Semantic Web Interest Group IRC Chat Logs for 2002-04-19

00:04:45 <bijan> I'm around.

00:07:02 <bijan> Hey jim.

00:07:22 <bijan> (/me just got back from 60 surprise birthday party for my advisor)

00:08:08 <bijan> (Apropos of absolutely nothing at all.)

00:14:40 <AaronSw> jhendler

00:15:31 <bijan> AaronSw

00:16:19 <sbp> bijan

00:17:54 <bijan> Rocky!

00:18:06 <bijan> Aaron: where's the rdfcore flame fest.

00:18:14 <bijan> I need to roast some marshmallows.

00:18:18 <AaronSw> heh

00:18:25 <AaronSw> it's more a floodfest than a flamefest

00:19:03 <bijan> Denotation of datatypes?

00:28:58 <bijan> Hahah!

00:29:00 <bijan> Datatyping in this case is

00:29:00 <bijan> wanted, by some users, in order to provide a lexical form constraint

00:29:00 <bijan> on the literal. (I think this is kind of dumb as well, but that's

00:29:00 <bijan> what they want.)

00:29:14 <bijan> I [heart] Pat Hayes!

00:29:58 <bijan> If we

00:29:58 <bijan> tell them that the MT says one thing, but don't worry about that

00:29:58 <bijan> because we all know that *really* it doesnt mean that, you know, ha

00:29:58 <bijan> ha, that's just for those semantic geeks, then we are seriously

00:29:58 <bijan> misleading them. Because for any conforming inference engine (or

00:29:59 <bijan> indeed for any conforming engine of any kind), that IS what it means.

00:30:03 <bijan> Gotta love it!

00:33:55 <Seth> hmmmm ... not understanding this .. is PatH saying that if we have

00:34:34 <Seth> ex:age rdfd:datatype xsd:integer. Jane ex:age "25"

00:35:18 <Seth> that the MT does not sanction us storing Jan's age in a integer type in our computer memories?

00:36:49 <Seth> anyone got a quick answer to that one?

00:37:25 <bijan> For some interpretation of your statement yets.

00:37:31 <bijan> You have to use a different idiom.

00:38:03 <bijan> More precisely, i believe that says, acording to the MT that Jane's age is a string with two characters.

00:38:17 <Seth> so what ideom would i use?

00:38:33 <bijan> You could store it as an integer type in your computer memory if you were using that integer type to represetnt the two character string.

00:39:04 <bijan> I think the dlex one.

00:39:24 <bijan> Colloquially, Jane's age is something, and that something has the dlexical form of "25"

00:39:51 <bijan> > ex:age rdfd:datatype xsd:integer .

00:39:51 <bijan> > ex:age rdfd:datatype xsd:string .

00:39:51 <bijan> > Jane ex:age _:x .

00:39:51 <bijan> > _:x rdfd:lex "10" .

00:39:53 <bijan> I think.

00:39:58 <bijan> See http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2002Apr/0293.html

00:40:04 <bijan> This is non-normative.

00:40:44 <bijan> And Id on't know about "sanction". It's user defined if you interpret the first as an integer. The MT meaning is that it's a string.

00:40:52 <Seth> oh i see, thanks bijan

00:41:27 <bijan> No problem.

00:41:36 <bijan> The rule to remember is literals always denote themselves.

00:41:39 <bijan> And that they are *string* literals.

00:42:12 <Seth> ... well dont get me started on that one

00:42:21 <Sojah247> !list

00:42:37 <bijan> Well, it's what the MT says. Whether that's what it *should* say is different.

00:42:45 <bijan> Pat seems convinced that there are clear use cases.

00:43:03 <bijan> One thing will be to design syntaxes that are less cumbersome.

00:43:27 <bijan> It's quite relieving actually, at least it has a concrete definite meaning.

00:43:41 <bijan> Rather than all the speculation we've tossed around on #rdfig.

00:44:02 <bijan> Not only is it clear, concrete, and definite, but it's possible to express all the things we'd want to express.

00:44:10 <bijan> Even if somethings a little more verbosely.

00:46:45 * Seth sides with Stickler

00:47:52 <Seth> if the graph states explicitdly that something is a number, i want to store it as a number in my computer so that if i want to add or subtract to it i can withoug further inference or transfromation of my graph.

00:50:47 <Seth> ie all literals that are infered to be integers are stored as integers, regardless of whether there is a node in the graph denoting the integer value or not. That is just stupid ... to force people to use the other ideom when all the knowledge about it is available, just to comply with some remote mt document.

00:54:38 <bijan> You argument is bogus. The MT was made that way to reflect user needs.

00:54:47 <bijan> You may not have those needs, but so what?

