Semantic Web Interest Group IRC Chat Logs for 2001-12-12

00:00:26 <sbp> DanC, try http://rdfig.xmlhack.com/index.html - you lost 2/3 of the quote

00:02:35 <DanC> nothing essential

00:03:03 <DanC> sigh... out of time for the day

01:06:21 * sbp manages to get his homepage generated fairly smoothly by CWM from N3 source and N3 transform

01:14:16 <AaronSw> wow, cool

02:34:29 <sandro> let's see!

02:36:08 <AaronSw> He probably did:

02:36:20 <AaronSw> :Homepage log:outputString """[his whole homepage]""" .

02:39:06 <sbp> heh, nope

02:39:46 <sbp> I tried to convert as much of the data as possible into proper triples... obviously, a lot of the larger text chunks were impossible to split up

02:40:06 <sbp> But the lists, address data, headings, navbar, and metadata etc. were all quite easy to do

02:42:34 <AaronSw> Please share.

02:42:54 <sbp> I'm still hacking on it... got distracted

02:46:30 <sandro> How long does it take to generate the html?

02:46:43 <sbp> about 5-10 seconds

02:46:59 <sbp> actually, probably less than that - I also get the --think output

03:04:44 * tansaku looks around for danbri

03:09:21 <tansaku> hi Aaron - I'll ask you here since I sort of mentioned this to some people here before

03:09:44 * tansaku is coming from swhack where he was already talking to Aaron

03:10:17 <AaronSw> ok

03:10:19 <tansaku> So the root of all this is wanting to be able to describe meta-data relative to different authors

03:10:38 <AaronSw> authors of metadata, right?

03:10:43 <tansaku> right

03:11:18 <tansaku> e.g. I think that this webpage is, like(sic), really related to the keyword britney - say 5/10

03:11:24 <AaronSw> so meta-meta-data

03:11:30 <tansaku> webpage --> URI

03:11:40 <tansaku> and you think that it is only related 1/10

03:11:44 <tansaku> meta-meta-data

03:11:52 <tansaku> eventually we'll get to meta^data

03:12:04 <AaronSw> :-)

03:12:05 <tansaku> or should that be meta^N:date

03:12:09 <tansaku> anyway

03:12:21 <tansaku> so as we talked about before, describing it like this:

03:12:51 <tansaku> <ng:related_to ng:meta_author="sam" ng:strength="5/10" >Britney</ng:related_to>

03:12:55 <tansaku> is not legal RDF

03:13:08 <tansaku> so I was speaking to Blanu and he was suggesting something like this

03:13:39 <tansaku> assume that we've given the statement "blah keyword britney" the

03:13:39 <tansaku> URI "X".

03:14:09 <tansaku> where blah is the original resource we're refering to

03:14:23 <tansaku> blanu said let's represent that like this: "X: blah keyword britney"

03:15:01 <tansaku> and then add new statements like thi

03:15:03 <tansaku> this:

03:15:04 <tansaku> Y: X strength 5/10

03:15:04 <tansaku> Y accordingTo Sam

03:15:04 <dc_rdfig> Label Y not found.

03:15:14 <tansaku> sorry dc-rdfig

03:15:25 <AaronSw> The obvious way to represent <ng:related_to ng:meta_author="sam" ng:strength="5/10" >Britney</ng:related_to> is as <ng:related_to ng:meta_author="sam" ng:strength="5/10" rdf:value ="Britney" />

03:15:47 <tansaku> is that legal?

03:15:51 <AaronSw> yep

03:16:05 <tansaku> hmm

03:16:32 <tansaku> I like that because its succinct, and I was beginning to get worried about the processing overhead on all this XML

03:16:57 <AaronSw> it boils down to:

03:17:08 <AaronSw> <x> ng:relatedTo _:something

03:17:15 <AaronSw> _:something metaAuthor "sam"

03:17:21 <AaronSw> _something strength "5/10"

03:17:24 <AaronSw> in triples

03:17:31 <tansaku> blanu was suggesting that statement-level control over how metadata is a good thing

03:17:44 <tansaku> and I guess what you're suggesting is also just statements

03:18:07 <tansaku> which leaves me confused about how the triples relate to this nested structure we see

03:18:28 <tansaku> in things like <rdf:description rdf:about="#A0">

03:18:42 <AaronSw> well:

03:19:03 <AaronSw> <rdf:description rdf:about="foo">

03:19:03 <AaronSw> <foo:bar>baz</>

03:19:04 <AaronSw> </>

03:19:05 <AaronSw> is just:

03:19:13 <AaronSw> <foo> foo:bar "baz" .

03:19:22 <tansaku> I'll have to write that down

03:19:36 <AaronSw> if you spit it into the rdf validator, you can get the ntriples back

03:19:41 <tansaku> or at least cut and paste :-)

03:19:57 <AaronSw> :-)

03:21:14 <tansaku> there are two foo's in your above statement - do we distinguish between the two of them?

03:21:30 <AaronSw> two foos?

03:21:34 <AaronSw> <foo> is a URI

03:21:48 <AaronSw> foo:bar is an abbreviated uri meaning http://foo.org/something/goes/here/bar

03:21:54 <AaronSw> it's just namespaces

03:22:46 <AaronSw> does that make sense?

03:22:46 <tansaku> okay, but say I want to describe the URI http://britney.com but I was to use a predicate from http://ng.com/something/predicate

03:23:07 <AaronSw> <http://britney.com> <http://ng.com/something/predicate> "foo" .

03:23:08 <tansaku> I understand the namespaces - I'm just trying to translate blanu's example into triples

03:23:36 <tansaku> and not doing immensely well

03:24:26 <tansaku> okay - so then blanu's thing becomes <rdf:description:about="http://britney.com">

03:24:48 <tansaku> <ng:relates>Britney</>

03:24:55 <tansaku> </>

03:25:07 <AaronSw> it's easier if we ignore the serialization for now

03:26:14 <tansaku> sure - it's just that I think I need to understand the transformation to make sense of all this (I have different bits of syntax floating in my head)

03:26:46 <tansaku> I understand blanu's example perfectly - but I can't see how to perform the transformation

03:27:09 <AaronSw> can you write it out in triples?

03:27:24 <tansaku> it was above, but here it is again -

03:27:45 <tansaku> X: blah keyword britney

03:27:45 <tansaku> Y: X strength 5/10

03:27:45 <tansaku> Y accordingTo Sam

03:27:46 <dc_rdfig> Label X not found.

03:27:46 <dc_rdfig> Label Y not found.

03:27:57 <AaronSw> what are those X and Ys?

03:28:12 <sbp> reification, dear fellow...

03:28:18 <tansaku> well blanu said that X is a URI assigned to "blah keyword britney"

03:28:27 <AaronSw> ugh, reification

03:28:31 <tansaku> indeed

03:28:36 <AaronSw> ok, that makes sense

03:28:40 <tansaku> I understand the concept perfectly

03:28:53 <tansaku> I'm just trying to work out the long-hand representation

03:28:59 <AaronSw> yeah, so that makes sense.

03:29:05 <AaronSw> so it'd be:

03:29:24 <AaronSw> x rdf:subject blah; rdf:predicate keyword; rdf:object britney .

