Semantic Web Interest Group IRC Chat Logs for 2003-08-07

07:02:04 <GNUPredator> GNUPredator is now known as GNUPredator-work

08:37:19 <arnarl> hi

10:37:40 <sbp> sbp is now known as d8-}

10:37:44 <d8-}> d8-} is now known as sbp

10:50:18 <Mutiny> Lets throw 680mb of data at rdflib and see how it goes..

10:51:47 <sbp> oh my goodness

10:53:16 <danbri> what's the dataset?

10:54:51 <Mutiny> musicbrainz's data dump

10:55:14 <swh> Mutiny: is that RDF, or did you convert it?

10:55:24 <Mutiny> it's rdf

10:55:44 <swh> neat... I think I'l have that

10:56:58 <swh> doesnt seem to be there anymore

10:57:51 <Mutiny> its at ftp://ftp.musicbrainz.org/pub/musicbrainz/data

10:58:30 <swh> thanks

10:58:59 <Mutiny> danbri: i don't suppose you have any pointers on querying a big ole dataset in rdf/xml like this, do you? ;)

11:00:15 <swh> Mutiny: how many triples?

11:01:56 <Mutiny> i have no idea

11:02:06 <Mutiny> i'm just going to say.. lots.

11:02:08 <swh> rdfdump | wc -l ?

11:02:21 <danbri> mutiny, only the swad-e reports i tihnk already mentioned, linked from http://www.w3.org/2001/sw/Europe/reports/intro.html

11:02:23 <swh> if its <10million it shouldn't be a problem

11:02:46 <swh> 100million or more is uncharted teritory AFAIK

11:02:56 * danbri would like to try 3store with results of a FOAF harvest

11:03:15 <swh> danbri: thats pretty small isnlt it?

11:03:27 <danbri> per http://planb.nicecupoftea.org/archives/000050.html there's quote a lot out there, but way < 10 million

11:03:42 <Mutiny> if line number is any indication, it's ~11m

11:03:43 <danbri> however Jim Ley's sql db is creaking, so foafnaut.org stuff needs a new backend

11:03:52 <danbri> -> http://swordfish.rdfweb.org/discovery/2001/08/codepict/scutterplan.rdf

11:04:01 <swh> Mutiny: roughtly halve that as its rdf/xml

11:04:05 <danbri> 2658 rdf files

11:04:20 <danbri> swh, we're not just loading them, but doing data-merge on uniquely identifying properties

11:04:35 <danbri> is that easy in 3store? or need app-specific code?

11:04:44 <swh> danbri: 3store doesnt really like lots of small files (it will take a while to load) but should be fine

11:04:46 <Mutiny> hmm. <rdfs:seeAlso rdf:resource='file:scutterplan.rdf'/>

11:05:01 <swh> danbri: are the merges queries? or do you use some funky logic?

11:05:14 <danbri> swh, see http://rdfweb.org/mt/foaflog/archives/000039.html for overview

11:05:43 <danbri> basically, we need to merge nodes (either in advance or when answering queries) based on owl:InverseFunctionalProperty

11:05:57 <danbri> (so typically this is assuming all schemas loaded up ok)

11:06:17 <Mutiny> danbri: are these just foaf files or whatever rdf files you could find?

11:06:26 <danbri> mostly foaf, but a mix...

11:06:33 <swh> danbri: ahh, we dont support that post-load - we plan to support owl:sameAs, but it wont be mega fast

11:07:11 <danbri> yup, speed's the thing... I reckon Jim's db would be ok except we're trying to defrag the data, since typically a person or image etc is mentioned in multiple files as a bnode

11:07:19 <danbri> also mattb's note at http://www.hackdiary.com/archives/000021.html

11:07:43 <swh> we have a paper on the generic problem, referential integrity: http://eprints.ecs.soton.ac.uk/archive/00006649/

11:08:51 <swh> danbri: hmmm... I dont see how you could garbage collect all those bnodes post load

11:08:55 <swh> there kind there

11:09:33 <swh> were planning to gather all the transitive etc. entailments for owl:sameAs and then backchain from uris at query time

11:09:40 <swh> gonna be expensive, so it will be optional

11:10:11 <swh> we have a massive problem with duplicate individuals in areUK computer science data

11:10:18 <swh> our

11:12:02 <danbri> dups: exactly

11:13:39 <swh> ...actaully, thinkin about it the backchaining is only expensive when the uri is a variable, for known uris tis quite cheap

11:18:42 <swh> horay, 3store now passes all the positive entailment tests - that I plan for it to

11:19:54 <danbri> congrats!

