Semantic Web Interest Group IRC Chat Logs for 2002-02-27

00:02:05 <sbp> it's important that these myths be crushed at every turn

00:02:36 <jhendler> hmm, the business week article doesn't seem to be showing up in the "W3C in the Press" page

00:03:11 <jhendler> yes, but at the same time, a certain amount of hype doesn't hurt when one is actually trying to change the world...at least they spelled "semantic" right...

00:03:43 <sbp> on the contrary, I believe that hype when trying to change the world is the single biggest obstacle to overcome. The code should speak for itself

00:04:13 <sbp> hype can be damaging in that it leads people to misunderstand the basic principles and philosophies behind the new technologies

00:04:15 <jhendler> hmm, but if no one looks for/at it, it doesn't go far - a certain "coevolution" helps

00:05:03 <jhendler> yes, over hype is bad, but as someone pointed out to me yesterday when I was complaining about some stuff in the article, the people who matter will read it and figure out who to talk to for the truth.

00:05:35 <sbp> agreed. Most intelligent people can tell good journalism from mindless hype

00:07:59 <sbp> but the general users matter too :-)

00:08:10 <jhendler> hmm, but we are talking Business Week :->

00:08:23 <jhendler> where will they find intelligent readers?

00:08:31 <sbp> lol

00:09:10 * jhendler gotta go, g'night.

00:11:13 <sbp> 'night

00:42:58 <acalbazan> in using rdf, does a resource have to be a flat .htm file or can it be a targer that requires parameters? e.g. - http://www.asdf.com/?resource=1?

00:44:17 <jonb> a resource is anything identified by a URIref

00:45:19 <jonb> RDF is not limited to making assertions about HTML files

00:46:26 <jonb> nite all

02:16:10 * danbri thanks mnott for the US Govt data links; adds to http://rdfweb.org/people/danbri/2001/09/foafcorp/intro.html for future exploration

02:18:05 <mnot> no worries

02:20:03 <danbri> btw, Josh (from theyrule.net) gave me a data dump of the companies data. We have it in RDF now. Soon as I get time will publish... meantimes, a taster...

02:20:05 <danbri> SELECT ?corp1, ?corp2, ?person WHERE (fc::name ?cp1 ?corp1) (fc::name ?cp2 ?corp2) (fc::board ?cp1 ?board1) (fc::board ?cp2 ?board2) (fc::member ?board1 ?member) (fc::member ?board2 ?member) (fc::name ?member ?person) USING fc for http://xmlns.com/foaf/corp#

02:20:37 <danbri> danbri@fireball:~/public_html/2002/02/java/rewriter > java -cp jars/rewrite-helpers.jar Squish2SQL sql/test3.squish

02:20:42 <danbri> SELECT DISTINCT b2.value AS corp2, b3.value AS corp1, b4.value AS person from triples a1, triples a2, triples a3, triples a4, triples a5, triples a6, triples a7, resources b2, resources b3, resources b4 WHERE a1.predicate = '-1296757095' AND a2.predicate = '-1296757095' AND a3.predicate = '2098112321' AND a4.predicate = '2098112321' AND a5.predicate = '-425569083' AND a6.predicate = '-425569083' AND a7.predicate = '-12967

02:20:44 <danbri> AND a1.subject=a3.subject AND a2.subject=a4.subject AND a3.object=a5.subject AND a4.object=a6.subject AND a5.object=a6.object AND a6.object=a7.subject AND b2.key=a2.object AND b3.key=a1.object AND b4.key=a7.object ;

02:21:30 <danbri> ...this works quite nicely for querying PostgreSQL, using some stuff Libby's been working on.

02:22:12 * danbri rewriting some of this in Ruby, making a little progress towards a Squish / rdf query test suite as a side effect

02:25:16 <AaronSw> hey danbri

02:38:54 <mnot> Hmm, perhaps you could be an expert witness for the Enron case, and/or identify other companies with similar practices ;)

03:12:33 <AaronSw> DanCon, mucho thanks for the debating skillz

03:12:37 <AaronSw> they have served me well

03:21:35 <sbp> served you well: no, they haven't

03:21:47 <AaronSw> in what sense?

03:22:20 <AaronSw> They have been the sweet taste on a serving of ill, I suppose.

03:22:26 <sbp> lol

03:22:26 * sbp won't even bother :-)

05:15:47 * sbp updates http://infomesh.net/2001/cwm/

05:23:39 <mnot> You can throw on http://www.mnot.net/sw/cwm_xpath if you want; any feedback appreciated

05:27:14 <sbp> wow... interesting

05:27:19 <sbp> care to chump it?

05:27:36 <mnot> sure

05:27:43 <mnot>http://www.mnot.net/sw/cwm_xpath/

05:27:43 <dc_rdfig> A: http://www.mnot.net/sw/cwm_xpath/ from mnot

05:27:55 <mnot> A:|cwm built-in for XPath

05:27:55 <dc_rdfig> titled item A

05:28:28 <mnot> A: right now, it specialises in HTML scraping using XPath expessions (was bespoke coded for DanCon ;).

