijan> Of course

21:41:03 <bijan> We're all good marxist, yes?

21:41:20 <kspace> of course.

21:41:34 * DanC feels a hole in his eductaion; doesn't really know what a marxist is

21:41:42 <bijan> Actually, Id on't see at all why there needs to be anything, basic or not, that defines the "naive" meaning of RDF document

21:41:52 <bijan> I mean, is "naive" meaning differnet than, well, the meaning?

21:42:02 <timbl> ./me brb

21:42:19 <kspace> what is the use of having a less-specific meaning of an ontology?

21:42:25 * bijan debugs some owl, or an owl reasoner, or both

21:42:47 <bijan> kspace: it's not clear that naive meaning is *less* specific

21:43:03 <bijan> "naive meaning" in tim's sense

21:43:42 <kspace> ahh

21:43:43 <kspace> brb

21:43:50 <bijan> Or were you implying that my use of "meaning" in constrast to NM was less specific,a ndthus less useful

21:43:53 * bijan goes back to the debugging

21:45:10 * timbl back

21:45:35 * bijan listens

21:48:11 * DanC contemplates the null hypothesis for sw-meaning. suppose we asked the TAG to with draw the issue and closed the sw-meaning list.

21:48:40 * sandro finally catches up to real-time, after 40 minutes (with distractions).

21:48:43 <bijan> I certainly am cool with the former. The latter may still be useful.

21:48:53 <timbl> You ask wheteher there is any need for the term "naive meaning" - why is different from the term "meaning".

21:49:00 <timbl> I agree.

21:49:10 <timbl> We need a naive statement of what the meaning is.

21:49:18 <bijan> I disagree.

21:49:19 <bijan> Or, ok

21:49:37 <bijan> "The meaning of a document is what it means."

21:49:47 <bijan> I mean, there are tons of naive statements possible

21:49:54 <timbl> If we simply askwhat the meaning is and there are philosophers areound, then we lauch into a history of the philosphy of language and semantics from Amcient Greece to Wittgenstein.

21:49:56 <bijan> I don't see the value of naivety in this matter

21:50:21 <timbl> Naively, we all agree on this call what a transitive property is.

21:50:34 <bijan> Y'know, I would strongly appreciate, in general, that gratutious slams of philosopher were much less frequent in these discussions

21:50:42 <bijan> They, in effect, tell me off

21:50:45 <bijan> And I don't appreciate that.

21:50:51 <timbl> I'm sorry.

21:50:56 <bijan> No problem

21:51:03 <timbl> I'm not slamming sphilsophers.

21:51:36 <timbl> I do want to avoid the engineering of a data format from being held up by too much philosophical discussion - mor ethan we need.

21:51:47 <bijan> Sure.

21:51:55 <bijan> But less than we need seems to be an equal danger

21:52:09 <bijan> And promulagting bad specs is perhaps worse than no specs

21:52:17 <timbl> yes.

21:52:35 <timbl> So. lets get back to whatyou objected to.

21:52:56 <timbl> I think one part which you didn't object to was taht the meaning of the RDF document was the logical conjucntion ofthe meaning of the statements.

21:53:03 Topic now Semantic Web hack-n-chat http://rdfig.xmlhack.com/ latest logs http://ilrt.org/discovery/chatlogs/rdfig/latest wiki http://esw.w3.org/topic/FrontPage

21:53:03 Users on #rdfig: logger Davey ear1grey datum dmiles_afk libby JohnBlack fink mdean_ d2m ChrisDodo timbl Ol bitsko Cardinal karl_ aharth eikeon dajobe darobin xover grove_ bijan josek MarkB mdupont shellac nmg iwaim kao sbp mea_culpa dc_rdfig grove jordan xower golbeck JibberJim merriam eikco dngor tav deltab ericP _joshua grault sandro grumbles mmealling GabeW dhmspector ericm Wack Oblomov kham DanC

21:53:04 <NickServ> If this is your nickname, type /msg NickServ IDENTIFY <password>

21:53:04 <NickServ> Password accepted - you are now recognized

21:53:04 <MemoServ> You have no new memos

21:53:04 <ChanServ> [#rdfig] This channel is logged and blogged: http://logicerror.com/rdfIRCWelcome

21:53:05 <timbl> we can getit down to what an rdf statement means.