11:20:06 <swh> thanks - now for the negatives :/

11:22:19 <dajobe> " that I plan for it to"

11:22:41 <swh> dajobe: yes, 3store doesnt understand floats - so it fails that one

11:22:54 <dajobe> that's datatypes, who cares?

11:23:06 <swh> and I chose to fail the one that disagrees wit hthe spec and the literal bnode stuff

11:23:28 <dajobe> the later is fine as I recall

11:23:28 <swh> dajobe: you guys aparently ;)

11:23:30 <dajobe> you are wrong

11:23:45 <swh> dajobe: about?

11:26:22 <swh> oh dear, looks like I fail loads of the neg enatilment tests :(

11:26:24 <dajobe> "the literal bnode stuff"

11:26:42 <dajobe> I don't even know what "one that disagrees wit hthe spe" is

11:26:49 <swh> what am I wrong about it?

11:26:56 <dajobe> the test is correct

11:26:58 <dajobe> and you shoul dpass it

11:27:46 <swh> rdfms-seq-representation-Manifest-rdf-test002

11:27:51 <swh> ('scuse mangled uri)

11:28:19 <swh> the spec says you only have to enatil <rdf:_1> <rdf:type> <whatever it is>

11:28:28 <swh> if <rdf:_1> is in the document, and its not

11:29:06 <swh> c.f. http://lists.w3.org/Archives/Public/www-rdf-comments/2003JulSep/0076.html

11:35:49 <swh> lunchtime

11:35:54 <swh> swh is now known as swh_lunch

12:45:35 <eric> eric is now known as ericP

12:50:55 <DanC-AIM> hi from Montreal.

12:53:03 <dajobe> hi DanC-AIM

12:53:32 <dajobe> I just noticed that knowledge technologies had (re-?)joined Extreme Markup this year

13:08:33 <DanC-AIM> Really? KT was a spring thing.

13:09:16 <dajobe> the home page says so

13:10:20 <swh_lunch> swh_lunch is now known as swh

13:16:31 <DanC-AIM> Anybody heard from EricM this week? Anybody know if brian has what he needs for this week's rdfcore telcon?

13:16:33 <DanC-AIM> -- Dian Kennedy annouces idealliance reorg

13:16:36 <DanC-AIM> Sponsoring open source work

13:17:25 <MikeM> I was just looking for him myself...

13:17:57 <DanC-AIM> Offering remote collaboration infrastructure so the conversation can continue between ftf events

13:18:44 <DanC-AIM> -- Ciancarini et. Al.: "on the integration of RDF and Topic Maps"

13:19:02 <DanC-AIM> University of Bologna

13:20:48 <DanC-AIM> (speaker apologises for her english, which is actually more clear than many native speakers I know)

13:21:51 <DanC-AIM> Example of RDF/XML and XTM syntax for <p> author "john smith"

13:22:17 <DanC-AIM> Rdf syntax: 3 lines. Xtm: 20 or so

13:22:46 <DanC-AIM> "rdf and xtm have been accepted by different communities"

13:22:48 <MikeM> I wish people wouldn't apologize for their English. Its always better than my ability to speak their native language....

13:23:27 * danbri apologies for his english

13:23:39 <danbri> no, haven't seen ericm, he's on the road re some conference?

13:23:46 <DanC-AIM> Quite.

13:24:21 <DanC-AIM> He's in California somewhere, yeah.

13:25:21 <DanC-AIM> Spkr cites Moore, Lacher/Decker, [one other. Missed it]

13:25:42 <DanC-AIM> Somebody wanna chump this paper for me, please?