05:28:28 <dc_rdfig> added comment A1

05:28:30 <AaronSw> """\ is a bit weird

05:28:47 <mnot> A: requires PyXML (recent; after the 4Suite fold-in)

05:28:47 <dc_rdfig> added comment A2

05:28:59 <mnot> A: comments, suggestions appreciated.

05:29:01 <dc_rdfig> added comment A3

05:29:40 <mnot> AaronSw: that gives you multi-line comments without the leading double-newline (it may no longer be necessary; it's an ingrained habit from python1 days)

05:30:06 <AaronSw> i figured as much, but for the very first one you have a double newline too

05:30:35 <mnot> ah, so I do. like I said... ;)

05:31:05 * sbp adds a note about it to /2001/cwm/

05:31:23 <sbp> comments: mail it to TimBL :-)

05:31:42 <sbp> really, we need some way for CWM to be able to interactively add built-ins modules

05:31:54 <mnot> yes, that would be nice

05:32:18 <sbp> of course, it's not too much hassle to modify the source code... but then you get CVS merging problems etc.

05:32:25 <mnot> I should have some time to work on cwm_uri soon

05:33:02 * mnot wishes the APIs for built-ins were better documented

05:33:32 <sbp> it's pretty straightforward, though

05:34:01 <sbp> [some documentation would still be nice, I agree]

05:34:02 <mnot> somewhat, yes.

05:34:20 <sbp> cwm_uri: there's no way that you can return a list with the current CWM

05:34:25 <sbp> it won't let you

05:34:48 <sbp> there's a "TODO" in the llyn.py source code to that effect

05:34:50 <mnot> that's what I was afraid of. I got excited when I saw cwm_string taking lists, but hadn't had a chance to try it the other way. *sigh*

05:35:21 <sbp> ugh, I should add this info to the CWM guide again... so much to add!

05:35:53 <mnot> well, I may have more time to try to get my head around cwm/llyn (not sure if mnot has enough lifetime left for that, tho ;)

05:37:05 <sbp> heh, heh

05:41:54 <sbp> oh! my mistake: Tim seems to have implemented it

05:43:39 <mnot> really?

05:44:00 * mnot dances the dance of someone with something to do

05:44:14 <mnot> can you point me at something?

05:47:24 <sbp> at what?

05:48:30 <mnot> are there any tests for it, etc?

05:48:46 <sbp> not yet. I'm just making one right now...

05:51:53 <redmonk> redmonk is now known as rm

05:53:24 <sbp> pff, it's totally borked

05:53:31 <sbp> NameError: global name 'store' is not defined

05:53:54 <mnot> urgh

05:54:17 <rm> rm is now known as rm_lrkr

05:54:22 <sbp> ah, I think I can fix it

05:55:48 <sbp> Hmm... it's just sitting there

05:57:19 <sbp> nope, it doesn't like it when I "fix" it :-)

05:57:35 <mnot> heh

05:57:46 <mnot> neither does my cat

05:58:56 <mnot> ah, well. It may be doable without returning a list, I suppose; just (scheme, authority, path, query, fragment) may be good enough

05:59:18 <sbp> Hmm... I can get it to return an empty list. That's no good, of course...

05:59:20 <mnot> (with some more complexity behind authority and query)

06:00:34 <mnot> is it possible to return complex data structures (dictionaries-of-dictionaries) and have The Right Thing happen? Or does one have to go through and make individual statements in the model?

06:02:34 <sbp> in builtins? you can only returns strings, literals, or bNodes

06:03:21 <mnot> that's what I was afraid of.

06:04:19 <mnot> so you'd have to construct the bNodes, link them together, insert the statements, and then return the root of the tree?

06:05:26 <sbp> you'd have to do a lot of llyn.py editing :-)

06:05:47 <mnot> ah. That's what I was *really* afraid of ;)

06:05:56 <mnot> thanks.

06:06:25 <sbp> it's not so bad...

06:06:28 <sbp> I managed to get:-

06:06:29 <sbp> "(http://example.org/ blargh)" a <#Result> .

06:06:31 <sbp> from:-

06:06:38 <sbp> { ("http://example.org/#blargh" "#") string:split :x } log:implies

06:06:38 <sbp> { :x a :Result } .

06:07:50 <mnot> interesting...

06:08:10 <mnot> so perhaps a uriSplit that just produces a list...

06:09:26 <sbp> well, it'd be better to have a real DAML list returned...

06:11:14 <mnot> oops, I didn't see the quotes there... yes, it would (I thought you'd gotten it working)

06:12:31 <sbp> nah

06:12:48 <sbp> the best I can do is:-

06:12:48 <sbp> ( )

06:12:48 <sbp> a <#Result> .

06:12:58 <mnot> very zen ;)

06:14:02 <sbp> heh, yeah...

06:33:10 <mnot> sbp: forgot about this one: http://www.mnot.net/sw/cwm_date/

06:33:27 <mnot> also for dancon, IIRc

06:33:47 <sbp> ooh! I could do with one of them, but just using the normal Python time module stuff

06:34:06 <mnot> what, working with the 9-tuples?