00:55:39 <bijan> And, in any case, you can always do whatever you freaking want to, but you lose something. You lose the guarantee's of the MT. You may not want them, but so?

00:56:20 <bijan> even your statement of your principles is problematice, "all literals that are inferred to be integers are stored as integers". Inferred *how*?

00:56:40 <Seth> to me its a case where the MT is being (how shall i phrase it) illogical.

00:56:41 <bijan> In the MT, in RDF, no literal can be infered to be an integers. They are *not* integers.

00:57:05 <bijan> Well, just another case where the way the world is to you isn't the way the world is.

00:57:27 <bijan> That or you're deliberately misuing terms to rhetorical effect.

00:57:42 <Seth> {age range integer} => that the object or age is an integer

00:57:57 <bijan> Pat Hayes point is perfectly consitent and coherent and proveable so.

00:57:58 <Seth> or=of

00:58:32 <bijan> Well, look, that's just not what it means (or, at least, probably not what it means).

00:59:04 <Seth> im sure PatH's point is ... but it is illogical to say {age range integer} and then to require an extra bnode to actually mean that

00:59:06 <bijan> Preferring it to mean that is perfectly acceptible, but that's *hardly* the only view.

00:59:35 <bijan> There is no "that".

00:59:52 <bijan> The meaning is well specified, it just doesn't mean what you want it to mean.

00:59:59 <bijan> That's a very different point.

01:00:22 <bijan> What you should do is figure out if the people who want it to mean what it (currently) means have a reasonable case.

01:00:57 <bijan> As I understand it, your way would mean that those people *could not express* what they want to in RDF>

01:01:05 <Seth> yeah and doent mean what Strikler wants it to mean .. and probably a lot of other programmer types that will be wanting to add and subtract nmbers and wanting to know when they can and cant do that ligitimately

01:01:08 <bijan> This way *everyone* can express the things they want to express.

01:01:27 <bijan> But YOU CAN DO THAT>

01:01:30 <bijan> Just not with that idiom.

01:02:20 <Seth> but your right .. im sure PatH formjula are quit right and mean exactly what he is saying that they mean ... and that a group of us dont like it mitht mean that he wrote his formulas wrong

01:03:08 <bijan> Whatever. we're stopping now.

01:04:11 <Seth> i know i can do that, by putting in the extra bnode (right?) ... but the data that i read on the web, some of it's gonna have that bnode and other of it aint .. and that is just going to contribute to non interoperability .. and that is bad (right?)

01:04:50 <bijan> We were stopping. But no, it isn't.

01:04:56 * jhendler says hi - was tied up in some email flaming

01:05:11 <bijan> The idea is that the people who don't "put that bnode" in *meant* it to mean something different.

01:05:19 <bijan> Hehehe.

01:05:22 <bijan> Join the irc flaming!

01:05:39 <Seth> well i would bet you anything that they did, if they did use xsd:integer !!

01:06:28 <bijan> Seth, your positing people not reading specs, not using tools, and not being smacked. For me, the semantic web entails all these things.

01:06:55 <jhendler> esp. the being smacked - well, or maybe that's just the WG chairs!

01:06:57 <bijan> So we stop for real this time.

01:07:00 <bijan> Hehe.

01:07:06 <bijan> I'm a Semantic Smacker!

01:07:12 <bijan> Semantic Web Smacker.

01:07:19 <bijan> Pat Hayes is our god.

01:07:38 * sbp mentions something about giving someone a good swhack

01:07:39 <Seth> hey, dont mind you smacks ... now i dont have to join the flame fest ... got it all off my chest at your expense :)

01:07:43 <jhendler> please, let's not go too far (or do you mean the G-d of the semantic web smackers!)

01:07:51 <bijan> Heheh.

01:08:15 <sbp> Pat Hayes is to the Semantic Web folk as G****d is to the validator.w3.org folk

01:08:21 <bijan> I pray for the miracle of bread and smackeral!

01:08:33 <Seth> lol

01:08:44 <jhendler> bijan, if these guys will let us sneak it in, I wanted to discuss some Parka stuff - starting to get some experiments going, worth thinking about what would be interesting to see if it can perform as an RDF-DB

01:09:08 <bijan> I'm hip.

01:09:20 <bijan> ron got all the wordnet rdf in, I believe.

01:09:34 * Seth goes back to mentographing his python rdf api

01:09:55 <bijan> Though I have to talk to dajobe about raptor not expanding entities for some weird reason (though, of course, this may be a bug inthe wordnet rdf)

01:09:55 <jhendler> I read your email about the rdf:li's in the chat logs - agree that if we could work out how to remove (or at least) limit those, would make a good test -- seems like we have a couple million triples in there.

01:10:09 <bijan> WE removed them.