03:30:13 <AaronSw> similarly with y

03:30:18 <tansaku> that makes sense

03:30:34 <tansaku> but can that be transformed into the even longer format

03:30:39 <AaronSw> right

03:30:41 <AaronSw> so like:

03:30:44 <tansaku> with about=""etc

03:30:53 <AaronSw> <rdf:Description rdf:about="x">

03:30:53 <AaronSw> <rdf:subject>blah</></>

03:32:21 <tansaku> doesn't it come out to:

03:32:22 <tansaku> <rdf:Description rdf:about="x">

03:32:22 <tansaku> <rdf:subject>blah</>

03:32:22 <tansaku> <rdf:predicate>keyword</>

03:32:22 <tansaku> <rdf:object>britney</>

03:32:23 <tansaku> </>

03:32:30 <AaronSw> approximately

03:32:43 <tansaku> you were just short-handing?

03:32:53 <AaronSw> yeah

03:32:58 <tansaku> so then to achieve what we were talking about before we'd do this

03:33:40 <tansaku> <rdf:Description rdf:about="x">

03:33:40 <tansaku> <rdf:subject>http://britney.com</>

03:33:40 <tansaku> <rdf:predicate>ng:keyword</>

03:33:40 <tansaku> <rdf:object>britney</>

03:33:40 <tansaku> </>

03:33:46 <tansaku> followed by

03:34:13 <tansaku> <rdf:Description rdf:about="y">

03:34:13 <tansaku> <rdf:subject>x</>

03:34:13 <tansaku> <rdf:predicate>ng:strength</>

03:34:13 <tansaku> <rdf:object>5/10</>

03:34:13 <tansaku> </>

03:34:39 <tansaku> and we might replace x with #A0 in order to refer within the same XML document

03:35:21 <tansaku> but have to assign a concrete URI to x such as http://www.ng.net/somewhere/a_statement01 in order to make the pieces independent

03:35:59 * tansaku feels sorry for taking up so much space

03:36:19 * tansaku feels sorry for having killed sbp

03:37:07 <tansaku> So my final niggling doubt (which I'll try to answer for myself) is about how the long hand form that we see

03:37:42 <tansaku> the ones that don't include all three subject predicate and object - what they mean in terms of triples

03:37:59 <tansaku> e.g.

03:37:59 <tansaku> <rdf:description rdf:about="X">

03:37:59 <tansaku> <X:keyword>Britney</>

03:37:59 <tansaku> </>

03:38:03 <sbp>http://purl.org/net/sbp/

03:38:03 <dc_rdfig> A: http://purl.org/net/sbp/ from sbp

03:38:13 <tansaku> thanks sbp

03:38:14 <sbp> A:|A Homepage Of Sean B. Palmer

03:38:14 <dc_rdfig> titled item A

03:39:02 <tansaku> but I think I worked out that what was confusing me is that we can represent the triple two ways in this long format

03:39:29 <sbp> A:Now generated from [http://www.w3.org/DesignIssues/Notation3|Notation3] source: [http://infomesh.net/sbp/home.n3|home.n3], by an N3 transformation: [http://infomesh.net/sbp/hometoxhtml.n3|hometoxhtml.n3]

03:39:29 <dc_rdfig> commented item A

03:39:32 <sbp> thanks for what?

03:40:00 <tansaku> sbp: sorry, I thought you were recommending a page to me to help with by stuff - my mistake

03:40:08 * tansaku looks embaressed

03:40:44 <sbp> Heh, heh... sorry to paste over whilst you were speaking. But you just appeared to be rambling to yourself, so... :-)

03:40:52 <tansaku> thanks Aaron - you really helped me out, as always

03:40:57 <AaronSw> no prob

03:41:02 <tansaku> sbp: your analysis was perfect

03:41:39 <tansaku> and my summary ramble: is that there are two ways to use the long format - 1. to state a triple

03:41:52 <tansaku> 2. to make a statement about a triple (i.e. reification)

03:42:13 <AaronSw> yep

03:43:59 <sbp> A:Data vs. documentation appears to be about 50:50. I did quite well to get it to that level, I reckon. Maintainance of the page is a little trickier, but I can run .sh scripts to get the transformation going

03:43:59 <dc_rdfig> commented item A

03:44:19 <AaronSw> Ah, so that's what you wanted the script for.

03:44:34 <sbp> Yep :-)

03:45:41 * tansaku wanders off to try and work out how to get Jena to do all this reification stuff ....

03:47:16 <AaronSw> reification is just like a normal triple

03:55:09 <tansaku> so I just created a long-hand version of the stuff above and ran it through the RDF validator

03:55:15 <tansaku> it seemed to like it okay

03:55:43 <tansaku> I can't point you to that page - hmmm - but I just want to check one thing

03:56:03 <tansaku> <?xml version="1.0"?>

03:56:03 <tansaku> <rdf:RDF xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#"

03:56:03 <tansaku> xmlns:dc="http://purl.org/dc/elements/1.1/"

03:56:03 <tansaku> xmlns:ng="http://www.neurogrid.net/ng/elements/1.0/">

03:56:03 <tansaku> <rdf:Description rdf:about="#A0">

03:56:03 <tansaku> <rdf:subject>http://britney.com/</rdf:subject>

03:56:05 <tansaku> <rdf:predicate>dc:subject</rdf:predicate>

03:56:07 <tansaku> <rdf:object>Britney</rdf:object>

03:56:09 <tansaku> </rdf:Description>

03:56:12 <tansaku> <rdf:Description rdf:about="#A1">

03:56:14 <tansaku> <rdf:subject>#A0</rdf:subject>

03:56:15 <tansaku> <rdf:predicate>dc:subject</rdf:predicate>

03:56:18 <tansaku> <rdf:object>Britney</rdf:object>

03:56:19 <tansaku> </rdf:Description>

03:56:21 <tansaku> <rdf:Description rdf:about="#A2">

03:56:23 <tansaku> <rdf:subject>#A1</rdf:subject>

03:56:25 <tansaku> <rdf:predicate>ng:accordingTo</rdf:predicate>

03:56:28 <tansaku> <rdf:object>Sam</rdf:object>

03:56:29 <tansaku> </rdf:Description>

03:56:32 <tansaku> </rdf:RDF>

03:56:33 <tansaku> in the above - it's fine to refer to namespaces outside the angle brackets - like ng:accordingTo ?

03:56:51 <AaronSw> no, it's not

03:56:57 <tansaku> oops

03:57:03 <AaronSw> you'd need to do:

03:57:17 <AaronSw> <rdf:predicate rdf:resource="http://longURI.org/foooooo" />

03:57:31 <tansaku> or

03:57:57 <tansaku> <rdf:predicate rdf:resource="ng:accordingTo" /> if the namespace was defined properly?

03:58:02 <AaronSw> nope

03:58:06 <AaronSw> that's no good

03:59:04 <tansaku> <rdf:predicate rdf:resource="http://ng.net/elements/accordingTo" /> then? no short-hand?

03:59:10 <AaronSw> yep

03:59:17 <tansaku> ludy hell

03:59:24 <AaronSw> heh heh

04:00:36 <tansaku> although you said earlier I could use the short hand <ng:related_to ng:meta_author="sam" ng:strength="5/10" rdf:value ="Britney" />

04:00:46 <tansaku> and put each triple on a single line?

04:02:00 <sbp> Try <rdf:Description rdf:subject="x" rdf:pred[...] />

04:02:34 <tansaku> is there an example showing all the different description formats on the web - I've been having a little look, but ...

04:02:53 <sbp> description formats?

04:03:00 <AaronSw> you're probably best off ignoring the XML bit until much later

04:03:18 <AaronSw> BTW, you can type N-Triples into http://infomesh.net/2001/cwm/

04:03:20 <AaronSw> and it'll give you back xml

04:03:20 * AaronSw disconnects

04:03:34 <sbp> good point!