06:34:23 <sbp> pardon?

06:34:58 <sbp> I just want something that can give me an ISO formatted time, or a time using some format that I choose

06:35:09 <mnot> ah, strftime and friends

06:35:50 <sbp> yep

06:35:51 <mnot> that should be pretty easy.

06:36:32 <mnot> I'll rearrange that module so that it has an explicit name for the isodate stuff, and then does python time.* stuff as well

06:36:35 <sbp> e.g. "%Y%-m-%d %H:%M:%S" date:date ?x .

06:37:24 <mnot> wouldn't it have to be a list, like ("%Y", "1999"), date:date ?x . ?

06:37:39 <mnot> (much like string:scrape)

06:37:46 <sbp> yep

06:38:24 <sbp> excellent. Could you do a try/except clause on the isodate import, too?

06:38:36 <sbp> so that I don't have to get isodate just to run CWM... :-)

06:38:43 <mnot> fair enough.

06:38:52 <mnot> it's free to distribute

06:39:08 <mnot> hmm

06:39:24 <mnot> perhaps a separate cwm_time module that aligns with python's time module woudl be better

06:41:42 <sbp> yep, I agree

06:44:36 <mnot> hmm, what you want isn't actually part of the library, you really want strftime(gmtime()) or strftime(localtime()), because built-ins can't produce lists.

06:44:46 * mnot wonders what to call it, then

06:45:18 <mnot> gm_strftime and local_strftime? Are they too ugly?

06:45:35 <sbp> yes

06:45:50 <mnot> format_gmt and format_local?

06:46:11 <sbp> no underscores... ugh

06:46:25 <sbp> just formatGmt and formatLocal should be cool

06:46:26 <mnot> ah. formatGMT and formatLocal then

06:46:32 <mnot> fair enough

06:47:03 <sbp> (good idea to capitalize GMT)

06:48:13 <sbp> Gotta run

06:49:21 <mnot> OK, I'll play with that and see where it goes. Later

08:18:19 <libby> yes! eating our own dogfood :)

08:18:41 <dajobe> ugh

08:18:54 <dajobe> did he post his slides url?

08:18:55 <libby> (says Steve Bratt, W3C COO)

08:22:16 <mnot>http://www.mnot.net/sw/cwm_time/

08:22:16 <dc_rdfig> B: http://www.mnot.net/sw/cwm_time/ from mnot

08:22:29 <mnot> B:|time format parsing for CWM

08:22:29 <dc_rdfig> titled item B

08:23:22 <mnot> B:originally discussed as mirroring Python's time.* functions, but that didn't seem such a good idea on reflection; many are based on localtime, which introduces machine-specific context.

08:23:22 <dc_rdfig> added comment B1

08:24:10 <mnot> B:So this ends up being just versions of strptime and strftime.

08:24:10 <dc_rdfig> added comment B2

08:25:13 <libby> dajobe, no I dont think so yet

08:34:11 <mnot> hi graham!

08:36:18 * dajobe takes pics of plenary

08:36:40 <libby> hey graham

08:40:14 <gk> Hi Mark, Libby

08:41:02 <mnot> how's this tech plenary shaping up so far?

08:42:17 <gk> So far,mostly admin type stuff .. had presentation from the new COO, now the TAG session just started, but still talking about what it is rather than into the real meat.

08:43:08 <libby> dancs up now

08:44:11 * dajobe looks at blurry pic of libby typing

08:44:21 <gk> We just saw all the architectural poroblms of the web fly by in 2 seconds ...;-)

08:45:04 <mnot> yep, that's the TAG all right ;)

08:49:21 <gk> DanC's just attributed Jon Postel's "Be conservative..." motto to Brian Carpenter

08:50:04 <gk> (Brian C probably was the first to document it in an RFC, bit it's widely held to be Jon Postel who promoted it.)

08:52:50 <gk> DanC: local principle + large scale = new effects (or somethibng like that) Interesting thought., I think.

09:03:30 <libby> tag is answering a q about reusing other groups standards and interdependencies you get. they think they might look at this

09:04:27 <libby> ...they can maybe influence things

10:05:11 <libby> demos of SW technologies coming up :)

10:07:03 <libby> .http://www.w3.org/Talks/2002/02/27-tpsw/

10:07:04 <xena> Error 403 Sorry, Forbidden. We're sorry but the link you tried to access is forbidden. Please step through the following points to find and hopefully fix the problem: If you typed the URL by hand then please make sure that it is exactly as it should be. Also check the capitalization and that you are using forward slashes ( / ). If you are in doubt then you can always try to find what you are

10:07:05 <xena> looking for from our home page. If this does not help then you may want to try our search engine to look for specific keywords. If the link you have is very old or you picked it up from a book etc. then it may not be available anymore. In that case please mail the full URL to our Webmaster explaining where you got the link and what information you are looking for. Certain areas of our site are under

10:07:06 <xena> .. a very large amount of text.