13:26:09 <dajobe> name?

13:26:15 <DanC-AIM> /me considers trying to present the ESW collaboration facilities

13:26:16 <dajobe> Ciancarini

13:26:22 <MikeM> chump?

13:26:31 <dajobe>http://www.mulberrytech.com/Extreme/Proceedings/html/2003/Presutti01/EML2003Presutti01.html

13:26:32 <dc_rdfig> A: http://www.mulberrytech.com/Extreme/Proceedings/html/2003/Presutti01/EML2003Presutti01.html from dajobe

13:26:37 <dajobe> A:|Metadata on the Web: on the integration of RDF and Topic Maps

13:26:37 <dc_rdfig> Titled item A.

13:26:49 <dajobe> A:Ciancarini et al. at Extreme Markup Languages 2003

13:26:49 <dc_rdfig> Added comment A1.

13:27:07 <dajobe> MikeM: just what I did

13:27:11 <DanC-AIM> Logger_1, pointer?

13:27:12 <DanC-AIM> See http://ilrt.org/discovery/chatlogs/rdfig/2003-08-07#T13-27-11

13:28:06 <DanC-AIM> Oh crud... I can't make a pipe... Somebody link that pointer from A: please?

13:28:23 <MikeM> figured it was something like that...

13:28:47 <dajobe> A:see [http://ilrt.org/discovery/chatlogs/rdfig/2003-08-07#T13-27-11|chat]

13:28:47 <dc_rdfig> Added comment A2.

13:29:02 <DanC-AIM> Thx, dajobe

13:29:21 <MikeM> what kind of device are you using

13:29:57 <DanC-AIM> This presentation is really good. I wish conference proceedings included slides

13:31:00 <DanC-AIM> Sidekick. Cf http://dm93.org/z2001/WearableGizmo

13:32:10 <DanC-AIM> There's *just* enough gprs bandwidth in the room to make this work. Comes and goes, but the software deals gracefully.

13:33:14 <MikeM> I'm trying to find a real keyboard for my Nokia 3650 so I can do the same...

13:34:47 <MikeM> I thought I'd never use it when I bought it, but that camera in the phone thing is actually quite useful...

13:35:05 <DanC-AIM> Ooh... Using isDefinedBy to model tm:subjectIndicatorRef

13:35:06 <DanC-AIM> A:see also: [http://esw.w3.org/topic/TopicMaps esw:TopicMaps]

13:35:07 <dc_rdfig> Added comment A3.

13:37:13 <DanC-AIM> The camera on this thing is almost completely useless.

13:39:42 * darobin will know who to ask next time he forgets his charger

13:41:33 <DanC-AIM> Mapping of tm names to rdf is slick.... Wanna look at it with N3 glasses... Bet it would look even slicker.

13:41:43 <sandro> Oooooh. That's a good use for my new beloved digital camera -- to take pictures of people's slides. Does that offend people?

13:42:28 <DanC-AIM> Wow! A gui editor that groks rdf and xtm schemas

13:42:52 <MikeM> I've noticed Japanese guys doing it for years....

13:43:46 <DanC-AIM> Q+ to ask: is the editor OpenSource?

13:43:57 <dajobe> I took pictures of the slides saying where they were on the web, before

13:50:27 <DanC-AIM> Q - non-binary relationships

13:50:59 <DanC-AIM> A - used an idea from related work... Express associations ala Ogievetsky 2001

13:51:05 <DanC-AIM> Q - where can we get it to play with it?

13:51:54 <DanC-AIM> A - the software sorta works... We're dealing with java versions... We're considering a web interface

13:52:17 <DanC-AIM> Q - how do you configure the converter?

13:52:28 <DanC-AIM> A - not a gui. Xml file.

13:53:20 <DanC-AIM> (applause again)

13:56:31 <DanC-AIM> -- Mason "fine-grain pub mgmt. Under Topic Map control" (with Le)

14:06:02 <DanC-AIM> Danbri, what do you think of my trying to get 5 minutes in front of the audience to invite this crowd to play in the esw wiki, #rdfig, and say a few words about ScheduledTopicChats?

