Semantic Web Interest Group IRC Chat Logs for 2002-06-26

00:13:36 * Seth grins with danbri

01:20:48 * jhendler notes it is busy in here tonight

01:23:23 <sbp> Be vewwy vewwy quiet; we're hunting twipples

01:27:53 <AaronSw> sandro needs a dirc upg

01:32:49 * timbl_ added time and date built-ins to cwm a couple of days ago.

01:33:08 * timbl_ has been offline ever since

01:33:49 * timbl_ thanks to Mark Nottingham for the original files.

02:11:44 * jhendler wonders "who was that masked man?" He came, hacked cwm, and rode off into the sunset...

02:17:16 <Seth> sbp, d'ya find any twipples ?

03:14:19 <AaronSw> Anyone have ideas on documents that deserve a persistent home on the web?

06:42:57 <mnot> hmm... #rdfig seems to be weighted towards the first half of the alphabet tonight...

06:48:54 <mnot> LOL: <sbp> Be vewwy vewwy quiet; we're hunting twipples

06:48:54 <dc_rdfig> Label LOL not found.

06:50:02 <mnot> AaronSw: what's the context for your previous question (re: persistent docs)

06:50:12 <AaronSw> ah, i was just about to bring that up

06:50:32 <AaronSw> "We're building a home for information that's useful to the Web community. Links to pages on this server should never break and the data should never go away."

06:51:08 <mnot> I should be rich and live in Stockholm... doesn't mean it's going to happen.

06:51:15 <mnot> who's "we"?

06:51:23 <AaronSw> heh

06:51:46 <AaronSw> so i'm trying to build a mirror network for the website

06:51:56 <AaronSw> do you have any experience in that sort of thing?

06:52:00 <mnot> nah...

06:52:10 <AaronSw> i figure i can just give people cvs/rsync/tgz access and have them update it regularly

06:52:18 <mnot> which site?

06:52:31 <AaronSw> it'll be at web.resource.org

06:53:38 <AaronSw> persistence for the people

06:54:14 <mnot> where/who will be the mirrors? Is this a P2P-ish thing?

06:54:38 <AaronSw> well, the next step is to find some willing mirrors

06:55:01 <mnot> Is the goal redundancy/reliability? Or just performance/load sharing?

06:55:39 <AaronSw> reliability. i'm not too worried about the load that comes from serving up rdf schemas...

06:55:53 <mnot> Ah, it becomes clear.

06:56:06 <mnot> Mirrors it is, then.

06:56:51 <mnot> is trust of your mirrors implied?

06:57:11 <AaronSw> yeah, i guess...

06:57:23 <mnot> then it's more-or-less straightforward.

06:57:57 <AaronSw> what would you recommend? rsync and round-robin DNS?

06:58:06 <mnot> not round-robin

06:58:23 <AaronSw> oh?

06:58:29 <mnot> rsync is so cool and underappreciated... those guys should get an award or something

06:58:37 <mnot> and unison on top of it kicks ass

06:59:17 <mnot> There are some primitive open source request directors out there... lemme see..

07:00:10 <AaronSw> hm, unison looks cool. anyone actually use it?

07:00:22 <mnot> I do all the time. It's fantastic.

07:01:33 <AaronSw> wouldn't a request redirector require that the server is up? that's not very helpful if your goal is reliability

07:01:50 <mnot> not at the HTTP level, in the DNS

07:01:57 <AaronSw> ah, i see

07:09:37 <mnot> I haven't done a roundup of these for a while, but some that popped up a while back include: http://eddie.sourceforge.net/ http://www.stanford.edu/~schemers/docs/lbnamed/lbnamed.html

07:10:13 <mnot> eddie is from a bunch of hackers from Melbourne, so it can't be all bad ;)

07:10:42 <AaronSw> :-)

07:11:42 <AaronSw> yikes, written in erlang

07:12:09 <mnot> So, the question that I have is: how do you communicate your persistence policy in a machine-understandable way?

07:12:11 <mnot> yep

07:12:47 <AaronSw> That's a good question. Got any ideas?

07:13:01 <mnot> well, it should be RDF, of course

07:13:14 <AaronSw> maybe this can be one of our schemas :)

07:13:26 <mnot> somebody needs to go and define a sensible schema for it. Tim had some really nice writings about this in designissues

07:13:33 <mnot> (of course)

07:13:36 <AaronSw> i was also thinking about hosting a connegy media type archive

07:13:51 <mnot> what would be in it?

07:14:19 <AaronSw> Same kind of stuff as in http://www.isi.edu/in-notes/iana/assignments/media-types/ except with HTML and RDF versions, probably.

07:14:20 <mnot> re: communicating persistence: the real question to me is how to associate the assertions with a particular resource

07:14:23 <mnot> ah

07:14:34 <mnot> that would be good to have

07:15:04 <AaronSw> hm, this sounds vaguely like aboutEachPrefix... or your replacement

07:15:14 <mnot> you can associate with HTTP headers, or a separate file. If it's in a file, how do you discover that file/site metadata?

07:15:41 <mnot> yes, I still want to rewrite that in rules, but time is a problem.

07:15:48 <mnot> how do you discover it, tho?

07:16:20 <mnot> A while back I thought it should be using OPTIONS * on the site, but these days I lead more towards conneg against the most generic resource on the site - /

07:16:34 <mnot> s/lead/lean/

07:16:41 <AaronSw> http://site.com/metadata.rdf ;-)

07:16:41 <dc_rdfig> A: http://site.com/metadata.rdf from AaronSw

07:16:47 <AaronSw> oops.