10:07:18 <libby> is the slides but ot public yet

10:07:23 <libby> by the looks of it

10:07:43 <libby> s/ot/not

10:31:23 <libby> ooh, pretty rss view of W3C sitemap, will access contraol. v nice

10:32:54 <JibberJim> I'd be more impressed if they started coping with the huge numbers of 404s you get links off of w3.org.

10:40:46 <libby> looks like the 'jema' tool is cool too

10:40:52 <libby> unclear how compleet it is

11:09:46 * danbri admires the 'The Machine and the World' slide in http://www.w3.org/Talks/2002/02/27-tpsw/ (wonders whethers others can see it yet)

11:10:59 <JibberJim> I can see it, I could see it when Xena couldn't, I think Xena has the problem.

11:19:16 <libby> hm I had probs before but can see it now

13:14:19 * DanC starts noodling on a generic SQL->RDF query tool...

13:14:25 <DanC> .google python sql

13:14:25 <xena> python sql: http://www.jython.org/docs/javadoc/com/ziclix/python/sql/package-summary.html

13:15:13 <libby> what's it going to do danc?

13:15:34 <DanC> umm... take SQL queries (in HTTP GET requests) and return the results as RDF. I think.

13:15:49 <DanC> kinda like my IMAP query gizmo

13:15:55 <libby> generic sql queries?

13:16:22 <libby> so mapping of an sql db table strcture to rdf?

13:16:31 <DanC> yeah, just SQL syntax: select name, address from stuff where age>10.

13:16:35 <libby> didnt see the imap thin

13:16:39 <libby> g, sorry

13:16:56 <DanC> well, it's not like I did a big announcement of the IMAP thing

13:17:16 <libby> well I tryto look at stuff on here, but dont always manage it

13:17:53 <DanC> notes on imap thingy: http://www.w3.org/2000/04/maillog2rdf/email.html

13:19:11 * DanC tries to get python talking to mysql...

13:19:38 <libby> ta. looks nice

13:20:09 <libby> very nice.

13:41:41 <danbri> EricP claims to have gotten such a thing working (at the Hong Kong meeting); I've not tried it myself yet. A Perl thing that takes a mapping of RDBMS table / field names to RDF classes, properties and rewrites RDF queries accordingly.

13:42:01 <DanC> oh yeah...

13:42:41 <DanC> well, I don't want to tackle the query re-writing. I think that makes my idea easier.

13:42:53 <DanC> of course, if his code works, that's even easier.

13:43:05 * DanC doesn't really like reading perl, meanwhile...

13:44:22 <danbri> you just pass in the sql?

13:44:40 * danbri re-reads, realises 'yes'

13:44:51 <JibberJim> The Annotea database maps Algae to SQL to the RDF store.

13:47:02 <danbri> yup, that's mainly for a generic RDF table, I think DanC's looking at getting data from 'classic' SQL tables too

13:47:29 <libby> that would be very cool indeed

13:47:38 <JibberJim> Ah!

14:06:50 * MikeM just found out he's been volunteered to go to the next AC meeting....

14:25:00 * jonb reading logs ... wonders about converting SQL to RDF

14:25:17 <jonb> is that the same as converting XQuery results to RDF?

14:44:33 <acalbazan> are topic maps taking the place of rdf?

14:46:23 <DanC> dunno. why do you ask?

14:49:56 <acalbazan> Well.. I guess I am not sure if one provides something that the other does not.. Topic maps appear to do the same thing that rdf does - describe addressable resources... But topic maps are "newer" than rdf..

14:50:19 <DanC> BLURB: a quick-n-dirty SQL RDF python query service

14:50:19 <dc_rdfig> C: a quick-n-dirty SQL RDF python query service from DanC

14:50:23 <DanC> C:design notes...

14:50:23 <dc_rdfig> added comment C1

14:50:30 <libby> topc maps are about 10 years old no?

14:50:42 <libby> though xtm newer

14:50:56 <DanC> C:kinda like [imap query thingy|http://www.w3.org/2000/04/maillog2rdf/email.html]

14:50:56 <dc_rdfig> added comment C2

14:51:31 <DanC> are you building topicmap stuff, acalbazan? I'm still using RDF in stuff I build.

14:51:33 <acalbazan> ok.. i am looking at xtm...

14:53:08 <acalbazan> DanC: I am considering building with rdf..

14:53:38 <DanC> C:aha... [python distutils: home install scheme|http://www.python.org/sigs/distutils-sig/doc/inst/alt-install-prefix.html]

14:53:38 <dc_rdfig> added comment C3

14:54:11 <DanC> C:trying [MySQL for Python|http://sourceforge.net/projects/mysql-python]

14:54:11 <dc_rdfig> added comment C4

14:54:52 <DanC> what are you thinking of building, acalbazan?

14:56:38 <acalbazan> content management tools for my site...

14:59:03 <acalbazan> currently, we have all of our content items and content hierarchies (categories and topics) stored in data tables.. i find this somewhat rigid, especially since there is a need to serve content in several different formats..

