Semantic Web Interest Group IRC Chat Logs for 2001-10-15

00:44:55 * AaronSw returns

00:47:05 <AaronSw> Heheh.

00:47:24 * DanC waves from the family room...

00:47:37 <DanC> we're sitting down to watch "Star Wars Episode I"

00:47:50 <AaronSw> Ooh, did the DVD come out?

00:48:02 <DanC> dunno; we're just watching it on VHS

00:48:20 <AaronSw> enjoy ;)

00:53:54 <AaronSw> hi rillian

00:54:03 <rillian> howdy

00:54:11 <rillian> seems you're around alot :)

00:54:23 <AaronSw> Hehe, it does, doesn't it. :)

00:54:54 <rillian> kinda refreshing, actually

00:55:06 <rillian> to many of my friends have lives

00:55:41 <AaronSw> Ugh, lives! :-) Actually it's good practice for mega multitasking ;)

00:56:21 <rillian> well, the problem with irc is you have to be online, and if front of console

00:56:50 <rillian> jokes about neural links aside, that's a pretty limiting requirement

00:57:00 <rillian> unless you're doing that most of the time anyway

00:57:24 <AaronSw> It's easier with cable modem, 802.11 and IRC bouncers

00:58:18 <DanC> IRC bouncers?

00:58:33 <rillian> IRC bouncers

00:59:16 <DanC> oh. now that I've seen the words twice, I understand completely.

00:59:16 <AaronSw> Yeah, this cool thingy that lets me channel multiple machines into one IRC connection

00:59:19 <deltab> programs that relay irc connections

00:59:31 <AaronSw> and when no machines are connected it logs it until i reconnect

00:59:50 <rillian> AaronSw: do you have one of those you actually like?

01:00:06 * rillian keeps thinking they're a good idea, but the client integration must suck

01:00:09 <AaronSw> Well, I just started using dircproxy yesterday and it works quite nice.

01:00:20 <AaronSw> Client integration is seamless -- it just pretends to be an IRC server.

01:00:40 <rillian> so it just dumps the log from when you left for each channel?

01:01:06 <AaronSw> Yep.

01:01:14 <DanC> naming gets screwed up, I expect. e.g. "copy address of window" probably doesn't give irc://irc.openprojects.net/rdfig like it should.

01:01:31 <AaronSw> Ah, true.

01:01:33 <rillian> I guess that works except for timestamps :)

01:01:54 <AaronSw> Yeah, so I got it to put timestamps in front of logged messages.

01:02:48 * DanC catches up with the scratchpad...

01:05:19 <rillian> the other problem of course, it actually reading the backlog. But I guess that's optional :)

01:05:53 <AaronSw> Yep ;)

01:06:19 * rillian had some ideas for irc log browser at one point

01:10:08 <AaronSw> Hmm, perhaps you can write them up and we'll talk dajobe into implementing them ;)

01:11:33 <rillian> really?

01:11:51 <rillian> the real problem is how to auto-track conversation threads

01:11:57 <rillian> I wasn't sure how well I could do that

01:12:10 <deltab> yeah, I've been thinking about that too

01:12:16 <rillian> it's easy with channels that actually maintain the topic for that purpose

01:12:20 <rillian> but that's uncommon

01:12:37 <rillian> I wondered if you could track keywords and who's-talking-to-who

01:14:21 <AaronSw> Yeah, I like the "topic-keyword: message" convention, but you still need ordering, etc.

01:25:40 <bijan> Well, chump bots almost do that.

01:25:48 <bijan> By forcing you to use a prefix.

01:25:52 <rillian> what's a chump bot?

01:26:03 <AaronSw> Ah, true SemantiChump would solve that.

01:26:16 <bijan> A weblog publishing bot.

01:26:27 <AaronSw> see http://rdfig.xmlhack.com

01:26:34 <bijan> dc_rdfig:view 1

01:26:34 <dc_rdfig> C: Brief survey of the tragically fragmented Internet instant messaging world (http://www.newircusers.com/icqchat.html)

01:26:46 <bijan> So, that's a topic.

01:27:00 <bijan> You can add comments "under" it by starting your comment with the C: label.

01:27:42 <bijan> If you had a good client, i.e., one that could spawn a window for each thread of conversation that was *natural* to use, that could solve the threading problem.

01:27:54 <bijan> (these are all human intervention based ideas of course :))

01:28:41 <bijan> Similarly, I heard suggested having *every* comment labeled, with some way of doing post-facto immediate classification.

01:28:59 <bijan> More likely is to enable mode shifts.

01:29:21 <bijan> I.e., I stay in the "thread" i'm in until I say otherwise, either temporarily or permanantly.

01:29:42 * bijan goes back to writing his HTML editor.

01:30:00 <bijan> And to moarning HTML 3.0...

01:30:04 <bijan> er..

01:30:05 <bijan> mourning.

01:30:07 <bijan> Too.

01:30:53 <AaronSw> Heh.

01:31:12 * bijan will say it again: HTML 3.0 *rocked*.

01:31:28 <xbill> Why did it get ditched?

01:31:39 <AaronSw> Because it *rocked*.

01:31:39 <bijan> Give me footnotes, banners, and horizontal tabs *any day*.

01:31:45 <bijan> The evil of Netscape and Microsoft.

01:32:10 <xbill> Somehow that doesn't tell me anything.

01:32:14 <bijan> They killed it.

01:32:49 <xbill> Why would they kill it? (The thing about NS & M$ killing it only raises more questions than it answers).

01:33:00 <AaronSw> Ooh, admonishments. We should just adopt this for XHTML 2.0.

01:33:06 <bijan> Yes!

01:33:38 <AaronSw> Have you seen what the XHTML 2.0 WG is doing... hmm, I wonder if it's public

01:33:51 <bijan> From what I recall, the main "problem" with HTML 3.0 is that it didn't include all the sucky HTML extensions of Microsoft and espeically Netscape.

01:34:17 <rillian> bijan: one of the problems with threading is that irc conversations naturally drift a lot more than email

01:34:21 <bijan> It's a *very* clean HTML, athough it does include math...

01:34:37 <rillian> and I don't think a different client ui would change that much

01:34:41 <bijan> ...which isn't really a consumer item.

01:34:42 <rillian> without turning it into email

01:34:51 <xbill> How exactly do those heavyweight corp thingies corrupt standards processes to destroy things like HTML 3.0?

01:35:01 <rillian> or rather, something like those silly web forums where everything is a 3-line post

01:35:20 <bijan> rillian: it works quite well for the scratchpad and other bot based weblogs.

01:35:20 <AaronSw> As I recall the story was that no mortal humans could ever write documents in the beautiful language of HTML 3.0 and so it was ditched.

01:35:28 <bijan> Aaron: Huh?

01:35:40 <bijan> HTML 3.0 is *very* simple.

01:35:54 <bijan> xbill: by being rich and powerful and the implementors of the browsers.

01:35:54 <AaronSw> Well, that's just what somebody told me...

01:36:00 <bijan> That's bull.

01:36:07 <rillian> my idea was based more on visuallization. tracking when people come and go, and trying to track interactions through some rough way

01:36:10 <bijan> HTML 3.0 is very very simple. Non frames!

01:36:42 <bijan> The HTML 3.0 spec is a very nice read, too.

01:36:48 <rillian> how is it better than 2.0? That's what I think of as the golden age :)

01:36:57 <xbill> I'm not sure I understand the big picture but I have the vague notion that the standard is not very meaningful when no one implements it.

01:37:06 <bijan> It's a nice extension to 2.0.

01:37:22 <bijan> banners let you have frame like nav bars without frames.

01:37:31 <bijan> I.e., the banner doesn't scroll.

01:37:40 <bijan> Footnotes! Footnotes! I'd kill for footnotes.

01:38:03 <AaronSw> Good think DanC, isn't listening or he'd be moaning in pain ;)

01:38:05 <bijan> The horizontal tab thing was very nice too. It let you align things by "named" position.

01:39:49 <bijan> List headings were kind of nice, if not critical.

01:40:25 <bijan> FIG is interesting, thoguh I don't quite get it.

01:40:45 <bijan> Additional features include a static banner area for corporate logos, disclaimers and customized navigation/search controls. The LINK element can be used to provide standard toolbar/menu items for navigation, such as previous and next buttons. The NOTE

01:40:48 <bijan> element is used for admonishments such as notes, cautions or warnings, and also used for footnotes.

01:40:55 <bijan> I'd love to have all of those.

01:41:01 <bijan> I'd trade away frames in a second.

01:41:13 <xbill> HTML is very difficult for me to use because of the lack of math.

01:41:20 <bijan> HTML 3.0 has math!

01:41:28 <xbill> Essentially I end up using LaTeX.

01:41:43 <xbill> Well, a communication channel is not useful is no one is listening on it.

01:42:03 <bijan> The design of HTML math owes a lot to LaTeX's math mode, which has been found to be effective for a wide variety of mathematical typesetting. Where practical, HTML math uses tag names matching LaTeX commands

01:42:26 <bijan> Oh well.

01:42:39 <xbill> And no one implemented it. Yay.

01:42:43 <bijan> :)

01:42:57 <bijan> It never became a recommendation, I think.

01:43:19 <bijan> I was thinking of implementing it and trying to encourage it's use :)

01:43:51 <bijan> My pitch: "Yes, you too can write pages that most browsers can't use and write browsers than can't read most web pages!!!!"

01:43:56 <xbill> So how did M$ and NS kill this again?

01:44:08 <bijan> "But, um..y'know, it'll be cool. And stuff."

01:44:41 <xbill> They don't like it, and why did that make a difference to the W3C?

01:45:22 <bijan> Well, 1) they're w3c members, 2) it's really really pointless to produce a recommendation that has *no hope* of being even approximated by the (then) big two.

01:45:27 <xbill> (I personally resented HTML 3.0 getting ditched because it seemed to me at the time they killed it because it had math in it.)

01:45:39 <bijan> That certainly was part of it, I'll bet.

01:45:52 <rillian> amen. I always thought it was criminal html has never had math support

01:45:57 <rillian> given its origins

01:46:10 <AaronSw> bijan, remember that HTML 3.0 was an IETF project.