04:03:49 * AaronSw reconnects

04:03:59 <sbp> under "Paste 'n' Go"

04:04:03 <sbp> wb

04:04:24 <tansaku> cool - thanks

04:04:45 <tansaku> although the XML bit is crucial as I'm trying to build a system that passed bits of xml around

04:05:34 <sbp> Aaron is the second most likely person to dissuade you from using XML RDF

04:05:44 <AaronSw> Who's the first?

04:05:46 <sbp> me

04:05:51 <AaronSw> Hmm.

04:05:55 <sbp> :-)

04:06:04 <AaronSw> I think I'm firster than you.

04:06:18 <sbp> I dunno; I'm pretty firster

04:06:28 <tansaku> you reckon I should leave things in N3 or something and pass them around like that?

04:06:34 <AaronSw> Yeah!

04:08:02 <tansaku> I read with interest from the N3 page "This is not designed as an alternative to RDF's XML syntax which has the fundamental advantage that it is in XML"

04:08:19 <AaronSw> Yeah, that's just to trick you.

04:08:21 <sbp> N3/NTriples (mix) is only better than XML RDF if you're writing it, reading it, or parsing it

04:08:30 <AaronSw> :-)

04:08:32 <sbp> for all other purposes, use XML RDF

04:09:15 <deltab> note that is *the*, not *a*, advantage :-)

04:09:16 <tansaku> fair enough - but what about the rest of the people using blanu's RDF/XML based tristero

04:09:28 <sbp> heh, good point deltab

04:09:41 <tansaku> in fact the XML parsing speed sucks so badly I was looking for an alternative

04:10:09 <tansaku> I guess you don't have a Java N3 parser yet?

04:10:18 <sbp> .google Java N3 parser

04:10:19 <_xena_> Java N3 parser: http://www.agfa.com/w3c/euler/Parser.html

04:10:22 <AaronSw> I'm convincing him to use N-Triples, tansaku

04:10:37 <tansaku> aha, maybe I'm getting ahead of the game for once :)

04:11:23 <sbp> "_xena_"? Shouldn't that be (if __name__==) "__xena__"

04:11:29 <AaronSw> Heh heh heh.

04:13:48 <tansaku> looks like I have some reading to do - catcha later

04:14:01 * tansaku mumbles something about python fanatics

04:14:07 <AaronSw> Heh heh heh.

04:14:10 <sbp> heh, heh

04:34:52 * AaronSw disconnects

04:34:55 * AaronSw reconnects

04:53:43 <SeanP> SeanP is now known as sbp

05:01:13 <AaronSw> @ http://blogspace.com/swhack/weblog/2001/12/11/

05:01:28 <AaronSw> oops

05:01:29 <AaronSw>http://blogspace.com/swhack/weblog/2001/12/11/

05:01:29 <dc_rdfig> B: http://blogspace.com/swhack/weblog/2001/12/11/ from AaronSw

05:01:40 <AaronSw> B:|World Wide Web History Extravaganza!

05:01:41 <dc_rdfig> titled item B

05:02:05 <sbp> lol @ "@ URI"

05:02:26 <AaronSw> B:We use the Google Groups archives to dig up some of the WWW history.

05:02:26 <dc_rdfig> commented item B

05:02:37 <tansaku> here's a test for you - what's wrong with this N3 statement: [["Britney" is dc:subject of "http://britney.com/"] has ng:strength "5/10"] ng:accordingTo "Sam" .

05:03:13 <AaronSw> Put <> around URIs not ""s...

05:03:19 <AaronSw> get rid of the has

05:04:02 <sbp> [ ] can only have po inside it

05:04:15 <AaronSw> Ah, yeah.

05:04:44 <AaronSw> He's looking for N3 reification, which is sorta missing.

05:04:53 <sbp> { { "BritneyBrat" is dc:subject of <http://britney.com/> } ng:strength "5/10" } ng:accordingTo [ :named "Sam" ] .

05:05:04 <AaronSw> lol

05:05:13 <tansaku> aha, no reification in N3?

05:05:24 <sbp> I reckon they should use '' for reification

05:05:28 <AaronSw> Well, {} is reification, but it doesn't convert to XML so well.

05:05:34 <sbp> ' ' "BritneyBrat" is dc:subject of <http://britney.com/> ' ng:strength "5/10" ' ng:accordingTo [ :named "Sam" ] .

05:05:43 <AaronSw> at least CWM doesn't convert it to xml so well

05:05:43 <sbp> Hmm... mega confusing. That doesn't work

05:05:49 <AaronSw> Yeah, it is.

05:06:04 <sbp> <q></q>

05:07:05 <tansaku> sbp: po?

05:07:13 <sbp> predicate/object pair

05:08:14 <tansaku> { { "BritneyBrat" is dc:subject of <http://britney.com/> } ng:strength "5/10" } ng:accordingTo [ :named "Sam" ] .

05:08:28 <tansaku> doesn't parse at http://www.w3.org/DesignIssues/Notation3.html

05:08:45 <tansaku> should the { be [?

05:08:50 <sbp> well it does according to me

05:08:56 <tansaku> weird

05:08:57 <sbp> no

05:09:19 <sbp> try using http://infomesh.net/2001/cwm/

05:09:49 <DanC> tansaku, are you familiar with english grammar? stuff like independent clauses, dependent clauses and such? if so, think of {} as an independent clause (with a subject) and [] as a dependent clause.

05:10:23 <sbp> ta da: http://swag.webns.net/n3tordf?text=@prefix+%3A+%3C%23%3E+.%0D%0A%0D%0A%7B+%7B+%22BritneyBrat%22+is+%3Asubject+of+%3Chttp%3A%2F%2Fbritney.com%2F%3E+%7D+%3Astrength+%225%2F10%22+%7D+%3AaccordingTo+%5B+%3Anamed+%22Sam%22+%5D+.

05:10:35 <DanC> Fred said that Jose likes beans ==n3==> :Fred :said { :Jose :likes :beans }.

05:10:37 <AaronSw> The %%%%s are blinding.

05:10:56 <tansaku> gives me an error

05:10:56 <sbp> don't stare straight at them, Aaron

05:11:04 <tansaku> DanC: thanks, that helps

05:11:09 <AaronSw> I get an error at that page, sbp

05:11:17 <AaronSw> notation3.BadSyntax: Line 8: Bad syntax (expected '.' or '}' or ']' at end of statement) at ^ in: "...h "5/10" } :accordingTo [ :named "Sam" ]^

05:11:17 <AaronSw> "

05:11:37 <AaronSw> hmm, my browser must've chopped off the period

05:11:37 <sbp> Hmm... try adding the dot at the end

05:11:43 <AaronSw> that's better

05:12:13 * DanC thinks it's kinda funny that the parser diagnostics in notation3.py have survived so long.

05:12:36 <sbp> yeah, pretty robust

05:12:42 * AaronSw thinks it's kinda funny that there are diagnostics.

05:12:52 <DanC> I expected somebody to turn them into something emacs-happy by now, so that you could use M-x next error

05:12:53 <tansaku> sbp: can you just paste here the full N3 you used to generate that

05:13:32 <sbp> depending on your browser, chop "http://swag.webns.net/n3tordf?text=" out, and use "about:" or "data," instead

05:13:47 <AaronSw> .py import urlparse

05:14:01 <AaronSw> hmm

05:14:42 <DanC> just to complete a thought:

05:14:45 <DanC> Fred said that Jose likes beans ==n3==> :Fred :said { :Jose :likes :beans }.