07:16:57 <mnot> DanC will whack you upside the head for that...

07:17:41 <mnot> That violates opacity; if we say "you have to dedicate /metadata.rdf on your site to our special format", we're intruding on other ppl's namespace

07:17:58 <AaronSw> A:example of how to communicate metadata for a site, ala [http://www.robotstxt.org/|robots.txt]

07:17:59 <dc_rdfig> added comment A1

07:18:01 <AaronSw> yeah, i know

07:18:07 <mnot> I know you knew ;)

07:18:52 <mnot> that's why I hated WS-Inspection so much... they really should know better, and they don't need to be backwards compatible.

07:19:06 <mnot> but conneg doesn't get good mindshare, unfortunately.

07:19:38 <AaronSw> what, you propose conneging for an rdf file? that's not right

07:19:47 <AaronSw> maybe meta-link headers...

07:20:42 <mnot> sure... you do conneg on http://www.example.org/ and ask for application/site-metadata+rdf or somesuch; it gives you metadata about the site (since by definition all other resources on the site are contained by /, it works here), but you don't intrude into their namespace.

07:21:09 <mnot> Of course, that representation might very well have a Content-Location of /metadata.rdf, but that's up to the site owner.

07:22:31 <AaronSw> no, the metadata is not http:equivalent to the site

07:22:33 <mnot> meta-link works too, but it's not as efficient; the client has to request /, hope that they get back text/html (OK, this is pretty likely), parse out the LINK, and then re-request the metadata. Two round trips.

07:22:58 <AaronSw> i was thinking meta-link HTTP *headers*

07:23:32 <mnot> You're saying that the ap/site-md+rdf version isn't a representation of /?

07:24:01 <danbri> mornin'

07:24:06 <mnot> hey

07:24:31 <mnot> AaronSw: so responses contain HTTP headers linking to the metadata?

07:24:47 <AaronSw> yes

07:24:51 <AaronSw> yes

07:25:25 <mnot> that would be good for opportunistic discovery

07:25:36 <mnot> but some ppl (esp. with big sites) wouldn't like the extra traffic

07:25:58 <mnot> and it's still inefficient if a client needs to proactively find the metadata

07:26:32 <mnot> perhaps a number of possibilities; a) conneg on / b) HTTP response header c) HTML LINK header

07:29:17 <mnot> re: the Excel discussion yesterday, this is interesting: http://www.knownow.com/products/whitepapers/KNDatasheet_ExcelConnector.pdf

07:56:21 <AaronSw> Interesting: http://www.root-servers.org/

07:56:29 <AaronSw> These are the people I need to take over in order to fix ICANN.

07:57:48 <mnot> .. E, IIRC, is located in the middle of Moffet Field, in an underground bunker (lots of ways for chump to trip up on that one...)

07:58:02 <mnot> colocated with MAE West

07:58:26 <AaronSw> cool

07:58:41 <AaronSw> I wonder if we can get jhendler to take over D

07:58:52 <mnot> heh

07:58:53 <AaronSw> Is there any process for changing the root servers?

07:59:15 <AaronSw> Where does the DoC get its power over them?

07:59:20 <mnot> there are changes over time, I remember some a few years back (when I was closer to this stuff in the ISP world)

08:00:00 <mnot> that would be something interesting to look into

08:03:56 <mnot> see: http://www.isi.edu/in-notes/rfc2870.txt

08:06:23 <AaronSw> hm: ftp://ftp.rs.internic.net/domain/named.root

08:07:02 <mnot> yep, that's the seed

08:08:09 <mnot> IIRC it's limited to 13 because that's how many will fit into a UDP packet

08:08:46 <mnot> argh, bedtime

08:10:08 <AaronSw> heh

08:10:27 <AaronSw> i wonder if i can just hijack ftp.rs.internic.net... hm, is that a static ip

08:11:02 <mnot> it's very rarely downloaded... only when there's a change, and that's a manual process.

08:11:37 <mnot> (i.e, it's pre-web ;)

08:12:07 <AaronSw> OK, i thought it might be that way.

08:12:42 <AaronSw> So I don't see where DoC gets its authority? Do these 13 people really believe in ICANN?

08:13:13 <mnot> s/people/legal entities/

08:58:06 <jonas> jonas is now known as aigan

10:07:12 * aigan is away: annat att göra...

10:12:58 <dmiles> dmiles is now known as dmileZzZ

10:51:33 * aigan is back (gone 00:44:20)

11:12:33 <lilo> [GlobalNotice] Hi all. Apologies for interrupting your work. Currently #openprojects is +m. It's sad that on a network that's here to be a friendly forum for projects, a vocal minority can so control the main administrative channel. Please don't let those people be the only people expressing their viewpoints. Thanks.

13:17:44 <rreck> morning

13:20:26 <mhgrove> morning man

13:20:48 <mhgrove> can you do me a favor and put a link to the ConvertToRDF tool on the main mindswap page?

13:21:24 <rreck> yeah just please email the url you mean

13:21:38 <mhgrove> sure, thanks!

13:31:41 <rreck> you didnt tell me what to call it

13:35:30 <mhgrove> doh, um just call it ConvertToRDF - a tool to convert delimited text files to RDF/DAML

13:35:33 <mhgrove> or something like that

13:38:09 <rreck> just realize you could do it too, you have the access

13:39:55 <mhgrove> oh really?