14:59:18 <DanC> hmm... interesting.

14:59:20 * danbri wonders what XTM APIs and query languages look like

15:01:46 <JibberJim> Has anyone done any work on manipulating the SVG (with DOM) produced by the RDF validator as a way of editing/creating RDF like RDFAuthor but in a friendlier (read not JAVA!) manner?

15:02:35 <libby> theres an xtm query list, which has been a bit quiiet recently: it's on yahoogroups

15:02:51 <libby> I think the archives are closed but anyone can join

15:04:35 <danbri> the RDF validator's SVG is just the SVG output from GraphViz, available separately...

15:05:38 <JibberJim> Yes I realise, I was meaning more using SVG+DOM as an editor as opposed to using Cocoa, particularly if anyone has looked at taking the SVG and getting the triples back.

15:06:08 <danbri> Not from that particular SVG dialect, to my knowledge...

15:06:16 <JibberJim> they have from others?

15:07:05 <danbri> Good question... I think closest is the stuff Charles and others were doing re marking up SVG so RDF can be extracted, but I've not seen any specific work on extracting from a nodes'n'arcs diagram.

15:08:40 <JibberJim> I think a RDFAuthor-esque editor would be pretty easy to build on top of the SVG, and it would be much nicer than RDFAuthor as it would need no installation and just be reliant on having SVG.

15:13:27 <acalbazan> are there any rdf apis for java?

15:17:05 <libby> one or two

15:17:12 <libby> jena and stanford

15:17:15 <libby> probably others

15:18:34 <libby> stanford: http://www-db.stanford.edu/~melnik/rdf/api.html

15:19:12 <libby> jena: http://www.hpl.hp.com/semweb/ somewhere

15:39:20 <gk> Libby, I was just following up a couple of the responses to Dave Reynold's Q about provenance in RDFIG mailing list...

15:39:41 <gk> ... one response suggested a close relationship with someof the library work, but I don't quite see it yet.

15:40:16 <gk> ... (But it did lead me to some of the ABC work)

15:40:53 <libby> right, dan did a lot on that: we were talking about hisapproach yesterday briefly about reification

15:41:12 <gk> Here? ... is it in the logs?

15:41:14 <libby> danbri that is.

15:41:22 <gk> (sure)

15:41:45 <libby> yeah, but not a lot, just a pointer to a perl file that translate dbetween reification and reification as links

15:42:19 <libby> abc stuff has changed a lot since the beginning; I dont think danbri is involved much any more

15:42:59 <danbri> The first draft came out of discussiosn between DC and INDECS folks about rights metadata; we then worked on it a bit in the Harmony project (for which Libby is now ILRT's PI).

15:43:13 <libby> had a most interesing discussion with Martin Doer and a bloke called Nicola form IEEE standrad upper ontology last year

15:43:16 <libby> or rather, I listened

15:43:29 <danbri> I dropped out of any work on the-namespace-called-ABC, which Jane and Carl are still working on.

15:43:39 <libby> I'm concentrating on implemating examples of it where possible, rather than designing abc

15:44:08 <danbri> I do think it was a good name for a base vocabulary though...

15:44:43 <libby> that stuff is _so_ hard

15:44:57 <libby> brain-hurty hard

15:45:37 <gk> ... I found some bits of the discussion but not the link ... the program converts a linked RDF document into inline reifications of the statements it contains??

15:46:09 <libby> er possibly, or maybe the other way round! cant remember

15:46:29 <DanC> hmm... I asked timbl to explain his approach to defaults (and provenance, by the way) to webont. he hasn't gotten around to it. priority++

15:46:37 <gk> OK -- it's the presumed equivalence I was checking (under appropriate circumstances)

15:46:56 <libby> yeah. would be v useful I think

15:47:47 <gk> DanC, FYI, I'm about half way through doing a worked out MT for contexts without quotation. Using N3 (syntax) as a starting point.

15:47:53 <libby> danbri's harmony example: http://ilrt.org/discovery/2001/06/rdfperl/current/t/harmony.pl

15:49:20 * gk Hi DanBri, shame you can't be here.

15:49:54 <danbri> yeah, bummer... I'm working on a Ruby Squish parser as therapy :)

15:50:15 <libby> occupational therapy ;)

15:51:43 <gk> If possible, I'd be interested to chat with someone who understands what's been done by the library++ folks to understand if their work on provenance is in any way overlapping with what I'm contemplating.

15:52:14 <DanC> contexts: interesting, gk

15:53:00 <gk> Yup... one of those topics I never quite let go of, between more pressing things-to-do

15:53:56 <gk> Which, interestingly, all started when the CC/PP group were talking about how to deal with conflicting profile information from differewnt sources.

15:54:00 <DanC> my approach to it was quoting, as I guess you're aware.

15:55:18 <gk> Yes... I'm trying to see if I can make sense of the "other way" (i.e. use without assertion).