01:46:18 <bijan> Ah right!

01:46:27 <deltab> maybe those features can be resurrected in XHTML modules

01:46:30 <xbill> HTML's usefulness for scientific paper archives is exactly 0 until it has math.

01:46:31 * bijan notes that he's *looking* at the spec :)

01:47:01 <bijan> The HTML 3.0 draft, issued in 1995, attempted to build upon HTML 2.0 with the addition of features such as tables and greater control of text flow around graphics.

01:47:06 <bijan> Although some HTML 3.0 features were widely adopted by browser developers, many were not. In some cases, alternative approaches implemented by browser developers became more widespread than the "official" tags. The HTML 3.0 draft has now expired, and is th

01:47:13 <bijan> therefore no longer an official standard.

01:47:27 <bijan> In May 1996, W3C released the HTML 3.2 draft, which was designed to reflect and standardize generally-accepted practices. Therefore, HTML 3.2 includes the HTML 3.0 tags that were adopted by browser developers such as Netscape and Microsoft, as well as

01:47:37 <bijan> widely-supported extensions to HTML.

01:47:51 <xbill> The Great Satan strikes again, corrupting standards and os on.

01:48:11 <xbill> (well, and OS's too, but I meant "so")

01:48:15 <bijan> The w3c take:

01:48:22 <bijan> What happened to HTML 3.0?

01:48:24 <bijan> HTML 3.0 was a proposal for extending HTML published in March 1995. The Arena browser was a testbed implementation, and a few other experimental implementations have been developed including UdiWWW, Emacs-W3, etc.

01:48:30 <bijan> However, the difference between HTML 2.0 and HTML 3.0 was so large that standardization and deployment of the whole proposal proved unwieldy. The HTML 3.0 draft has expired, and is not being maintained.

01:48:48 <bijan> I find that a *tad* disingenuous, given the complexity of subsequent HTML.

01:49:13 <bijan> But, maybe I'm wrong.

01:49:20 <bijan> Math layout is tricky.

01:50:12 * DanC tries to catch up...

01:50:22 <DanC> HTML 3.0 wasn't an IETF project. It was Dave Raggett's project.

01:50:23 <bijan> Hey dan! I hope I've not totally misrepresented the history :)

01:50:28 <DanC> He did implement it. Grab Arena.

01:50:38 <xbill> I think I have arena.

01:50:52 <bijan> There seems to be two others?

01:51:06 <bijan> UdiWWW?

01:51:22 <AaronSw> Well, there was an HTML-WG, and he was writing an IETF draft, so why wasn't it an IETF project?

01:51:52 <DanC> well, you could call it an IETF project, in that sense. But the WG never endorsed it.

01:51:57 <bijan> Wow, arena looks pretty cool.

01:53:13 * bijan hit a frustration with Amaya: No obvious ways to switch from XHTML viewing to HTML viewing when the XHTML turns out to be not well formed.

01:53:21 <DanC> the timeline went something like: libwww release, HTML psuedo-spec, Mosaic 1.x, 2.x, HTML+ drafting starts, HTML 2.0 drafting starts (May 1994), ...

01:53:45 <bijan> While I endorse catestrophic failure, I also want work arounds! :)

01:54:27 <deltab> the lack of workarounds could be considered a catastrophic failure :-)

01:54:33 <DanC> workaround: just go into "view source" mode, fix it, and synchronize, no?

01:54:46 <DanC>http://www.w3.org/MarkUp/html-spec/

01:54:47 <dc_rdfig> A: http://www.w3.org/MarkUp/html-spec/ from DanC

01:54:57 <DanC> A:HTML 2.0 stuff, including history of HTML

01:54:57 <dc_rdfig> commented item A

01:54:58 <bijan> Eh..tha'ts not convenient...espeically when I don't own the XHTML :)

01:55:30 <AaronSw> The idea is to give you incentives to notify the author. ;-)

01:55:39 <DanC> A:hmm... should be augmented by milestone browser releases. Didn't Kevin Hughes have a timeline with that stuff?

01:55:39 <dc_rdfig> commented item A

01:56:08 <DanC> the idea is actually to concentrate on implementing emerging standards, and not bother with bugward compatibility, which takes 5x as long.

01:56:34 <bijan> sure...but amaya handles "regular" HTML too.

01:56:46 <deltab> hmm, how about it automatically making a local copy which you can change and them send the diff to the author?

01:56:49 <bijan> So if I take off the doctype, it can render the page.

01:56:51 <DanC> i.e. it's a sad fact that 99% of the content in the web is dreck, but I don't want the Amaya developers to spend one minute bothering with that dreck.

01:57:06 <DanC> doctype isn't necessary.

01:57:17 <bijan> The 1 minute I want is a button "Browse as broken html" :)

01:57:30 <deltab> Tidy could be used as a preprocessor

01:57:39 <DanC> yes, I suggested integrating tidy into Amaya...

01:57:39 <bijan> That woudl also work.

01:57:58 <xbill> One of the driving notions behind HTML was to enable one to follow hypertext links from a paper to a paper it cited, right?

01:58:09 <xbill> e.g. for scientific publications?

01:58:19 <bijan> The page in question had an img tag without the />

01:58:29 <bijan> Making it not well formed XML.

01:58:35 <bijan> But it was all perfectly fine HTML.

01:58:39 <DanC> my Aug 200 suggestion for integrating tidy into Amaya: http://lists.w3.org/Archives/Public/www-amaya/2000JulSep/0237.html

01:58:50 <bijan> But amaya stops hard on a well-formedness error.

01:59:10 <bijan> Which is a Good Thing....except if you just want to read the page ;)

01:59:47 <bijan> If it didn't *already* do HTML, that would be fine. I wouldn't want them to do a thing.

02:00:07 <bijan> But Amaya does do HTML...that's why changing hte doctype enabled amaya to render this page.

02:02:10 * DanC is getting a little edgy about the lack of progress on the RDF primer.

02:02:24 <bijan> Hmm?

02:02:53 <DanC> a couple weeks ago, a subgroup of the RDF WG started working on a primer. But not much has come of it.

02:03:05 <AaronSw> Perhaps we should round up some volunteers. Bijan, want to write something for the primer? ;)

02:03:08 <DanC> hmm... there's a big contribution this evening, though.

02:03:57 <bijan> Aaron: No! :)

02:04:09 * bijan is dying on the next rdf and prolog article :)

02:04:51 <DanC> bummer... scroll bar gets very small. lots of text.

02:05:00 <DanC> sigh... things-versus-their-names confusion: "the creator of [the particular Web page we're talking about] is "John Smith" "

02:05:16 <DanC> I don't think a 10 character string created a web page.

02:05:21 * DanC is away: family time

02:05:25 <bijan> Heheh. You missed the great Literal Debate :)

02:07:05 <bijan> Ooo, I like DanC's amaya proposal.

03:52:41 <AaronSw> Hi stefan

04:35:15 <GabeW> AaronSw - I like your description of semweb as enabling one to do a "JOIN" (in the SQL sense) of two "websites"

04:35:30 <AaronSw> Ah, thanks.

04:35:45 <AaronSw> I find that brings the point home to people used to RDBMS-style stuff

04:36:00 <GabeW> as we were describing the sw to look in #infoanarchy, I like the db view of the semweb a lot

04:36:27 <AaronSw> Yeah, it makes the most sense to me.

04:36:49 <AaronSw> Internet databases are cool stuff... let me see if i can find that url

04:37:03 <GabeW> I used that metaphor to describe the semweb (or at least RDF) to my business partner and he got it really quickly

04:37:43 <GabeW> (esp. in the context of a Metaverse-type scenario where the relationships between things are as various as the data living the universe)

04:39:29 <AaronSw> Hmm

04:39:35 <GabeW> hmm?

04:39:58 <AaronSw> Metaverse as in "Snow Crash"?

04:40:02 <GabeW> yeah

04:40:03 <GabeW> yup

04:40:21 <GabeW> I mean, there are a thousand ways of thinking about that

04:40:39 <AaronSw> Indeed. It's all a matter of your point of view.

04:40:48 <GabeW> I guess we take a very islands of data interacting with other islands of data sorta view

04:41:09 <GabeW> You could take a straightforward avatar/graphics view of things..

04:41:44 <GabeW> maybe Ive read the book too many times

04:42:58 <AaronSw> Heheh.

04:43:00 <AaronSw> It's a good book.

04:43:03 <GabeW> its fun

04:43:42 <GabeW> I mean, hey, I'd like to skate like that ;-)

04:44:21 <AaronSw> Yeah! :-)

04:44:22 <AaronSw>http://www.mrqe.com/

04:44:22 <dc_rdfig> B: http://www.mrqe.com/ from AaronSw

04:44:30 <AaronSw> B:|Movie Review Query Engine

04:44:30 <dc_rdfig> titled item B

04:44:52 <AaronSw> B:Never seen this before, but a good example of what the Web would be liked if every page had nice metadata attached to it

04:44:53 <dc_rdfig> commented item B

04:45:03 <AaronSw> B:Perhaps we can convince them of an RDF export?

04:45:03 <dc_rdfig> commented item B

04:45:46 <GabeW> B: But isn't their business based on providing that data with banner ads?

04:45:46 <dc_rdfig> commented item B

04:46:17 <AaronSw> I don't see any banner ads... oh, I must be filtering them with my ad blocker

04:46:23 <GabeW> yeah, they're there

04:46:28 <GabeW> annoying as hell too

04:47:19 <GabeW> I mean, that's actually an interesting general point - many businesses make money online by providing limited access through the narrow window of their ad-attached website

04:47:46 <GabeW> I mean, collection of quality data often costs $$

04:47:58 <GabeW> (not neccesarly in this case, but in the general case)

04:48:15 <AaronSw> Yep.