05:14:48 <DanC> contrast with:

05:15:29 <DanC> Fred likes something blue ==> :Fred :likes [ :color :blue ].

05:15:45 <AaronSw> '@prefix : <#> .\015\012\015\012{ { "BritneyBrat" is :subject of <http://britney.com/> } :strength "5/10" } :accordingTo [ :named "Sam" ] .'

05:16:03 <tansaku> sbp: thanks dude - although that still doesn't work in http://www.w3.org/DesignIssues/Notation3.html - hmmm

05:16:10 <tansaku> but works at CVM

05:16:11 <AaronSw> it's an older version

05:16:15 <sbp> yep

05:16:21 <AaronSw> at DesignIssues, that is

05:16:39 <tansaku> DanC: aha - makes sense

05:17:06 <tansaku> sbp/Aaron: but say that I want to have multiple people referring to the same thing

05:17:19 * DanC collects one jedi-master point for giving tansaku an N3-aha

05:17:35 <tansaku> say { "sucks" is :description of <http://britney.com/> }

05:17:42 * sbp notes that the topic is still "Semantic Star Wars"

05:17:46 <tansaku> is according to like 150 people

05:17:59 * tansaku ducks to avoid a TIE Interceptor

05:18:04 <DanC> { "sucks" is :description of <http://britney.com/> } :accordingTo :Fred, :Bob, :Susie.

05:18:41 <tansaku> and so when each person thinks it sucks to different percentile points

05:19:11 <DanC> oh... that probably won't fit on one line.

05:19:42 <sbp> Yes it will: N3 doesn't recognize linebreaks as part of the syntax

05:19:44 <DanC> it would be more naturally expressed in something that handled n-ary relations.

05:19:45 <tansaku> { { "sucks" is :description of <http://britney.com/> } :strength "5/10" }:accordingTo :Fred.

05:19:45 <tansaku> { { "sucks" is :description of <http://britney.com/> } :strength "6/10" }:accordingTo :Bob.

05:19:46 <sbp> :-)

05:20:46 <tansaku> we can't reify { "sucks" is :description of <http://britney.com/> } as like #A0 and then have other statements that refer to it?

05:21:33 <sbp> yes... but formulae in N3 can't have ID's. You can say that something is equivalentTo them, though

05:21:57 <tansaku> so I'd start with

05:22:00 <DanC> well, you can say: :S1 = { "sucks" is ...}. and then use :S1. but... well... using = is pretty klunky.

05:22:11 <sbp> [ a :Rating; :by :Bob; :body <#A0>; :strength "5/10" ] .

05:22:32 <tansaku> hmm, I'm glad I'm logging this

05:22:41 <AaronSw> don't worry, logger is too

05:22:41 <sbp> it's logged to the Web anyway

05:22:43 <AaronSw> logger, hello

05:22:50 <sbp> logger, where am I?

05:22:52 <sbp> See http://ilrt.org/discovery/chatlogs/rdfig/2001-12-12#T05-22-50

05:22:52 <DanC> logger, pointer?

05:22:52 <DanC> See http://ilrt.org/discovery/chatlogs/rdfig/2001-12-12#T05-22-52-1

05:22:56 <AaronSw> heh heh heh

05:22:59 <sbp> :-)

05:23:24 * DanC gives sbp a high-five.

05:23:30 <tansaku> I mean I have all this already represented in the database and I was just trying to create an RDF/XML tristero interface so other people could access it and we could plug in different interfaces

05:23:45 <DanC> oooh... yes, please.

05:24:14 <DanC> if you want to do it in RDF/xml, you'll probably need to stay away from {}-ish stuff, though.

05:24:18 <sbp> interesting how logger edits the logs... "05:22:52 <sbp> See http://ilrt.org/[...]"

05:24:33 <tansaku> but the Jena RDF/XML parser is really slow, and as I tried to describe everything the XML got bigger and bigger

05:24:57 <tansaku> so I was looking for a compact representation that could be generated and parsed slightly faster

05:26:18 <sbp> Is there a speed-test comparison chart for the top XML/RDF parsers somewhere?

05:26:28 <sbp> .google speed comparison RDF parsers

05:26:29 <_xena_> speed comparison RDF parsers: http://www.xml.com/pub/a/Benchmark/exec.html

05:27:29 <tansaku> so just to return to the N3 for a second I could do something like this:

05:27:38 <tansaku> :A0 = { "sucks" is :description of <http://britney.com/> }

05:27:38 <tansaku> [ a :Rating; :by :Bob; :body <#A0>; :strength "5/10" ] .

05:28:09 <tansaku> .google speed comparison RDF parsers Jena

05:28:10 <_xena_> speed comparison RDF parsers Jena: http://ilrt.org/discovery/chatlogs/rdfig/2001-10-19.html

05:28:15 <tansaku> bugger

05:28:33 <tansaku> aa, although a different log apparently

05:28:34 <sbp> heh, it's great when that happens

05:29:11 <DanC> I'm often suprised at how highly google rates discussions that I thought were relatively obscure.

05:29:31 <DanC> hmm... say that in N3.

05:30:59 <DanC> this log:forSome :x. { :Google :thinks { :x :veryRelevantTo [] }. { :Dan :thought { :x a :RelativelyObscure } }.

05:31:01 <sbp> _:DanC a :Person; :nick "DanC"; :surprisedAt [ a :GoogleIdiom; :overrates ?x ] . ?x a :Discussion; :thoughtIrreleventBy _:DanC .

05:31:23 * tansaku thinks "oh joy"

05:31:37 <tansaku> CVM won't parse

05:31:46 <tansaku> :A0 = { "sucks" is :description of <http://britney.com/> }

05:31:46 <tansaku> [ a :Rating; :by :Bob; :body <#A0>; :strength "5/10" ] .

05:31:47 <DanC> hmm... "often" is difficult to capture.

05:32:08 <DanC> no? what diagnostic do you get, tansaku?

05:32:13 <sbp> yeah... forOften

05:32:55 <tansaku> ah, I think it needed a full stop after the first line - I see no linebreaks, everything separated by periods

05:33:31 <tansaku> woo - that's big in xml

05:34:16 <DanC> er... cwm writes out something pretty goofy for {...} when you ask it to convert it to RDF/xml.

05:34:43 <DanC> hm... the cgi connected to /DesignIssues/Notation3 doesn't seem to like "literas" as subjects.

05:35:03 <tansaku> I just want to use something like - <ng:related_to ng:meta_author="sam" ng:strength="5/10" rdf:value ="Britney" />

05:35:07 <sbp> CWM could model the contexts per model.n3, but wisely chooses not to...

05:35:24 <tansaku> somehow it feels more comfortable

05:35:51 <tansaku> I mean I think I could get Jena to generate that - maybe - and it would be quite fast to parse

05:36:14 <DanC> hm... jena will generate propAttrs?

05:36:19 <tansaku> the Java N3 parser that xena suggested was just that - no generation

05:36:38 <tansaku> DanC: maybe not - I've been surprised by some of its limitations

05:36:42 * DanC watches evolution slowly paint the screen...

05:37:09 <tansaku> I'm beginning to think it would be faster to write my own parser and generator using the new java1.4 regex

05:42:46 <tansaku> hmmm

06:15:17 <AaronSw> I heard that!

06:15:19 <AaronSw> I am not old.

06:15:54 <sbp> Heard what?

06:16:20 <AaronSw> Oops.

06:16:31 <AaronSw> I was refering to:

06:16:32 <AaronSw> <sbp> [off] Now that the old man's gone, we can swear, and throw parties, and get drunk!