21:53:08 <bijan> hey logger

21:53:17 * timbl waves to logger

21:53:19 <dajobe> (I'll repair things later/tomorrow)

21:53:25 <bijan> thanks again!

21:54:14 <timbl> But I think you objected to my saying that the meaning of a statement s p oo was that the relationship denoted by p held between the things denosted by s and o.

21:54:15 <bijan> (meaning document == meaning of the conjunctin of the statements, sure)

21:54:25 <timbl> s/oo/o

21:54:40 <bijan> hmm.

21:54:45 <bijan> yes

21:54:51 <bijan> I might.

21:55:02 <bijan> Let me think

21:55:06 <timbl> maybe others protested things like (a) what do you mean by denote (and vice-versa) anyway? and (b) why is p special?

21:55:31 <bijan> A more than B

21:55:42 <bijan> In the formulation above, P isn't especially distinguished

21:55:51 <timbl> it was

21:55:56 <timbl> denoted by p

21:55:57 <bijan> Inthe ways I've objected to before

21:56:06 <bijan> "denoted by s and o"

21:57:07 <DanC> yes, timbl, so far, the choice of p over s and o is totally arbitrary. And a 4th isomorphic view is: "... that the relationship rdf-holds holds between s, p, and o"

21:57:11 <bijan> I fail to see the distinguishing

21:58:21 <bijan> (No one says that they all have the *same* role.)

21:58:29 <bijan> (that would be bizarre)

21:58:38 <timbl> So what is the difference>

21:59:39 <bijan> For one, on the standard logical reading, p is a predicate term, and has some syntactic restrictions, and gets mapped to a set of ordered pairs in the model theory

21:59:39 <timbl> ?

21:59:57 <bijan> s and o get mapped to sets of individuals

22:00:03 <DanC> the only difference between s, p, and o in your description so far, timbl, is that they're ordered, i.e. distinguishable. nothing you've specified so far distinguishes your design from the rdf-holds view.

22:00:33 <DanC> sets of individuals?

22:00:51 <bijan> D'oh

22:00:53 <bijan> Sorry

22:00:58 <bijan> To individuals

22:02:08 <DanC> I agree that distinguishing predicates leads to an interesting and useful system, but I can't, for the life of me, say why.

22:02:13 <timbl> so can I maintain that <foo> rdf:type owl:TransitiveProperty means that <foo> is a carrot?

22:02:29 <DanC> nothing you've said so far says otherwise, timbl.

22:02:33 <bijan> The kind of distinguishing you seemed to be after was some how getting more statements about P was somehow more important and told you more than getting further information about s or o

22:02:39 <timbl> I could make an ontology for <foo> and declare that it is such that any statement about it means that it is a carrot.

22:02:47 <bijan> And that's just not the case, afaict

22:03:41 <timbl> DanC, no I have said that p identifies th reltion which holds.

22:03:47 <timbl> That is what RDF needs to say.

22:04:18 <bijan> You mean that the statemnt gets mapped to an ordered pair

22:04:22 <DanC> but how is that different from saying that o identifies a relation which holds between s and p?

22:04:23 <bijan> not a set of ordered pairs

22:04:32 <timbl> Then, the web aprchitecture tells me (in way snot formalizes but it would be a good idea) that p is thatt rdf:type is that defined by the rdf namespace docuemnt as "type".

22:04:41 <bijan> But that's a function of the statement structure, not the p

22:05:30 <timbl> statement structure?

22:05:43 <DanC> yeah... what's statement structure?

22:05:46 <timbl> triple?

22:06:17 <bijan> One sec, phone

22:07:52 <DanC> btw, tim, "Then, the web arch tells me rdf:type means..." appeals to my intuition, but isn't responsive to the question "why is p special as opposed to s or o?"

22:07:56 <timbl> (and the rdf:type definition should tell me in english as well as idealy formally that the meaning of rdf:type is the binary relation of class membership. )

22:08:37 <timbl> isn't responive to the question "why is p special"?

