Semantic Web Interest Group IRC Chat Logs for 2001-09-05

00:05:02 Topic now Semantic Web Chat - http://www.w3.org/RDF/Interest/#irc

00:05:02 Users on #rdfig: logger tav` Bayta theran GabeW AaronSw lasDesk em deltab sandro ArtB [filsa] DanC_ bijan jang dc_rdfig eikeon

00:05:02 <ChanServ> This channel is logged and blogged: http://logicerror.com/rdfIRCWelcome

01:54:20 * AaronSw waves

02:03:50 * DanC_ tunes in for a bit

03:45:23 * AaronSw waves

03:59:27 <[filsa]> [filsa] is now known as filsa

03:59:37 * filsa waves...

04:00:15 <sbp> Hi filsa

04:00:25 <filsa> hi!

04:01:09 <filsa> so riddle me this...

04:01:21 <sbp> go on

04:02:03 <filsa> aaron suggested that i use structured properties

04:02:24 <filsa> but once i feed that rdf data into a rdf db, like redfoot

04:02:38 <filsa> that structure isn't captured anywhere, is it?

04:03:18 <filsa> is that structured property only for human benefit?

04:03:28 <filsa> is it possible to express an structured property in n3?

04:03:46 * filsa has had all day to digest feedback from this morning.

04:04:04 <AaronSw> Yes, filsa:

04:04:17 <AaronSw> :Message :to [ :name "bar" ; :email "baz" ] .

04:04:31 <AaronSw> it's all captured...

04:04:41 <AaronSw> :message :to _:genid

04:04:49 <AaronSw> _:genid :name "bar" .

04:04:54 <filsa> so a reddb like redfoot does capture it as well?

04:04:58 <AaronSw> yes

04:05:11 <sbp> damn anonymous nodes

04:05:13 <AaronSw> all rdfdbs will capture it

04:05:21 <AaronSw> err, most, i suppose

04:05:27 <filsa> qualitatively, though, what is the difference?

04:05:44 <filsa> is it a matter of convenience, or is there additional information that i gain/lose ?

04:05:59 <filsa> is there some sort of rule that says, i should use anon nodes here?

04:06:05 * filsa is full of questions

04:06:28 <DanC_> modelling email headers in RDF is remarkably subtle...

04:06:34 <AaronSw> anon nodes as opposed to what?

04:06:56 <filsa> as opposed to what i did. a straight list of nodes.

04:07:20 <AaronSw> Umm, you get that structured data about resources...

04:07:26 <AaronSw> it's like describing a new resource

04:07:28 <filsa> DanC_: agreed. i dont see any advantage in what i did w metamail vis-avis maillog2rdf.

04:07:37 <DanC_> e.g. the To header... the RFC822 syntax allows you to conclude, for example, that a message was *not* addressed to some address. To encode that in RDF, you have to use sequences (DAML style, in fact; RDF sequences are too broken).

04:07:39 <filsa> of course, i'm learning a hellalot

04:08:25 <filsa> okay, aaron, i see that. a new resource.

04:08:51 <filsa> but evidently i pull that structure out of thin air, cause the rdf vocab doesnt specify it.

04:09:18 <filsa> do you create an xml schema along side your rdf vocabs to handle the structure?

04:09:37 <AaronSw> Umm... no.

04:09:44 <filsa> should you?

04:09:50 <DanC_> er... how do I get a look at this thing? the chump entry leads me nowhere.

04:09:50 <sbp> no!

04:10:01 <filsa> sorry dns probs

04:10:03 <sbp> RDF takes the need for a "structure" in that sense away

04:10:15 <filsa> ... http://xmlns.filsa.net/metamail/0.1/ should work

04:10:26 <filsa> the .org address is broken pending netsol

04:10:44 <sbp> it maps on to the RDF graph, rather than a tree structure, so there's no need to validate the tree structure

04:11:13 <filsa> sbp: ok, but i'm trying to map what you say againstaarons recommendation to use a structured property.

04:12:00 <DanC_> filsa, do you have any example mail messages converted to RDF in this schema?

04:12:06 <filsa> yup

04:12:12 <filsa> ... demo is at http

04:12:23 <filsa> ... http://xmlns.filsa.net/metamail/demo

04:12:34 * DanC_ finds mbox2rdf.py materials...

04:15:01 <DanC_> looking at the .py code, it looks like you only spit out one address from the To header field

04:15:07 <AaronSw> filsa, RDF documents just describe things as best they can... The vocab doesn't need to strictly state exactly what must and mustn't be said.

04:15:17 <filsa> right. im not even close to handle the from and to headers properly.

04:15:30 <filsa> s/handle/handling/

04:16:12 <filsa> because i dont grok that sequencing stuff yet.

04:16:13 <DanC_> it's dc:title, not dc:Title, btw. (similarly for dc:date, etc.)

04:16:38 <filsa> yeah--aaronsw told me about the lower case for properties convention this am.

04:17:14 <DanC_> it's not just a matter of convention here; if you were making up the property names, it would be convention. But when you're using somebody else's property names, you have to use their exact spelling.

04:17:21 <filsa> ok

04:18:32 * filsa makes a list of things to fix

04:19:11 <DanC_> this sequence stuff... I wonder if I can explain it.

04:19:36 <DanC_> If I tell you: I have a brother, Jon, and another brother, Paul...

04:19:55 <DanC_> ... I haven't told you that I do *not* have a 3rd brother, William, right?

04:20:49 <filsa> right

04:21:35 <filsa> it's possible that you have a 3rd brother, but we haven't talked about it yet.

04:22:04 <DanC_> on the other hand, if I tell you my brothers are {Paul and Jon}, then any mention of "Dan's brother" is either talking about Paul or Jon.

04:23:05 <filsa> ok, it's a subtle difference, but im w you

04:23:27 <DanC_> to capture such stuff in RDF, I'd use two properties: brothers(DanC, {Paul, Jon}) on the one hand and brother(DanC, Jon) brother(DanC, Paul) on the other.

04:23:52 <DanC_> er... is prop(subj, obj) a reasonably familiar notation to you?

04:23:57 <filsa> yes

04:24:41 <filsa> so i'd need both brothers() and brother()...?

04:24:46 <DanC_> so the email To header is a lot like the brother/brothers situation.

04:24:47 <DanC_> yes.

04:25:12 * filsa wonders what the prop names for email To should be

04:25:12 <DanC_> if you want to convert rfc822 format to RDF without losing info, you need the brothers() style doodad.

04:26:10 <filsa> how about the name <address> thing?

04:26:20 <filsa> To: DanC <danc@wherever.org>

04:26:40 <filsa> i wasnt sure how to model that either.

04:26:52 <DanC_> that's another hairball; ready to go there, or shall we chew on the names for To props for a bit?

04:27:06 <filsa> lets chew for a bit.

04:28:56 * DanC_ noodles... recipient/recipients... addressedTo ...

04:29:18 * filsa scans rfc822

04:29:38 <filsa> addressees

04:31:54 <filsa> i like recipient/recipients, but now i'm thinking about the cc group

04:32:18 <DanC_> frankly, the casual user is going to expect to(mid:foo@bar, mailto:joe@domain) , not to(mid:..., X), item(X, mailto:joe@domain). So the short/handy "to" name should be used for the easy one.

04:32:52 <DanC_> so how about brother:brothers :: to:addressedTo

04:32:58 <filsa> sounds good.

04:33:30 <DanC_> ok... now onto the To: DanC <danc@wherever.org> hairball...

04:34:56 <DanC_> the questions are: is the to(mid:.., ?) relation between messages and mailboxes? or between messages and people?

04:34:56 <filsa> name-addr = [display-name] angle-addr

04:35:35 <filsa> people, imo

04:35:54 <filsa> well....

04:36:48 <DanC_> I thought RFC822 would say messages and mailboxes, but I looked, and it's fairly careful to say messages are addressed to people.

04:37:06 <filsa> which section?

04:37:27 * filsa was looking at 822 3.4

04:37:30 <DanC_> calling my bluff, eh? it's been a while since I read it, but I think the section was called, ironically enough, "semantics"

04:38:40 <filsa> heh

04:38:44 <DanC_> hmm... [[ 6.2. SEMANTICS

04:38:44 <DanC_> A mailbox receives mail.

04:38:44 <DanC_> ]]

04:38:52 <deltab> uh, you do knopw that 822 is obsolete?

04:38:57 <DanC_> yes. sad news.