06:16:59 <sbp> blast him and his phantabulous ways!

06:30:07 <DanC> any idea how to tell evolution what browser to use?

07:07:50 <DanC> hmm... no bookmarklet support in galeon

07:11:49 <DanC> hmm... maybe there is...

08:46:46 <DanC>http://www.earlham.edu/~peters/hometoc.htm

08:46:46 <dc_rdfig> C: http://www.earlham.edu/~peters/hometoc.htm from DanC

08:46:54 <DanC> C:|Peter Suber

08:46:55 <dc_rdfig> titled item C

08:47:53 <DanC> C:I re-discovered his ["Logical Systems"|http://www.earlham.edu/~peters/courses/logsys/lshome.htm] course notes, 1st chumped [16 May|http://rdfig.xmlhack.com/2001/05/16/2001-05-16.html].

08:47:53 <dc_rdfig> commented item C

08:49:08 <DanC> C:I got curious about the author; discovered that he teaches "Computer Science and Legal Studies" which piqued my interest.

08:49:08 <dc_rdfig> commented item C

08:49:56 <DanC> C:an hour later, I find I'm reading all sort of stuff, not the least of which the [* Free Online Scholarship (FOS) Newsletter

08:49:56 <DanC> |http://www.earlham.edu/~peters/fos/index.htm]

08:49:56 <dc_rdfig> commented item C

08:50:17 <DanC> C:er... [FOS|http://www.earlham.edu/~peters/fos/index.htm]

08:50:18 <dc_rdfig> commented item C

08:54:28 <deltab> C:He's the original inventor of the self-modifying game [Nomic|http://www.earlham.edu/~peters/nomic.htm].

08:54:29 <dc_rdfig> commented item C

15:38:42 * DanC waves

15:38:51 <DanC> updating WebOnt membership records again.

15:39:06 <DanC> sigh.. this 200mhz machine really doesn't cut it. gotta upgrade soon!

15:42:55 <bijan> More WebOnt members!!!?!!

15:42:59 <bijan> Hey DanC.

15:43:22 <DanC> yes, more members

15:43:28 <bijan> Eeek!

15:43:36 <DanC> well, more folks from existing member orgs, at this point, I think

15:43:41 <bijan> You need a WebOntOntology.

15:44:04 <DanC> I have one. what do you think is keeping this 200mhz machine going for so long?

15:44:16 <DanC> I'm checking consistency between various membership records with cwm.

15:44:20 <bijan> :bob a woo:WebOntCoreMember. :sally a woo:WebOntNotSoCoreMember.

15:44:21 <bijan> heh.

15:44:28 <bijan> On a 200mhz machine! Brave man.

15:44:51 <bijan> Are you running CWM under Python >2?

15:44:57 <DanC> see http://www.w3.org/2001/sw/WebOnt/webont-roll.rdf

15:45:04 <DanC> 2.1 yes

15:45:23 <bijan> Ah! Cool.

15:46:44 * bijan fixing some stuff in n3parser.pl

15:46:59 <bijan> In anticpation of testing it against some real n3 files.

15:48:15 <bijan> Did you catch me plumbing the frightening depths of RDFStore.query(..) the other day?

15:48:37 <bijan> Boldly going where no one but tim has wanted to go before :)

15:53:10 <DanC> no.

15:53:18 <DanC> did you cross yourself before and after?

15:53:31 <bijan> I started an annotea commentary on llyn.py in an effort to document my journy.

15:53:44 <bijan> But I can't load the comments I write, and niether can anyone else :)

15:54:14 <bijan> I tried to get Tim to see the virtues of the State pattern.

16:07:54 <bijan> Hmm. is cwm.test() supposed to work? It uses a lot of old old old syntax.

16:08:40 <bijan> "bind default <blah blah blah>"

16:10:56 <bijan> Indeed, buggy old syntax :)

16:16:51 <bijan> Indeed, some things needed for the cwm.test() to run are commented out (e.g., the import StringIO statement)

16:20:45 <DanC> no, I'm afraid cwm.test() is vestigial

16:21:08 <bijan> Yep. Eliminate it! :)

16:21:26 <bijan> Actually, it's not *that* hard to get working. Just not very useful when you do :)

16:21:36 <DanC> should be replaced by a comment refering to retest.sh , after making sure the test data there doesn't have anything novel

16:22:09 <phaze3k> Hi people..

16:22:09 <bijan> retest.sh?

16:22:18 <DanC> where's jos when I need him? cwm is giving me surprising results... I really want an "oh yeah? why do you think so?" button... i.e. an euler proof

16:22:26 <bijan> Hehehe.

16:22:38 <DanC> retest.sh is the driver for the regression suite.

16:22:50 <bijan> Aha! but shell, not python based?

16:22:51 <DanC> swap/test/retest.sh

16:23:12 <bijan> Hmm. Would a eular proof help?

16:23:12 <DanC> yes, TimBL wrote it in sh

16:23:44 <DanC> I took timbl's sh code and factored out the common pattern, which should facilitate translation to python.

16:23:59 <bijan> I.e., won't there be stuff that eular wouldn't see do the forward chaining?

16:24:35 <DanC> actually, as cwm is, by design, a command-line tool, sh is a reasonable language for testing it, no?

16:24:50 <DanC> euler: not sure.

16:25:03 <bijan> Or, worse maybe, if the bug is earlier in the proof chain and a false positive (i.e., adding something that it shouldn't), eular might construct valid proofs on that buggy store.

16:25:31 <bijan> DanC: Sure! Except Python is an equally reasonable langauge for testing cmd line tools, and considerably more portable :)

16:27:05 <DanC> well, I don't have any motivation to eliminate a dependency on sh. But as I say, I did clean up the sh code that's there, which makes it straightforward to port to another language. contributions are welcome.

16:27:12 <bijan> Yep.

16:27:31 <bijan> I just put it on the "if I get round to it" list :)

16:27:45 * bijan trying to do a bit of cwm profiling.

16:27:46 <DanC> let's offer a bounty...

16:28:03 <bijan> Got a good test file to profile?

16:28:07 <DanC>http://www.w3.org/2000/10/swap/test/retest.sh

16:28:08 <dc_rdfig> D: http://www.w3.org/2000/10/swap/test/retest.sh from DanC

16:28:27 <DanC> D:retest.sh -- regression test driver for swap/cwm

16:28:27 <dc_rdfig> commented item D

16:28:35 <DanC> D:bijan noted it's not in python.

16:28:36 <dc_rdfig> commented item D

16:28:55 <DanC> D:I offer a 20 point bounty for a translation to python, sent to me, cc www-archive+n3bugs@w3.org

16:28:55 <dc_rdfig> commented item D

16:29:11 <DanC> ummm... we ran into a particularly slow test the other day...

16:29:29 * DanC thinks of keywords to grep

16:29:30 <bijan> Also poking around with cmd.py...be nice to have an n3 repl (not in irc :))

16:29:52 <DanC> logger, grep taking forever

16:30:13 <logger> I'm logging. I found 4 answers for 'taking forever'

16:30:14 <logger> 0) 2001-12-12 16:29:52 <DanC> logger, grep taking forever

16:30:15 <logger> 1) 2001-11-02 07:39:49 <DanC> but the last 80% is taking forever. I can't seem to decide what I'm after.

16:30:16 <logger> 2) 2001-03-15 06:02:05 <DanC> man... cwm is hopeless on any realistic data set. I chopped rfc-index.n3 down to about 20 entries, and it's still taking forever.