13:41:44 <rreck> YEAH

13:41:50 <rreck> but dont i already did

13:44:58 <mhgrove> oh ok, i was just putzing around iama, good thing you said something

13:45:01 <rreck> did you chump the link ?

13:45:25 <ronwalf> Hello

13:45:30 <rreck> cool more troops

13:45:53 <ronwalf> rreck: Don't all of our tools except for Parka output rdf?

13:46:07 <mhgrove> i dont know how to chump

13:47:30 <dajobe> do /msg

13:47:37 <dajobe> um, /msg dc_rdfig help

13:49:02 <mhgrove> ok lets try this heh

13:49:12 <mhgrove>http://www.mindswap.org/~mhgrove/ConvertToRDF

13:49:12 <dc_rdfig> B: http://www.mindswap.org/~mhgrove/ConvertToRDF from mhgrove

13:49:37 <mhgrove> B:| Tool for converting delimited text files to RDF/DAML

13:49:37 <dc_rdfig> titled item B

13:49:44 <mhgrove> heh cool

13:50:20 <mhgrove>http://www.mindswap.org/~mhgrove/RIC/RIC.html

13:50:20 <dc_rdfig> C: http://www.mindswap.org/~mhgrove/RIC/RIC.html from mhgrove

13:50:54 <mhgrove> C:| RDF Instance Creator (RIC) - A visual RDF instance authoring tool

13:50:55 <dc_rdfig> titled item C

13:51:17 <mhgrove> ronwalf: i think they do...both of mine do i would assume

13:51:45 <ronwalf> mhgrove: I think we're refering to Parka in this case

13:52:45 <mhgrove> yeah that is the only thing i can think of that doesn't output rdf, aditya's stuff definately does

13:56:22 <rreck> im trying to ask if there is an obligation

13:58:08 <mhgrove> nah

13:58:52 <rreck> but otherwise im building to the human web

13:59:26 <ronwalf> Is parka really on the web? That's the interface to parka

14:00:23 <rreck> there is a C api, but the wrapper i use is like 10 lines of C wrappered by perl

14:00:26 <ronwalf> And the interface to parka can go and dump stuff out in a flat file, or you might be able to define a service in rdf to describe searching through it

14:00:50 <rreck> ok i can dump the assertions to a file now

14:01:12 <ronwalf> I think parka will need to be 'fixed' eventually.

14:01:22 <rreck> yeah i need to talk to amy

14:01:30 <rreck> did you get the linux to compile ?

14:01:49 <ronwalf> I don't think she's going to be able to do much until about four weeks from now

14:01:55 <ronwalf> rreck: I didn't, Amy did

14:01:59 <rreck> i need to convert my lab workstation to redhat

14:02:11 <ronwalf> She did funky makefile stuff, though

14:02:20 <rreck> i have the current distribution only set to build under solaris

14:02:31 <rreck> i need to add her changes

14:02:40 <ronwalf> Yeah, we need to get a configure script running

14:02:55 <rreck> the version i am running on mindswap is legacy

14:03:11 <rreck> i dont run the version she compiled and need to talk to her

14:03:27 <rreck> then we can integrate the changes

14:04:10 <ronwalf> Yeah, there was some confusion as to which on was newer.

14:04:19 <ronwalf> The CVS version is aparently old and out of date

14:04:25 <rreck> well i have the distribution ready for download

14:04:49 <ronwalf> This is the most recent one?

14:04:49 <rreck> well someone already fixed it to work under 2.8 so i think we will go with that

14:05:01 <rreck> it took some doing

14:05:22 <rreck> so i want to see how that compares to amy's version

14:05:42 <rreck> amys version comes from another branch ported to linux

14:07:10 <ronwalf> ah, ok. I'll bug her to put it somewhere

14:07:11 <rreck> that PIQ query tool seems like it would be perfect if added to mike/aditcha's tools

14:07:21 <rreck> well its more i need a braindump

14:07:38 <rreck> she can tell me in 2 minutes what would take forever to grok alone

14:07:44 <ronwalf> Where's your version sitting?

14:07:55 <rreck> one sec

14:08:05 <mhgrove> i think PIQ is hard to use

14:08:14 <ronwalf> PIQ?

14:08:25 <mhgrove> i want to write my own layer to put on top of PARKA, then aditya and i can use the same thing

14:08:40 <mhgrove> i fact that seems like a good project for today since I finished the new ver of RIC

14:08:46 <mhgrove> and i gave up on the searching heh

14:09:25 <mhgrove> PIQ = PARKA Interface Query or something like that

14:09:37 <rreck> yeah you should see it

14:09:42 <mhgrove> its a visual PARKA query creator, does it via graphs and stuff, its complex

14:09:58 <rreck> well to cut to the chase its java

14:10:17 <mhgrove> *nods* yeah i noticed it was an applet

14:10:26 <rreck> ron i sent you the url for parka in the other window

14:10:27 <mhgrove> i would love to get a hold of that code

14:10:33 <rreck> well lets ask jeff

14:10:44 <mhgrove> he still at UMD?

14:10:44 <ronwalf> Yeah, I got it

14:10:45 <rreck> you want me to, that's why i showed you

14:11:09 <rreck> no but i think we can kindly ask him

14:11:20 <mhgrove> sure, if nothing else I want to see how he interface w/ PARKA so I can figure out how to do it

14:12:09 <mhgrove> cause right now i have no idea how that would work out

14:12:49 <mhgrove> unfortunately probably wraps the java around some native methods which gets nasty quickly

14:13:01 <rreck> like any other db weould basically. open a socket and a db call

14:13:17 <ronwalf> Hm, makedepend is barfing out on the same config file as Amy had problems with.