04:39:01 <filsa> yeah.

04:39:53 <filsa> section 6.2 doesnt live in rfc2822. :(

04:40:05 <DanC_> here's the bit I remember: [[ The name reference is optional and is usually used to

04:40:06 <DanC_> indicate the human name of a recipient. ]]

04:41:29 <filsa> that's right.

04:41:42 <DanC_> ok... so back to the example of To: DanC <danc@wherever.org>

04:41:47 <filsa> i see that the relevant sections of 6.2 are in rfc2822 3.4

04:42:17 <filsa> [[ name-addr = [display-name] angle-addr

04:42:20 <DanC_> does that say "this message is addressed to a person called DanC, available at mailto:danc@wherever.org"? or...

04:43:23 <filsa> rfc2822 3.4: [[ A mailbox receives mail. It is a conceptual entity which does not

04:43:23 <filsa> necessarily pertain to file storage. For example, some sites may

04:43:23 <filsa> choose to print mail on a printer and deliver the output to the

04:43:23 <filsa> addressee's desk. Normally, a mailbox is comprised of two parts: (1)

04:43:23 <filsa> an optional display name that indicates the name of the recipient

04:43:24 <filsa> (which could be a person or a system) that could be displayed to the

04:43:26 <filsa> user of a mail application, and (2) an addr-spec address enclosed in

04:43:26 <DanC_> does it say "... addressed to the mailbox mailto:danc@wherever.org , which is owned/used by someone/something called DanC"?

04:43:28 <filsa> angle brackets ("<" and ">"). There is also an alternate simple form

04:43:30 <filsa> of a mailbox where the addr-spec address appears alone, ]]

04:44:15 <filsa> that section of the spec seems to indicate the former.

04:44:37 <filsa> like a snail mail address, 'Phil Suh' at <4540 California St SF CA>

04:46:02 <filsa> (the idea of printing out an email and putting it on someone's desk is conceivable, but really dated)

04:46:07 <deltab> an optional display name that indicates the name of the recipient

04:46:07 <deltab> (which could be a person or a system) that could be displayed to the

04:46:33 <DanC_> I can see both sides... the "A mailbox receives mail" supports the latter, but the bit about recipient/person/system suggests the former.

04:47:42 <DanC_> the "mailbox receives mail" view is sorta simpler to model, but again, loses info.

04:47:54 <DanC_> lemme see if I can illustrate that with an example...

04:48:07 * filsa likes examples

04:48:45 <DanC_> To: Bill <b@dom>, Bob <b@dom>

04:50:06 <DanC_> per the mailbox-receives-mail view, that's to(M, mailto:b@dom), userNamed(b@dom, "Bill"), userNamed(b@dom, "Bob")

04:50:29 <DanC_> and addressedTo(M, {b@dom})

04:50:44 * filsa nods

04:51:24 * deltab nods

04:51:53 <DanC_> per the other view, that's to(M, P1), to(M, P2), mailbox(P1, b@dom), name(P1, Bill), mailbox(P2, b@dom), mailbox(P2, Bob)

04:52:03 <DanC_> and addressedTo(M, {P1, P2})

04:52:45 <filsa> wow

04:53:43 * DanC_ got sloppy with the quotes and such, but hopes the gist got thru

04:54:39 <filsa> no, that's great.

04:55:26 <filsa> is the P1 and P2 part of the rdf schema, or are they anon nodes?

04:55:35 * filsa hopes that makes sense

04:55:44 <DanC_> most likely anon nodes (which is orthogonal to most schema issues, btw)

04:55:50 <filsa> ok.

04:56:29 <filsa> so in the schema i'd model to, addressedTo, name, mailbox.

04:58:44 <DanC_> you like the P1/P2 view? it's rare that folks don't go "ew... that's too complicated" when faced with thorough models.

04:58:52 <filsa> i like thorough.

04:58:59 <deltab> the latter is how humans would probably see it, but the MTA sees the former

04:59:16 <filsa> deltab: hmm.

04:59:41 <filsa> i guess it depends on what you're doing.

04:59:48 * deltab nods

05:00:28 <filsa> as for complicated models, i find you trade simplicity for power and flexibility.

05:01:07 <DanC_> my intuition says that using the P1/P2 form is going to result in more valuable/reusable data. But I don't have much experience to back that up, and I agree that it depends on what you're doing; i.e. distinctions that aren't exploited are a waste.

05:02:00 <deltab> I agree that P1/P2 would be more useful, but current software doesn't see it that way

05:02:07 <filsa> well, they are wasteful maybe in the short term. but i find that building for growth is a good thing. there's no way to anticipate what the uses would be.

05:02:38 <DanC_> I'm increasingly swayed by the "you aren't going to need it" principle of Extreme Programming.

05:02:53 <filsa> deltab: your MTA-view/human-view insight prompts me to ask: is it useful to model human perspective in RDF?

05:03:18 <DanC_> model human perspective: I think so, yes.

05:03:21 <deltab> filsa: that I leave to wiser minds than mine :-)

05:03:23 <filsa> DanC: i've got a programming partner who helps balance my "cover-everything" view.

05:03:34 <DanC_> ah... good.

05:03:38 <filsa> when we do XP he makes me leave all that out.

05:04:35 <DanC_> a practical point: the name(P1, "Bill") and mailbox(P1, mailto:b@dom) properties are ones that lots of other folks are using too; e.g. foaf:

05:04:59 <filsa> good to know.

05:05:06 <filsa> another thought

05:05:16 <filsa> how do i interconnect my schema w other schemas?

05:05:29 <DanC_> not quite as widespread as dc:title and such, but you're likely to find other agents that grok foaf:name and foaf:mailbox; and I hope W3C will standardize names for those relations before we all get old.

05:05:42 <filsa> i tried to use subpropertyof

05:05:48 <DanC_> interconnect: either by just using those other terms, or by subPropertyOf connections.

05:06:18 <deltab> I was thinking it would be along the lines of "myschema:name = foaf:name ."

05:06:46 <filsa> deltab: can i do that now, if so, how? in a comment?

05:07:08 <DanC_> yes, you can use = (i.e. daml:equivalentTo) in stead of subPropertyOf to make the links. But that's hardly worth the bother: if you mean the same thing, why make up a new name? just use the name that's already there.

05:08:07 <filsa> subPropertyOf I grok; however 'just use those other terms'... in the RDF instance files, or the RDF schemavocab?

05:08:14 <DanC_> its one thing to discover, post-hoc, that two terms are used for the same relationship; it's another to litter the semantic web with aliases consciously.

05:08:18 <deltab> DanC_: what if you design your schema before you find another one with the same concept?

05:09:02 <DanC_> just use: in instances (and in the schema, where appropriate; e.g. domain/range)

05:10:13 <filsa> okay.

05:11:20 <filsa> well, i got a long list of stuff to work on :-)

05:11:28 <DanC_> so... an example to recap:

05:11:32 * filsa nods

05:12:05 <DanC_> To: <box@dom>, Fred <fred@d2>, Bob <bob@xyz>

05:12:22 <DanC_> Message-Id: m1@d

05:13:01 <DanC_> turns into: to(mid:m1@d, mailto:box@dom), to(mid:m1@d, mailto:fred@d2), ...

05:13:17 <DanC_> ... to(mid:m1@d, mailto:bob@xyz), ...

05:13:47 <DanC_> ... addressedTo(mid:m1@d, {X, R1, R2}), ...

05:14:57 <DanC_> ... mailbox(X, mailto:box@dom), mailbox(R1, mailto:fred@d2), mailbox(R2, mailto:bob@xyz), ...

05:15:13 <DanC_> ... name(R1, "Fred"), name(R2, "Bob).

05:15:16 <DanC_> .

05:15:50 <filsa> that's not too bad, really.

05:16:07 <DanC_> where {X, R1, R2} is short for an RDF collection. (or DAML list)

05:16:39 <filsa> gotcha. i think i might try both models and see how useful they are.

05:17:21 <filsa> the plan is to get the mbox info into a redfoot db, and build an interface around it.

05:18:39 <DanC_> yes... build it and see if they come ;-)

05:19:28 <filsa> i'm in a spot where i have a little time to work on learning this stuff...

05:19:54 <filsa> ...i've got a gig which is nothing but building metadata vocabs and stores for lots of financial data...

05:20:05 <filsa> so i'm getting ready for that.

05:20:21 <filsa> your help is really great. thanks.