22:08:52 <timbl> why: becasue the spec [should] say so.

22:09:01 <DanC> argument by assertion. bzzzt.

22:09:25 <timbl> I thought you found a stement to the effect that p identifes a relaton.

22:10:00 <timbl> It is true that you can make a triad of systems where s is considered to be teh binary relation betwen all the propertyes and obejcts for which s p o holds.

22:10:04 <DanC> yes, it's in the RDF model theory. but it's totally arbitrary. It could be rewritten to say that o identifies a relation without any observable impact on RDF software, tests, etc.

22:10:41 <timbl> so you agre eits in the spec.

22:10:48 <timbl> so no bzzzt

22:11:03 <timbl> arguing by quoting the spec is not a tilt condidtion.

22:11:58 <DanC> but the spec contains no justification why p is special, and indeed, could be re-written so that p is not special with no observable impact.

22:13:59 * bijan back

22:14:02 <DanC> appealing to that spec would be like saying that P is more important than DL in HTML because it's earlier in the spec.

22:14:44 <DanC> the HTML spec makes no claims about the importance of P vs DL.

22:14:54 <bijan> Plus, in rdf p identifies an individual which has a relation to an extention

22:15:04 <bijan> So much more like option 4

22:15:08 <timbl> Yes, so why is it special?

22:15:14 <DanC> it's not.

22:15:23 <bijan> Yes, it's really not

22:15:28 <timbl> Why isthe verb difefrent from the subject or objet in a sentence forthat matter?

22:15:36 <DanC> in the RDF specs that are on the REC track, p is not special.

22:15:40 <bijan> You elided from "special" to "different"

22:15:48 <bijan> We all agree "different"

22:16:14 <bijan> The verb of a sentence isn't special, either

22:16:25 <DanC> I dunno why verbs are different (or if they are).

22:16:37 <timbl> It could be that the difference is topological.

22:16:53 * bijan notes that they're clearly different syntactically in many natural languages, etc. etc.

22:17:12 <DanC> topological? elaborate?

22:17:43 <timbl> When you have the (naive) appoarch that you start with soem siomple ontologies and define other ontolgies in terms of them, then you end up with a web in which closre=pt gives you a small set of statements.

22:18:23 <DanC> ah... now this seems promising...

22:18:30 <timbl> Doing the same thing reading specs you getthe same effect. As you follow back down p and t you get a finite set of specs, which are grounded in egnglish we we assume we all \agree on :-)

22:18:50 <DanC> ... derive the utility of the closure=pt behaviour from the observed and expected deployment of RDF data/schemas/ontologies.

22:19:14 <bijan> Well, if you put that into the spec, you'll presumably get that behavior

22:19:17 <timbl> So the algorithm: if you don't know what it means, look up P, an dif P is rdf:type look up T, terminates

22:19:19 <DanC> grounded in english... er... I think you'll do better not to go there.

22:19:23 <bijan> So why not observe exisiting behavior

22:19:43 <bijan> I remember a survey of DAMl+OIL ontologies and RDFS schemas

22:19:58 <bijan> There were plenty of High Property docuemnts (i.e, lots of properties)

22:20:09 <bijan> Often with deep property trees.

22:20:27 <bijan> Plus, given that niether rdfs nor owl is *particularly* expressive wrt properties

22:20:42 <bijan> It's perhaps no surprise that you'd get fewer statements about them.

22:21:04 <bijan> But that's suggests that less effort and interest will go into thos defining statements about the properties

22:21:12 <timbl> I have to go soon.

22:21:22 <bijan> Indeed, I'd think that the *more* somoene says "definitionally" about a term

22:21:31 <bijan> The more they'd want me to get those statements

22:21:39 <bijan> At least so my naivite sez

22:22:08 * DanC finds the carrot example intuitively compelling; wishes he could understand how to generalize it

22:22:21 <bijan> In any case, I'd be a bit perturbed if we built something in that depended on a somewhat inessential expressive asymamatry in current languages

22:22:26 <bijan> Surely that's premature optimization

22:23:02 <Davey> Davey is now known as D[a]vey

22:23:09 <bijan> (And for standardization/recommendation purposes, I'd wait for actual experience to kick in. Hence, no need to put things in current specs."