04:48:33 <AaronSw> Luckily we've found enough people willing to release data in RDF

04:48:50 <AaronSw> Hmm, IMDb seems to have taken down their free database service :-(

04:48:52 <GabeW> I mean, its sort of another dimension in open source

04:49:00 <GabeW> open source data

04:49:52 <GabeW> I wonder how different the economics are for "open source data" vs. "open source software"

04:50:11 <AaronSw> open source software is more fun ;)

04:50:48 <GabeW> well, that's cuz you like it - some people probably prefer the collection and sifting of datalike we enjoy writing code

04:51:24 <GabeW> I think my cousing had a job where he had to listen to old records and categorize them (essentially create metadata for them, IIRC)

04:51:39 <GabeW> I think he digged it because he got to listen to good music all day long

04:51:48 <AaronSw> I would've yought that was extremely unlikely until I heard that you enjoy fixing systems that have been r00ted. Now I know anything is possible.

04:52:02 <AaronSw> that being people who enjoy sifting thru data

04:52:02 <GabeW> did I say that?

04:52:14 <AaronSw> I think so, unless my memory blew up on me again.

04:52:17 <AaronSw> Or something similar.

04:52:18 <GabeW> I think I did

04:52:41 <GabeW> Actually "fixing systems that have been r00ted" almost always means reinstalling everyting - and that's not fun

04:52:55 <AaronSw> B:Also little known (and well hidden) is that the famous [IMDb|http://www.imdb.com] provides all of their data as [plain text files|http://us.imdb.com/interfaces] although they have draconian licensing schemes.

04:52:55 <dc_rdfig> commented item B

04:52:56 <GabeW> But its the detective part - finding out how and exactly what happened (to the extent it makes sense to )

04:53:32 <AaronSw> Ahh, I can sort of see that.

04:53:52 <GabeW> In some areas, you have to pay really skilled people a lot of money to create "metadata" (if you will) - like in the area of legal primary materials.

04:54:36 <GabeW> And while there are efforts to cull the efforts of people who'd like to contribute their "secondary" material, I imagine that secondary material is often of vastly different quality

04:55:15 <GabeW> vastly varying quality, that is

04:56:29 <AaronSw> Yeah, I don't think there's a whole lot of distinction between data and metadata.

04:56:38 <GabeW> I mean, if you are collecting relatively structured, "bright-line" meta data (like names of songs on a CD or authors of academic papers), then its not so hard to collect metadata, but try getting 5 experts, much less 100 non-experts to agree on what the movie "Terminator" is about

04:56:53 <GabeW> yeah, right - metadata is just a subclass of data

04:57:31 <GabeW> I'm focusing too much on metadata, but I think I'm getting my point across</dead-horse-floggin>

04:58:13 <AaronSw> :-)

05:55:07 <rillian> AaronSw: nifty about the imdb. didn't know that

05:55:30 <AaronSw> Yeah, they hide it well.

06:00:36 <rillian> oh, I think I did see this once before

06:00:41 <rillian> surprised it's still here :)

06:03:30 <AaronSw> Yeah, perhaps they just forgot to turn it off. ;-)

06:05:20 <AaronSw> Time for sleep... g'nite

07:10:37 <rillian> night

09:20:40 <dajobe> A:|HTML 2.0 materials

09:20:40 <dc_rdfig> titled item A

13:15:09 <AaronSw> good morning, good morning, good morning-ah!

13:15:09 Disconnected from irc.openprojects.net (ERROR :Closing Link: logger[tatooine.ilrt.bris.ac.uk] by benford.openprojects.net (Ping timeout for logger[tatooine.ilrt.bris.ac.uk]))

13:16:02 Topic now semanticWebChat: Join the radicalDataFaction! | http://www.w3.org/RDF/Interest#irc

13:16:02 Users on #rdfig: logger libby jang deltab larsbot chihchun +dajobe tim-gone oierw` DanC bijan xbill AaronSw sandro kifbot lasDesk dc_rdfig Iron_SpermWhale em

13:16:03 <ChanServ> This channel is logged and blogged: http://logicerror.com/rdfIRCWelcome

13:51:26 <AaronSw> In relation to the discussion last night about threading IRC conversations, perhaps we can emulate email to some extent.

13:51:31 <AaronSw> I've always liked this: http://web.lfw.org/ping/criticons/

13:56:11 <larsbot> do they use markers in front of their msgs to indicate hierarchy?

13:57:59 <AaronSw> Yeah.

14:08:54 <larsbot> hmmm. interesting idea

14:20:23 <larsbot> what is the gain, though?

16:56:35 * danbri waves beardily and heads off to the pub with ILRT RDF folk

16:56:43 <danbri> danbri is now known as danb_zzz

18:28:40 <AaronSw> beardily?

18:29:10 <AaronSw> larsbot, the gain is the ability to quickly skim a long discussion thread

18:29:39 <AaronSw> It makes the discussion far more valuable

18:41:03 <jhendler> Aaron - I'm thinking of having some students work on SW meets email (archive organization based on cmobination of metadata and SW markup) - perhaps this would also work with IRC threading.

18:41:28 <AaronSw> Ooh, that'd be quite interesting.

18:41:38 * bijan notes that the automatic catelog advice based on simple text analysis stuff *BELONGS* here :)

18:41:54 <AaronSw> Have you seen the work filsa is doing with RDF mail archives?

18:41:54 <larsbot> AaronSw: well, _do_ you gain that?

18:42:15 <larsbot> the demo site you pointed to wasn't that convincing

18:42:24 <AaronSw> Well, it's never been deployed to my knowledge, so I can't say for sure.

18:42:29 <AaronSw> But that's the hope.

18:42:40 <AaronSw> Obviously, in real life you'd tweak the software to improve things.

18:42:41 <larsbot> I'd be happy to try it out, but I'm kind of skeptical

18:42:54 <larsbot> sure

18:43:39 <larsbot> btw, is there an RDF bot that you can speak some RDF syntax to

18:43:48 <larsbot> and have it build the RDF model for you as you are talking with it?

18:43:56 <AaronSw> There's SW bot, who knows N3.

18:44:03 <larsbot> so you can talk n3 to it?

18:44:07 <AaronSw> Yep.

18:44:13 <larsbot> can you show me?

18:44:16 * larsbot is curious

18:44:18 <AaronSw> Sure... just a sec...

18:46:26 <AaronSw> Hmm, something's wrong...

18:46:44 <larsbot> the famous demo effect :)

18:46:50 <AaronSw> Heh

18:47:05 <AaronSw> self.sock.connect(server, port) fails... that shouldn't happen

18:47:31 <larsbot> what platform is it?

18:47:42 <AaronSw> Python on Mac OS X... it's worked for me before.

18:47:46 <larsbot> I'm having the same problem on redhat with kernel 2.2.19-somethng

18:48:00 <AaronSw> Perhaps a Python update?

18:48:10 <larsbot> haven't done one, but maybe I should

18:48:40 <AaronSw> Ahh, i'm running the wrong app

18:48:49 <larsbot> :-)

18:49:34 <larsbot> are you using irclib.py, BTW?

18:49:52 <AaronSw> ircAsync.py

18:50:14 <larsbot> interesting. where does that live?

18:50:34 <larsbot> irclib.py has no async support, so...

18:50:51 <AaronSw> It lives with swBot... http://dev.w3.org/cvsweb/2000/scribe-bot/

18:52:07 <AaronSw> OK, join me in #rdfbot

18:52:34 <larsbot> momenot

19:03:21 <larsbot> so, as I was saying, it would be easy to make a bot that could

19:03:29 <AaronSw> So lars was saying that swBot was very cool and I explained to him the relationship between N3, cwm and RDF.

19:03:35 <larsbot> do ltm instead of n3, and tolog instead of the logic stuff

19:04:06 <larsbot> whether that would make anyone any happier than swbot I don't know

19:04:10 <AaronSw> Heh.

19:04:15 <larsbot> except that it would make me happy, as it'd be a cool hack :)

19:04:33 <AaronSw> Hmm, I forgot to introduce myself. I'm Aaron Swartz, <http://www.aaronsw.com/> And you?

19:04:41 <larsbot> I guessed that

19:04:54 <larsbot> I'm Lars Marius Garshol <http://www.garshol.priv.no>

19:05:27 <AaronSw> Pleased to meet you.

19:05:42 * bijan notes the sniffing of uris.

19:05:46 <larsbot> the same :)

19:05:57 <AaronSw> The sniffing of URIs?

19:06:15 <larsbot> guess he's making some kind of dog analogy :)

19:06:19 <bijan> Like dogs sniffing each other's butts in greeting!

19:06:41 <AaronSw> Thanks for sharing. ;)

19:06:46 <bijan> One the internet, no one knows that you're a hyperintelligent result of cruel experiments on canines...

19:06:52 <bijan> ...until you give yourself away.

19:06:58 <AaronSw> lol!

19:07:09 <AaronSw> I always knew there was something special about bijan.

19:07:20 <larsbot> :-)

19:07:35 <bijan> er..actually, I was referring to you and lars :)

19:07:51 <bijan> You don't see *me* sniffing URIbutts!

19:08:33 <AaronSw> Well, you suggested the analogy, which implied dog metaphors were a mode of thought for you.

19:08:42 <AaronSw> :-)

19:09:06 <bijan> Actions speak louder than analogies, woof.

19:09:09 <bijan> Oops.

19:10:32 <larsbot> this cwm thing, it's just a playground for you RDF people, right?

19:10:32 <AaronSw> Well, bijan is a dog, lars is a bot, and I'm well... we don't know.

19:10:39 <larsbot> an Sw?

19:10:45 <AaronSw> Heh.

19:10:48 <AaronSw> cwm: Yeah, pretty much.

19:11:04 <AaronSw> Tim Berners-Lee's playground, mostly -- we're just happy users.

19:11:34 <larsbot> I see

19:11:46 <bijan> I know.

19:11:48 <bijan> What aaron is.

19:11:55 <larsbot> let's hear :)

19:11:55 <bijan> And "happy" is perhaps strong.

19:12:04 <bijan> But I won't *say*.

19:12:07 <AaronSw> Heheh.

19:12:16 <bijan> Oh ohoho. *Users*.

19:12:24 <bijan> I thought you meant "readers of the source code" :)

19:12:35 <AaronSw> Right.

19:13:08 <larsbot> the one tends to lead to the other, I guess

19:15:57 <larsbot> btw, aaron, what's your involvement with RDF?