01:10:19 <bijan> There's like 7 predicate types now :)

01:10:38 <jhendler>http://www.cs.umd.edu/projects/plus/Parka/aaai97.ps

01:10:38 <dc_rdfig> A: http://www.cs.umd.edu/projects/plus/Parka/aaai97.ps from jhendler

01:10:42 <jhendler> cool!

01:10:45 <bijan> I read that.

01:11:02 <jhendler> A:|Performance of Parka-DB on some large datasets (including WordNet)

01:11:03 <dc_rdfig> titled item A

01:11:03 <bijan> It's half a presentation paper for the sardinia thing, if we can plug in new numbers.

01:11:17 <bijan> Id idn't see that you *compared* it to some other wordnet engines.

01:11:25 <bijan> Is that right?

01:11:37 <jhendler> A: question to wrestle with - can we make Parka-DB into a backend for RDF triple stores?

01:11:37 <dc_rdfig> added comment A1

01:11:48 <bijan> (Have you seen OilEd?)

01:11:54 <bijan> A:Preliminary findings: yes.

01:11:54 <dc_rdfig> added comment A2

01:12:54 <bijan> A:Tricky bit #1: A lot of RDF involves large numbers of generated predicaties, i.e., from rdf:lis. (See the chat logs for #rdfig for example...one day can have over 400...or even over 1000!)

01:12:54 <dc_rdfig> added comment A3

01:13:12 <jhendler> agree with the yes, although I'd point out that to really use Parka's power, we need to get the class/subclass stuff to work with its inheritance algorithms (perhaps as a switch on a query so one can turn it off when not needed). Without that it's just a clever Relational DB

01:13:30 <bijan> A: But traditional large scale KBs tend to have small numbers of predicates, and thus optimize for them, or have hard limits.

01:13:30 <dc_rdfig> added comment A4

01:13:47 <bijan> Yep.

01:14:01 <bijan> but is hyponym stuff inheritancy?

01:14:06 <bijan> It's certainly pointer following.

01:14:59 <bijan> A:Using DAML lists can help. I have a few other tricks. And dajobe mentioned a rdf:li supertype/class that RDF Core is working on, or something.

01:15:00 <dc_rdfig> added comment A5

01:15:54 <bijan> A:(DAML lists help because you use daml:first and daml:rest, rather than a genided predicate for each element in the list. Essentially, cons cells vs. integer indexs for arrays.)

01:15:54 <dc_rdfig> added comment A6

01:16:39 <bijan> Hmm. From that paper, jim, I see you *do* have hyponyms in Parka Wordnet.

01:16:40 <jhendler> (although a clever relational DB for triples - so still RDF value)

01:16:47 <jhendler> bijan - did Ron figure out how to find the KB size info -- be interesting to know how many triples wordnet was in RDF - our old parka version had about 385,000

01:17:04 <bijan> Good. That gives us a comparison case for "old style parka" and "parka rdf"

01:17:06 <bijan> Yes.

01:17:08 <bijan> He did.

01:17:21 <bijan> rreck: 60618901 Apr 18 18:59 wordnet_big

01:17:21 <bijan> rreck: 60 megs of assertions

01:17:21 <bijan> rreck: (##returned 273644 statements)

01:17:45 <bijan> I don't know if that's all the RDF wordnet files yet.

01:17:49 <bijan> I'm sure it's more than one.

01:18:05 <jhendler> yes, we definitely had the hyponyms - Our best query was (hmm, cannot cut and pste from postscript)

01:18:48 <bijan> Need to conver to pdf.

01:19:04 <deltab> or text

01:19:07 <bijan> Yeah, that's the one that kills all the standard wordnet tools.

01:19:21 <bijan> And they specifically warn against on their online version :)

01:19:28 <jhendler> "Find all senses of TREE and all their HYPONYMS which are member meronyms of all senses of genus citrus."

01:19:44 <bijan> Yep.

01:20:02 <bijan> Ron ran (in regular wordnet) something like, "all the hyponyms of animal"

01:20:04 <jhendler> Took about 3.3 seconds on a single processor (Sun 5ish machine I think), .5 seconds on 16 processor machine

01:20:15 <bijan> Last I heard, it was still running :)

01:20:27 <jhendler> I don't think danbri's stuff handled that level of complexity - don't know if he put in the member meronyms.

01:20:37 <bijan> So maybe it just hung :)

01:21:19 <bijan> I will say is, pace the entity expansion, it really was trivial to run these files through raptor, to n-triples and do a simple conversion to parka format.

01:21:40 <jhendler> odd, the animals one only took 4 seconds in the old encoding - possible he did it wrong (it does have 7950 answers though!)

01:21:40 <bijan> Nice not to even have to THINK about the wordnet format.