05:20:41 <DanC_> you're more than welcome. it's nice to have a sounding-board for the ideas.

05:20:53 <DanC_> I have been meaning to think this thru carefully for a while.

05:21:20 <DanC_> parting shot: I don't think email subject maps well to dc:title. dc:description, maybe, but not title.

05:21:33 <filsa> okay, point taken.

05:21:46 <filsa> feedback is definitely welcome. i'm still feeling my way around.

05:22:11 <deltab> dc_rdfig:description rarely matches either

05:22:11 <dc_rdfig> Not understood: description rarely matches either

05:22:36 <DanC_> a bit of justification: titles are sorta stand-alone names for things; mail messages aren't usually referred to by subject alone. Folks always say "my message to frobnits-list@xyz with subject 'all about weebles'"

05:24:27 <DanC_> deltab, email:subject isn't a subproperty of dc:description? explain?

05:24:38 <filsa> hmmm... there is a dc:subject

05:25:07 <filsa> ah, maybe it's not good either: 'Typically, a Subject will be expressed as keywords, key phrases or classification codes that describe a topic of the resource. Recommended best practice is to select a value from a controlled vocabulary or formal classification scheme'

05:25:40 <deltab> Subject header fields used in Internet mail and Usenet don't often match the contents

05:26:12 <DanC_> BLURB: Modelling email in RDF

05:26:12 <dc_rdfig> A: Modelling email in RDF from DanC_

05:26:18 <filsa> true. i'll have to decide whether to ditch dc:title altogether.

05:26:23 <DanC_> DanC_ is now known as DanC

05:26:52 <DanC> A:this is a sort of bookmark pointer into the logs

05:26:52 <dc_rdfig> commented item A

05:27:35 <DanC> deltab, usage errors are a different kettle of worms altogether.

05:28:12 <deltab> Jon Udell wrote something about it somewhere that I can't find right at the moment

05:28:14 <DanC> you're saying users don't use the email subject field well, not that the email subject field, when used as designed, doesn't match the dc:description property.

05:28:30 <deltab> yes

05:28:45 <DanC> I don't need Jon U. to tell me that 80% of life is drek. I know that already. ;-)

05:29:23 <deltab> he was pointing out that on an intranet nntp server, you can set policies for what subjects contain

05:29:48 <deltab> so that they actually become useful again

05:30:31 * filsa is on phone

05:30:42 <DanC> at the KT2001 conference, a guy from Lotus gave 2 features of collaborative discussion support software that were brain-dead simple to implement and made all the difference:

05:31:21 <DanC> (1) keep a descripton of the forum visible to users, and

05:31:53 <DanC> (2) each forum should have one or moderators, which, again, should be visible (their names, that is) to users

05:33:39 * AaronSw waves goodnight

05:33:44 <filsa> nite!

05:33:46 * filsa is back

05:34:06 * DanC was gonna work on a proof theory for cwm's logic this evening, but is happy with the way the email modelling discussion turned out in stead.

05:34:10 <deltab> aha! http://software-carpentry.codesourcery.com/Groupware/report.html

05:34:25 * filsa was happy too.

05:34:51 * AaronSw declares success, goes home ;-)

05:35:30 <filsa> ya, i'm going to call it quits for the night as well.

05:36:11 <filsa> i'll regroup tomorrow night, and hopefully have another rev of the schema.demo before the weekend.

05:36:18 <filsa> (meetings all day tomorrow ugh)

05:37:48 <deltab> in particular http://software-carpentry.codesourcery.com/Groupware/report.html#54

05:38:33 * DanC reads nifty article...

05:38:38 <DanC>http://software-carpentry.codesourcery.com/Groupware/report.html

05:38:38 <dc_rdfig> B: http://software-carpentry.codesourcery.com/Groupware/report.html from DanC

05:38:50 <DanC> B:|Internet Groupware for Scientific Collaboration

05:38:51 <dc_rdfig> titled item B

05:39:00 <filsa> deltab: hmm. thats a great example.

05:39:06 <DanC> B:by John Udell, 2000

05:39:06 <dc_rdfig> commented item B

05:39:32 <DanC> B:nifty stuff on the meeting-scheduling-problem, near and dear to my heart.

05:39:32 <dc_rdfig> commented item B

05:40:02 <filsa> i'm hopeful that smart mail archiving software would enable a moderator to annotate subjects like udell's example

05:42:06 <deltab> I've seen a MUA that showed the first couple of non-quote lines in a message in the overview

05:42:47 <filsa> outlook does that.

05:42:57 <deltab> ah, that's probably it

05:43:40 <deltab> of course it'd be nice if the authors gave good subject lines in the first place

05:45:39 <filsa> it'd be nice not to have to pay taxes too ;-)

05:51:05 * filsa waves goodnight

05:51:10 <filsa> filsa is now known as [filsa]

06:16:53 <DanC> A:"Currently at version 3.2, Amaya isn't nearly stable enough or complete enough for

06:16:54 <dc_rdfig> commented item A

06:16:54 <DanC> real work, but it's an intoxicating glimpse of the way things ought to be."

06:17:04 <DanC> A:... real work, but it's an intoxicating glimpse of the way things ought to be."

06:17:04 <dc_rdfig> commented item A

06:44:47 <DanC>http://software-carpentry.codesourcery.com/entries/second-round/track/Roundup/

06:44:47 <dc_rdfig> C: http://software-carpentry.codesourcery.com/entries/second-round/track/Roundup/ from DanC

06:44:53 <DanC> C:|Roundup

06:44:53 <dc_rdfig> titled item C

06:45:19 <DanC> C:has an interesting "hyperdatabase layer" that looks awfully RDF-ish

06:45:19 <dc_rdfig> commented item C

07:34:03 <DanC>http://www.dsmit.com/cons/

07:34:03 <dc_rdfig> D: http://www.dsmit.com/cons/ from DanC

07:34:13 <DanC> D:Cons -- a Make replacement

07:34:13 <dc_rdfig> commented item D

07:34:35 <DanC> D:see also: Ant (pointer, anyone?)

07:34:36 <dc_rdfig> commented item D

07:34:43 <DanC> D:re SemanticWebMake

07:34:44 <dc_rdfig> commented item D

07:35:31 <deltab> ant - http://jakarta.apache.org/ant/

07:35:57 <DanC> D:that's [ant|http://jakarta.apache.org/ant/] (thx deltab)

07:35:57 <dc_rdfig> commented item D

09:37:07 <dajobe> hi em_ecdl - how's Darmstadt?

09:37:21 <em_ecdl> hey!

09:37:39 <em_ecdl> darmstadt is overcast and rainy :(

09:38:00 * em_ecdl but enjoying warm 802.11 connection inside

09:38:08 <dajobe> he he

09:38:26 * em_ecdl working on ecdl keynote

09:38:47 * em_ecdl geting a bit carried away with circles and arrows

09:38:51 <dajobe> yeah, we want to get the details when you do them

09:38:56 <danbri_> hi eric, dave!

09:38:59 <dajobe> (Although I'm sure Aaron will be URL-hunting on w3.org)

09:39:02 <em_ecdl> haye danbri!

09:39:25 <em_ecdl> I'm a little surprised AaronSw hasn't chumped anything yet :)

09:39:37 <dajobe>http://www.ecdl2001.org/

09:39:37 <dc_rdfig> E: http://www.ecdl2001.org/ from dajobe

09:39:43 <dajobe> E:|5th European Conference on Research and Advanced Technology for Digital Libraries

09:39:43 <dc_rdfig> titled item E

09:39:49 <em_ecdl> danbri, Jane H. was under the impression you were comming to ECDL?

09:40:15 <danbri_> She was under the impression I was going to try to come to ECDL; which I did, unsuccessfully.

09:40:58 <danbri_> There's a harmony/cidoc meeting later in the week I'd hoped to attend, ABC meets CRM, but too busy. Libby's going.