22:23:30 <DanC> I don't think it's premature optimization. That view seems to lead to no semantic web happening at all; it would have led to no web happening at all 10 years ago.

22:23:39 <bijan> ?

22:23:44 <timbl> Need to put thinsg into current specs.

22:24:27 <bijan> DanC: you think letting practice drive standardization on this matter is, uh, semantic web killing?

22:24:30 <bijan> I fail to see how.

22:24:40 <timbl> I don't want a spammer to gt away with saying that <> From timbl means that the sun sets in the west.

22:24:41 <DanC> i.e. 10 years ago, we could have said "sure, ange-ftp identifiers are just as good as URIs" but the result would have been no web at all.

22:25:22 <bijan> uh..I totally fail to see the analogy

22:25:52 <bijan> I mean, 10 years ago, you could have said, "Hey! Two way links are *really* essential"

22:26:00 <DanC> well, it's just my intuitions anyway. I don't have any real argument to make.

22:26:29 <DanC> lots of people *did* say 2 way links were critical. remember them?

22:27:00 <bijan> Do you see the analogy *i* made? yours was bout not-doing, my was about doing

22:27:21 <bijan> I say, "we don't need to pick a behavior pattern yet"

22:27:42 <bijan> I assume that's like saying, "we don't need an identifier scheme (yet? there's the disanalogy)"

22:28:01 <DanC> yes, that's like saying we don't need URIs to start the web.

22:28:04 <DanC> I think.

22:28:25 <bijan> "We don't need to pick a naive behavior pattern yet"?

22:28:31 <timbl> Fortunately, I think people are well established building a sem web out of properties and classes. if you start a project inwhich you rotate s p an.d o... but i have to go

22:28:40 <timbl> timbl is now known as tim_away

22:29:13 <bijan> Again, it's well established, then what's the fuss? What do the currents spec block? Where's the projected failure?

22:29:24 * bijan wraps

22:31:16 * DanC isn't sure what "wraps" means, but supposes it means something like "wanders off"

22:32:03 <bijan> Did I mention that I deserve credit for carrying on this conversation while inter-debugging an owl reaonser and a sws owl ontology?

22:32:15 <bijan> Working in raw RDF/XML? :)

22:32:21 <DanC> I think so. I hereby award you 1 credit. :)

22:32:37 <bijan> After having rising at 5am?

22:32:58 <bijan> And walked 5 miles uphill bothways in a freakish hurricane-blizzard?

22:33:09 * bijan isn't going to get that second credit... :)

22:33:34 <DanC> fwiw, I'm OK with the null hypothesis.

22:33:43 <bijan> Interesting

22:34:16 <DanC> I said as much in my intro to sw-meaning: quoth MSM: if I don't publish a paper on gravity, it'll still work.

22:34:27 <bijan> Indeed.

22:34:37 <bijan> I believe I made this point way back before the tech plenary

22:34:42 <bijan> it's been my view all along

22:34:54 * danbri remembers you saying that, yes

22:34:58 <bijan> I find it rather soothing, actually

22:35:01 <bijan> hey danbri!

22:35:05 <danbri> hello

22:35:12 <bijan> Long time!

22:35:25 * danbri sharing his 802.11 with other starbucks 'customers' (incl. ericp)

22:35:29 <danbri> yup!

22:35:32 <bijan> heh

22:35:39 * bijan definitely found a reasoning bug in pellet

22:35:41 <bijan> Again.

22:35:54 <DanC> I have a pet-pieve around "we need". People often use it when I'd rather they said "I really want" or "I find the lack frustrating..."

22:35:57 <bijan> Amazing what bugs can lurk in the face of all these test cases!

22:36:02 <bijan> OH YEAH!

22:36:07 <bijan> Drives me freaking *nuts*.