14:13:30 <ronwalf> I think I will look at this later

14:13:32 <rreck> you have to follw those instructions i include

14:13:36 <ronwalf> I'm off to class

14:13:38 <mhgrove> ron, you make it sound so easy, just open a socket and talk to it hhaha

14:13:40 <rreck> later

14:13:44 <mhgrove> have fun ;)

14:48:59 * DanC browses cia factbook in RDF... http://www.daml.org/2001/12/factbook/uz

14:49:18 <DanC> hmm...

14:49:19 <DanC> # Warning: ID=A112256 on statement ignored

14:49:19 <DanC> # Warning: ID=A112274 on statement ignored

14:50:07 <DanC> <NS0:irrigatedLand rdf:ID='A112256'>40000</NS0:irrigatedLand>

14:50:27 <DanC> <rdf:Description rdf:about='#A112256'>

14:50:27 <DanC> <rdf:type rdf:resource='http://www.daml.org/2001/12/factbook/factbook-ont#Estimate'/>

14:50:27 <DanC> <NS0:year>1993</NS0:year>

14:50:27 <DanC> </rdf:Description>

14:51:01 * DanC wonders how timbl would suggest saying that in N3

14:53:00 * timbl_ ctaches up

14:54:51 <timbl_> I wouldn't say that in N3.

14:54:58 <timbl_> 0.5 ;-)

14:56:10 <kao> whats that supposed to mean? an annotated property?

14:56:26 <timbl_> Maybe if you *had to*, #A112256 = { what:ever NS0:irrigatedLand "40000" }; a log:Truth.

14:57:08 <timbl_> Maybe if you *had to*, #A112256 = { what:ever NS0:irrigatedLand "40000" }; a <http://www.daml.org/2001/12/factbook/factbook-ont#Estimate>.

14:57:14 <timbl_> I should say.

14:57:41 <timbl_> If you in a mode to belive fact book estimates, then you would use the data.

14:58:11 <timbl_> The annotation disclaims the data, so you can't just add triples, you have to cut the data out and quote it.

14:58:35 <timbl_> Maybe you don't need any ID for it.

14:59:05 <DanC> hmm... disclaims... or perhaps justifies it?

14:59:05 <timbl_> { what:ever NS0:irrigatedLand "40000" } a <http://www.daml.org/2001/12/factbook/factbook-ont#Estimate>.

15:00:47 <timbl_> Well, I might want to runa rule which says that if {s p o} is an estiamte, then spx is true for some 0.5*o < x < 2.0*o

15:01:47 <DanC> I suppose it does disclaim it, in this context. But it seems very close to the citation idiom: "2 out of 3 dentists recommend trident [AMA, 1997]".

15:02:03 <DanC> er... s/AMA/ADA/

15:02:09 <timbl_> So you run yoru machine with teh axiom that NSO:Estimate is a subclass of log:Truth. If it screws up, you try it without.

15:03:25 * DanC wonders how sandro would say it in ptl

15:03:38 <timbl_> { what:ever NS0:irrigatedLand "40000" }; a log:Truth; annot:from :CIA.

15:03:40 * DanC can't find much lat/long info in the fact book. bummer.

15:04:09 * timbl_ bets the CIA has it on file somewhere.

15:04:15 <DanC> does cwm still only see rules from log:Truth's?

15:05:00 <timbl_> yes....you either have to infer the rule itself, or that it is in a log:Truth, for cwm to process it using --think for example.

15:05:45 <timbl_> cwm doesn't facts in log:truths without an axiom to that effect.

15:05:54 <DanC> ok.

15:09:17 * sandro shows up. DanC, do you want me to tune in?

15:09:43 <DanC> I suppose.

15:13:47 <sandro> I'm having trouble understanding the use case. :(

15:14:54 <DanC> "In 1993, the CIA estimated the irrigated land of Uzbekistan at 40000"

15:15:23 <DanC> timbl suggested something ala: { :uz :irrigatedLand "40000" } a cia:Estimante.

15:16:06 <DanC> ptl doesn't have {}'s, right?

15:16:14 <sandro> That seems okay to me. Do you want me to put that n3 into ntriples, using ptl?

15:16:27 <DanC> yes, I want that.

15:16:27 <sandro> right -- PTL tells us how to flatten n3.

15:16:39 <timbl_> voila

15:16:43 <sandro> working.....

15:16:53 <timbl_> What is your current method of flattening n3?

15:16:57 <sandro> [ star trek computer sounds... ]

15:17:16 * DanC laughs, can't help seeing lots of blinking lights...

15:18:04 <sandro> It's documented here, Tim: http://www.w3.org/2002/05/positive-triples/pt1.n3 I'm contemplating puting it into cwm, but also trying to see if it really works.

15:21:14 <DanC> pt1:true a pt1:Conjunction; <--hmm... why a conjuction, not a sentence?

15:23:50 <sandro> Just to keep the structure more list-like. Maybe I should call it pt1:theNullConjunction (which is known true).

15:25:19 <timbl_> You talk about "false". cwm deosn't.

15:25:51 <sandro> I talk about it, but it's no more reachable than it is in cwm.

15:25:58 <timbl_> modus tollens is *not* available just now, as the logic of not is introduced.

15:31:54 <sandro> I think it's appropriate for PTL to not allow stating a contradiction, and thus it avoids negation, but if a reasoner has negated facts from other (non-RDF) source, I don't think there's a problem with it using it.