09:41:21 <em_ecdl> ok, i'll tell her

10:01:00 <dajobe>http://www.schemas-forum.org/workshops/ws3/presentations/baker/index.htm

10:01:00 <dc_rdfig> F: http://www.schemas-forum.org/workshops/ws3/presentations/baker/index.htm from dajobe

10:01:07 <dajobe> F:|Schemas and the Semantic Web, Tom Baker

10:01:07 <dc_rdfig> titled item F

10:01:22 <dajobe> F:a good intro IMHO

10:01:22 <dc_rdfig> commented item F

10:03:02 <dajobe> F:more [http://www.schemas-forum.org/workshops/ws3/programme.html|interesting stuff] in the programme especially multilingual issues (topic of workshop)

10:03:02 <dc_rdfig> commented item F

10:13:41 * danbri_ donates some RDF to whitehouse.org

10:14:10 <danbri_> (http://rdfweb.org/people/danbri/rdfweb/whitehouse.png)

10:14:27 <danbri_> thought it might get us some nice web-of-trust publicity, though doubt they'll upload it

10:37:56 <danbri_>http://groups.yahoo.com/group/rdfweb-dev/message/127

10:37:56 <dc_rdfig> G: http://groups.yahoo.com/group/rdfweb-dev/message/127 from danbri_

10:38:18 <danbri_> G:|RDFWeb student project: Whitehouse.* web-o-(dis)trust

10:38:19 <dc_rdfig> titled item G

10:39:53 <danbri_> G:An [RDFization|http://rdfweb.org/people/danbri/rdfweb/whitehouse.rdf] ([png|http://rdfweb.org/people/danbri/rdfweb/whitehouse.png]) of the [whitehouse.org denouncement| http://www.whitehouse.org/news/2001/090401.asp] of [whitehouse.gov|http://www.whitehouse.gov/].

10:39:53 <dc_rdfig> commented item G

10:41:00 <danbri_> G:But who to believe? If the Semantic Web can't help with this sort of scenario, we might as well all go home now. And we better be prepared for people to use RDF to tell fibs.

10:41:01 <dc_rdfig> commented item G

10:41:38 <danbri_> G:See my [mechanised skepticism notes|http://www.skepsis.no/english/maillist/classification/98/msg00025.html] for more context.

10:41:38 <dc_rdfig> commented item G

10:43:09 * danbri_ remembers that Jim Hendler might be looking for student projects...

10:45:53 <danbri_> G:Ugh, I tried to [use rdf reification|http://www.skepsis.no/english/maillist/classification/98/msg00025.html]; that was a mistake.

10:45:53 <dc_rdfig> commented item G

10:45:55 <danbri_> Hi jang

10:50:54 <dajobe> danbri or danbri_ re: the 700K http://www.w3.org/TR/xquery-operators/ - is it worth reviewing any particular bit?

10:52:09 <danbri_> I skimmed it looking for bits that might reflect well into binrary relations between datatype nodes in an RDF graph. Some parts looked like they'd do that easily. Haven't got my head around the entire work though.

10:53:30 <dajobe> so looking at the relations, comparisons with an RDF-hat on

10:56:56 <danbri_> yes; if they give URI names to relationships <= < etc amongst XML datatypes, we might want to simply re-use those rather than re-invent. One might imagine an appendix to their doc saying 'these [...] we consider to also be RDF properties', for processing in an RDF context.

10:58:33 <dajobe> skimming again(!), bascially just things like operation(data type, data type) can give a URI for that, which we can use

10:59:50 <danbri_> It's been a little while since I read it, but that's the idea I had in mind, yeah. Although quite how datatypes look in RDF isn't exactly nailed down: wrapped by resource nodes versus just literal nodes; if they're wrapped as resource nodes then having relations between them seems easy.

11:00:02 <danbri_> I think thats how DAML+OIL prefers to do it

11:01:36 <deltab> there's an extra < near the end of http://rdfweb.org/people/danbri/rdfweb/whitehouse.rdf

11:02:16 <danbri_> thanks; fixed. Odd that the validator didn't catch it.

11:02:33 <danbri_> now if only we could get bill clinton to digitally sign that file ;-)

13:14:41 <AaronSw> C:Roundup is written in Python and is [now available|http://web.lfw.org/ping/roundup.tar.gz]

13:14:41 <dc_rdfig> commented item C

13:16:07 <danbri> hi aaronsw

13:16:13 <AaronSw> Hi danbri

13:16:27 <AaronSw> I loved the whitehouse thing!

13:16:43 <danbri> thanks

13:20:27 <danbri> C:(from the [creator|http://web.lfw.org/ping/] of the [Crit|http://crit.org] Web annotation system).

13:20:28 <dc_rdfig> commented item C

13:23:39 <AaronSw> C:I've been a big [?!ng|http://web.lfw.org/ping/] fan for a while now.

13:23:39 <dc_rdfig> commented item C

13:26:47 <AaronSw> C:[Malkovich Malkovich Malkovich?|http://web.lfw.org/jminc/http://rdfig.xmlhack.com/index.html]

13:26:47 <dc_rdfig> commented item C

13:41:47 * AaronSw waves to em, demands URL ;)

13:42:27 * em_ecdl waves to AaronSw, urls shortly

13:42:43 * danbri_ waves

13:44:03 * AaronSw wonders what the time zone is there...

13:44:40 <danbri_> an hour further away from east coast US than the UK (he says, verbosely)

13:46:44 <AaronSw> Hmm... 3:00?

13:47:03 <AaronSw> PM

13:59:59 * AaronSw heads off...

14:01:37 <em_ecdl> 4:00

14:19:08 <DanC>http://www.erights.org/

14:19:09 <dc_rdfig> H: http://www.erights.org/ from DanC

14:19:21 <DanC> H:ERights.Org -- Open Source Distributed Capabilities

14:19:22 <dc_rdfig> commented item H

14:19:28 <DanC> H:sigh... so much to study

14:19:28 <dc_rdfig> commented item H

14:20:04 <DanC> H:index under WebOfTrust, DigitalRightsManagement

14:20:04 <dc_rdfig> commented item H

14:20:13 <DanC> H:aka web of trust, digital rights management

14:20:13 <dc_rdfig> commented item H

14:33:43 <danbri_>http://www.w3.org/TR/2001/REC-SVG-20010904/

14:33:43 <dc_rdfig> I: http://www.w3.org/TR/2001/REC-SVG-20010904/ from danbri_

14:34:00 <danbri_> I:|Scalable Vector Graphics (SVG) 1.0 Specification (W3C Recommendation 04 September 2001)

14:34:00 <dc_rdfig> titled item I

14:34:34 <danbri_> I:SVG is a Rec now; time to get implimenting...

14:34:34 <dc_rdfig> commented item I

14:34:55 <dajobe> RDF is a wreck^H^H^H^H^HREC too

14:35:09 * em_ecdl wonders if I should use the SVG workflow example in the talk...

14:35:44 * em_ecdl seems appropriate if time allows... all circles and arrows are done in SVG with Adobe 3.0 beta plugin... works suprizingly well

14:35:48 <DanC> I:time to get implementing was at CR; this thing is implemented. They're dotting the last i's and crossing the last t's on full test suites.

14:35:48 <dc_rdfig> commented item I

14:35:51 <danbri_> I:The [metadata section|http://www.w3.org/TR/2001/REC-SVG-20010904/metadata.html] may be of interest

14:35:51 <dc_rdfig> commented item I

14:36:04 * em_ecdl goes back to work

14:38:19 <danbri_> I:Yes, time to get SemWeb-meets-SVG implementing even more than [be|http://www.rdfviz.org/][fore|http://www.w3.org/2001/08/rdfweb/svg-foaf].

14:38:19 <dc_rdfig> commented item I

14:39:52 <DanC> H:this [Capability-based Financial Instruments|http://www.erights.org/elib/capability/ode/overview.html] overview seems to discuss the technology at the level I'm interested in it.

14:39:52 <dc_rdfig> commented item H

14:43:56 <DanC> H:using game theory to model security is very appealing...

14:43:56 <dc_rdfig> commented item H

14:49:39 <jang> capability-based systems are just the ticket for distributed apps

14:51:11 <DanC> The "sematic web layer cake" thesis is that proof-based access control is the way to go. I'm noodling on how it compares to capability-based systems.

15:13:24 <DanC> H:ooh... vats... persistence too... it just keeps getting better. (I like the way they did objects based on lambda.)

15:13:24 <dc_rdfig> commented item H

15:15:43 <jang> capabilities and proof-based are almost the same thing, dan...

15:16:00 <jang> particularly if you include the notion of a capability for secured introductions

15:16:15 <DanC> same thing: that would be good. care to elaborate?

15:16:52 <jang> my intuition is that you can do proof-based security using capabilities.

15:18:43 <jang> hey, that's paper's good

15:18:57 <DanC> yes; it's gonna cost me an hour or so today ;-)

15:19:14 <DanC> (tell the chump that you like it, pls, btw)

15:19:26 <jang> first time I've seen a treatment of introduction mechanisms written down explicitly

15:22:47 <jang> that's definitely worth a bookmark. A whistle-stop tour (de force!)