15:58:33 <jonb> get to work :-)

16:01:20 <jonb> i would be thrilled to see contexts in the MT

16:04:10 <gk> So would I :-)

16:04:42 <gk> (But not yet as a standard -- it's not yet well-enough understood by this community, methinks)

16:05:15 <jonb> nahh... contexts are traditional spaces

16:05:40 <jonb> goes waaay back to the late 1970s

16:11:08 <gk> Anticipating that... I added the "this community" just before I hit the return key... by which I meant the network protocol standardization community. Of course, there are some folks who do understand, but that understanding is not yet pervasive here.

16:11:36 <jonb> fair enough, but had to slip that in :-)

16:11:48 <gk> :-)

17:16:29 <bijan> Anyone handy have a clue about N3 list syntax?

17:21:09 <DanC> hi bijan.

17:21:23 <DanC> I've been accused of having n3 clues.

17:22:22 * DanC turns on timestamps...

17:25:37 <DanC> C:aha... [Python Database API v2.0|http://www.python.org/topics/database/DatabaseAPI-2.0.html]

17:25:37 <dc_rdfig> added comment C5

17:30:43 <bijan> Ok, is (:a [a :foo] :b) a three element list?

17:31:06 <sandro> What else might it be?

17:31:13 <bijan> Syntax error :)

17:31:16 <DanC> yes.

17:31:26 <DanC> yes, it's a 3elt list

17:31:27 <bijan> Yes it's a syntax error!!

17:31:30 <bijan> Oh, too bad.

17:31:43 <DanC> too bad? is rdfn3.g not clear?

17:31:59 <bijan> Oh, I wasn't looking at that. I should.

17:32:29 <bijan> I'm just trying to add list sugar to CWMClone and it's a little more involved than I initially hoped.

17:32:52 <DanC> see http://www.w3.org/2000/10/swap/rdfn3-gram.html

17:33:10 <bijan> right.

17:33:10 <sandro> "@prefix : <#>. :x :y (:a [a :foo] :b)."

17:33:10 <sandro> works for me in cwm. no syntax error.

17:33:21 <bijan> No, I know. I was just hoping.

17:38:04 <DanC> there's a test case or two in swap/test/list*

17:38:29 <bijan> Hmm. Any good performance tests not involving builtins, btw?

17:38:53 <bijan> Though if it's only a few builtins I could implement them.

17:39:19 <bijan> The compiler doesn't handle builtins yet, maybe by the end of the day (at least a prototype)

17:41:57 <sandro> Um - what do you mean by builtins?

17:42:20 <bijan> CWM builtins.

17:42:36 <bijan> string:, math:, a good chunk of log:

17:43:32 <sandro> Ah - the vocabulary terms the cwm reasoner looks for.

17:44:00 <bijan> Er...

17:44:02 <sandro> So performance testing "not using builtins" would be cwm non-reasoning performace -- parsing and pretty-printing?

17:44:15 <bijan> The things which are implemented as BI_ classes, mostly.

17:44:27 <bijan> er..no.

17:44:41 <bijan> It would be *reasoning* performance, not calling out to python code.

17:45:16 <bijan> this log:forAll :x, :y. {:x :parent :y} log:implies {:y :childOf :x}.

17:45:21 <sandro> Ah -- more like the math "builtins" in prolog. Stuff that COULD be done with rules, but (for performance reasons) is not.

17:45:21 <bijan> Doens't use "any" builtins.

17:45:36 <bijan> log:forAll and log:implies seem like special syntax.

17:45:39 <bijan> Those are implemented.

17:45:53 <bijan> ("http://example.org/#blargh" "#") string:split ?x .

17:46:00 <bijan> string:split is a built_in.

17:46:08 <bijan> yes, they are "primative"

17:46:23 <bijan> Actually, I did implement string:split last night.

17:46:30 <bijan> As it stands the listy version could be a rule.

17:46:49 <sandro> log:forAll and log:implies are just the vocabulary of the inference component, while string:stuff is the vocabulary of a string-manipulation component.

17:46:51 <bijan> (I.e., the one that has the subject adn object as lists)

17:47:08 <bijan> well, yes, sorta.

17:47:18 <sandro> the syntax is "special" only to the component who treats it as its native vocabulary.

17:47:30 <bijan> Maybe, maybe not.

17:47:56 <sandro> Meaning you don't want to argue the point?

17:48:06 <bijan> I'm actually not particularly interested in honing down the precise definition of "built-in", there seems to be a common usage :)

17:48:31 <bijan> (Well, not yet, no).

17:48:32 <sandro> It was more the issue of log:implies being "special syntax" I was arguing.

17:48:40 <sandro> but fine. :-)

17:49:09 <bijan> Well, I mean implementationally, more.

17:49:17 <bijan> You don't get reasoning without log:implies.

17:49:43 <bijan> And sure, it could be "loadable" module, but the inference engine has to be senstive to in a *very* specific way.

17:49:47 <sandro> "reasoning" is just one area of information processing.

17:50:11 <bijan> Whereas all or most of the other builtins are reasonably generic (wrt to the inference engine)

17:50:28 <sandro> No more so than a walking robot has to be sensitive to instructions in its knowledge base (RDF file) about where it can walk.