19:16:34 <AaronSw> I'm on the RDF Core WG for the W3C.

19:16:44 <larsbot> right

19:16:53 <GabeW> the semantic web lives in aaron's head

19:17:04 <larsbot> not only there, I'd hope :)

19:17:26 <GabeW> we need RDF to point to neurons in his head

19:17:33 <larsbot> I've been trying to find out what the status of RDF Schema is

19:17:37 <larsbot> perhaps you can tell me?

19:17:51 <larsbot> been asking several times, but no luck

19:18:05 <AaronSw> :-)

19:18:12 <AaronSw> Hehe, GabeW.

19:18:22 <AaronSw> Well, what's unclear about RDF Schema's status.

19:18:43 <AaronSw> It's been a Candidate Recommendation for years, and now RDF Core is fixing it up.

19:18:51 <larsbot> ah, I see

19:18:59 <larsbot> so it's about to go R?

19:19:39 <AaronSw> That's a matter of some discussion.

19:19:46 <larsbot> I see :-)

19:19:58 <AaronSw> Some would like to fix the bugs and release it as a REC.

19:20:03 <larsbot> and others?

19:20:16 <AaronSw> Others would like to redo it to make it more in line with the new set of publications RDF Core is releasing.

19:20:22 <bijan> Yes!

19:20:28 <bijan> Please!

19:20:39 <larsbot> because RDF itself is being cleaned up slightly

19:20:47 <bijan> Massively, please!

19:20:49 <larsbot> they would like to see RDF schema remain in sync with RDF?

19:20:54 <AaronSw> bijan, why are you so enthusastic about that?

19:21:20 <AaronSw> larsbot, it's more than just that -- it's that most of the spec is covered by the model theory, that Web Ont is doing similar work, etc.

19:21:31 <bijan> Well, I'd really like a synched, uptodate, largely bug free, comprehensive RDF set of specs.

19:21:36 <AaronSw> RDF Schema itself is probably pretty unaffected by the changes we're making to RDF.

19:21:42 <bijan> For the "core".

19:21:44 <AaronSw> bijan, Yes!

19:22:05 <larsbot> hmmmmmm.

19:22:09 <larsbot> I think I see

19:22:15 * DanC tries to catch up...

19:23:37 <bijan> So, actually, I don't *really* care if things get released willy nilly :) I just want a date where stuff is largely together, largely coherent, and largely bug free that I can point to and say: That's the RDF I'm going to use and build stuff on.

19:24:05 <bijan> Since, to retro back to the mid 90's, RDF is sorta a "platform".

19:29:35 <larsbot> one thing I am curious about is how topic maps are perceived

19:29:39 <larsbot> by the RDF crowd

19:30:42 <AaronSw> Heh... they're the _other_ guys which are doing the same stuff but with different words and we need to invest on conversions between us.

19:30:47 <AaronSw> How do you guys think of us?

19:31:04 <larsbot> I could repeat your words :)

19:31:35 <larsbot> I'm not polite company :)

19:32:20 <larsbot> it seems to me that RDF is a lot simpler and lower-level that TMs

19:32:44 <larsbot> which means certain things become easier and others harder

19:32:52 <larsbot> this is all conjecture, though; I've never used RDF

19:32:58 <AaronSw> Hmm, it seems to me that TM is a lot simpler and lower-level than RDF... but I haven't researched it more than the presentations at RDF meetings.

19:33:21 <larsbot> RDF has a model consisting of one thing: the triple

19:33:27 <larsbot> topic maps are more complicated

19:33:32 * jhendler thinks that larsbot should also check out DAML if he's interested in seeing something that isn't lower level and simpler than TMs (of course, /me thinks everyone needs to check out DAML)

19:33:44 <AaronSw> Topic maps, from what I've seen have the node and the link...

19:33:44 <larsbot> I've looked at DAML

19:33:54 <AaronSw> jhendler was DAML Project Manager. ;)

19:33:57 <larsbot> node being topic, link being association, right?

19:34:07 <AaronSw> Yes, I think so.

19:34:08 <larsbot> I've looked at DAML, and AFAICT it's a schema language

19:34:17 <larsbot> well, topic maps have more than that

19:34:22 <larsbot> topics have names

19:34:23 <AaronSw> DAML: Heheh! I said the same thing, and no one has conviced me otherwise.

19:34:24 <dc_rdfig> Label DAML not found.

19:34:34 <larsbot> topics have occurrences

19:34:39 <larsbot> topics have subject indicators

19:34:49 <larsbot> associations are also more complex than triples

19:35:01 <larsbot> there are major similarities, but TMs are more complex

19:35:23 <jhendler> Well, I'm not sure I agree, DAML is like a schema language (and extends it), but comes with a more formal semantics and, more importantly IMHO, a bunch of terms for associating things from different ontologies/schemas

19:36:07 <jhendler> For example, I can say things are equivalentto each other, transitive, inverse, etc - that voacabulary is very powerful, and not found elsewhere in the SW

19:36:15 <AaronSw> That's a schema language to me.

19:36:19 <larsbot> ditto

19:36:40 <larsbot> in fact, topic maps have those same things simply by way of

19:36:44 <jhendler> well, depends on definition - most schema langauges cannot refer to things in other schemas, and cannot place restrictions on the schemas.

19:36:44 <larsbot> how the model is put together

19:36:57 <larsbot> they aren't features, they are just consequences of the structure

19:36:58 <larsbot> of the model

19:37:05 <AaronSw> Hmm, sounds like I need to research topic maps more.

19:37:19 <larsbot> in fact, RDF is like TMs with only topics and occurrences

19:37:24 <larsbot> no names, no assocs, no subject indicators

19:37:39 <jhendler> yes, topic maps are the only thing I've seen in the XML world with the idea that you can map terms from different schemas together, etc. Problem is they have a fairly ad hoc vocabulary and not enough of an underlying model.

19:37:56 <larsbot> I agree 300 %

19:38:03 <larsbot> the model is being created, though

19:38:10 <jhendler> example, some of the connector types are explained only by example - no underlying semantics makes it hard to predict how they go together.

19:38:19 <larsbot> connector types?

19:38:49 <larsbot> AaronSw: I'm finishing an RDF-topic maps comparison paper as we 'speak'

19:38:55 <larsbot> you could have a look if you wish

19:38:56 <AaronSw> Great!

19:39:17 <jhendler> sorry - am not plugged into TM terminology - what is the term for the various thngs that are graphed as circles with X's in them and the like.

19:39:24 <AaronSw> Heheh.

19:40:01 <jhendler> meanwhile, we are hoping in the W3 SW Coordination Group to see if we can get TM and webont (or RDFS) activities more lined up.

19:40:11 <larsbot> that would be really good

19:40:25 <larsbot> circles with Xs in them? topics, perhaps?

19:41:19 <jhendler> would be nice to end up with one ontology/meta-schema approach to the world instead of two -- and the formal semantics of the DAML stuff could be applied (dare I say it in this forum) to a non-RDF-based model -- although would need to be a model that had much of the same expressivity

19:42:10 <larsbot> ok, here's the paper: <http://www.garshol.priv.no/tmp/paper.zip>

19:42:19 <AaronSw> A zip?

19:42:26 <larsbot> yeah, containing HTML + PNGs

19:42:41 <larsbot> whoa, wait a sec

19:43:43 <larsbot> didn't think :-) <http://www.garshol.priv.no/tmp/paper.html>

19:43:58 <larsbot> it's kinda rough, but might still help

19:44:11 <AaronSw> thanks

19:44:18 <larsbot> jhendler: I agree it would be nice

19:44:28 <larsbot> I could see some of the DAML stuff going into a TM schema lang

19:44:46 <larsbot> would be nicer still if the TM and RDF models were unified, though

19:45:58 <AaronSw> larsbot, images aren't showing up.

19:46:04 * larsbot check

19:46:27 <AaronSw> Heheh: src="file:/home/larsga/cvs-co/doc/articles/tm-rdf-daml-oil/things.png"

19:46:37 <larsbot> ach!

19:46:44 <larsbot> my XSLT stylesheet is playing tricks with me

19:47:15 <larsbot> fixed! (thanks!)

19:47:22 <jhendler> lars - I'm in complete agreement - if TM/RDF unified, then DAML sits on top neatly (i.e. DAML is RDF and serializes to XML as RDF does).

19:48:26 <larsbot> the thing is: how can we unify TMs and RDF without breaking either?

19:48:35 <AaronSw> Lars, this looks paper like it could really help with the unification...

19:48:51 <larsbot> if so I'd be pleasantly surprised :)

19:50:09 <AaronSw> Are you ready to "chump" it?

19:50:13 <larsbot> chump?

19:50:23 <AaronSw> Yeah, post the URL to http://rdfig.xmlhack.com/

19:50:37 <sbp>http://lists.w3.org/Archives/Public/www-rdf-interest/2001May/0208

19:50:37 <dc_rdfig> C: http://lists.w3.org/Archives/Public/www-rdf-interest/2001May/0208 from sbp

19:50:38 <larsbot> this is a temporary URL

19:50:51 <sbp> C:|Re: Do these folks know about RDF?

19:50:51 <dc_rdfig> titled item C

19:51:35 <larsbot> this paper is still being cleaned up

19:51:45 <sbp> C:"This is not AI. That was one of those "six months away" things with people holding down the skirts up which they were blowing smoke. If the question is "will it scale?" the answer is that a version of it already has. When the Web Tiles are as cheap as smart credit cards they will practically "paper the planet"."

19:51:46 <dc_rdfig> commented item C

19:51:52 <AaronSw> larsbot, That's what I thought

19:52:03 <larsbot> once it's clean I'd be happy to, though I guess I'd better check

19:52:14 <larsbot> whether the GCA likes having XML 2001 papers posted 2 months

19:52:17 <larsbot> before the conference starts

19:52:28 <AaronSw> Sean, what brought C up again?

19:52:58 <sbp> Ah, so many things!

19:53:08 <sbp> dataplesh, for one

19:53:39 <AaronSw> Hmm.