01:22:44 <bijan> Hmm. Maybe I'll violate the wordnet online request... :)

01:22:51 <jhendler> agreed, as a quick hack, we should change the subclass and class and rdf:type assertions to their parka equivalents as we do the conversion.

01:23:38 <jhendler> i.e. when we go from ntriples to parka. In longer run we need to get into the parka code and make it recognize those things directly

01:23:52 <bijan> Yes.

01:24:02 <jhendler> in that way we could use the inheritance features - also makes for more complex queries

01:24:22 <bijan> *But* I think ntriples is a perfectly good format. Better than directly hooking up an RDF/XML parser.

01:24:48 <bijan> Hmm. In parka, isn't the inheritence a feature of a slot of the underlying frame?

01:25:05 <bijan> I mean, is it just a tweak to get it to recognize the rdf/s versions?

01:26:55 <jhendler> I think it is just a tweak, modulo that there are some differences in how Parka's language handled class and subcass relations (needed to assert them both ways, ditto for instance data) so some things that are 1 assertion in RDF would become 2 in Parka. This was all built into the Parka KB builder, so it should be "automagic" once we find the right place in the code

01:27:51 <bijan> Aha. Good.

01:28:00 <jhendler> We also need to write some interesting piece of code to handle namespaces correctly - RDFDB just uses the whole strings, which wastes a lot of space -- Parka will do same unless we do something intelligent - I was thinking that the front end coudl "hash" on them or something

01:28:23 <bijan> Why not just intern the full uris?

01:28:36 <bijan> (What's parka-db written in, anyway?)

01:28:55 <bijan> (I say, "just intern" as if it were written in CL or Prolog or Smalltalk :)

01:28:59 <jhendler> Parka's written in C (Ansi standard C, not C++, for efficiency)

01:29:09 <bijan> RDFDB really uses strings?

01:29:10 <bijan> Wow.

01:29:12 <bijan> Daft.

01:29:30 <jhendler> The original version was written in *LISP (The connection machine Lisp -- ahh for the good old days of 32000 processors)

01:29:37 <bijan> Heheheheh.

01:30:20 <jhendler> I don't know if it uses strings per se, but when it stores foo:X it stores it as "http://...#x"

01:31:06 <jhendler> btw, off the Parka subject, I'm hiring a 16 year old high school hacker to come work with us this summer - tired of Aaron being the only whiz kid around...

01:32:02 <bijan> What, I'm not good enough? :)

01:32:12 <bijan> I can act younger than aaron!

01:32:53 <bijan> Granted, I have to stretch a bit to overcome my natural gravitas...but I'm much more flexible now :)

01:32:57 <jhendler> (long pause while I think about how to react to that)

01:33:30 <bijan> HEHEHEHEH.

01:33:37 <bijan> Which proves my point! :)

01:33:39 <jhendler> looked through the other papers, don't seen the wordnet comparison. I think I had it in one of my old Parka talks.

01:33:57 <bijan> Well, this is what minions are for, y'know.

01:33:59 <jhendler> (aaron has obviously gone elsewhere bored by the Parka talk)

01:34:10 <AaronSw> jhendler!

01:34:20 <bijan> That meddling kid.

01:34:23 * AaronSw just got back

01:34:33 <jhendler> oh, guess he must have some sort of "aaron" detector running :->

01:34:42 <AaronSw> always

01:34:55 <AaronSw> it makes quite a noise, i could hear you all the way downstairs

01:35:17 <jhendler> Thought it was that Borg-like implant to the chat group

01:35:28 <AaronSw> no, that's still in beta

01:35:35 <jhendler> lol

01:36:07 <AaronSw> they tried it on the other guy and he started sleeptyping

01:36:20 <bijan> Hmm. didn't the service guy come yet aaron?

01:36:22 <jhendler> I've done that most of my career!

01:36:26 <AaronSw> but worse, he would act out the /me actions

01:36:44 <jhendler> it's those "emoticons" that get you

01:36:47 <AaronSw> major security hole

01:36:51 <bijan> The one that was going to hook up the "aaron" detector to eletrodes attached to your...nasel hair?

01:37:11 <bijan> So, if Iwere to say aaron, you'd get a little jolt.

01:37:17 <bijan> AARON a big jolt.

01:37:28 <AaronSw> .wn nasel

01:37:34 <bijan> And arron a *huge* jolt (extra points for misspelling :))

01:37:35 <AaronSw> I think you mean nasal.

01:37:40 <bijan> Yes.

01:37:46 <bijan> But i have a detector too.

01:37:46 <jhendler> .wn .wn?

01:37:50 <AaronSw> .wn wn

01:37:52 <bijan> To stop me from using such words.