14:06:37 <libby> +1 DanC-AIM

14:06:41 <danbri> sounds fun, though I'm off out shortly, if you were thinking of having people show up here now

14:08:23 <DanC-AIM> I'm not targetting a real-time demo. Not enough bandwidth here.

14:59:02 <DanC_jam> 'sup, AaronSw?

15:00:15 <AaronSw> oops. hi there

15:00:24 <AaronSw> still at Extreme Markup or whatever?

15:05:04 <libby>http://esw.w3.org/topic/ImageDescription

15:05:05 <dc_rdfig> B: http://esw.w3.org/topic/ImageDescription from libby

15:05:30 <libby> B:ImageDescription wiki page: two meetings times proposed

15:05:30 <dc_rdfig> Added comment B1.

15:05:43 <libby> B|:ImageDescription wiki page: two meetings times proposed

15:05:48 <libby> B:|ImageDescription wiki page: two meetings times proposed

15:05:49 <dc_rdfig> Titled item B.

15:06:32 <libby> B1:[http://www.timeanddate.com/worldclock/fixedtime.html?day=13&month=8&year=2003&hour=13&min=0&sec=0&p1=0|1300UTC, 13th Aug 2003 on EXIF+RDF]

15:06:32 <dc_rdfig> Replaced comment B1.

15:07:13 <DanC_jam> I'm in Montreal, but I've stepped out of the ExML conference to get some bandwidth to prepare for a telcon.

15:07:25 <libby> B:[http://www.timeanddate.com/worldclock/fixedtime.html?day=27&month=8&year=2003&hour=14&min=30&sec=0&p1=0|1430UTC, 27th Aug 2003 on SVG and image description]

15:07:26 <dc_rdfig> Added comment B2.

15:08:05 * DanC_jam wonders about the use of wiki pages for negotiating schedules. you can't change what people have agreed to by editing the page

15:08:24 <libby> B:see [http://ilrt.org/discovery/chatlogs/foaf/2003-08-06.html#T14-12-19|extensive scheduling discussions in #foaf yesterday]

15:08:25 <dc_rdfig> Added comment B3.

15:08:35 <libby> true

15:08:44 <libby> I'm not sure of a better way...

15:08:51 <DanC_jam> I guess the wiki has a change history, so you could tell if Fred said "I'll be there at 9am" and somebody else scribbled that out and wrote "Fred will be there at 10am"

15:09:30 <libby> bijan: all welcome.

15:09:32 <libby> oops

15:09:37 <libby> B:all welcome.

15:09:37 <dc_rdfig> Added comment B4.

15:09:39 <bijan> I knew that :)

15:09:43 <libby> :)

15:09:52 <libby> heh

15:09:57 <AaronSw> for that kind of thing, something like a QuickTopic discussion might be better

15:10:08 <DanC_jam> QuickTopic?

15:10:42 <AaronSw> it's a site that lets you set up free mini-threads, date-ordered lists of name, comment pairs

15:10:52 <AaronSw> accessible thru web and email

15:11:25 <libby> "for that kind of thing" = scheduling?

15:11:30 <karlcow> the wiki is fine for meeting date most of the time. I have used it often and it has worked quite well.

15:11:48 <karlcow> The CraoWiki community is managing that entirely like that

15:11:51 <AaronSw> well, for making sure people can't edit each other's stuff specifically

15:11:55 <karlcow> with a kind of dated space

15:12:34 <karlcow> For example, we had a weblogger meeting yesterday with 40 people - http://wiki.crao.net/index.php/MousseNavire/2003-08-06

15:14:25 <libby> karlcow, are you Karl Dubost?

15:17:01 <karlcow>http://semanticweb.forpoets.org/archives/001421.htm

15:17:01 <dc_rdfig> C: http://semanticweb.forpoets.org/archives/001421.htm from karlcow

15:17:15 <karlcow> C:|FOAF, Flocking, and the Semantics of Starlings

15:17:16 <dc_rdfig> Titled item C.