17:50:46 <bijan> Hmm. I said I"m not arguing this point. It doesn't really matter. From the internal point of view the distinction is reasonable COMMUNICATIVE.

17:51:03 <bijan> I.e., if I ask for performance tests with "no" builtins, Sean knows what I mean :)

17:51:14 <sandro> Sorry. I'll stop. Get back to coding. Sorry.

17:51:43 <AaronSw> Heh, sandro and bijan.

17:52:19 <bijan> Heh, Aaron.

17:52:35 * sandro is just kinda lost in big-picture stuff....

17:52:50 <AaronSw> sandro and bijan on irc, A-R-G-U-I-N-G...

17:52:52 <AaronSw> sorry, i'll stop ;)

17:53:00 <bijan> That would be good.

17:59:16 <sandro> (You need another sylable in "on irc", Aaron.)

17:59:41 <AaronSw> chat on irc, perhaps

17:59:57 <sandro> yeah. or type. or something. :-)

18:00:05 <sandro> or "sit"

18:02:00 <AaronSw> too bad C-O-D-I-N-G doesn't fit

18:02:44 <sandro> H-A-C-K-I-N-G

18:03:11 <AaronSw> ah, thanks

18:21:41 <bijan> DanC, http://www.w3.org/2000/10/swap/test/list/ is inaccessible to me (FYI). I do see a few other list tests in /test

18:25:16 <sbp> bijan: use CVS

18:25:36 <bijan> Ah, I prolly have it already then.

18:28:49 <DanC> C:aha! got a database connection from python.

18:28:49 <dc_rdfig> added comment C6

18:28:52 <DanC> logger, pointer?

18:29:01 <DanC> logger_1, pointer?

18:29:02 <DanC> See http://ilrt.org/discovery/chatlogs/rdfig/2002-02-27#T18-29-01

18:29:16 <DanC> C:see [notes|http://ilrt.org/discovery/chatlogs/rdfig/2002-02-27#T18-29-01]

18:29:16 <dc_rdfig> added comment C7

18:30:07 <danbri>http://lists.w3.org/Archives/Public/www-archive/2002Feb/0041.html

18:30:07 <dc_rdfig> D: http://lists.w3.org/Archives/Public/www-archive/2002Feb/0041.html from danbri

18:30:19 <danbri> D:|Jan and Dan offchannel chat on reification

18:30:19 <dc_rdfig> titled item D

18:30:59 <danbri> D:Chumped so I can find it again, and because I try out a new 'argument' on Jan. Not sure he was impressed though...

18:31:00 <dc_rdfig> added comment D1

18:33:19 <danbri> D:Short version: Q: can an rdf/xml document that draws on non-referring URI-strings ever be true?

18:33:20 <dc_rdfig> added comment D2

19:19:06 <AaronSw>http://www.linuxjournal.com//article.php?sid=5854

19:19:07 <dc_rdfig> E: http://www.linuxjournal.com//article.php?sid=5854 from AaronSw

19:19:19 <AaronSw> E:|Import This: The Tenth Internation Python Conference

19:19:19 <dc_rdfig> titled item E

19:19:24 <AaronSw> E:good summary of TimBL's talk

19:19:24 <dc_rdfig> added comment E1

22:05:03 <jhendler>http://www.businessweek.com/print/premium/content/02_09/b3772108.htm?mainwindow

22:05:03 <dc_rdfig> F: http://www.businessweek.com/print/premium/content/02_09/b3772108.htm?mainwindow from jhendler

22:05:42 <jhendler> F:| Sem Web article from Business Week

22:05:42 <dc_rdfig> titled item F

22:06:05 <jhendler> F: free version available from Business Week on-line

22:06:05 <dc_rdfig> added comment F1

22:06:16 <jhendler> F: See all the misquotes for yourself

22:06:16 <dc_rdfig> added comment F2

22:06:36 <jhendler> F: and note that we'll have the sem web done by 2005.

22:06:37 <dc_rdfig> added comment F3

22:32:01 <bijan> F:Note: I got a *lot* of flak for saying in my first [Prolog and RDF|http://www.xml.com/pub/a/2001/04/25/prologrdf/] article that the SW is AI, even though I have a nice, modest conception of what that means.

22:32:02 <dc_rdfig> added comment F4

22:32:20 <bijan> F:From this article (and the SciAm one), *clearly* I should have gone overboard!

22:32:20 <dc_rdfig> added comment F5

23:00:19 <bijan>http://www.swi.psy.uva.nl/projects/SWI-Prolog/Manual/db.html

23:00:20 <dc_rdfig> G: http://www.swi.psy.uva.nl/projects/SWI-Prolog/Manual/db.html from bijan

23:00:36 <bijan> G:|Interesting stuff about SWI-Prolog's various database mechanisms

23:00:39 <dc_rdfig> titled item G

23:01:37 <bijan> G:I had forgotten about this. I've been overusing assert/retract in [CWMClone|http://www.unc.edu/~bparsia/sw/cwmclone/cwmclone.html]. This will help with a bunch of things, including tracking variables.