19:53:47 * sbp smiles at the thought of RDF being turing complete

19:54:02 <sbp> s/turing/Turing

19:54:24 <AaronSw> Heh.

19:54:55 <sbp> we could have RDF documents that are XML RDF parsers! yeah!

19:55:43 <AaronSw> Huh?

19:56:25 <sbp> TMs can complete any programming activity, so they can darn well parse XML RDF, as long as you had something to interpret the TM

19:56:53 <sbp> just as XSLT can

19:57:13 <sbp> Gotta run

19:57:20 <AaronSw> c'ya

19:57:30 * em joins and catches up

19:58:20 <AaronSw> Hi em!

19:58:29 * em scans over tm and rdf discussions

19:58:36 <em> hey AaronSw :)

19:58:50 * em loads http://www.garshol.priv.no/tmp/paper.html

19:59:17 <em> larsbot... you still around? are you looking for comments?

19:59:23 <jhendler> hmm, is he asking if the authors of the Sci Am article know about RDF?

19:59:30 <larsbot> comments would be very nice

19:59:50 <em> how about questions first?

19:59:59 <AaronSw> jhendler, yeah, those wackos... probably just skimmed the material on the RDF site before writing

20:00:01 <AaronSw> ;-)

20:00:04 * jhendler shoot - my irc drops some of my messages - that was supposed to be in response to C:

20:00:23 <larsbot> em: questions are fine

20:00:42 * em likes the pictures as they raise some additional questions

20:00:44 <jhendler> C: Authors of the Semantic Web article felt we were being restrained compared to our usual statements about the potential of the SW

20:00:46 <dc_rdfig> commented item C

20:01:01 <larsbot> right :) currently working on more of them

20:01:11 <bijan> The Sci Amer one?

20:01:20 <em> i've been noodling on tm & rdf integration for a while... so appologies in advance if i'm a bit off in the weeds here

20:01:52 <larsbot> no prob

20:01:54 <AaronSw> restrained? everyone's quoted that article as how crazy SW folks are!

20:02:09 * jhendler issue of whether SW is for real drives me crazy - it's already here and starting to evolve - It's like hearing from the people who told me this web thing would never catch on.

20:02:11 <bijan> I hereby reference it as an example of how crazy SW folks are.

20:02:23 <bijan> And it's *very* crazy :)

20:02:31 <jhendler> web was years old before media, comedians, and all discovered it!

20:02:38 <em> larsbot... why is it your assuption that the two images in figure ?? (One relationship in topic maps.. The same relationship in RDF) actually are indeed the same thing just in two different modeling languages?

20:03:09 * sandro thinks jhendler needs a big, bold, quote at the top of w3.org/2001/sw, defining in the SW in terms that behave as he describes. Perhaps XML is part of the semantic web?

20:03:20 <larsbot> em: because to a human (me) they mean the same thing

20:03:21 * em find this article very helpful btw

20:03:26 <larsbot> thx

20:03:41 <larsbot> if the human (me) is getting something wrong, please let me know

20:03:55 <em> but it seems that you've left a bit of additional information (e.g. role or type) out of the rdf representation?

20:04:26 <larsbot> that's because topic maps and RDF don't model things the same way

20:04:33 <larsbot> aaron gave me an example earlier

20:04:42 <larsbot> :Dogs :get :fleas, he said

20:04:43 <em> agreed (in part)... but I think there closer that this

20:04:57 <larsbot> get([dogs] : getter, [fleas] : gotten), we'd say

20:05:00 * DanC is back (gone 17:59:39)

20:05:02 <AaronSw> larsbot, that's a bad example, really -- but the canonical swBot one.

20:05:19 <larsbot> I know -- it gets the point across, though

20:05:30 <larsbot> em: you think TMs and RDF are closer than this?

20:05:32 <larsbot> how so?

20:05:33 <em> you have some additional information here that didn't get transfered (namely the roles)

20:05:45 <AaronSw> Yeah, the major difference is that TM can have n-tuples where RDF is just triples

20:06:02 <AaronSw> you can sorta fake it with anon-nodes though...

20:06:09 <larsbot> well, I'm not sure that's the right way to see it

20:06:10 <em> exactly

20:06:22 <larsbot> if you were to model companies and employees in RDF

20:06:29 <larsbot> would you mess around with roles?

20:06:38 <larsbot> I'm fairly sure you wouldn't

20:06:42 * em has to run but will write this up and add this to chump

20:06:48 <AaronSw> :Comapany :CEO :John ; :CEO a :Employee .

20:07:02 <AaronSw> :CEO :subClassOf :Employee, actually

20:07:11 <em> yes... somethign like this is what i was thinking

20:07:22 <larsbot> right, but you'd say that CEO is a property of the company, right

20:07:25 <em> i think you'll find this a lot closer than you might have first thought...

20:07:39 <larsbot> there's something you're not telling me :-)

20:07:42 <AaronSw> Isn't CEO a property of the company?

20:07:48 <larsbot> in RDF it is

20:07:58 <larsbot> in topic maps there are no properties

20:08:05 <sandro> Some CEOs act like the company is their property, not the other way around.

20:08:13 <bijan> Faugh, marxist TM claptrap!

20:08:14 <AaronSw> Heheh

20:08:16 <larsbot> now we're getting there

20:08:27 <larsbot> in topic maps you'd assert an association between the two

20:08:33 <larsbot> of type employed-by, I guess

20:08:39 <larsbot> the company playing the role of employer

20:08:47 <larsbot> the CEO playing the role of employee

20:08:54 <larsbot> or perhaps CEO, being a subclass of employee

20:09:37 <larsbot> it means the same thing, but it's said in a different way

20:10:03 <larsbot> employed-by([steve] : ceo, [ontopia] : employer)

20:10:04 <larsbot> or

20:10:22 <larsbot> employed-by([steve] : employee, [ceo] : position, [ontopia] : employer)

20:10:31 <AaronSw> There is no escape... RDF will assimilate you. ;-)

20:10:41 <larsbot> let's not go there :)

20:10:51 <larsbot> am I making sense, or not?

20:11:00 <AaronSw> Yep, you're making sense.

20:11:08 <AaronSw> You can say things multiple ways.

20:11:17 <larsbot> if you look at the RDF-to-topic map-mapping section

20:11:22 <larsbot> things may become even clearer

20:11:54 <larsbot> RDF property gives you assoc type + roles in the TM

20:11:58 <larsbot> provided someone tells you what they are

20:12:33 <AaronSw> Hmm, in your examples, Ontopia employs your mailbox and not you. ;-)

20:12:52 <AaronSw> Oh, I mean their homepage employs your mailbox.

20:13:01 <larsbot> not so

20:13:14 <AaronSw> 1 : {mailto:larsga@ontopia.net,

20:13:17 <AaronSw> http://psi.ontopia.net/example/employed-by,

20:13:17 <AaronSw> http://www.ontopia.net}

20:13:24 <larsbot> yeah

20:13:38 <larsbot> what URI would you choose to indicate me in RDF?

20:14:00 <AaronSw> Perhaps a tag: or an anon-node.

20:14:14 <larsbot> right. so you wouldn't use "real" URIs?

20:14:23 <AaronSw> [ :homepage <http://www.ontopia.net> ] :employs [ :email <mailto:larsga@ontopia.net> ] .

20:14:27 <AaronSw> A tag is a real URI.

20:14:50 <larsbot> yeah, but I mean URIs referring to network-retrievable resources

20:15:07 <AaronSw> tag:me@aaronsw.com,2001-10-15:Largsa

20:15:17 <AaronSw> Why do they have to be network-retrievable?

20:15:31 <larsbot> I'm not saying they have to; I'm just asking

20:15:38 <deltab> larsbot: you're not network-retrievable, are you?

20:15:49 <AaronSw> exactly

20:15:51 <AaronSw> it's confusing...

20:15:51 <larsbot> not until someone implements RFC 1437, no

20:16:00 <larsbot> what is confusing?

20:16:28 <AaronSw> Referring to something with a network retrievable entity

20:16:38 <larsbot> not necessarily

20:16:51 <larsbot> in RDF I guess it would be

20:16:54 <larsbot> in topic maps you can say

20:17:02 <larsbot> a) this resource is what the topic is about

20:17:09 <larsbot> b) this resource explains what the topic is about

20:17:45 <larsbot> I'm so TM-brainwashed I didn't consider that this might be confusing in RDF

20:17:57 <larsbot> one thing to change in the paper :-)

20:18:00 <AaronSw> Yes, this was a source of much confusion when the TM fellows came to explain things to us.

20:18:10 <larsbot> SRN & MB?

20:18:14 <larsbot> I can imagine :)

20:18:14 <AaronSw> Yep.

20:18:30 <larsbot> did you learn anything from their presentation?

20:18:44 <AaronSw> Some, I think... I've been thru two (three?) of them by now.

20:18:54 <AaronSw> Yes, subject indicators (a) are generally done with triples off of anon-nodes (bNodes)

20:19:08 <larsbot> well

20:19:22 <larsbot> that's inside their model of topic maps

20:19:28 <larsbot> not everyone buys that model

20:19:50 <AaronSw> Hmm.

20:20:30 <larsbot> <http://www.y12.doe.gov/sgml/sc34/document/0242.htm>

20:21:06 <larsbot> <http://www.y12.doe.gov/sgml/sc34/document/0240.htm>

20:21:10 <larsbot> see point 2 in the last one

20:21:27 <larsbot> N243 is the SRN&MB model

20:21:43 <AaronSw> Yes, we've heard about some of the TM turmoil

20:21:49 <larsbot> turmoil? :-)

20:22:16 <AaronSw> Are you on the opposite side of N243?

20:22:33 <larsbot> yes

20:22:40 <larsbot> though I'm still trying to work out what I really think of it

20:22:57 <larsbot> it does bring us a lot closer to RDF

20:23:03 <larsbot> but it seems to be doing too many things at once

20:23:18 <larsbot> :n243 :name "PMTM4", btw

20:23:25 <AaronSw> Too many names!