15:22:55 <jang> ohw do I talk to teh chump? I've not seeen the code

15:23:03 <jang> and these sorts of things pass me by

15:24:52 <danbri_> prefix what you say with, in this case, H:

15:25:06 <danbri_> H:I imagine Jan would like this stuff...

15:25:06 <dc_rdfig> commented item H

15:25:11 <danbri_> for example...

15:28:34 <jang> H:indeed. Some day, all systems will be built this way.

15:28:34 <dc_rdfig> commented item H

16:00:55 * sbp pops in for a quick go on the chump

16:01:06 <sbp>http://infomesh.net/2001/rdfwiki/

16:01:06 <dc_rdfig> J: http://infomesh.net/2001/rdfwiki/ from sbp

16:01:15 <sbp> J:|RDFWiki

16:01:15 <dc_rdfig> titled item J

16:02:08 <sbp> J:It's like any conventional [http://c2.com/cgi-bin/wiki?WikiWikiWeb|Wiki], except that all of the WikiNames have URIs (when exported), and all of the data is stored as RDF.

16:02:08 <dc_rdfig> commented item J

16:02:41 <sbp> J:Instead of just putting some information under each WikiPage, you have to supply a predicate (which can be a WikiName or a URI), and an object (which can be a WikiName or a Literal).

16:02:41 <dc_rdfig> commented item J

16:03:00 <sbp> J:The Wiki can be exported as N-Triples (and only N-Triples at the moment).

16:03:00 <dc_rdfig> commented item J

16:03:16 <danbri_> J:Excellent work Sean! :-)

16:03:16 <dc_rdfig> commented item J

16:03:27 <DanC> J:sounds like Enquire!

16:03:27 <dc_rdfig> commented item J

16:03:52 <sbp> J:It's based a little on ENQUIRE, and a little on Seth's current rants

16:03:52 <dc_rdfig> commented item J

16:03:59 <sbp> cheers DanBri

16:04:20 <dajobe> J:added to RDF resource guide, RSS 1.0 feed and syndicated to the world.

16:04:20 <dc_rdfig> commented item J

16:04:34 <sbp> wow, that was quick

16:04:48 <DanC> er... is this RDFWiki running as a service I can play with?

16:05:11 <sbp> I tried to get Aaron running it as a service... but he forgot to set one of the variables

16:05:18 * DanC enjoys the 404 message at http://infomesh.net/rdfwiki/?go=1

16:05:23 <sbp> I have it going locally though, I'll get you the URI...

16:05:36 <sbp> :-)

16:06:02 <sbp> very non-cool URI: http://pE0s08a06.client.global.net.uk:81/cgi-bin/rdfwiki.cgi

16:08:24 * sbp fixes the links to run over the Web...

16:08:49 <danbri_> J:My old [RDFWebWikiGlue|http://groups.yahoo.com/group/rdfweb-dev/message/61] notes might be worth comparing; I didn't give enough thought to how typed Wiki links might be RDFized...

16:08:49 <dc_rdfig> commented item J

16:10:06 <sbp> heh, "WikiIsm"

16:10:33 <sbp> Ah, that's interesting: http://xmlns.com/example/rdfwiki#wikiName

16:10:58 <sbp> I use 'tag:infomesh.net,2001-08-07:WikiName' internally for that

16:13:48 <sbp> Gotta run

16:32:10 <dajobe> sbp's wiki now out on meerkit wire service

16:32:15 <dajobe> s/kit/kat/

16:48:35 <DanC> a random thought... this E language is really cool... and it's not XML, and it never should be...

16:49:17 <DanC> ... I wonder if Yapps would be as interesting as XML as a knowledge exchange format interoperability layer...

16:49:44 <bijan> Hmm?

16:49:45 <DanC> ... where a Yapps language is an LL(1) grammar over regular tokens.

16:49:59 <bijan> Ah.

16:50:19 * bijan thinks danc is working his way back to sexpr :)

16:50:50 <DanC> i.e. screw tags, attributes, DTDs, and schemas; gimme regexps, LL(1) grammars. Maybe the attribute-grammar stuff is part of the interoperability magic too.

16:51:07 * bijan also notes that most prologs have some helpful stuff in this realm.

16:51:13 <bijan> yep, you want sexprs :)

16:51:19 <DanC> no, sexprs are, evidently, not a good match for human writing-system capabilities.

16:51:46 <DanC> by sexprs, you mean the (atom (atom...)) notation, right?

16:51:59 <bijan> Well...yes. :)

16:52:10 <bijan> But I'll take prolog style.

16:52:18 <bijan> You want Mexprs actually.

16:52:35 <DanC> sexprs under-utilizes our word/punctuation capabilities (morphemes?) and relies too much on our ability to match parens.

16:53:05 <bijan> Let me dismiss that, mostly because I don't htink it worth arguing...

16:53:13 <bijan> ...but you should check out Common Lisp.

16:53:25 <DanC> I'm intimately familiar with Common Lisp.

16:53:33 <bijan> The combination of reader macros, sexprs, and regular macros seems close...

16:53:35 <bijan> Oops.

16:54:04 <DanC> ok, the common lisp reader, with reader macros, is pretty close to yapps. but the reader macros aren't quite as powerful as regexps.

16:54:14 <bijan> Nope.

16:54:21 <bijan> Check out rebol, too, if you haven't.

16:54:26 <bijan> It has an LL parser built in.

16:54:43 <DanC> I've breezed by rebol; it is cool; I didn't realize it had built-in LL.

16:54:53 <bijan> Yes. check out "dialecting".

16:55:06 <bijan> What it *doesn't* have, IIRC, is the tokenizer stuff.

16:55:09 <DanC> sorta like tcl on steroids?

16:55:26 <bijan> Eh....depends on how you view tcl. I'd say no.

16:55:31 <DanC> (not sure what I meant by that tcl remark. never mind)

16:55:33 <bijan> It's more like scheme meets python.

16:55:35 * sandro doesn't think we need more languages like that. We need more human-centric declarative languages.

16:56:09 <sandro> .... languages where it's clear what the underlying triple model is.

16:56:28 <bijan> I think what a lot of these systems tend to skimp on is the tokenizing phase, alas.

16:56:31 <DanC> yes to human-centric declarative languages. But I'm saying: I wonder if LL(1)/regexp were as widely deployed and celebrated as XML; wouldn't that be more fun and useful?

16:56:54 <bijan> Prolog surely does (well, it blurs the tokenizing/parsing phase at a lot of points).

16:57:03 <sandro> I wholeheartedly agree, DanC. That's where I was (am?) trying to go with blindfold grammars.

16:57:06 <bijan> sexprs ask the user to tokenize.

16:57:17 <bijan> CL uses reader macros for that.

16:58:08 <DanC> hmm... strategy for blindfold in the current W3C context should include some mechanism for mapping from XML schemas to LL(1)/regexp and back.

16:58:42 <DanC> hmm... I guess you're using triples rather than XML schemas. I see.

16:58:50 <sandro> Yes. I think the right way is by mapping them both (in whatever syntax) to a grammar ontology.

16:59:28 <bijan> Dan: The rebol parsing stuff: http://www.rebol.com/docs/core23/rebolcore-15.html

16:59:34 <sandro> (that is, to instance data using some ontology for grammaatical information)

17:00:15 <DanC> well, I don't share your tastes in implementation technology (C/C++ ... blarg! I hope never to debug another malloc()/free() mess in my life) (and Yapps is so fun and easy!) but I can see the blindfold strategy now.

17:00:31 <bijan> I suspect that it's more obviously cool to those deep into rebol.

17:01:00 <bijan> Though section 6 (http://www.rebol.com/docs/core23/rebolcore-15.html#sect6.) is somewhat illuminating.

17:02:00 <bijan> section 7, too.

17:02:21 * sandro laughs (re C/C++).

17:02:36 <sandro> Yes, Yapps does kind of make me jealous.

17:03:07 <bijan> embedd python into your program! :)

17:09:45 <DanC> yes, that rebol parsing stuff is nifty

17:10:07 <sandro> I want to produce code which is highly portable and embeddable. I don't see that being true of the python subsystem -- although that may be more a political problem than a technical one.