01:37:54 <AaronSw> .wn wordnet

01:37:55 <xena> WordNet defined as:

01:37:56 <xena> - n : a machine-readable lexical database organized by meanings [syn: {WordNet}]

01:38:00 <bijan> so I don't type 'em :)

01:38:17 <bijan> Ah...yes, the original DWLB had a wordnet hookup.

01:38:27 <jhendler> cool - didn't know Xena had that in her

01:38:30 <bijan> Only it could enter the term as a chump item.

01:38:33 <jhendler> .wn moose

01:38:33 <xena> moose defined as:

01:38:34 <xena> - n : large northern deer with enormous flattened antlers in the male; called elk in Europe and moose in North America [syn: {elk}, {European elk}, {Alces alces}]

01:38:43 <bijan> So we would pub Words of the Day, plus commentary :)

01:39:01 <jhendler> wow! I gotta get me one of these in my word processor.

01:39:59 <bijan> Yeah, i started on an editor, I think I told you about this jim, that if you hovered on a word it would pop up the various wordnet senses, you would select one, and it would encode it in SHOE (or some such.

01:40:40 <bijan> Way way way back when.

01:40:49 <jhendler> interesting - should hook that up to Aditya's RDFeditor - but instead of wordnet you could choose which ontologies to load

01:41:07 <bijan> Yep.

01:41:17 <jhendler> biologists use UMLS, rocket scientists the NASA thesaurus, etc.

01:41:58 <jhendler> btw, National Cancer Institute bioinformatics group is going to start releasing its monthly vocabulary in DAML as well as their current XML thing.

01:41:58 <jhendler> '

01:42:08 <bijan> Cool!

01:48:43 <jhendler> Anyone know much about Zope?

01:49:33 <bijan> I know that 1) It seems neat when you read the specs.

01:49:45 <bijan> 2) The documentation causes your eyeballs to explode.

01:49:52 <bijan> 3) They're changing everythign about it.

01:49:58 <jhendler> yes on 2.

01:50:02 <bijan> 4) The no starch books is supposed to be quite good.

01:50:20 <bijan> But I know people who have set it up and survived.

01:50:29 <jhendler> Aaron obviously knows too much about Zope - exploded when he tried to tell us :->

01:50:37 <bijan> Heheheh.

01:50:58 <sbp> heh, heh

01:51:09 <jhendler> this HS kid was telling me about Zope, but he seemed to see connection to Sem Web ideas that I didn't see immediately

01:51:28 <jhendler> not sure if he meant for managing SW projects, or something deeper

01:51:44 <bijan> i'm very skeptical.

01:51:47 <jhendler> problem is I went to look at the docs and my eyeballs exploded.

01:51:58 <bijan> Zope is an ODB with a moderately complicated object model.

01:52:06 <bijan> It's meant for Content managemetn.

01:52:17 <bijan> It's somewhat html centric.

01:52:40 <bijan> But it has a fairly reasonable isntalled base.

01:52:47 <bijan> So if you hooked up semweb stuff, people might use.

01:52:57 <bijan> It was a fairly decent driver for XML-RPC.

01:52:59 <bijan> FWIW.

01:53:11 <jhendler> hmm, could be interesting - we used to bill Parka as realizing the OO model on an RDBMS - speed of relations with modeling power of OODB. Of course, that was when we were still thinking of trying to sell it :->

01:53:16 <bijan> Uche might be in a better place to do that sort of thing.

01:53:30 <bijan> Heh.

01:53:59 <bijan> You could build something reasonably like a SemanticZope ontop of parka, natch.

01:54:20 <bijan> Some of the ideas in the (still uncompleted) Expert System backed Website article might be relevant.

01:54:34 <bijan> Alas, prolly the *unwritten out* ideas :)

01:54:53 <jhendler> I was thinking that a good use case of the SW would be to help do searches on the open source stuff...

01:55:11 <jhendler> if you don't know the names of the projects, you're often hosed.

01:55:26 <bijan> Hmm. I wonder if sourceforge data is available...

01:55:34 <bijan> I think you can export some of it...

01:55:41 <bijan> But maybe only for your own projects?

01:55:59 <jhendler> try searching for "an open source java-based browser" or such, and you get lots of hits that are useless

01:56:14 <jhendler> .googler open source java-based browser

01:56:24 <bijan> Hmm. But you can navigate down various categories on sourceforge.

01:56:34 <bijan> A scraper might help there.

01:56:58 <bijan> SourceForge.net Statistics

01:56:58 <bijan> Hosted Projects: 38,195

01:56:58 <bijan> Registered Users: 403,214

01:57:01 <jhendler> yes, and we have the scraper to do it...