22:36:09 <bijan> I froathe

22:36:29 <bijan> The other one, "All you need to do to achieve popularity is foo. Hop to it"

22:36:43 <bijan> Minority programming languages are *innodated* with this sort of thing

22:37:03 <bijan> "Smalltalk would take over the world, if it had a Visual Basic/Python/Java like syntax"

22:37:20 <bijan> or "Without such a syntax, Smalltalk will die."

22:39:25 * bijan bemoans the datatype syntax required to express cardinalities

22:41:38 * DanC wishes more people could have seen the wisdom of the rdfs:format solution to datatypes ;-)

22:41:51 <bijan> Argh: com.hp.hpl.jena.datatypes.DatatypeFormatException: 1 is not a Number

22:42:38 <bijan> yeek.

22:42:41 <DanC> "I'm quite concerned that the rdf:datatype proposal

22:42:41 <DanC> is too complex to deal with in a timely manner" -- yours truely, Oct 2002. http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2002Oct/0031.html

22:42:59 <bijan> Does the w3c validator handle datatypes?

22:43:10 <DanC> umm... I think so; good question.

22:43:17 <bijan> _:jARP84497 <http://www.w3.org/2002/07/owl#cardinality> "1" .

22:43:33 <bijan> From: <owl:cardinality rdf:datatype="&xsd;nonNegativeInteger">1</owl:cardinality>

22:43:47 <bijan> That looks like a big ole "no" :(

22:44:13 <DanC> "It currenlty uses version 2-alpha-1."

22:44:55 <bijan> Argh, caught between too much and too little

22:44:59 <DanC> "Note: this online service has been updated and now supports the Last Call Working Draft specifications issued by the RDF Core Working Group, including datatypes."

22:45:16 <DanC> it claims to handle datatypes. 1st sentence on the page

22:45:40 <bijan> Argh. This smells like I have to file a bug report :(

22:46:12 <bijan> Doesn't ntriples do the ^^datatype thing on datatyped literals?

22:46:44 <bijan> oh...

22:47:08 <bijan> <sigh/> missing "#"

22:47:46 <DanC> ah. good. I got the validation service to show me some evidence of datatype support.

22:47:46 <bijan> Well, worked for one, but noth te other

22:48:14 <bijan> I'm getting it validating me without the datatype, afaict

22:48:28 <bijan> _:jARP84519 <http://www.w3.org/2002/07/owl#cardinality> "1" .

22:48:39 <bijan> From: <owl:cardinality rdf:datatype="&xsd;#nonNegativeInteger">1</owl:cardinality>

22:48:48 <DanC> yes, n-triples does ^^.

22:48:50 <bijan> With: <!ENTITY xsd "http://www.w3.org/2001/XMLSchema">

22:48:55 <DanC> I just got the validator to spit out "1"^^http://example/blah#nonNegativeInteger

22:49:03 <bijan> Woohoo!

22:49:06 <bijan> How?

22:49:17 * bijan musta screwed somethingup

22:49:19 <DanC> gave it this input: [[

22:49:21 <DanC> <?xml version="1.0"?>

22:49:21 <DanC> <rdf:RDF xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#"

22:49:21 <DanC> xmlns:dc="http://purl.org/dc/elements/1.1/"

22:49:21 <DanC> xmlns:owl="http://example/owl#">

22:49:21 <DanC> <rdf:Description>

22:49:22 <DanC> <owl:cardinality rdf:datatype="http://example/blah#nonNegativeInteger">1</owl:cardinality>

22:49:24 <DanC> </rdf:Description>

22:49:24 <bijan> Though silently dropign stuff sucks

22:49:26 <DanC> </rdf:RDF>

22:49:28 <DanC> ]]

22:50:38 <bijan> Wow. I can't get that to work in my example.

22:54:20 <bijan> Ok, got your example to work

22:55:04 <bijan> D'OH!!!

22:55:11 <bijan> The NTriples output is what's broken!

22:56:07 <bijan> Hmm. ANd the web bug report form silently truncants the input :)

22:56:36 <bijan> Grr. ANd now I have to retype your example.

22:56:44 * bijan notes that this isn't as easy as one might like