17:10:25 <sandro> On the other hand, here I am embedding XSB. Hrm. :-)

17:10:34 * bijan notes that twixt DCG and op(), prolog comes pretty close. Tokenization is rather tricky.

17:11:11 <DanC> python not portable/embeddable? seriously? python runs on *nix, windows, mac, and the java platform. it's routinely embedded. it's *way* more portable/embeddable than C/C++

17:12:08 <bijan> It's fairly standard practice, afaict, to generate a "knowledge base" language by giving varioius predicates precedence and associativity.

17:12:54 <bijan> I don't know how much trouble this gives folks on the interop front.

17:13:58 <sandro> One can only embed python in a GPL system, yes?

17:14:08 <bijan> No.

17:14:16 <bijan> Python isn't GPLed.

17:14:23 <bijan> At least it wasn't for the longest time.

17:14:56 <bijan> Indeed, there was a recent licence dustup in which the old python licence was considered *incompatible* with the GPL.

17:15:19 <sandro> Yeah, I'm reading the license now and I can't make sense of it. :-)

17:15:34 <DanC> speaking of licensing terms, rebol doesn't seem to be open source. bother.

17:15:43 <bijan> Nope :(

17:17:08 <sandro> Ah good, the python license seems quite non-viral. Nice.

17:17:17 <Seth> is python shaping up to be the language of choice for SWS apps ?

17:17:48 <sandro> Many of them. :-) I'm still feeling like I shouldn't go that route.

17:19:21 <Seth> let's put it this way .. if you were to recommend a language to develop sem web applications in that was the most compatable with other apps and was sufficiently efficient .. which language would you recomment today?

17:19:31 <bijan> It's the language that several people use and several interesting apps and frameworks are written in.

17:23:12 * sandro better stop talking about this, or he'll be paralyzed with language indecision again.

17:23:25 <Seth> lol

17:23:34 <bijan> Go for compatibility of data, not of "apps".

17:24:00 <sandro> Still, I need to code in something!

17:24:03 <bijan> If you want to write frameworks that others will use and enhance, either write them in the language they're using or make it easy to use from their language of choice.

17:24:21 <bijan> Python is a popular language on rdfig and is a reasonable lingua franca.

17:24:56 <danbri_> yes, but which one is _BEST_?

17:25:18 * bijan ignores danbri_, as is only proper.

17:27:46 <sandro> Anyway, I have to code in C today to interface with XSB. So this is irrelevant, and I can save myself.

17:35:33 <Seth> seth noodels on: does berkeley db interface easily with python

17:39:10 * DanC notes that rebol uses FastCGI for integration. http://www.rebol.com/news.html

17:39:16 <DanC> I always liked FastCGI

17:40:01 <DanC> py/berk: yup. via pickle/shelve, I think. A raw bsddb module is also available, I think.

17:40:18 <DanC> yup; just import bsddb

17:41:28 <Seth> bsddb sounds like unix ... me just got windoes

17:48:02 <DanC> well, try it?

17:51:34 * DanC sets up http://208.190.202.58/cgi-bin/rdfwiki

17:51:53 <DanC> IOError: [Errno 2] No such file or directory: '/var/www/rdfwiki/rdfwiki.pkl

17:53:26 <DanC> aha! "our Wiki has been initialized!

17:53:27 <DanC> "

17:55:53 <DanC> Sean? valuetype needs by-ref option

17:58:43 <DanC> J:RFE: support values by ref (in add triple, e.g.)

17:58:44 <dc_rdfig> commented item J

17:58:46 * AaronSw waves

17:59:03 <DanC> J:RFE: use rdfs:label when displaying property names

17:59:04 <dc_rdfig> commented item J

17:59:48 <DanC> J:I just set up a [temporary installation|http://208.190.202.58/cgi-bin/rdfwiki?RdfWiki]

17:59:48 <dc_rdfig> commented item J

18:00:54 <AaronSw> H:|ERights.org

18:00:56 <dc_rdfig> titled item H

18:01:08 <AaronSw> H:Another [Mark Miller|http://www.caplet.com/] production

18:01:08 <dc_rdfig> commented item H

18:02:29 <AaronSw> H:Mark also [thinks|http://smallstd.org/] "The world is filling up with bloated standards designed and expanded by committees run amok, as with XML and the w3c." and proposes "downward compatible", simplified standards.

18:02:30 <dc_rdfig> commented item H

18:02:50 <bijan> Wasn't there something on xml-dev on this sort of thing?

18:05:31 <DanC> who's Mark?

18:05:45 <DanC> oops.. I see the link now

18:06:07 <DanC> well, Mark's right; maybe I'll invite him to run for W3C TAG membership; along with John Udell.

18:06:14 <DanC> s/John/Jon/

18:06:19 <Seth> dan, how do i add a WikiName ... say for Seth ?

18:06:32 <DanC> not sure; ask sean.

18:13:14 <AaronSw> Jon Udell and Mark Miller for TAG? Woohoo!

18:14:51 <Seth> i see: Any time there are two or more capitalized words jammed together, Wiki will create a link.

18:15:04 <AaronSw> YepYep.

19:25:45 * sbp thanks DanC for the RDFWiki installation

19:30:31 <Seth> hi sean

19:30:37 <sbp> Hi Seth

19:30:48 <sbp> So, does RDFWiki meet the SemWebWiki requirements?

19:31:04 <Seth> its cool ... very close

19:31:14 <sbp> [Still needs a few improvements, but at least it works]

19:31:18 <Seth> wanna chat

19:31:27 <sbp> sure, O

19:31:38 <sbp> s/O/I've got about ten minutes to spare

19:51:30 <AaronSw> J:I've [installed RDFWiki|http://purl.org/swag/rdfwiki] on the SWAG website.

19:51:31 <dc_rdfig> commented item J

19:57:15 <AaronSw> J: RFE using URIs for subject

19:57:15 <dc_rdfig> commented item J

19:59:05 <sbp> well, a WikiName is a URI

19:59:21 <sbp> all URIs... O.K....

19:59:22 <AaronSw> yeah, but i mean arbitrary uris

20:00:47 <sandro> Are you happy with having Wiki names (and their URI forms) identify both the object and the page about the object?

20:01:26 <AaronSw> No.

20:01:33 <sbp> oh blimey, the Semantic Web Identification Problem... :-)

20:01:34 <AaronSw> You should pick one or the other.

20:01:56 <sbp> no, becuase the URI represents the resource, and the XHTML returned is the representation of the resource, in this context

20:02:38 <AaronSw> So then they're only identifying the object.

20:02:53 <AaronSw> Which is how I assumed it would work.

20:03:05 <sbp> yep, that's how it does work

20:04:42 <AaronSw> cool

20:04:47 <sbp> aha, Semantic Web Identification ProblEm: SWIPE

20:05:24 <sbp> "Sandro springs SWIPE on SW suckers"

20:05:43 <AaronSw> heheh

20:05:44 <sandro> I'm missing URIs for identifying values/objects in triples.

20:05:57 <sbp> rdf:value?

20:06:21 <sbp> log:resolvesTo?

20:06:29 <AaronSw> ugggh. don't use rdf:value for anything until it gets defined

20:06:44 <sbp> ah, a property with no definition...

20:06:53 <sandro> I can only have literals and WikiNames in the third position of the triple. (sorry for using the word value in a non-conformant way; ignore that.)

20:07:33 <sbp> oh, I see... a feature request! Yeah, I'm going to end up allowing URI/Literals/WikiNames in any position, aren't I?

20:07:58 <sbp> I'll put it in the next version... I just didn't want to make the input form too complicated

20:08:04 <sandro> I think your stretching "represents" to the point of comedy in saying a WikiWiki page about a car "represents" the car. :-)

20:08:25 <AaronSw> why do you say that, Sandro?

20:08:38 <AaronSw> The page doesn't represent the car. THe URI identifies it.

20:08:57 <AaronSw> At least that was my understanding of what Sean was saying.

20:09:17 <sbp> it's the old question of what a URI represents: the concept that the page stands for, the page itself, or what?

20:09:35 <sandro> Web representation of resources as about multi-media re-fomulations of information. Some bit of knowledge maybe presented in various different analogous forms. That's what "respresents" is supposed to mean on the web.

20:09:37 <sbp> since no one seems to be decided, I say we just choose, and use it consistently

20:11:08 <sbp> but HTTP in particular can do more than just return a bag of multimedia and HyperText. Go look at the range of HTTP response codes; note that they can be extended

20:11:15 <sandro> I think we *can* use the same identifier for a web page and the subject of the web page; I'm not sure what problems it might cause and/or solve to do so. I think it's a perfectly valid design approach for now.