15:32:36 <danja> 'scuse

15:32:38 <timbl_> You want to avoid p or not p as an axiom.

15:32:41 <danja>http://www.isacat.net/2002/irclogs/

15:32:41 <dc_rdfig> D: http://www.isacat.net/2002/irclogs/ from danja

15:33:04 <danja> D:| Strimmed version of the day's #rdfig log

15:33:04 <dc_rdfig> titled item D

15:33:12 <DanC> "We could

15:33:12 <DanC> express this in DAML+OIL, but the layering seems a little odd

15:33:12 <DanC> there" <-- then why do you daml:import daml at the top?

15:33:16 <danja> D: a bit lighter, easier on the Palm, but very slooww

15:33:17 <dc_rdfig> added comment D1

15:33:29 <danja> D: also http://www.isacat.net/2002/irclogs/last_year.asp is this date last year

15:33:29 <dc_rdfig> added comment D2

15:33:45 <sandro> Yeah, I guess I should take that out, DanC.

15:33:54 <danja> ciao

15:34:20 <DanC> hmm... pt1.n3 could sure use an example.

15:34:59 <DanC> e.g. do you write :st pt1:subject <http:...>. or ... pt1:subject "http:..." ?

15:35:07 <sandro> Indeed. I think that's my cue: there are two. The second one is: http://www.w3.org/2002/05/positive-triples/pt1-cia-ex1.rdf

15:35:48 <DanC> <Constant>

15:35:48 <DanC> <denotation rdf:resource="&cia;uz"/>

15:35:48 <DanC> </Constant>

15:35:51 * DanC mulls that over

15:35:52 <AaronSw> ouch

15:36:14 <AaronSw> I was hoping for "r: http://example.com/cia#irrigatedLand"

15:37:54 <timbl_>http://www.w3.org/2000/10/swap/time.n3

15:37:54 <dc_rdfig> E: http://www.w3.org/2000/10/swap/time.n3 from timbl_

15:38:14 <timbl_> E:|Schema for date and time built-ins for cwm

15:38:14 <dc_rdfig> titled item E

15:38:31 <sandro> I don't understand, Aaron.

15:39:02 <DanC> "the two-digit year" <- typo in month docs, methinks, timbl

15:39:29 <timbl_> oops... yes I did paste in year 7 times

15:39:50 <AaronSw> well, I thought the problem with reiification was that you could infer equivalences across literals

15:40:25 <AaronSw> err across symbols

15:41:01 <DanC> timbl_, pls publish http://www.w3.org/2000/10/swap/time.rdf too

15:41:04 <AaronSw> i.e. x rdf:subject foo . foo = bar . => x rdf:subject bar .

15:42:13 <sandro> This kind of reification does not protect the terms from such equalities; I beleive this is the right kind to use in rules.

15:43:45 * DanC tries to think of places where I have hard-coded stuff that this new time module would do generally

15:46:28 <DanC> the Constant/denotation stuff looks awkward, but not wrong. it seems to say "any constant that evaluates to what this refers to"

15:47:19 <DanC> i.e. it's not a use/mention bug, but it's not strictly syntactic, which kinda bugs me.

15:48:21 <DanC> it's kinda like: pt1:subject [ log:uri [ is log:uri of <foo> ]].

15:48:37 <DanC> no...

15:48:50 <DanC> it's like pt1:subject [ is log:uri of <foo>].

15:49:13 <sandro> I would simply say "C pt1:denotation D" means "C is a logical constant symbol, a syntactic entity, which has the denation D, a thing in the domain of discourse." BOTH are in domain of discourse of the RDF graph, however.

15:49:20 * timbl_ publishes time.rdf and ads time:gmTime and time:localTime

15:49:53 <DanC> how about a seeAlso from the time schema to some test files?

15:50:25 <sandro> Yeah, it's a lot like pt1:subject [ is log:uri of <foo>]

15:50:43 <DanC> and timbl, credit mnot, at least for inspiration

15:50:43 <sandro> It's needed, btw, so we can put variables in the sentences, too.

15:51:07 <timbl_> You say Sentence and Term are disjoint. But in N3, one option for a Term is a Sentence.

15:51:28 <DanC> that seems by design, timbl. ptl doesn't have {}

15:51:56 * timbl_ Credit mnot in schema , yes

15:52:03 <timbl_> then ptl can't flaten n3

15:52:12 <DanC> ?

15:52:26 <DanC> the whole idea of flattening is to write without {}, no?

15:52:39 <timbl_> Yes, but you have to have stg to flatten {} to.

15:52:59 <timbl_> So I would expect it to say thaht the subject of an atomic sentence was a conjunction...

15:53:46 * Seth waves to sbp

15:54:03 <DanC> I see what you mean, timbl

15:54:18 <timbl_> Mark's isodate.py and timegm.py are included now in the cwm needs.

15:55:01 <DanC> umm... did you check the license in isodate.py?

15:55:20 <timbl_> Mark said i could include it in thhe distribution.

15:55:25 <DanC> ok

15:55:26 <timbl_> in private mail.

15:55:33 <sandro> You're talking about in a nested case? { { a b c } d e } log:implies foo ?

15:55:49 <DanC> no, not a nested case; just a case of two conjuncts in the subject of the implication

15:56:06 <DanC> { x y z. p d q} log:implies whatever.

15:56:26 <sandro> that's a conjunction of two atomic sentences.

15:57:15 <DanC> does the copyright on isodate.py remain with mnot? can we put the W3C license on it? let's not let the license on cwm get all patchwork.