20:23:29 <AaronSw> :-)

20:23:36 <larsbot> yah :)

20:24:02 <larsbot> so what the last URI is saying is: work out what the relationship

20:24:05 <larsbot> between these two things are

20:25:26 <larsbot> I'm curious about what made you say "turmoil", though

20:25:28 <AaronSw> I see.

20:25:48 <danbri> danbri is now known as danb_zzzz

20:25:57 * danb_zzzz waves

20:26:03 <danb_zzzz> say hi to chaaaalz

20:26:35 <AaronSw> hi chaaalz

20:27:20 <AaronSw> sleepchatting is bad for your health, you know ;)

20:27:33 <danb_zzzz> wazzzzat you sayzzz?

20:27:59 <AaronSw> wazzzzout!

20:28:16 <AaronSw>http://pigdog.org/auto/mr_bads_list/shortcolumn/1914.html

20:28:16 <dc_rdfig> D: http://pigdog.org/auto/mr_bads_list/shortcolumn/1914.html from AaronSw

20:28:21 <AaronSw> D:|Things to Say When You're Losing a Technical Argument

20:28:22 <dc_rdfig> titled item D

20:28:40 <AaronSw> D:An oldie but a goodie. Probably chumped before, but so funny I just had to read it again.

20:28:40 <dc_rdfig> commented item D

20:29:44 * danb_zzzz finds a long reply from Pat Hayes in his inbox; postpones reply 'till tommorrow

20:30:08 * jhendler has anyone ever got a short reply from Pat Hayes :->

20:30:16 <AaronSw> Yes! :)

20:30:30 <danb_zzzz> does he type fast or think fast or both I guess.

20:30:55 <danb_zzzz> always nice to get a reaction anyway; was dissapointed not to have got a reply before, seeing as it was reasonably controversial.

20:31:15 <sandro> This latest from him is a doozie. But expected. I knew when I saw that one from danbri that we had the irresistable force and immovable object in very close proximity.......

20:31:33 * danb_zzzz ponderzzz a reply

20:31:58 <AaronSw> Can someone drop me a URI? is this on rdf-logic?

20:32:20 <danb_zzzz> rdf-rulez

20:32:36 <danb_zzzz> which needs to be properly announced at some point soon, btw.

20:33:12 <sandro>http://lists.w3.org/Archives/Public/www-rdf-rules/2001Oct/0043.html

20:33:12 <dc_rdfig> E: http://lists.w3.org/Archives/Public/www-rdf-rules/2001Oct/0043.html from sandro

20:33:14 <danb_zzzz> heh, that's the longest reply I've ever gotten from Pat on-list

20:33:19 <jhendler> yes, rdf-rules is going well, but somehow got jump started without the usual announcements.

20:34:00 <AaronSw> E:Heheh: """Wow, "RDF family"? That's a new term in my lexicon. Sounds like a TV series."""

20:34:01 <dc_rdfig> commented item E

20:34:24 <danb_zzzz> EricP launced it. I was a bit grumpy about the list scope so left it to churn. Actually turned out ok, though scope w.r.t. the other lists is rather vague.

20:36:02 <sandro> You have to know the difference between logical inference and logical implication to understand whether something is germaine to rdf-rules or rdf-logic. That's a bit crazy.

20:36:32 <sandro> That difference is subtle of often irrelevant.

20:36:37 <danb_zzzz> wanna write a faq entry for it?

20:36:38 <sandro> s/of/and

20:36:48 <sandro> bwahaha

20:38:11 <sandro> I asked the question and Drew and some others did reply. You just have to find where. :-) (alas, the subject isn't telling me. hrm)

20:40:21 <sandro> Logical inference is where a program takes one knowledge base and changes it (just by adding stuff, unless it's a non-monotonic logic) into another, in certain pre-programmed patterns. These transformations are inferences, and the patterns of them can be called inference rules.

20:41:44 <sandro> Logical implication is where you have a logical expression which, to be true, requires other things to be true.

20:43:43 <sandro> The rule "If Sam is outside, then the sun is shining" could be considered either way. If you imagine a program checking to see if Sam is outside, and if he is, then saying "The sun is shining", that's inference. If instead it reasons about it, that "we know Sam is outside" therefor it must also be true that "The sun is shining", that's implication. Implication would let us know Sam was inside if we knew it was raining.

20:44:23 <sandro> As far as I can tell, only inference rules are in-scope for rdf-rules, and logical implication discussions still belong on rdf-logic.

20:44:50 * danb_zzzz heads off; night all

20:44:53 <sandro> How was that?

20:47:06 <AaronSw> Hm, seems like the same stuff used in different contexts, no?

20:47:18 <AaronSw> Is there acutally a difference between the rules themselves?

20:57:42 <sandro> I suppose that's the source of the confusion. The same language works for both kinds of rules. Inference rules can sometimes have more expressive languages, though.

20:58:53 <larsbot> could you say it's different ways to use the same rules?

20:58:58 <sandro> But I think there are cases where it's very hard to tell the difference. I think n3 log:* means to be implication, but the cwm -think interface is very much like you'd expect for inference. Without negation, it's hard to find a test-case for the difference (or perhaps its impossible).

21:00:16 <sandro> I don't know, larsbot. I'm not familiar with the cases where the distinction becomes really important, which I think is when you get into non-classical logics, or something.

21:00:58 <larsbot> non-classical?

21:01:58 <sandro> It seems to me that we just want implication, and to use something standard like Resolution for our (one and only) inference rule. But there are people who seem to think there are much better and more interesting things out there (in places I don't understand yet).

21:02:21 * bijan wonders if sandro is talking about syntax version semantics.

21:02:43 <bijan> I.e., "inference" often refers to "syntactic" inference.

21:02:59 <bijan> Actually, I tend to use implication for syntax and entailment for semantics.

21:03:05 <sandro> non-classical ... ? I don't really know. One nice reference is http://www.earlham.edu/~peters/courses/logsys/nonstbib.htm

21:03:32 <bijan> Non-classical: deny the law of non-contradiction or of excluded middle, usually.

21:04:48 <larsbot> I think the answer is that I need to learn more logic

21:04:58 <larsbot> to be able to grasp what non-classical logic is

21:05:19 <bijan> In "classical" logic, you can't have both p and not-p simultaneously true.

21:05:25 <sandro> I don't see your syntax/semantics distinction, bijan. Formal logic is just manipulating syntax, and if the logics are sound, they tell us corresponding things about the world.

21:05:26 <DanC> non-classical: or go non-monotonic

21:05:29 <bijan> And every sentence is either true or false.

21:05:52 <larsbot> p and not-p: certainly

21:06:06 <bijan> sandro: monus ponens is a "syntactic" inference rule. It happens to be sound in some logics.

21:06:06 <larsbot> I thought everything went haywire if that were true

21:06:38 <sandro> So what's a "semantic" inference rule, Bijan?

21:07:03 <bijan> That if the members of the set (P, P->Q) are all true, so too, is Q

21:08:05 <bijan> syntactic modus pones is, Give the formula P, and the formula P->Q, you may conclude Q.

21:08:23 <sandro> That's like a one level up rule, a language where you can (and just did) express modus ponens?

21:08:35 <bijan> That these line up in FOSL is why it's sound.

21:08:57 <bijan> Both these rules are expressed in a metalangauge, yes.

21:09:20 <bijan> It's the difference between |- and |=

21:09:42 <sandro> But your "semantic inference" language is two levels up, right?

21:09:47 <bijan> Huh?

21:10:04 <bijan> P, P->Q entails Q

21:10:12 <sandro> a meta-meta-language.

21:10:12 <bijan> P, P->Q also implies Q

21:10:23 <bijan> It entials Q by version of the semantics.

21:10:44 <bijan> It implies it by virture of the inference rule Modus Ponens.

21:11:15 <bijan> In FOSL, with the standard interpretation of ->, Modus Ponens is sound.

21:11:31 <bijan> That is, when you have an implication, you have an entailment.

21:12:39 <lilo> [GlobalNotice] Hi all. Services will be back up in a moment, after some splits. Please bear with us.

21:13:20 <sandro> So my question is: why (assuming classical logic) would you ever want an inference rule other than than Resolution (or maybe Modus Ponens if you want to be a bit simpler).

21:13:40 <bijan> What do you mean?

21:14:10 <ChanServ> This channel is logged and blogged: http://logicerror.com/rdfIRCWelcome

21:14:10 <bijan> Assuming you want to limit yourself to sound rules, derivived rules are very convenient.

21:14:17 <bijan> Did you get that?

21:14:34 <sandro> I got "Assuming...."

21:15:05 <bijan> You may need double negation, or some proof strategies too.

21:15:18 <bijan> Modus Ponens alone won't do it.

21:15:19 <sandro> Derived rules - meaning you look for implications inside the sentence and pull them out as inference rules, then use them on the sentence?

21:15:30 <bijan> But otherwise, the version derivied rules make life easier and proofs shorter.

21:15:37 <bijan> s/version/various/

21:15:54 <bijan> I mean conjunction and disjuntion are just pleasent to use :)

21:16:20 <bijan> It's like asking, 'Why not just use the sheffer stroke?" Answer: "Becuase is super sucks in normal use."

21:16:24 * AaronSw returns

21:16:27 * sandro looks confused. Who said anything about conjunction and disjunction?

21:16:56 <bijan> I'm talking about the rules. A, B |- A&B, A |- AvB

21:17:22 <bijan> Perhaps I don't understand your question.

21:17:42 <bijan> I took it to be, "why have any but the minimum sufficient rules possible in a formal system."

21:18:09 <bijan> Answer: Because derived rules (i.e., strictly unnecessary but handy) are very convenient.

21:18:29 <sandro> And your answer is: "Because sometimes it's more efficient for computers and easier for humans to deal with"

21:18:39 <sandro> IE for convenience, pure and simple.

21:18:49 <bijan> Assuming that the rules are strictly derived, yes.

21:19:07 <bijan> You might add an unsound, but good, rule.

21:19:31 <bijan> That would enable more powerful, though sometimes faulty, inferences.

21:19:53 <bijan> You might add more powerful rules that are sound, but compromise decidability.

21:19:56 <bijan> Etc. etc.

21:20:15 <sandro> That seems like a bad idea, but I suppose if you know what you're doing, using a chainsaw while naked is okay.