16:30:17 <logger> 3) 2001-03-15 05:33:36 <DanC> cwm is chewing on the version with " changed to '. taking forever.

16:30:18 <bijan> Actually, cmd.py's interface is *really* slick. verra nice. I'm a little surprised that it *doesn't* have an irc backend.

16:30:18 <DanC> logger, hello?

16:30:41 <DanC> logger, grep slow

16:30:44 <logger> I'm logging. I found 146 answers for 'slow' (showing 0...4)

16:30:45 <logger> 0) 2001-12-12 16:30:41 <DanC> logger, grep slow

16:30:46 <logger> 1) 2001-12-12 16:29:11 <DanC> ummm... we ran into a particularly slow test the other day...

16:30:47 <logger> 2) 2001-12-12 05:36:42 <DanC> * DanC watches evolution slowly paint the screen...

16:30:48 <logger> 3) 2001-12-12 05:24:33 <tansaku> but the Jena RDF/XML parser is really slow, and as I tried to describe everything the XML got bigger and bigger

16:30:49 <logger> 4) 2001-12-10 12:14:20 <dajobe> java is *so* slow

16:30:53 <dajobe> lol

16:30:59 <dajobe> logger is so slow too

16:31:05 <danbri> logger, grep stuck

16:31:09 <logger> I'm logging. I found 47 answers for 'stuck' (showing 0...4)

16:31:10 <logger> 0) 2001-12-12 16:31:05 <danbri> logger, grep stuck

16:31:11 <logger> 1) 2001-12-04 22:00:49 <DanC> hmm... stuck here:

16:31:12 <logger> 2) 2001-12-04 01:40:00 <bijan> i think you're stuck with explicit precedence rules, alas.

16:31:13 <logger> 3) 2001-12-03 00:43:10 <sbp> * sbp notes that he has key generation working in CWM, but it getting quite stuck

16:31:14 <logger> 4) 2001-12-02 00:51:06 <danb_lurk> logger should provably not get stuck in loops...

16:31:14 <bijan> Have you used profile.py with cwm?

16:31:41 <DanC> I haven't done any profiling. I think timbl has.

16:32:16 <bijan> Hmm. My conversation with him was making me suspicious that premature optimization was going on in query...perhaps I was mistaken.

16:33:05 <bijan> Just *my* pre-profile bad intuitions were getting violated :)

17:09:43 <bijan> Hmm. Is a suffixless qname an ok xml element name?

17:12:45 <deltab> suffixless? like what?

17:13:46 <bijan> A:

17:13:46 <dc_rdfig>http://purl.org/net/sbp/

17:13:47 <dc_rdfig> A Homepage Of Sean B. Palmer

17:13:48 <dc_rdfig> (sbp) Now generated from [http://www.w3.org/DesignIssues/Notation3|Notation3] source: [http://infomesh.net/sbp/home.n3|home.n3], by an N3 transformation: [http://infomesh.net/sbp/hometoxhtml.n3|hometoxhtml.n3]

17:13:49 <dc_rdfig> (sbp) Data vs. documentation appears to be about 50:50. I did quite well to get it to that level, I reckon. Maintainance of the page is a little trickier, but I can run .sh scripts to get the transformation going

17:13:50 <bijan> Oops.

17:13:53 <bijan> Like that :)

17:14:35 <deltab> no: http://www.w3.org/TR/1999/REC-xml-names-19990114/#NT-QName

17:15:11 <bijan> Thanks!

17:20:41 * danbri grins at "SGML (ugh! but standard)" in timbl's http://groups.google.com/groups?selm=6484%40cernvax.cern.ch www writeup

17:21:43 <DanC> hi danbri.

17:21:58 <danbri> hiya

17:22:11 * danbri continues to be impressed by Google

17:22:51 <DanC> indeed.

17:23:14 <DanC> in fact, I'm pretty much completely dependent on them to function. makes me a bit nervous, since they have no competition.

17:24:04 <DanC> i.e. some MegaCorp could just decide google is worth more to them than it is to the current google owners...

17:24:18 <DanC> ... then MegaCorp could take google disneyland (ala altavista) and I'd be hosed.

17:25:30 <danbri> 5 years ago I was forever switching search engines; now I don't bother looking. I assume some others must be good too (copycats at least)...

17:28:24 <sbp> Gmrpgmrpmgrmgh - my mail keeps bouncing when sent to w3c addresses...

17:29:01 <jang> speaking of which, we're a step closer (yeah, right) to spamming being illegal in the EU :-)

17:29:19 <jang> they actually made that stick with unsolicited faxes

18:35:17 <hendler> wierd problems with openprojects today - lets me in - waits a little while, tells me my ident is in use and then kicks me off - sorry for the ping pong...

18:36:22 <sandro> Does someone have a full NTriples I/O for prolog? (like doing unicode)?

18:38:56 <bijan> Nope.

18:39:23 <bijan> What prolog has unicode?

18:41:32 <deltab> Visual Prolog, Amzi Prolog, Win-Prolog

18:41:46 <bijan> Is that exhaustive?

18:42:01 <bijan> (Just curious)

18:43:18 <deltab> no

18:44:17 <bijan> Ah, Hmm.

18:44:28 <bijan> Amzi unicode support seems limited to strings, not atoms.

18:45:36 <deltab> it seems SWI-Prolog is going to have Unicode support

18:45:47 <bijan> i was looking more for more general support, i.e., predicates for char tests etc.

18:45:48 <bijan> really?

18:46:13 <bijan> That's cool.

18:46:49 <deltab> so I gather from http://www.swi.psy.uva.nl/projects/SWI-Prolog/mailinglist/archive/1061.html

18:47:10 <bijan> oops! Looks like there's a reasonable string lib with Amzi.

18:49:13 <sandro> SInce prolog strings are just lists of integers anyway, you kind of don't need any special support for unicode in prolog. (I don't need it in atoms, since those are just URIs.)

18:49:37 <bijan> sandro, except many/most prologs have a "String" type as well.

18:49:45 <bijan> Which is more like a standard string.

18:49:59 <sandro> Ah.

18:50:23 <bijan> The hard bit of usign normal prolog strings is getting all the input/ouput/read/write functions aware of a new encoding.

18:50:46 <bijan> (And character testing, etc.)

18:51:15 <bijan> *But*, If you just want to stuff your unicode into a "string" it shouldn't be that hard.

18:52:45 <bijan> But, no, there is no prolog n-triples read/writer afaiK :)

18:53:04 <deltab> hmm, Unicode support has been in SWIPL's TODO for 19 months

18:53:06 <bijan> On the list, though.

18:53:35 <bijan> is there a todo list? Where?

18:54:03 <bijan> They must do *something* with the xml parser.

18:58:03 <bijan>http://www.logic.at/multlog/

18:58:03 <dc_rdfig> E: http://www.logic.at/multlog/ from bijan

18:58:11 <bijan> E:|MUltlog

18:58:11 <dc_rdfig> titled item E

18:58:21 <bijan> E:"MUltlog is a system which takes as input the specification of a finitely-valued first-order logic and produces a sequent calculus, a natural deduction system, and clause formation rules for this logic." WackY!

18:58:22 <dc_rdfig> commented item E

19:06:24 <blanu_> blanu_ is now known as blanu

19:06:57 <blanu> Can anyone enlighten me on how to do reification with N-Triples?

19:07:23 <sandro> Have you tried looking at the output for the validator? It should show you.

19:08:37 <hendler> I will avoid asking the loaded question of why anyone would want to...

19:08:40 <bijan> Hmm. "how do do reification"? Beyond just triplizing the M&S stuff?