15:57:27 <timbl_> What about { {xyz} d q} log:implies whatever ...

15:57:39 <DanC> slow down, pls, timbl.

15:57:51 <timbl_> licence: agreed .. I take Mark's message to allow W3C licence.

15:58:03 <DanC> but the range of pt1:subject is term, sandro, and terms don't include conjuctions.

15:58:08 * timbl_ needs boilerplate file in dist.

15:58:46 <DanC> see, e.g. KIFSink.py for a license blurb that just got an OK from reagle

15:59:06 <sandro> Why would you have a pt1:subject arc pointing to a conjunction? they point from atomicSentences and to terms (which are constants or variables).

15:59:18 <DanC> it is somewhat conventional to have a LICENSE file in a package.

15:59:44 <DanC> why point to conj: pls transcribe { x y z. p d q} log:implies whatever. to ptl

15:59:56 * timbl_ wonders whether LICENCE.html would be OK

16:00:11 <DanC> yes, LICENSE.html would probably be fine

16:02:32 <sandro> That sounds to me like you're asking to reify that sentence (with the "implies" predicate), which is unnecessary. See the other example for a conjunction: http://www.w3.org/2002/05/positive-triples/pt1-gf-ex1.rdf

16:04:20 * DanC can't parse pt1-gf-ex1.rdf; esp the <implication> tag

16:05:03 <DanC> hmm.. I guess I see that part now...

16:05:21 <sandro> ntriples: x pt1:implication y read as: the sentence x has an implication, which is the sentence y.

16:07:51 <DanC> ugh... so P log:implies Q isn't parallel to P R O when mapped to ptl

16:07:52 <DanC> ugh... so P log:implies Q isn't parallel to P R Q when mapped to ptl

16:07:52 <DanC> hmm... maybe I'm confused...

16:08:49 <DanC> hmm... <UniVar rdf:about="&var;grandfather"/> ... odd... &var;grandfather *refers to* a variable...

16:08:49 <sandro> pt1:implication is basically the same as log:implies.

16:09:47 <sandro> &var;grandfather is the URIRef of the variable. The variable is an object in RDF's domain of discourse.

16:10:19 <DanC> ok, perhaps we do need nesting to make the point; let's try { :Joe :says { :baseball a :Fun. :soccer a :NoFun } } log:implies { :Joe a :OldSchooler }.

16:10:31 <DanC> it's *a* URIref of the variable.

16:10:46 <sandro> Yes, "*a*" you are absolutely correct.

16:11:22 <DanC> in the Joe case, the object of the antecedent is a conjunction.

16:12:49 <sandro> Ah, I see what you're saying. No, the object of the antecedent is a Term. That term is a Constant. That Constant has a denotation, which happens to be a Conjunction.

16:13:21 <sandro> (not fun stuff on too-little sleep, with no pretty diagrams.)

16:14:22 <DanC> hmm... I can see saying the object is a constant, but it seems to me that constant *is* a Conjunction; it doesn't denote one. hmm...

16:14:26 <sandro> Arg, MACE has an arbitrary arity restriction on relations and functions.... How to work around it.....

16:14:39 <DanC> MACE?

16:15:17 <sandro> MACE is a companion to Otter, which finds models -- good for proving a formula is satisfiable (as we want the axiomatic sematics to be).

16:15:58 <sandro> It uses Otter's input language with some minor exceptions, like the arity one, which I hadn't noticed before.

16:16:01 <DanC> timbl_, I see that retest.sh has been updated, but I don't see timet1.n3

16:17:33 <timbl_> time/t1.n3 is tyhe source. timet1.n3 is the file should be in ref/

16:17:48 <DanC> er... maybe use lists? r(x,y,z) becomes r((x,y,z)), i.e. r(cons(x,cons(y,cons(z,nil)))...oops that's a 2-ary function; gotta curry it...

16:18:12 <DanC> ah; quite, timbl; I misread retest.sh

16:18:59 <DanC> r(x,y) becomes r(cons(x)(cons(y)(nil)))

16:19:52 <DanC> where cons is a function that takes a thing (the car) and returns a function that takes a list (the cdr) and returns a list.

16:20:11 <DanC> can functions return functions in MACE?

16:20:51 * DanC expects not

16:20:55 <timbl_>http://www.w3.org/2000/10/swap/LICENSE.rdf

16:20:55 <dc_rdfig> F: http://www.w3.org/2000/10/swap/LICENSE.rdf from timbl_

16:21:08 <sandro> I would think so. But the odd thing is "Mace does not accept function symbols with arity greater than 3 or relation symbols with arity greater than 4."

16:21:08 <sandro> and I don't think I'm actually violating that. I'm at four, but not over it. Time to binary search the code (no good error message).

16:21:29 <sandro> the file is pt1-axioms.otter

16:22:25 <DanC> which way does subLicense go? (is it documented in swap/pim/doc?)

16:22:44 <DanC> doesn't seem to be

16:22:46 <timbl_> F:|Code license for swpa code in RDF, invokes W3C license as sublicense using doc:sublicense

16:22:46 <dc_rdfig> titled item F

16:23:30 <timbl_> a SUb licence licenses a SUBset of the rights. Documented in schema.

16:23:38 <DanC> F:hmm... how about defining OpenSourceLicense in terms of log:content, string:scrape, and the like?

16:23:40 <dc_rdfig> added comment F1

16:24:16 * DanC can't find anything about subLicense in doc schema

16:24:52 <sandro> Ah, it's an arity-4 skolem function it's making. fooey.