21:20:40 <bijan> Hmm?

21:20:44 <AaronSw> chainsaw: That's a quote!

21:20:48 <sandro> (That's re: unsound. I have no problem with compsomising decidability.)

21:20:53 <bijan> unsound rules are great.

21:20:59 <bijan> Most forms of inference are unsound.

21:21:04 <sandro> Example...?

21:21:05 <bijan> Inductive inference.

21:21:11 <bijan> Abductive inference.

21:21:37 <sandro> Yeah -- things I don't understand. :-)

21:21:44 <bijan> Yes you do.

21:21:54 <bijan> you just don't have a theoretical explanation that makes you happy :)

21:22:18 <sandro> Is the proof that inductive inference is unsound from Godel?

21:22:23 <bijan> No.

21:22:43 <bijan> Erhm.../me isn't clear what proof is needed exactly.

21:23:13 <bijan> Take cartoon statistical generalization: This crow is black...that crow is black....n crows are black. All crows are black.

21:23:51 <sandro> That's a "good" but incorrect inference. That's not even induction, that's incorrect induction, yes?

21:23:54 <bijan> i trust the counterexample of an albino crow is obvious.

21:24:11 <bijan> It's generally considered a form of induction.

21:24:42 <bijan> But it's clearly both invalid and unsound.

21:24:44 <sandro> I remember from nth grade that you needed to prove P(n) -> P(n+1) for induction.

21:24:57 <bijan> Er...deductively invalid.

21:25:03 <bijan> That's mathematical induction.

21:25:09 <bijan> Which is a deductive technique.

21:25:26 <sandro> So were you saying the crow inference was a good?

21:25:46 <bijan> Ok, let's say that there are no non-black crows.

21:26:00 <bijan> Then the conclusion is correct.

21:26:05 <bijan> It is supported by the evidence.

21:26:10 <AaronSw> Sandro, don't forget the basis step! (/me has a test on mathematical induction tomorrow)

21:26:49 <sandro> I would never forget it, Aaron. :-) I aced that test, oh n+1 years ago.

21:27:03 <AaronSw> ;-)

21:27:21 <bijan> It's not sound because the conjoining the premises as the antecedent of a conditional wiht the conclusion as the consequent isn't a taugtology.

21:27:47 <bijan> An interpretation showing this is a world where theres N black crows and one white crow.

21:28:35 <sandro> Hey -- what's the difference between antecedent and premise? (or conclusion and consequent?) Is it this same inference vs. implication distinction?

21:29:02 <bijan> C1&C2&C3...&Cn->AllCrowsAreBlack

21:29:19 <bijan> That's a single sentence.

21:29:34 <bijan> Everything before the -> is the antecedant of the conditional.

21:29:52 <bijan> it, itself, is a single sentence, a long conjunction.

21:30:09 <bijan> AllCrowsAreBlack is teh consequent of the conditional.

21:30:26 <sandro> The conditional is itself also a sentence, of course.

21:30:35 <bijan> The corresponding argument is something like: C1. C2. C3...Cn. |- AllCrowsAreBlack.

21:30:39 <bijan> > C1&C2&C3...&Cn->AllCrowsAreBlack

21:30:40 <bijan> > That's a single sentence.

21:30:45 <bijan> ^^^^^^

21:30:52 <bijan> That that refers to the conditional :)

21:31:16 <sandro> Ah.

21:31:18 <bijan> In the argument there are n sentences which form the set of the premises, and one sentence which is the conclusion.

21:32:36 <bijan> erhmm.../me can recommend some handy logic texts :)

21:33:09 * bijan finds it handy to go over them now and again.

21:33:21 <sandro> In your "if the members of the set (P, P->Q) are all true, so too, is Q", would you call "P" and "P->Q" premises or antecedents?

21:33:57 <bijan> Premises.

21:34:19 <sandro> Good.

21:34:45 <bijan> entailment and implication are relationships *between* sentences (or sets of sentences)

21:35:15 <sandro> So it is (inference, premise, conclusion, www-rdf-rules) and (conditional, antecedence, consequent, www-rdf-logic).

21:35:22 <bijan> And folks are consistent about which is entailment and which is implication (i.e., you may see implication used for entailment)

21:35:38 <bijan> sandro: I have no idea :)

21:35:58 <sandro> You agree about the first three terms of each quad, though?

21:36:27 <sandro> Pat Hayes was using "implication" as you use "conditional", to name that kind of "rule".

21:36:40 <bijan> Yes, they are called implications too :)

21:36:51 <sandro> You mean "foks are INconsistent" dont you?

21:36:59 <bijan> Thus the "parodoxes of material implication"

21:37:01 <bijan> Er.. yes.

21:37:43 <bijan> I think there's some consensus in the philosophy/logic world to call the sentence "conditional", the syntactic inference "implication, and the semantic inference, "entailment"

21:37:53 <bijan> I know several books that try to keep that straight.

21:38:03 <bijan> Of coruse, it might be parochical to me.

21:38:05 <sandro> Yesterday I was flipping through Principia Mathematica and was really surprised at how similar the language, notation, and subject matter was. Long before computers, much of this was the same.

21:38:18 <bijan> About all of it was.

21:38:24 <bijan> PM really fixed the notation.

21:38:45 <sandro> The use of dots instead of parentheses is a little odd, though. :-)

21:38:55 <bijan> Yes. Except, thank god, for that.

21:39:14 <bijan> Quine's methods of logic still uses the dots.

21:39:21 <bijan> And Mike Resnick still teaches it at UNC.

21:39:33 <sandro> Well, the dots actually make it easier to match up the parens, I suspect.

21:41:37 <sandro> Its astoundingly hard to learn this terminology passively -- just reading. (which is why I go ahead and reveal my ignorance here, because if I get into to the conversation, I learn a lot more. A good little tidbit for theory-of-learning people. :-)

21:43:09 <sandro> I'm pretty sure the AI community (or at least Norvig/Russel & Pat Hayes, my main sources) use "entailment" to cover syntactic inference, but maybe I've just misunderstood.

21:43:40 <bijan> I wouldnt' be surprised.

21:43:42 <sandro> (this is of course why people go to *school* for this kind of stuff. :-)

21:43:43 <bijan> It's very common.

21:45:56 <bijan> And get *out* of schoool :)

21:49:37 <sandro> Hmmm -- Russell/Norvig (p158) -- "We want to generate new sentences that are necessarily true, given that the old sentences are true. This relationship between sentences is called entailment, and mirrors the relaton of one fact following from another. In mathematic notation, the relation of entailment between a KB and a sentence alpha is written KB |= a."

21:49:56 <bijan> There you go.

21:50:06 <bijan> They *don't* call implication "entailment"

21:50:11 <bijan> They callit "following from" :)

21:50:12 <sandro> And it then talks about an inference proceedure, and uses KB |- a for that.

21:50:47 <bijan> Turnstile = "syntactic", double turnstile = semantic.

21:51:25 <bijan> Oh, and I do tend to restrict "inference" to "following from".

21:51:44 <bijan> Too.

21:51:52 <sandro> And conditional ( --> ) is neither of those. It's inside the sentence, |- and |= are about sentences.

21:51:57 <bijan> Right.

21:52:25 <bijan> Actually, --> (or the horseshoe) typically mean *material* conditional.

21:52:38 <bijan> THere are lots of conditionals.

21:53:00 <sandro> Is material conditional difference from material impliciation?

21:53:01 <bijan> The fishhook of model logic (sometimes called!....implication :))

21:53:09 <bijan> Not usually.

21:54:28 <bijan>http://www.earlham.edu/~peters/courses/log/mat-imp.htm

21:54:28 <dc_rdfig> F: http://www.earlham.edu/~peters/courses/log/mat-imp.htm from bijan

21:54:52 <bijan> F:|A paper on the "Paradoxes of material Implication"

21:54:52 <dc_rdfig> titled item F

21:55:13 <bijan> There are many kinds of implication. We may use the "if..., then..." construction to indicate definitional, causal, or logical relations. Even among the logical relations that could go under the name of implication, not all are equivalent.

21:55:20 <bijan> We name the kind we use "material implication". Material implication is just one of many sorts of implication, and for that reason does not conform exactly to the English "if...then" or "implies".

21:55:40 <bijan> The problem is that if...then, in English, often means |-

21:56:00 <bijan> If you take this premise, that premise and the other premse, *then* (you can conclude) foo.

21:58:50 <sandro> Damn, I really though I understood a two-level difference -- in the sentence or out of the sentence -- but this three-way difference (between |- and |=) is not making sense.

22:04:35 <larsbot> does anyone know of an RDF graph that makes use of reification?

22:04:46 <larsbot> I don't mean an example, I mean an RDF graph that has actually

22:04:49 <larsbot> been used for something

22:05:02 <sbp> EARL instance data usually contains reified triples

22:05:06 <larsbot> EARL?

22:05:29 <sbp> Evaluation And Report Language - http://www.w3.org/2001/03/earl/

22:05:33 <larsbot> thx!

22:06:02 <bijan> Ok. first

22:06:03 <larsbot> where can I find instances?

22:06:22 <bijan> -> (material conditional), it's a sentential connective, defined by the truth table.

22:06:37 <bijan> (Or, if your an inferenctionalist, a set of rules of use :))

22:06:51 <sbp> um... littered over the W3C site, e.g. http://www.w3.org/2001/10/xag.n3

22:07:08 <bijan> As a sentential connective, it can only appear "inside" sentences.

22:07:11 <larsbot> this is an example, though

22:07:32 <bijan> But note that it's a *sentential* *connective*...it connects sentences, and thus expresses a relation between them.

22:07:45 <sbp> oops, and it's not reified (the reified version is at http://lists.w3.org/Archives/Public/www-archive/2001Oct/att-0006/02-xag.rdf

22:07:48 <sbp> )

22:08:35 <sandro> But it does so in a sentence. Yes, I think I understand material implication. But what's the difference between entailment |= and derivation |- ?

22:08:43 <bijan> |- is implication. What appears on the rhs can be *derived* from the LHS using varoiius inference rules.