01:57:19 <AaronSw> .google open source java-based browser

01:57:20 <bijan> Even nicer, we could define an ontology for the SF and let non sf stuff participate.

01:57:20 <xena> open source java-based browser: http://xml.coverpages.org/ni2001-03-23-a.html

01:57:41 <bijan> .google open source java-based browser that doesn't work

01:57:43 <xena> open source java-based browser that doesn't work: http://www.aim.com/help_faq/error_mess/linuxfaq.adp?aolperm=%3C%25=$header%25

01:57:48 <bijan> Heheh.

01:58:02 <jhendler> Aditya's been improving his scraper (he announced a new version in rdf-ig) - got some impressive results

01:58:09 <deltab> freshmeat has xml for projects, e.g. http://freshmeat.net/projects-xml/apache/apache.xml

01:58:18 <bijan> Great.

01:58:24 <jhendler> I thought the "doesn't work" was redundant w/open source and Java...

01:58:35 <bijan> Or non-open source and Java...

02:08:17 <jhendler> been fun all, but back to work.

02:08:20 <jhendler> g'night

03:08:01 <Seth> let me say this here and now when nobody is listening ... and also because nobody else will say it ...

03:09:33 <Seth> when Pat says stuff like "That will help to drive another nail

03:09:33 <Seth> into RDF's coffin, which might in the long run be the best thing for

03:09:33 <Seth> the world in general in any case.

03:11:26 <Seth> sure makes you think that he has a ~conflict of interest~ and that letting someone write your MT that doesnt believe in you is like letting the Prosecution write the closing arguments for the Defense.

03:11:41 <Seth> ..... well just had to get that off my chest .. whew!

03:25:17 <AaronSw> DanC's message of 18 Apr 2002 21:46:27 -0500

03:25:22 <AaronSw> is... puzzling\

09:48:04 <sbp-> sbp- is now known as sbp

10:09:34 <JibberJim> We need http://news6.thdo.bbc.co.uk/hi/english/sci/tech/newsid_1936000/1936949.stm for the foaf:image data, so the path chopping out is automatic...

10:11:13 <danbri__> cool. wonder what state of the opensource-art is for that kind of stuff...

10:13:35 <JibberJim> On the images.google.com find image stuff, what output do you want from it?

10:17:41 <danbri__> hmm, not sure. I was thinking about irc-interface to image cataloguing generally...

10:17:57 <danbri__> for starters, just finding a tool in whatever language that can scrape from images.google would be good

10:18:05 <danbri__> a sherlock plugin maybe?

10:18:12 <JibberJim> I've already got that...

10:18:54 <danbri__> what are you using or doing?

10:19:03 <JibberJim> !findimage Dan Brickley

10:19:04 <FillyJonk> Is http://www.w3.org/2001/sw/RDFCore/20010801-f2f/brian-art-daveB-danC.jpg Dan Brickley?

10:19:17 <danbri__> ugh, that's the _worst_ picture of me in google!

10:19:25 <danbri__> No, it isn't!

10:19:26 <dajobe> is that a challenge?

10:19:31 <danbri__> uhoh

10:19:34 <FillyJonk> Yes:0 No:1

10:19:34 <dajobe> that question has no yes/no answer

10:19:34 <FillyJonk> Is http://rdfwebring.org/~pldab/rdfweb/danbri.wot.gif Dan Brickley?

10:19:35 <danbri__> don't go there!

10:19:39 <danbri__> no

10:19:44 <danbri__> 1

10:19:53 <danbri__> 1

10:20:04 <FillyJonk> Yes:0 No:1

10:20:04 <FillyJonk> Is http://lists.w3.org/Archives/Public/www-rdf-interest/2000Aug/att-0120/01-C:_TEMP_nsmail9H.gif Dan Brickley?

10:20:21 <danbri__> 1

10:20:26 <JibberJim> no

10:20:34 <FillyJonk> Yes:0 No:1

10:20:34 <FillyJonk> Is http://lists.w3.org/Archives/Public/www-archive/2001Sep/att-0016/01-damlquery.gif Dan Brickley?

10:20:38 <danbri__> ah, I thought it wanted numeric codes

10:20:46 <danbri__> no

10:21:00 <JibberJim> it takes yes/no, and will keep going until it gets a yes, or it runs out of images.

10:21:04 <FillyJonk> Yes:0 No:1

10:21:04 <FillyJonk> Is http://blogspace.com/pictures/photo/med/4291/104-0465_IMG.jpg Dan Brickley?

10:21:04 <danbri__> It's requerying and scraping each time?

10:21:14 <JibberJim> nope. just scrapes the once.

10:21:17 <danbri__> ick

10:21:21 <dajobe> try http://images.google.com/images?q=danbri

10:21:21 <danbri__> yes

10:21:21 <danbri__> grudgingly

10:21:29 <danbri__> Very cute!