20:12:33 <sbp> yep, I remember seeing you come to that conclusion before

20:12:49 <sandro> I like the way you call N-Triples "RDF". I wonder if that kind of thing (calling any RDF format "RDF") will really catch on.

20:13:23 <sbp> Yeah, I call RDF-M&S-XML "XML RDF" when I can

20:13:31 <sandro> So I was just asking if you really meant to do that, and how you felt about it. :)

20:14:20 <sbp> I certainly think that if it has triples of URI references, then it is RDF... I'm not sure how one would define it perfectly, or even if one should (are NQuads RDF? dunno), but it seems more less confusing that way

20:14:58 <AaronSw> I think that the new specs need to make it clear that RDF is a concept/abstract syntax/model theory not an XML format.

20:15:25 <sbp> yeah, I very much agree

20:23:48 <rreck> i concur with my 2 cents

20:23:48 <sbp> cool, SMIL (Anim) and SVG both a Rec

20:23:48 <sandro> I happen to agree with you, Aaron, but I'm afraid it might alienate lots of XML folks if we're not careful.

20:23:48 <AaronSw> It seems we've already alienated a lot of XML folks...

20:23:48 Disconnected from irc.openprojects.net (ERROR :Closing Link: logger[tatooine.ilrt.bris.ac.uk] by carter.openprojects.net (Ping timeout for logger[tatooine.ilrt.bris.ac.uk]))

20:24:03 Users on #rdfig: logger [filsa] dajobe ArtB Bayta sbp davb bijan Seth rreck jang deltab lasDesk em sandro DanC dc_rdfig eikeon AaronSw

20:24:04 <ChanServ> This channel is logged and blogged: http://logicerror.com/rdfIRCWelcome

20:25:55 <sbp> shall I paste in what logger missed? Any objections?

20:26:07 <AaronSw> nope

20:26:16 <sbp> <AaronSw> I remember talking to someone who made XMLdbs and they said "RDF isn't XML -- it has it's own structure".

20:26:16 <sbp> *** dajobe has quit IRC (Ping timeout for dajobe[pc-62-30-66-121-az.blueyonder.co.uk])

20:26:16 <sbp> <sbp> heh

20:26:16 <sbp> <AaronSw> or something to that effect

20:26:16 <sbp> <sandro> I think that's very true. It's kind of like: C / C++ -> XML / RDF-XML.

20:26:17 <sbp> <sandro> Although not in syntax. Maybe it's more like C with some specialized libraries that change your style totally.

20:26:20 <sbp> <sbp> well, RDF isn't XML, but XML RDF is a valid (not in the strict "validate" sense) application XML

20:26:22 <sbp> <sbp> s/application/application of

20:52:53 <Seth> sandro, take a web page (any web page) and talk about it in a wiki ... the url of that page in the wiki REPRESENTS the original web page ... don't you agree?

20:54:23 <Seth> or rather the bits returned by the url of the page in the wiki, represents the original web page in that wiki

20:55:06 <Seth> take http://robustai.net/

20:55:47 <Seth> and http://wiki/?http://robustai.net/ as the url of the page in the wiki

20:56:53 <sandro> Seth, (1) I think that's a different (more complex) case, and (2) I think "represents" is a very difficult word. wordnet has 11 senses of it. I think the web means it in sense 9 and you just used it in sense 1 or 2. http://www.cogsci.princeton.edu/cgi-bin/webwn/?stage=1&word=represents

20:57:02 <Seth> the second url names the node in the wiki that represents the page returned by a get(http://robustai.net/)

20:58:31 * sandro now knows which Seth you are. :-)

20:58:32 <Seth> me thiks this case might be the most common usage ... especially in sem web wiki land

20:58:47 <Seth> same seth i always was

20:59:03 <sandro> I hadn't noticed you here before.

20:59:29 <Seth> usually im quiet

21:00:18 <Seth> anyway ive been using this 'represents' concept for quite some time .. its very useful if you dont confuse it

21:00:35 <Seth> especially with semantic web stuff

21:00:50 <sandro> I agree in semweb land we're using logical symbols a lot, so that's the kind of meaning of represents we're thinkng about. I think the original web folks were thinking more in terms of multiple presention of a document.

21:02:57 <Seth> thing i like about 'represents' is it partitions the world into spaces .. just like our mind and the external world .. so it works well for this kind of mental stuff that happens on the sem web

21:03:16 <DanC> doc reps: sort of; more like: representations of the state of an object. HTTP's roots are in Objective-C/smalltalk.

21:04:58 <bijan> DanC: really? That's interesting.

21:05:43 * DanC always assumes everybody knows the history of the Web; poor assumption.

21:05:58 <bijan> I knew that objc was the implementation language.

21:06:14 <bijan> But I was unaware that there was an explicit modeling on ObjC/Smalltalk.

21:06:37 * bijan is a big Smalltalker, so is interested in such details.

21:07:17 <AaronSw> DanC, is your POV on the history of the Web documented somehow?

21:07:21 <DanC> umm... I'm not sure what you mean by "explicit". But even today, the HTTP spec calls GET a method name, no?

21:07:30 <DanC> I'm hunting for sources, AaronSw.

21:08:25 <bijan> Hmm. Well, I meant to mean by "explicit modeling' what you meant by "roots are in".

21:09:01 <bijan> What I hoped you meant was something like "HTTP requests are supposed to be like Smalltalk messages only blah blah and blah."

21:09:55 * bijan is trying to remember the state of NeXT at the time...

21:10:03 <bijan> Was DCOM around then?

21:10:12 <DanC> no, DCOM wasn't around in '89

21:10:47 <bijan> Didn't think so. So there was none of the NeXTStep distributed stuff then.

21:10:58 * bijan is weaker on objc history.

21:11:46 <DanC> hmm... found some design notes that don't back up the OOP perspective: http://www.w3.org/Protocols/DesignIssues.html

21:11:53 <AaronSw> Wow, there's a lot of web history sites.

21:12:47 <bijan> Yes, that makes it seem RPC based.

21:12:55 <bijan> Which is what I've always assumed.

21:14:11 <DanC> I thought that design note showed why HTTP is *not* RPC based

21:14:13 * DanC reads again...

21:14:20 <bijan> Hmm.

21:14:23 * bijan reads again.

21:14:49 <bijan> Ah. It's supposed to be "like an internet protocol".

21:15:01 <AaronSw> Interesting, Jenny Raggett contributes to webhistory.org

21:16:25 <DanC> webhistory.org ... grumble... bad news, that one.

21:16:36 <AaronSw> eh?

21:16:48 <DanC> Jenny was involved a long time ago, but I doubt you'll find any recent contribution from her.

21:16:54 <bijan> hmm. what would you call an "internet protocol"? A dialogue/text message system?

21:17:06 <bijan> How does that differ from a RPC?

21:17:18 <bijan> I guess you don't have a generalized calling convention.

21:17:20 <DanC> internet protocols are ala SMTP/NNTP. 3-digit response codes and the like.

21:17:40 <AaronSw> Heh, this is a great photo of TimBL: http://www.webhistory.org/photos/timbl.jpg

21:17:51 <bijan> Right, I sorta subsumed them under RPC in this discussion, since they are sorta "procedural".

21:18:08 <bijan> But I'm not clear that that's a useful way to think about it.

21:18:29 <AaronSw> I infer that TimBL and webhistory.org went sour at some point?

21:21:12 <DanC> hmm... http://lists.w3.org/Archives/Public/www-history/ seems to be 99.99% spam

21:23:26 <bijan> Sigh, I really really really need to write a good tutorial on using DCG for web apps. Just so that *I* get it.

21:49:48 <sandro> It's a very odd style. You're talking about using them as output grammars, generating the page layout?

21:50:03 <bijan> Not in this case.

21:50:08 <bijan> that i have a handle on.

21:50:11 <bijan> Somewhat.

21:50:17 <bijan> See my article :)

21:50:34 <bijan> I sat down to transliterate DanC's Yapps N3 grammar.

21:50:47 <bijan> And it should be straightforward, but, as usual, I hang up on the details.

21:52:38 <sandro> My little attempts at n3 parsing in DCG's had unacceptable performance, but perhaps I was doing some little stupid thing.