16:31:11 <timbl_> You mean an parse the english of a licence? Ouch.

16:33:43 <DanC> no, I mean: "it's an OpenSource license iff <http://opensource.org/licenses-or-whatever> points to it"

16:34:15 <timbl_> I don't think that worked .... did we ever get pointer? I though they ran out of pointing power.

16:34:48 <DanC> this is the page I was thinking of... http://www.opensource.org/licenses/

16:34:49 <timbl_> I'm not sure th eW3C licnse is pointed to.

16:35:10 <timbl_> oh, good, it is

16:35:11 <DanC> oh frap; they don't point; they copy. http://www.opensource.org/licenses/W3C.php

16:35:33 <AaronSw> i wonder if opensourcelicenses are all licensed for them to copy

16:36:27 <timbl_> Well, they've slapped a copyright on their copy of our licence.

16:36:46 <jang> the copy versus pointing is obvious...

16:36:56 <jang> ...to prevent "endorsing" something that accidentally changes

16:37:10 <jang> although the breach of copyright is quite amusing

16:37:11 <timbl_> Well, they could do both.

16:37:13 <DanC> well, why not copy *and* link.

16:37:20 <jang> that would make more sense

16:37:48 <timbl_> Forytunately the copyright notice itself contains a URL in the text as a last resort.

16:37:58 <jang> maybe license text wants to be free, too?

16:38:29 <jang> btw. are there guidelines for using the w3c license?

16:38:41 <jang> ie, in a separate LICENSE file or at the top of every source file?

16:40:02 <jang> the reason I ask is that the little ntriples comparison utility we're using for test cases in rdfig should be properly licensed

16:40:30 <jang> rdfcore, not rdfig

16:41:17 <timbl_> Dan, you've been just quoting the W3C licence by URL and I think that should be good enough.

16:41:58 <jang> ok, that'll do. cheers.

16:43:24 <timbl_> There is the question as to whether schemata are code or docs. If orget what people have done to date.

16:43:34 <timbl_> Jo Reagle is the expert.

16:47:01 <jang> well, djb won a case on the basis that source code was protected under the first amendment,

16:47:08 <jang> so the distinction is certainly narrowing

16:47:44 <jang> it's particularly the case with high-level ontologies, which are intended to be machine-processable renderings of human understanding

16:51:43 <larsbot> em, you have time to chat now?

16:51:48 * larsbot has 45 free minutes

16:57:52 * sandro manages to redo his axiomatic semantics so the Skolem functions all have arity <= 3.

17:01:41 <sandro> Tim, DanC, if I want to write some python to generate PTL from n3, should I do it as --flat cwm, or in a separate program, or does it matter?

17:06:54 * larsbot wonders what PTL is

17:07:55 <AaronSw> PTL: http://www.w3.org/2002/05/positive-triples/

17:07:55 <dc_rdfig> Label PTL not found.

17:08:23 <larsbot> thanks :)

17:17:40 <DanC> I'd rather have it as --flat in cwm, sandro

17:18:07 <DanC> but I don't have a strong preference.

17:18:27 <DanC> a separate program that calls on notation3.py would be fine.

17:19:17 <DanC> ... using the yapps grammar stuff might be fun, though the grammar isn't quite debugged. (can't seem to left-factor it and still allow periods (or not) in all the relevant places)

17:19:28 <sandro> I

17:20:08 <sandro> That PTL document is still very drafty, which is why I haven't chumped it yet. (but it is my top priority now)

17:21:06 <DanC> hmm... seems to me it's worth chumping regardless what's inside, if we spent 10 minutes talking about it today.

17:41:43 <DanC> ugh... time/now.n3 isn't a good regression test

17:41:53 <DanC> ah... it's not in retest.n3 for that reason, I suppose

18:09:24 <sandro> It's not a question of whether it's worth chumping, but when.

18:13:39 * danbri looks at http://www.w3.org/2002/05/positive-triples/

18:13:52 <danbri> 'turing complete subset' of cwm/n3 logic...

18:14:08 <danbri> I didn't realise cwm/n3 logic was (the sort of thing that) was Turing complete

18:14:25 <danbri> cwm is, and can be coaxed into doing things that never complete, I s'pose

18:14:28 <danbri> But N3?

18:16:27 <DanC>http://www.w3.org/2002/05/positive-triples/

18:16:28 <dc_rdfig> G: http://www.w3.org/2002/05/positive-triples/ from DanC

18:16:37 <DanC> G:|Positive Triples Logic

18:16:37 <dc_rdfig> titled item G

18:16:42 <DanC> logger, pointer/

18:16:42 <DanC> See http://ilrt.org/discovery/chatlogs/rdfig/2002-06-26#T18-16-42

18:16:56 <DanC> G:see [discussion|http://ilrt.org/discovery/chatlogs/rdfig/2002-06-26#T18-16-42] of flattening, etc.

18:16:57 <dc_rdfig> added comment G1

18:23:04 <DanC>http://theory.lcs.mit.edu/tds/ioa.html

18:23:05 <dc_rdfig> H: http://theory.lcs.mit.edu/tds/ioa.html from DanC

18:23:14 <DanC> H:| IOA Language and Toolset

18:23:15 <dc_rdfig> titled item H

18:23:58 <DanC> G:by [MIT Theory of Distributed Systems|http://theory.lcs.mit.edu/tds/] folk, incl. S. Garland

18:23:59 <dc_rdfig> added comment G2

18:24:23 <DanC> G:hmm... useful for Web Services analysis?