10:21:34 <FillyJonk> Yes:1 No:1

10:21:34 <FillyJonk> Is http://blogspace.com/pictures/photo/orig/4286/104-0452_IMG.jpg Dan Brickley?

10:21:36 <danbri__> question is what next... hmmm

10:21:40 <JibberJim> yes

10:21:41 <JibberJim> yes

10:22:04 <FillyJonk> Yes:2 No:0

10:22:04 <FillyJonk> Found Dan Brickley - blogspace.com/pictures/photo/orig/4286/104-0452_IMG.jpg

10:22:28 <danbri__> I want to find a way of making little state-diagram structures for IRC bots, so we can model where in a conversation it is, and provide prompts etc in config files...

10:22:55 <danbri__> like TML quiz markup or the stuff the phone/xml folk are doing...

10:23:26 <danbri__> next part of this irc bot conversation might be the 'figuring out who else is in the photo' session, for eg.

10:23:36 <danbri__> Q: Are there other people in the photo?

10:23:36 <dc_rdfig> Label Q not found.

10:23:45 <danbri__> Q... do you want to catalogue them?

10:24:47 <danbri__> Select one of (i) homepage, (ii) mailbox (iii) hash-of-mailbox (iv) AlpiriKB name (v) dmoz category... to identify the individual

10:24:57 <danbri__> er, or (vi) URI, I guess ;-)

10:25:24 <danbri__> ...and loop through identifying each person with one or more property, then do other people

10:25:39 <danbri__> ...and then I guess ask for major noun terms for the picture

10:25:47 <danbri__> which is where we get into bot-to-bot protocols

10:25:54 <danbri__> .wn ketchup

10:25:56 <xena> ketchup defined as:

10:25:57 <xena> - n : thick spicy sauce made from tomatoes [syn: {catsup}, {cetchup}, {tomato ketchup}]

10:26:03 <danbri__> .wn glass

10:26:04 <xena> glass defined as:

10:26:04 <JibberJim> well FillyJonk also has an RDF parser (and some broken algae querying.) so I thought perhaps you could start off by loading up some foaf RDF about people, and then add in the depiction afterwards.

10:26:04 <xena> - n 1: a brittle transparent solid with irregular atomic structure

10:26:05 <xena> - 2: a glass container for holding liquids while drinking [syn: {drinking glass}]

10:26:06 <xena> - 3: the quantity a glass will hold [syn: {glassful}]

10:26:07 <danbri__> .wn geek

10:26:07 <xena> geek defined as:

10:26:07 <xena> - 4: a small refracting telescope [syn: {field glass}, {spyglass}]

10:26:08 <xena> - n : a person with an unusual or odd personality [syn: {eccentric}, {eccentric person}, {oddball}]

10:26:22 <danbri__> .wn meal

10:26:23 <xena> meal defined as:

10:26:24 <xena> - n 1: the food served and eaten at one time [syn: {repast}]

10:26:25 <xena> - 2: coarsely ground foodstuff; especially seeds of various cereal grasses or pulse

10:26:31 <danbri__> ...etc.

10:27:12 <JibberJim> but I want people to trace out the outlines.. they can't do that in IRC :-(

10:27:15 <danbri__> It'd be cute if we could have common conventions in bots for talking to each other and remote services... we all seem to reinvent bot plugins differently, and they don't talk to each other well.

10:27:30 <danbri__> up, up, up, right doh left, up, up... :)

10:27:42 <danbri__> outlines: sure... nor can they see the picture, for that matter.

10:27:59 <JibberJim> That's 'cos it's generally easier to add to your own bot, than do all the risky stuff of having bots talk to each other, it's too easy for them to go mad.

10:28:26 <danbri__> Having the bot conversation drive a reloading web page might be one way, and when there's enough info gathered, shift UI to WWW/SVG

10:29:18 <danbri__> ..."now please outline the ketchup in the photo. use ctrl-select on your pointer device to finish outlining"

10:29:25 <danbri__> ..."now please outline the geek in the photo. use ctrl-select on your pointer device to finish outlining"

10:29:40 <danbri__> ...are there more ketchups and geeks to outline? abort/retry/fail etc...

10:29:49 <danbri__> this'll be the tricky bit

10:30:03 <danbri__> Hmm, I wonder if two people could catatlogue the image at same time, outlining different bits

10:30:16 <danbri__> feeping creaturism...

10:31:07 <danbri__> re bot2bot: yeah, the w3c zakim and bwm's jema bots often end up chatting to each other.

10:31:21 <danbri__> I was thinking more of web services, which is where bot brains should probably live

10:38:56 <JibberJim> There's a jpeg to SVG conversion tool which must identify areas of pictures...