21:52:54 <bijan> Check the craft of prolog :)

21:53:10 <bijan> Yes, but I wanted to transliterate DanCs, not generate a parser per se.

21:56:27 <bijan> Yapps is interesting in that it should be straightforward to write a Yaaps <--> DCG translater, even popping up stubs for the semantic bits.

21:57:01 <bijan> I just totally suck at that sort of thing :)

21:57:32 <sandro> That's true of all perser generators, yes? Of course you need the semantic bits to be declarative to map the a (pure) DCG.

21:58:49 <sandro> (blindfold is supposed to output DCGs as an easy way to do output grammars.)

22:06:20 <bijan> Well, LL(1) grammars should have a straightforward DCG expression, I take it.

22:06:52 <bijan> And the attributing in Yapps feels like extra variables in DCGs

22:06:55 <bijan> For at least some of them.

22:07:43 <bijan> I mean, take this rule:

22:07:46 <bijan> rule list<<scp>> : "\\(" term<<scp>> * "\\)"

22:07:58 <sandro> Yes. But DCG's don't do lookahead. (They do have prolog's more powerful and dangerous backtracking, though)

22:08:35 <bijan> list(scp) :- ['('], term(scp), [')'].

22:08:46 <bijan> (er.., i'm not sure off hand how to express the * in one line.

22:09:34 <bijan> No lookahead, right.

22:09:44 <sandro> Unless I'm missing something, you don't. You need to rewrite the Kleene operators (+,*,?) using other rules. (ie transform from EBNF to BNF.)

22:10:08 <bijan> Yep.

22:10:25 <sbp> J:[http://infomesh.net/2001/rdfwiki/rdfwiki1_1.txt|Version 1.1] now incorporates some of the RFE's above, including URIs for subjects and objects, and using rdfs:label on the output

22:10:25 <dc_rdfig> commented item J

22:10:27 <sandro> Doing that by hand is a bit tedious and error-prone.

22:10:50 <DanC> RFE's... cool! collaboration is fun!

22:11:10 <bijan> Yep, which is why a translator would be cool :)

22:11:24 <sbp> :-)

22:11:45 <AaronSw> nice, sbp

22:11:48 <sbp> I'll get Aaron to re-install it at SWAG... I'll have to break it to him gently...

22:11:48 <bijan> I wonder if there's an ebnf dcg lib around.

22:11:55 <sbp> oh, there he is! Damn

22:11:58 <AaronSw> :-)

22:12:01 <sbp> :-)

22:12:03 <AaronSw> Can you just send patches?

22:12:32 <AaronSw> oh, i guess i'll do it myself.

22:12:34 <bijan> This is sorta what I meant by "DCG for web apps". All the examples on the web are NLP stuff.

22:12:46 <sbp> it might be easier just to reinstall, rather than doing diff

22:13:00 * sbp apologizes for talking through bijan and sandro...

22:13:07 <bijan> Quite all right! :)

22:13:38 <sbp> it's only a little bit less rude on IRC than it is for real :-)

22:14:01 <bijan> Asketh and google shall retrievith:

22:14:04 <bijan> The file bnf.pl contains a Prolog-parser for the EBNF-notation used to present

22:14:05 <bijan> the concrete syntax in the version 0.99 Draft of the CASL summary

22:14:24 <sandro> Cool. :-)

22:14:37 <DanC> CASL... that's a larch successor, no?

22:15:45 <bijan> I am active in the design of the algebraic specification language CASL (Common Algebraic Specification Language) within the Common Framework Initiative

22:15:52 <Seth>http://robustai.net/mentography/semWebWikiTheory.gif

22:15:52 <dc_rdfig> K: http://robustai.net/mentography/semWebWikiTheory.gif from Seth

22:16:16 <Seth> K:|First very rough draft of the theory of sem web wiki

22:16:16 <dc_rdfig> titled item K

22:16:58 <bijan> The Common Algebraic Specification Language, CASL, is an algebraic specification language based on order-sorted first-order logic with predicates, total and partial functions, developed by the Common Framework Initiative, CoFI.

22:18:34 <Seth> K: which tries to arrive at some way to manage uri and representations

22:18:34 <dc_rdfig> commented item K

22:21:15 <bijan> The bummery bit is ending up with more non-terminals.

22:21:43 <bijan>http://cs.wwc.edu/Research/Compiler/PrologTools/EBNF-BNF.html

22:21:43 <dc_rdfig> L: http://cs.wwc.edu/Research/Compiler/PrologTools/EBNF-BNF.html from bijan

22:22:01 <bijan> L:|EBNF to BNF, in Prolog.

22:22:01 <dc_rdfig> titled item L

22:22:44 <AaronSw> OK, SWAG is updated

22:24:33 <AaronSw> with the new rdfwiki

22:25:34 <bijan> L:Actually, the [Prolog tools|http://cs.wwc.edu/Research/Compiler/PrologTools/] have a number of useful and interesting goodies.

22:25:35 <dc_rdfig> commented item L

22:26:43 <AaronSw> J: RFE (big): Give the database rules and have it think about stuff

22:26:44 <dc_rdfig> commented item J

22:28:48 <Seth> J: Give it smushing rules and have it smush prince nodes

22:28:48 <dc_rdfig> commented item J

22:34:35 <bijan> Regarding the EBNF standard, ISO/IEC 14977:1996 "Information Technology --

22:34:35 <bijan> Syntactic Metalanguage -- Extneded BNF", I have mentioned that I was one of

22:34:35 <bijan> the reviewers of that document in 1995. There are a good number of

22:34:35 <bijan> ambiguities in that notation. As far as a I know, no one is using that

22:34:35 <bijan> notation for programming languages. Considering that many reviewers didn't

22:34:36 <bijan> notice that two pages were missing from the document ... it might suggest

22:34:38 <bijan> that many people didn't take it seriously.

22:34:40 <bijan> Hehe.

22:35:56 <sandro> I didn't know there was a standard. (thus the need to come up with a more precise syntax, as I did.)

22:36:26 <bijan> Looks like a crappy standard, though, if this is to be trusted: http://suo.ieee.org/email/msg00457.html

22:36:28 <Seth> maybe it was because: "The official ISO/IEC 14977:1996(E) document is unfortunately not freely available online and interested people will have to order a paper copy from

22:37:34 <Seth> ... which sure turned me off

22:37:48 <bijan> Er...that's standard for formal standards.

22:38:13 <bijan> E.g., the PDF of the ANSI smalltalk standard is $20.

22:38:34 <bijan> The common lisp hyperspec is freely available because on of the CL companies ponied up for the advert/good will value.

22:46:15 * bijan notes that, as is usual I take it, the ISO draft of EBNF is available: http://www.cl.cam.ac.uk/~mgk25/iso-14977.pdf

22:51:39 * bijan really hopes that ISO EBNF isn't as bad as it's made out to be. That would be a shame.

22:54:11 <sandro> It looks like the made the logical choices -- requiring comma between elements -- which people seem to dislike using.

23:33:53 <AaronSw>http://www.saint-andre.com/elegant-html/

23:33:54 <dc_rdfig> M: http://www.saint-andre.com/elegant-html/ from AaronSw

23:34:02 <AaronSw> M:|How to Create Elegant HTML

23:34:02 <dc_rdfig> titled item M

23:34:31 <AaronSw> M:Seems like a pretty interesting change from the usual tutorials -- he has people writing pages that are well-formed XML.

23:34:31 <dc_rdfig> commented item M

23:35:00 <AaronSw> M:But on the other hand, he uses frames.

23:35:00 <dc_rdfig> commented item M

23:52:02 <AaronSw>http://reviews2.alexa.com/review?type=3&url=www.w3.org/

23:52:03 <dc_rdfig> N: http://reviews2.alexa.com/review?type=3&url=www.w3.org/ from AaronSw

23:52:13 <AaronSw> N:|Alexa Reviews

23:52:14 <dc_rdfig> titled item N

23:52:32 <AaronSw> N:Alexa allows you to view and [write|http://reviews2.alexa.com/cgi-bin/reviews.cgi?url=www.w3.org] reviews for arbitrary websites.

23:52:32 <dc_rdfig> commented item N

23:54:52 <AaronSw> N:Part of their [web metadata|http://info.alexa.com/data/details?url=http://www.w3.org] system

23:54:52 <dc_rdfig> commented item N

23:56:35 <AaronSw> N:And the whole thing [runs on XSL|http://client.alexa.com/details/xsl/]

23:56:36 <dc_rdfig> commented item N