15:19:51 <libby> karlcow, if you are Karl Dubost, I hope you can make the meeting described in http://rdfig.xmlhack.com/2003/08/07/2003-08-07.html#1060268704.312959, since you were talking about that stuff on www-rdf-interest...if not, maybe we should reschedule

15:20:48 <karlcow> I will be able.

15:21:00 <karlcow> libby: Danbri asked me already and I said yes

15:21:04 <karlcow> :)

15:21:42 <libby> phew

15:21:43 <libby> :)

15:21:58 <karlcow> ities and misunderstandings. """

15:22:16 <libby> danbri is out and I wasn;t sure

15:22:18 <karlcow> C: surprising.

15:22:18 <dc_rdfig> Added comment C1.

15:25:58 <DanC-AIM> byebye for now

15:26:09 <DanC-AIM> byebye

16:08:05 <dajobe> FYI 9141172 in that musicbrainz dump

16:08:14 <dajobe> plus illegal XML -- inside comments, that I had to strip out

16:08:26 <dajobe> so will be ok for swh, <10M

16:10:14 <swh> dajobe: im planning to add it to out main kb this evening (it will take a while, so I dont want to do it while people are using it)

16:10:30 <swh> ive never tested >10M before, should be interesting

16:10:58 <swh> rdf test case results online: http://triplestore.aktors.org/rdf-test/

16:11:01 <dajobe> wee, that was fast parsing

16:11:07 <swh> dajobe can disagree with my comments now :)

16:11:22 <dajobe> 66240 triples/second parsing

16:11:29 <swh> dajobe: pretty good

16:11:42 <dajobe> I/O bound I expect

16:11:54 <swh> what were you doing with the triples?

16:12:02 <dajobe> counting them

16:12:05 <swh> ah

16:12:09 <dajobe> rapper --count --ignore-errors mbdump.rdf

16:12:09 <swh> still pretty fast

16:12:24 <dajobe> ... after stripping out the ---- crummy XML comments

16:12:41 <swh> will I have to manually remove them?

16:12:51 <dajobe> bunzip2 < mbdump.rdf.bz2 |sed -e 's/----*//' > mbdump.rdf

16:13:03 <swh> heh, i just ran vi on it ;)

16:13:16 <dajobe> that summary seems fine

16:13:20 <dajobe> given you don't do datatypes etc.

16:13:53 <swh> yes. we kinda do integers, but absolutly no float, and we discard the type information after parsing

16:15:30 <dajobe> I should mention those numbers; I want to kill the rdf parser 'bloat' fiction

16:17:15 <swh> ive found raptor to be pretty efficient

17:57:13 <cam_olin> does anybody here have experience with DanC's iCal => RDF .pl script?

18:01:44 * DanC_jam waves, but with no time to chat

18:02:58 <DanC-AIM> Off to pepper's talk...

18:07:58 <DanC-AIM> Phtphtpht. No steve.

18:12:15 <sandro> Why not?

18:38:55 <DanC-AIM> He didn't come to the conference at all.

18:40:22 <DanC-AIM> Aargh! I had all the context for calling that owl telcon swapped in; then I rushed out of the room to see pepper's talk. Lose. Now my laptop won't wake up.

18:50:06 <DanC-AIM> (missed most of a talk on coroutines)

18:50:57 <DanC-AIM> -- Quin reviews xquery implementation status

18:58:01 * GNUPredator listens

19:07:05 <DanC-AIM> Liam demos sgv output from xml query... Draggable thumbnails... Animated links

19:09:16 <DanC-AIM> All zooming around in mozilla on linux

19:12:18 <DanC-AIM> Q - what's novel in xquery w.r.t xslt?

19:13:50 <DanC-AIM> A - [disclaimer... Personal answer] xquery has separately compilable modules

19:13:52 <DanC-AIM> ... Difference of approach in emphasis. Xquery works on collections of documents

19:15:09 <GNUPredator> DanC-AIM: URI? where can I get this?