22:09:02 <bijan> I.e., repeated application of stuff like Modus Ponens.

22:09:33 <bijan> |= is entailment. What appears on the rhs is *true* if everything on the LHS is also true.

22:10:06 <bijan> |- for first order logic is *sound*, that is, every |- has a corresponding |=

22:10:27 <bijan> It's also *complete*, every entailment has an implication.

22:10:33 <sandro> |= deals with abstract things, not sentences?

22:10:49 <bijan> No, insofar as we're talking about sentences, it's sentences.

22:11:17 <bijan> you can substituted 'proposition' for sentences *Throughout*, if you prefer.

22:11:56 <bijan> I'm using "sentence" in a very non-committal way, i.e., *not* as "sentences of english", but as 'wwfs of the formal langauge"

22:12:06 <sandro> But |= is abstract and |- is what computers (etc) might actually do?

22:12:15 <bijan> No, computers can use entailment.

22:12:19 <bijan> What's "abstract" about it?

22:12:23 <bijan> They're both *formal*.

22:12:53 <bijan> using a truth table to check for entailment is *easier* in most cases than trying to get a proof.

22:13:09 <bijan> Using models is all on the |= side of things.

22:13:15 <bijan> Hence the term "model theory"

22:13:30 <sandro> How would the code checkin KB |= a be different from the code checking KB |- a ?

22:13:53 <sandro> (you started to answer that while I was typing it)

22:14:41 <bijan> yes, assuming enough simplicity, you'd use a truth table for the first.

22:15:07 <bijan> How can I show that P. P->Q. |= Q?

22:15:18 <bijan> use a truth table.

22:15:33 <bijan> There will be no line in the truth table where P and P->Q are T and Q is F

22:15:34 <sandro> And the other?

22:15:47 <bijan> I generatae a proof.

22:15:59 <bijan> In this case Q follows by modus ponens.

22:16:28 <bijan> Where modeus pones has to be specified using metavariables. (Typcially)

22:16:40 <bijan> MP = p. p->q |- q

22:16:52 <bijan> (lower case letters are sentential variables)

22:17:10 <bijan> P. P->Q |- Q is a substitution in for MP.

22:17:16 <bijan> Thus the relationship hold.

22:17:17 <bijan> s

22:17:51 <bijan> Of course, usually, you'd use backward chaining.

22:17:58 <bijan> I want Q.

22:18:06 <bijan> One way to get Q is MP.

22:18:20 <bijan> But to use MP, I need a conditional with Q as the consequent.

22:18:26 <bijan> Aha! I have P->Q

22:18:29 <bijan> Good.

22:18:40 <bijan> But for MP I need the antecedent (in this case) P

22:18:45 <bijan> Ahas! I have P!

22:18:52 <bijan> Thus I can use MP to dervie Q

22:18:54 <bijan> YES.

22:19:06 <sandro> And while we're at it, could be shown to be true either way. That is, (empty kb) |= (p ^ p==>q) ==> q could be shown by a truth table, and (empty kb) |- (p ^ p==>q) ==> q could be shown by some symbol maniputions (write it in conjunctive normal form first, etc). Is that right?

22:19:24 <bijan> Pace the details of the latter, yes.

22:19:38 <sandro> Pace?

22:19:48 <sandro> grrr.

22:19:53 <bijan> "assuming you fill in the right details" :0

22:19:57 <sandro> My sentence should have been:

22:20:10 <bijan> I.e., conditional proof, or axioms.

22:20:24 <sandro> And while we're at it, (p ^ p==>q) ==> q could be shown to be true either way. That is, (empty kb) |= (p ^ p==>q) ==> q could be shown by a truth table, and (empty kb) |- (p ^ p==>q) ==> q could be shown by some symbol maniputions (write it in conjunctive normal form first, etc). Is that right?

22:20:43 <bijan> Sandro: it's not *shown true* either way.

22:20:49 <bijan> In the first, it's shown to be true.

22:20:56 <bijan> In the second, it's shown to be a theorem.

22:21:26 <bijan> Without the semantics, the proof doesn't say anthing about meaning, truth, or falsity.

22:21:32 <sandro> And if that |- has been proven sound, then you can conclude it's true...?

22:21:46 <bijan> Yes.

22:22:01 <bijan> It's both a theorem and a taugtolgoy.

22:22:23 <sandro> Wow. This is kind of making sense.

22:22:32 * bijan notes that he *does* teach logic :)

22:22:42 <sandro> And well, I might add. :-)

22:22:46 <bijan> Thanks! :)

22:26:50 <sandro> I have the feeling the description logics processors (eg FaCT, Classic) do entailment (|=) stuff instead of derivation (|-) stuff; does that seems right to you?

22:27:14 <sandro> (They seem to talk a lot more about "finding a model" than "showing the proof")

22:27:40 * sandro is rethinking about how MACE and OTTER fit together, too.

22:28:09 <bijan> I don't really know a lot (anything) about them. I wouldn't be surprised.

22:29:23 <sandro> boy... this gets so complicated. how to make it work for Joe Visual Basic..... I think pretty much all programmer have an intuitive sense of the meaning of boolean expressions, and that can be used to get to what we need. I think.

22:29:54 <bijan> Note we haven't at all talked about the complexities added by predicates and quantifiers :)

22:30:33 <sandro> Or functions. Yeah. Let's skip that for now. :)

22:30:33 <bijan> Which don't change the basic notions *per se* (at least of entailment vs. implication) but things get trickier.

22:31:00 <bijan> When a system is sound and complete, you can shift from one mode to the other without fear.

22:31:23 <bijan> If you get the entailment, you know there's a proof.

22:32:21 <sandro> And FOL is the most expressive we can get for which there is a known sound & complete inference proceedure?

22:32:44 <bijan> Hmmm.

22:32:59 <bijan> It's not very clear what you're asking.

22:33:22 <bijan> since "most expressive" is a little odd.

22:33:38 <bijan> Nothing which can express the peano axioms is complete.

22:33:48 <bijan> FOL is complete

22:34:01 <bijan> FOL can't express teh peano axioms (and thus, not the whole of arith)

22:34:17 <bijan> There is room in between the two.

22:34:38 <bijan> There are also (for example) decidable higher order logics.

22:34:48 <bijan> Just not proper supersets of FOL.

22:35:07 <bijan> (Though that whole areas is a bit murky)

22:53:32 <sandro> Is there any superset of FOL which is complete?

22:57:44 <bijan> Sure.

22:58:01 <bijan> Remember you can give up soundness, too.

23:06:09 <larsbot> time to zzzz

23:21:03 * AaronSw installs TiVo Ethernet card.

23:21:06 <AaronSw> It's Aliiive!

23:23:54 <sandro> Well, I did ask "sound and complete" a minute ago -- that's still what I mean. I have no interest on giving up on soundness on the SW.

23:25:20 <rillian> AaronSw?

23:25:28 <AaronSw> How'd you guess?

23:25:34 * rillian grins

23:25:46 <rillian> can I ask you some macosx questions?

23:25:53 <AaronSw> It's funny, you always join right as I come back to my computer after a long break.

23:26:07 <AaronSw> Sure, but they're probably more appropriate in #swhack.

23:26:11 <rillian> talent runs in my family

23:26:19 <AaronSw> Heh.

23:31:27 <bijan> Soundness is way overrated.

23:36:54 <sandro> I can understand giving up on soundness in exploring a situation, but in contractual communication it's basically the same as lying. So perhaps this is a situation of confusing use cases.

23:38:29 <bijan> Erhm....unless you mean something very specific by "in contractual communication", no.

23:39:20 <bijan> That a formal system isn't sound doesn't mean that you can't use, say, varous heuristics and checks to up your chances of being right :)

23:39:34 <bijan> After all, unsound doens't necessarily mean *fallacious*.

23:40:36 <bijan> If you reason with uncertainty (and even if all your internal stuff is sound) once you make the "threshhold" jump, you've made an unsound move.

23:40:56 <bijan> Yet this is hardly startling, surprising, or necessarily bad.

23:41:14 <bijan> It just introduces a new class of error.

23:42:15 <bijan> Oh, Paul Wilson has some interesting stuff about "detouring" through invalidity when doing code transformations.

23:42:17 <sandro> Hmm. Okay, I can see that. As long as you realistically estimate the odds and consequences of such an error. It hadn't occured to me that you'd be able to do that.

23:43:24 <bijan> There's some very cool phil of math stuff that poinst out that some short, easy probabilitic proofs will give you a result that's has a better probabilty of being correct than you are at *not* introducing an error in the deductive proof.

23:43:41 <bijan> We're *always* fallible. Always.

23:44:03 <bijan> A bug in your program will screw you just as hard and maybe more than an unsound rule :)

23:44:26 <sandro> Yes, I do understand that.

23:44:36 * DanC is working on a SWAD plan, wonders if chatting about it here would help...

23:44:45 <bijan> SWAD?

23:45:00 <sandro> Well, if you understand the ways in which the rule might screw you at least. People know how software can be....

23:45:05 <sandro> Sure, DanC.

23:45:20 <sandro> SWAD = W3C Semantic Web Advanced Development (Group)

23:45:31 <bijan> Ah!

23:45:35 <bijan> WSWADG

23:45:37 <bijan> :)

23:46:18 <sandro>http://www.w3.org/2000/01/sw/

23:46:18 <dc_rdfig> G: http://www.w3.org/2000/01/sw/ from sandro

23:46:34 <bijan> Sandro: Trust me, understanding how most useful unsound rules might fail is *way* easier than understanding how Windows *does* fail.

23:46:44 <bijan> I mean, beyond "often" and "annoyingly" :)

23:47:02 <DanC> G:| W3C Semantic Web Advanced Development

23:47:02 <dc_rdfig> titled item G

23:47:57 <AaronSw> G:|W3C Semantic Web Advanced Development

23:47:57 <dc_rdfig> titled item G

23:48:03 <AaronSw> Oops.

23:49:08 * DanC tweaks plan a bit before sharing, lest we waste time discussing noise...

23:50:14 * sandro wonders if he should add himself and some others to that page.

23:53:05 <DanC> yes, your name should be there, sandro