18:24:24 <dc_rdfig> added comment G3

18:24:32 <DanC> phpht.

18:24:39 <DanC> H:by [MIT Theory of Distributed Systems|http://theory.lcs.mit.edu/tds/] folk, incl. S. Garland

18:24:40 <dc_rdfig> added comment H1

18:24:46 <DanC> H:hmm... useful for Web Services analysis?

18:24:47 <dc_rdfig> added comment H2

18:24:51 <dajobe> you can delete it with G3:""

18:25:03 <DanC> H:hmm... useful for Web Services analysis?

18:25:03 <dc_rdfig> added comment H3

18:25:13 <DanC> G3:""

18:25:14 <dc_rdfig> deleted comment G3

18:25:32 <DanC> G2:""

18:25:32 <dc_rdfig> deleted comment G2

18:29:03 <MarkB> H:nah, it's got the word "theory" in it, so clearly isn't suitable from a time-to-market perspective 8-/

18:29:04 <dc_rdfig> added comment H4

18:31:55 <lilo> [GlobalNotice] Hi all. I have a website up at http://lilo.sargasso.net/ explaining my current situation. I'm asking for help on a personal basis to keep working as OPN head of staff and to finish setting up the nonprofit corp. Contributions have been a bit light for the past couple of days, so if you're thinking of helping, please take a look, and thanks.

18:32:49 <sandro> danbri, are you familiar with the classic result that Horn Clauses are Turing Complete? Does that provide the middle ground? (or do you want to wait for the demo...?)

18:36:49 <Seth> Does anyone know where there is a n3 and also a ntriples file for the rdfs schema ?

18:55:33 <DanC> H:interesting... their "hello world" example is a bank account, just like the paper trail bank.

18:55:33 <dc_rdfig> added comment H5

19:01:11 <dmileZzZ> dmileZzZ is now known as dmiles

19:28:26 <CaptSolo_> hi all :)

19:31:26 <rreck> im looking for good queries to do on SEC Edgar data, ive loaded a couple days worth

19:59:07 <DanC> ARGH! not only are my ssh tunnels going nutso, but ftp/http stuff too! I suspect this NAT router box.

19:59:21 <DanC> rreck, find companies that share officers

20:17:34 * DanC substitutes scp for imap-over-ssh to get my mail...

20:28:44 <Seth> does anyone have anything better than /swap/xml2rdf.py to parse rdf/xml in python ?

20:54:33 <DanC> sax2rdf.py

20:54:52 <DanC> wow... cwm bug report from somebody I have never heard of!

20:54:58 <DanC> Sampo Syreeni

20:55:03 <DanC> (see www-archive)

20:55:13 <AaronSw> yeah, that was cool

20:55:19 * jordan thinks cwm rules

20:55:25 <AaronSw> And he was able to figure out the right process for sending in bugs.

20:56:15 <DanC> "student/math+cs/helsinki university"

20:56:37 <DanC> and he uses tel: URIs in his .sig. a man after my own heart!

20:57:37 <DanC> sigh... followed by clueless ASP question, under a subject of "XML problem"

20:58:21 <DanC> hoot! (more from my unkown senders inbox...) "STRICTLY COMFIDENTIAL"

20:58:53 <DanC> now... is that somebody cleverly avoiding spam filters? or just a typo?

20:59:21 <AaronSw> Maybe he meant .COMFIDENTIAL

21:04:02 <DanC> say... did you guys see this YADS stuff, 29May to uri@w3.org? http://www2.elsevier.co.uk/~tony/yads

21:04:54 <DanC> frap; galeon crashed

21:04:59 <DanC> I think I'm gonna call it a day

21:14:45 <jordan> what can I use to render RDF to XHTML? rdfweb.org seems to do this

21:17:16 <danbri> computer programs! specifically, rdf apis and query languages. Take a look at Jena or Redland, they're pretty hands toolkits that do this.

21:19:24 <jordan> redland does this? hmm.. i'll look into it

21:20:53 <Seth> danbri, are there win32 binaries for readlands yet?

22:53:31 <JosD>http://www.cs.gc.cuny.edu/~sartemov/Recent_Lectures_slides/NYAS.pdf

22:53:31 <dc_rdfig> I: http://www.cs.gc.cuny.edu/~sartemov/Recent_Lectures_slides/NYAS.pdf from JosD

22:54:11 <JosD> I:|Logic in Computer Science: History and Recent Developments

22:54:12 <dc_rdfig> titled item I

22:54:31 <JosD> I:Sergei N. Artemov CUNY Graduate Center

22:54:31 <dc_rdfig> added comment I1

22:54:59 <JosD> I:May 7, 2002 Talk at the New York Academy of Sciences

22:54:59 <dc_rdfig> added comment I2

23:06:59 <jordan>http://www.webreference.com/xml/column13/

23:06:59 <dc_rdfig> J: http://www.webreference.com/xml/column13/ from jordan

23:07:10 <jordan> J:|Creating RSS files for your website

23:07:10 <dc_rdfig> titled item J

23:08:34 <JosD> I:...Logic of Proofs = classical logic + additional atoms proofPolynomial:Formula

23:08:34 <dc_rdfig> added comment I3

23:21:41 <JosD> I:...proofPolynomial constructed from atomic proofs by elementary computable operations corresponding to modus ponens, proof checker and nondeterministic choice

23:21:41 <dc_rdfig> added comment I4

23:27:22 <jordan> it seems that http://www.w3.org/RDF/Validator/ fails on all RDF but its own example