19:09:23 <sandro> Sure you will, Hendler. :-) (Answer: because RDF doesn't have nesting.)

19:10:24 <hendler> When I first pitched basing DAML on RDF, I had to reconcile myself to the fact that people owuld think it meant Î had an opinion on reification...

19:10:46 <blanu> Well, I have some statements about statements in my database. I need to serialize these to N-Triples.

19:12:03 <bijan> right, so generate the appropriate quartet (or triple) of triples, a la M&S.

19:12:30 <bijan> (If you're going the reification route)

19:14:29 <blanu> What is M&S? And what is the alternative route to reification for serializing statements about statements?

19:14:57 <bijan> RDF Model and Syntax.

19:15:28 <bijan> And there is no alternative, if you're sticking with RDF1.0

19:16:21 <blanu> Well I just need something that can be serialized in N-Triples. I'm not familiar with anything suitable which isn't RDF1.0.

19:16:54 <bijan> Ok, so what's the sticking point about serializing to n-triples a standard M&S reification?

19:16:56 <blanu> Can you give me an example of a reified statement in N-Triples?

19:17:29 <blanu> Ah, because I don't know how to do that. Also you implied there is an alternative, which I find interesting.

19:18:13 <bijan> Section 4 of M&S contains the stuff on reification.

19:18:45 <bijan> For each statement you reify you need to specifiy 4 triples (or three if you're willing to infer type):

19:18:59 <bijan> (For the statement named http://foo):

19:19:42 <bijan> <http://foo/> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/1999/02/22-rdf-syntax-ns#Statement>.

19:19:57 <bijan> (etc. etc. for rdf:subject, rdf:object, rdf:predicate)

19:20:33 <bijan> Bit tedious to type it all out, not having namespaces.

19:21:26 <blanu> Ah, I see. I find that quite tedious in general. But if that's how it must be.

19:21:58 <bijan> Well, that's what is RDF1.0 reification :)

19:22:07 <bijan> Which is what you asked for as far as I can gather :)

19:22:39 <blanu> Couldn't you get the same effect in the format allowed for an optional indentifier tab to be included in the triple (although that would make it a quintuple)?

19:22:49 <blanu> Yes, thank you, that's exactly what I wanted to know.

19:23:15 <blanu> Although I wish the syntax for doing that was more convenient.

19:23:29 <bijan> Er..wouldn't it make it 4-tuple rather than a 5-tuples?

19:24:04 <bijan> There's two issues (at least). Quoting and referring to statements.

19:24:24 <bijan> Which somewhat go together.

19:24:45 <bijan> If you just had a tag on triples, then you couldn't talk about a triple without asserting it.

19:24:49 <bijan> Which is often undesirable.

19:24:56 <blanu> bijan: Ah, yes, 4-tuples I meant to say.

19:25:04 <hendler> *sigh* - SW-CG agenda item (Jan) - Find co-chair for Webont; Find co-team for Webont ...

19:25:21 <bijan> AFAIK, a RDF1.0 reified statement is *not* asserted.

19:25:26 <bijan> But I'm unclear on that :)

19:26:16 <blanu> You could just assume that tagged statements aren't asserted.

19:26:43 <bijan> Eh...that opens up a whole new kettle of not very RDFy fish :))

19:27:23 <bijan> I mean, there are loads of standard solutions to all this, but they aren't (yet) in RDF.

19:28:46 <blanu> Not really. By using this crazy syntax you are in effect producing a triple with a tag. By using the crazy syntax you are avoiding asserting it. If you instead had a syntax which allowed for untagged asserted statements or tagged unasserted statements, you would achieve the same effect but with a simpler syntax.

19:30:48 <bijan> Not really? Nothing like that is in RDF.

19:31:51 <blanu> I mean it doesn't really open a whole kettle of fish. It's just a simpler syntax for doing reification. It doesn't change the RDF model at all.

19:31:55 <bijan> I'm leary about arguments that go "in effect" meaning "well, for some purposes this is equivalent"

19:32:13 <blanu> I mean it would alter the model in exactly the same way.

19:32:18 <bijan> Yes it does. quads instead of triples.

19:33:26 <blanu> You don't actually put a quad into the model, you would just allow a statement to be serialized as 4 values on a line instead of 12 values spread over 4 lines.

19:33:35 <bijan> Sure.

19:33:47 <bijan> But no one's done that :)

19:34:00 <bijan> n-triples isn't likely to do that.

19:34:34 <bijan> I mean, I can invent all sorts of convenient syntaxtical/serialization shortcuts. if all you want is to be able to serialize your knowledge base, there's lots of things to do.

19:34:45 <blanu> Right, and I think that's too bad because the N-Triples syntax is the simplest one and it's unnecessarily complex without any benefit. So I wish there was a simpler syntax.

19:34:52 <bijan> But I presume you want to write something that others can read.

19:35:15 <bijan> Actually, quads would *complicate* the syntax, because you'd add productions :)

19:35:29 <bijan> You mean "simpler for people to write " :)

19:36:09 <blanu> It's both simpler for people to read and simpler to write code to deserialize metastatements.

19:36:41 <bijan> *only* if your metastatements have a special representation in your datastore.

19:36:55 <blanu> Exactly so.

19:37:04 <bijan> Well, that's a rather special case.

19:37:17 <bijan> And people aren't supposed to read n-tripels (i.e., that's not a design goal)

19:38:01 <bijan> If i'm a plain jane rdf1.0 processor, having some quads I have to explode into a refication is more work, albeit not a heck of a lot, than just reading endless triples.

19:38:46 <blanu> Hm, I think you're optimizing for design goals which have little usefulness in actual applications.

19:38:59 <bijan> I'm not optimizing for anything, actually.

19:39:12 <bijan> Not being involved in *any* of these specs :)

19:39:24 <blanu> But hey, why argue. I have no hope of the w3c adopting my syntax so I might as well get back to coding.

19:39:32 <bijan> Good point.

19:52:06 <dajobe>http://www.coginst.uwf.edu/users/phayes/w3-rdf-mt-current-draft.html

19:52:06 <dc_rdfig> F: http://www.coginst.uwf.edu/users/phayes/w3-rdf-mt-current-draft.html from dajobe

19:52:18 <dajobe> F:|RDF Model Theory (draft) 2001-12-12, Pat Hayes

19:52:18 <dc_rdfig> titled item F

19:52:55 <dajobe> F:as usual, I've checked it [http://cvs.ilrt.org/cvsweb/redland/rdfcore/mt/|into CVS] so you can see the differences

19:52:55 <dc_rdfig> commented item F

19:53:19 <dajobe> F:to previous drafts which were published at the same URL

19:53:20 <dc_rdfig> commented item F

21:00:03 <DanC>http://www.coginst.uwf.edu/users/phayes/w3-rdf-mt-current-draft.html

21:00:03 <dc_rdfig> G: http://www.coginst.uwf.edu/users/phayes/w3-rdf-mt-current-draft.html from DanC

21:00:09 <DanC> G:new model theory draft

21:00:09 <dc_rdfig> commented item G

21:00:19 <dajobe> already there DanC

21:00:29 <dajobe> see F:

21:00:35 <DanC> G:issues: (a) datatypes. explicitly called out by the editor, but could be linked to issues list

21:00:40 <dc_rdfig> commented item G

21:00:40 <DanC> where? sorry for rechump

21:00:48 <DanC> F:issues: (a) datatypes. explicitly called out by the editor, but could be linked to issues list

21:00:48 <dc_rdfig> commented item F

21:01:43 <DanC> F:I'd like to see more motivation for the use of blank nodes. it just says "seems more in keeping with [RDFMS]" which doesn't do our extensive discussions justice.