23:01:37 <dc_rdfig> added comment G1

23:02:36 <AaronSw>http://logicerror.com/phoneBOFs

23:02:36 <dc_rdfig> H: http://logicerror.com/phoneBOFs from AaronSw

23:02:42 <AaronSw> H:|W3C Phone BOFs

23:02:42 <dc_rdfig> titled item H

23:02:54 <AaronSw> H:a proposal from Aaron Swartz and Sandro Hawke

23:02:54 <dc_rdfig> added comment H1

23:03:22 <AaronSw> H:We feel this will be a useful phone-based off-shoot of Interest Group discussion fora like #rdfig and www-rdf-interest

23:03:22 <dc_rdfig> added comment H2

23:03:39 <AaronSw> Send us feedback. :)

23:13:28 <bijan> Hey libby.

23:14:58 <libby> hey bijan!

23:15:07 <bijan> Long time no chat! How are you?

23:15:42 <libby> i'm having fun in cannes :)

23:15:55 * bijan is *very* jealous!

23:16:20 <libby> I'm listening to the W3C all star band :)

23:16:24 * jhendler notes it is after midnight in Cannes - Libby up late

23:16:29 <libby> it's very nice indeed

23:16:33 <bijan> Thank you for curing my jealousy.

23:16:37 <bijan> :)

23:16:37 * libby highly recommends it

23:16:43 <libby> heheh

23:16:56 * bijan decompiles some rules.

23:17:13 * libby wishing she could find recorduing features on ibook

23:17:29 <jhendler> who is playing?

23:17:56 * bijan points libby to his sweetie's music: http://www.mp3.com/ZoeMulford

23:17:58 <jhendler> Is DanC on guitar? (we know it isn't Tim on piano, despite what Bus Week says)

23:19:08 <libby> ian on guitar! danc on guitar

23:19:26 <AaronSw> oh, man... record! record!

23:19:28 <bijan> *Everyone* gave me a hard time about my mild SW is AI! Danbri, dave beckett, sean, aaron, jan.

23:19:43 <bijan> But Tim, well, *he* gets special treatment :)

23:20:08 <libby> does anyone know how to record on ibook?

23:20:18 <bijan> Do you have an external mike?

23:20:18 <AaronSw> it should have built in microphone

23:20:34 <bijan> sound control panel?

23:20:39 <bijan> Might let you record a bit.

23:20:41 <jhendler> how were semantic web demos this AM? (Libby, I think iTunes has this built in)

23:20:42 <AaronSw> you can use the classic SimpleSound, or get a recording program off of version tracker

23:21:27 <libby> okeydokey, thank

23:21:29 <libby> s

23:22:13 <bijan> jhendler, how's the sw classes going?

23:26:24 <jhendler> bijan - this term I'm teaching robotics not SW, will do SW again next Fall.

23:26:51 <bijan> Ah, robotics, yum.

23:26:54 <jhendler> however, class went well, have hired four of my former students to be SW hackers for me this term

23:26:54 <bijan> Doing any lego stuff?

23:27:06 <bijan> oo, any prologers?

23:27:15 <bijan> Want to set them to working on CWMClone?

23:27:33 <jhendler> probably no Lego this term, trying to do a more "traditional" AI robotics course - reactive and hybrid architectures...

23:28:05 <jhendler> no prologers (we're Lisp'ers here in the US) -- we are using some cwm, but not CWMClone right now.

23:28:27 <bijan> Er..

23:28:30 * bijan in the us.

23:28:41 <bijan> Which lisp do you use?

23:28:50 <jhendler> check our running RDF (chumped yesterday) and expect some markup tools soon

23:28:55 <bijan> Ever play with KnowledgeWorks?

23:29:04 <bijan> I would have been surprised ifyou were using CWMClone.

23:29:09 <bijan> As it's not quite done :)

23:29:24 <jhendler> bijan - oops, just assumed you were in UK since you were using prolog :->

23:29:38 <bijan> Even after I told you I was in North Carolina :)

23:29:51 <jhendler> saw discussion of CWMClone the other day, pointed a student to read about it... North Carolina isn't in the UK??

23:29:59 <bijan> Sorry, no!

23:30:06 <bijan> I wish, well for my sake.

23:30:10 <bijan> I wouldn't wish it on the UK!

23:30:18 <bijan> Jesse Helms, MP!

23:30:43 <jhendler> lol. Sorry, I got confused about who was where - I should have remembered that Liddy Dole is in your part of the world.

23:30:52 <bijan> Ouch, please!

23:30:55 <bijan> Don't remind me.

23:31:02 <jhendler> beats Helms!

23:31:12 <bijan> Ouch, please!

23:31:15 <bijan> Don't remind me.

23:31:17 <jhendler> oops, gotta go - back later.

23:31:20 <bijan> Ta

23:33:42 <jhendler> Ta? Ain't that some of that thar British talk - thought you said you was an Amayricken!

23:33:55 <bijan> But I like to mess you up.

23:48:09 <jonb> night all