19:15:23 <DanC-AIM> -- Walsh "RDF Twig: accessing RDF Graphs in XSLT"

19:15:48 <DanC-AIM> Get what? Liam's stuff? Dunno. You might mail liam@w3.org

19:16:26 <GNUPredator> ok

19:16:29 <GNUPredator> i will

19:18:25 * GNUPredator wonders if he is being impusive

19:18:54 <GNUPredator> hey DanC-AIM , are you at a presentation?

19:21:56 <dajobe> he's at extreme markup, follow the links yourself from http://www.extrememarkup.com/extreme/

19:25:32 <DanC-AIM> Thanks, amy. I reached timbl

19:26:06 <amy> great

19:26:20 <amy> bye

19:26:36 <DanC-AIM> (I'm bummed that my laptop won't wake up. I really don't want to reboot it and lose all that context)

19:28:45 <swh> started assering the musicbrainz rdf 2 hours ago, still going :/

19:28:58 <swh> speed is down to 700 triples/sec :(

19:29:10 <swh> at 13M triples total

19:29:20 <dajobe> eek

19:29:25 <dajobe> you've got problems

19:29:35 <dajobe> 7M to go?

19:29:36 <swh> yeah, I knew the import code is really inefficient

19:29:58 <swh> no, I only had 7M when I started - so about 4M to go

19:30:01 <DanC-AIM> Walsh: hard to say not in rdf. Rdftwig has filterNot, which makes cwa, cuz I needed it.

19:30:09 <timbl_> This is into memory or disk?

19:30:19 <swh> dajobe: nearly done a draft impl. report, can I mail you a copy when its ok?

19:30:23 <dajobe> sure

19:30:23 <swh> timbl_: disk

19:30:32 <dajobe> MySQL underneath

19:30:44 <dajobe> 7M: ah, I thought you said 10

19:30:55 <swh> I'm doing the INSERTs one at a time which is patelogical and done do any caching

19:31:03 <dajobe> Doh!

19:31:07 <swh> 10 was peak, but someone removed some stuff

19:31:16 <DanC-AIM> Walsh: rdftwig supports rdql (I.e. Has hook to jen's rdql stuff)

19:31:23 <swh> dajobe: its fine if you only have a few M triples :)

19:32:06 <dajobe>http://www.mulberrytech.com/Extreme/Proceedings/html/2003/Walsh01/EML2003Walsh01.html

19:32:07 <dc_rdfig> D: http://www.mulberrytech.com/Extreme/Proceedings/html/2003/Walsh01/EML2003Walsh01.html from dajobe

19:32:20 <dajobe> D:|RDF Twig: accessing RDF graphs in XSLT, Norman Walsh (at Extreme Markup 2003)

19:32:20 <dc_rdfig> Titled item D.

19:32:31 <dajobe> logger_1, bookmark

19:32:31 <dajobe> See http://ilrt.org/discovery/chatlogs/rdfig/2003-08-07#T19-32-31

19:32:41 <dajobe> D:see [http://ilrt.org/discovery/chatlogs/rdfig/2003-08-07#T19-32-31|chat]

19:32:41 <dc_rdfig> Added comment D1.

19:32:56 <dajobe> hmm, still broken images, heyho

19:33:40 <DanC-AIM> /me bites bullet, removes laptop battery to reboot

19:33:55 <Morbus> now, that's odd.

19:33:57 <Morbus> i saw your /me.

19:34:14 <Morbus> ah. AIM gateway, perhaps.

19:34:17 <dajobe> yeah

19:34:36 <Morbus> i'm trying the trillian beta. got paranoid ;)

20:17:45 <swh> dajobe: sent oyu a draft, the imports still going :)

20:17:50 <swh> swh is now known as swh_away

22:15:17 <GNUPredator> GNUPredator is now known as GnuMonkey

23:22:22 <GnuMonkey> GnuMonkey is now known as md-ZZZz

23:44:25 * danbri can't see any of the images in http://www.mulberrytech.com/Extreme/Proceedings/html/2003/Walsh01/EML2003Walsh01.html

23:44:43 <danbri> 404s. Same w/ one of the other talks I tried to look at :(