21:01:43 <dc_rdfig> commented item F

21:01:55 <DanC> F:I've been asked for more justification of this point in SWAD meetings.

21:01:55 <dc_rdfig> commented item F

21:02:36 <DanC> F:issue: "since such value spaces as integers cannot be fully axiomatized in first-order logic.

21:02:37 <DanC> " seems to appear out of nowhere. Delete it or explain it better.

21:02:37 <dc_rdfig> commented item F

21:02:45 <DanC> F:" seems to appear out of nowhere. Delete it or explain it better.

21:02:46 <dc_rdfig> commented item F

21:03:35 <DanC> F:issue: "Herbrand Lemma. Any RDF graph has a satisfying interpretation.

21:03:36 <DanC> " this needs discussion, as we discovered in a telcon a couple weeks ago.

21:03:36 <dc_rdfig> commented item F

21:03:44 <DanC> F:" this needs discussion, as we discovered in a telcon a couple weeks ago.

21:03:44 <dc_rdfig> commented item F

21:04:01 <DanC> F:I think you can say false things in RDF. e.g. using ont:disjointFrom

21:04:02 <dc_rdfig> commented item F

21:04:53 <DanC> F:"tidying the resulting set" under 2.2 <- editorial left-over.

21:04:53 <dc_rdfig> commented item F

21:07:30 <DanC> F:"namespace entailment" <- don't like this term. layered entailment, vocabulary entailment, maybe. hmm...

21:07:31 <dc_rdfig> commented item F

21:08:25 <sbp> the term "namespace" should be banned from RDF Core publications, except by reference to namespaces in XML, IMO :-)

21:08:34 <DanC> yup

21:10:36 <DanC> F:RDFS reserved vocabulary <- actually, there is a semantic constraint for some of those other terms: isDefinedBy rdfs:subPropertyOf seeAlso

21:10:36 <dc_rdfig> commented item F

21:11:45 * sbp goes searching for QName structure stuff, and finds food for TAG in: http://lists.w3.org/Archives/Public/www-rdf-logic/2001Aug/0078

21:13:42 <DanC> G:|oops.. rechump.

21:13:42 <dc_rdfig> titled item G

21:15:09 <sbp> Hmm... so in general, W3C specification should be reviewed by DI, WAI PF, TAG, and I18N before they go to Rec.? Lots of dependencies

21:15:11 <DanC> logger, pointer?

21:15:11 <DanC> See http://ilrt.org/discovery/chatlogs/rdfig/2001-12-12#T21-15-11

21:16:47 <sbp> I wonder if sorting out the XPointer vs. RDF tension is also under the scope of TAG?

21:17:01 <DanC> I expect so

21:17:53 <sbp> Since XPointer only applies when you know that the MIME type of a resource is ...xml, it prevents content negotiation. I don't like it...

21:18:37 <sbp> I would have thought that a genericity to fragIDs should have been possible to create, with a little ironing of the specifications here and there, but XPointer made me think twice about that

21:20:30 <hendler> hmm, maybe the TAG should get one of these model theories done for the whole web (if nothing else, that would keep the model theory guys busy for a while!!)

21:20:42 <sbp> heh, heh, heh

21:27:21 * dajobe wonders if #rdfig is becoming normative

21:27:31 * sandro laughs

21:27:34 <dajobe> lol

21:29:13 <sbp> :-)

21:30:14 <dajobe> actually, I don't like that word - it is probably un-necessary jargon

21:30:48 <sandro> which word? normative?

21:31:07 <dajobe> yeah

21:31:31 <sandro> what's the non-jargon term? "governing?" "defininitve"

21:31:49 <hendler> power-hungry?

21:32:13 <dajobe> informative, but not part of the document, or required to be read in order to understand it

21:32:44 <sandro> isn't that non-normative?

21:32:52 <dajobe> oops

21:33:02 <dajobe> see; jargon!

21:35:33 <hendler>http://www.w3.org/2001/tag/

21:35:34 <dc_rdfig> H: http://www.w3.org/2001/tag/ from hendler

21:35:45 <hendler> H:|Technical Architecture Group

21:35:46 <dc_rdfig> titled item H

21:36:04 <hendler> H: TAG members elected/appointed - Dan Connolly to be a member!

21:36:05 <dc_rdfig> commented item H

21:36:10 <hendler> H: Way to go Dan

21:36:10 <dc_rdfig> commented item H

22:21:10 <DanC> hmm... I finally isolated the part of the rule that is giving me surprising results (in WebOnt membership).

22:21:19 <DanC> I don't understand why it's doing what it's doing.

22:28:42 <DanC> but I'm kludging around it for today...

22:29:21 <dajobe> as you said in http://lists.w3.org/Archives/Public/www-webont-wg/2001Dec/0060.html - "N3 scares me to. ;-)"

22:29:27 <dajobe> s/to/too

22:29:34 <DanC> yup

22:30:45 <dajobe> how many members has webont now?

22:31:10 <DanC> um... 47 plus a couple more that I'm adding now

22:32:02 <hendler> I make it out to be 49 or 50 depending on one pi mail we are waiting on follow up to.

22:32:30 <DanC> I still don't see any intro from Shimizu. Am I missing something?

22:40:00 <DanC> jim, I'm not scraping the ALT bits formally just now; I have published them, though. Remind me some time to scrape them.

22:42:11 * DanC added 2 new Ibrow folks, bringing us to 49

22:44:18 * DanC runs rules to check consistency between nominations, web page, and mailing list subscriptions again...

22:45:51 <DanC> hmm... i have to provide it web access credentials every time I run it. so making it a long-running agent presents certain challenges.

22:46:52 <DanC> I'd like to just sign something that authorizes it to run for, say, 18 months. But the recent PCA paper suggests that nonces and such are needed for proof-based access control. Hmm...

22:50:22 <dmiles> bijan: you here?

22:51:42 <dmiles> bijan: the tabling mechanism for SWI-prolog .. works :).. as long as SWI doesnt crash.. just sent bug over the mailing list.

22:52:11 <hendler> DanC - we've played a little bit with using cwm as an auth server - basically you would put a file in a trusted place with the rules about who could access, your system would query it for a token, and then that token gets decoded to go in the web access stuff.

22:52:45 <hendler> one of my ugrads has done a very primitive hack of this -- it seems to work.

22:53:26 <hendler> it gets around nonces and stuff by doing the external appeal for token, which I guess violates the PCA papers assumptions (since it introduces an external query)

22:54:19 <DanC> hmm... interesting. any details available?

22:54:31 <DanC> I don't suppose he's checking proofs, then?

22:55:50 <hendler> she, and we're not doing proof checking yet. The implementation she did is too ad hoc at this point - but she's working to improve it. It served mainly as a proof of concept, and is currently too weak to be used in a secure system (too easy to fake the auth tokens), but that's in part because we used no signatures or anything

22:56:46 <hendler> we're using it so far to check things like "who is allowed to page me" and things like that

22:57:24 <hendler> it's part of a hacked up service thing using jabber that some of the students in my class are doing - when they finish the projects, we'll post their stuff to the wiki.

22:58:46 <hendler> in fact, we want to keep the wiki open, but to have some sort of control over access, were thinking of using that for the proof checking stuff, but don't think it will be ready for prime time any time too soon (since students all go away from Dec 18 - Jan 28)

22:59:08 <hendler> family time - gotta go.