10:44:08 <deltab> URI scheme 'dcl' spotted: http://dcl.sourceforge.net/

10:51:17 <JibberJim> The discovery of paths, could maybe done automatically...

10:51:45 <JibberJim> - http://jibbering.com/2002/4/shepherds.svgz is automatically generated from the normal shepherd image http://jibbering.com/imgs/shepherds.jpg

11:20:44 <danbri__> Chaals was pretty interested in this (for web accessibility apps etc -- generating bump maps of images for visually impaired users, etc), but I've never found any tools to testdrive

11:20:56 <danbri__> do you url for the tool that does jpg2svg?

11:21:21 <JibberJim> - http://www.celinea.com

11:21:39 <danbri__> my browser (mozilla) doesn't like your .svgz url

11:21:59 <danbri__> Content-Type: application/octet-stream

11:22:09 <danbri__> that's why... is there an unpacked copy?

11:22:42 <JibberJim> My browser doesn't like it either!

11:24:02 <JibberJim> Should be okay now?

11:25:13 <danbri__> I tried gunzip'ing it, didn't work. ah, that's better...

11:26:09 <danbri__> works now. the picture looks great -- not sure if there are identifiable sub-regions for scripts etc to work with though

11:26:17 <danbri__> reminds me of some filters in Gimp, photoshop etc

11:26:50 <JibberJim> If you tweak the settings, you can get quite big blocky regions.

11:31:38 <JibberJim> Have a look at - http://jibbering.com/2002/4/test.svg - click on the bit bits you want to add to the path - does that work?

11:37:21 <danbri__> that's one of the strangest svg docs I've ever met! Yes, feels very close to useful. One would need to edit the resulting paths, and be able to toggle mistkaes off again; and ideally see the underlying areas to get a better idea where to click

11:37:27 <danbri__> but very impresive...

11:37:41 <danbri__>http://www.celinea.com

11:37:42 <dc_rdfig> B: http://www.celinea.com from danbri__

11:38:06 <danbri__> B:|Celinea -- a raster to vector image convertor (does SVG output too :)

11:38:06 <dc_rdfig> titled item B

11:38:46 <danbri__> B:Seems to be windows only and no sign of source, but output is pretty impressive

11:38:46 <dc_rdfig> added comment B1

11:38:58 <danbri__> logger, pointer

11:38:58 <danbri__> See http://ilrt.org/discovery/chatlogs/rdfig/2002-04-19#T11-38-58

11:39:42 <danbri__> B:We were discussing [http://ilrt.org/discovery/chatlogs/rdfig/2002-04-19#T11-38-58|image metadata tools], specifically for generating rdfweb/codepiction SVG paths that outline people in photos.

11:39:42 <dc_rdfig> added comment B2

11:40:48 <danbri__> Jim, you seem hesitant to feed URLs to the chump bot... in general any URLs we discuss are worth discussing, makes things much easier to find in future...

11:41:03 * danbri__ finds it easier than HTMLing pages of links to stuff, anyway

11:41:28 <JibberJim> I just generally don't think them worthwhile...

11:42:16 <JibberJim> I probably should as I find interesting things in there...

12:03:39 <rreck> maybe the hesitation is that once fed the thing will always be looking for handouts

12:11:03 <JibberJim> Also everytime you search on anything in google, you end up with links to the logs, they need devaluing not having even more ingo put in them...

12:19:45 <rreck> ok i have a simple question

12:20:04 <rreck> the order of ntriples doesnt matter.. is that correct ?

12:20:51 <bijan> Depends.

12:20:57 <bijan> Ordinarily, yes, it doesn't

12:39:43 <libby> hey jibberjim, I didn't realize how cool your bot was. neat :)

12:41:25 <maxf> Hi libby (et al)!

12:45:00 * danbri__ spends way to long writing to www-rdf-interest

12:45:11 <danbri__> (trying to get in habit of showing up there, not just irc...)

12:46:02 <danbri__> hi maxf :)

12:46:30 <libby> hi maxf, didn't know you were about

12:46:50 <maxf> just turned up.

12:47:01 <libby> ahha

12:48:04 <bijan> Why oh why does everyone write in Word?

12:48:22 <bijan> Worse, why do they write in word when they want their output to look LaTeX generated?

12:48:37 <danbri__>http://lists.w3.org/Archives/Public/www-rdf-interest/2002Apr/0212.html

12:48:37 <dc_rdfig> C: http://lists.w3.org/Archives/Public/www-rdf-interest/2002Apr/0212.html from danbri__

12:49:00 <bijan> surely it is easier to *use latex* than to format a paper, in Word, so that it vaguely looks like LaTeX?

12:49:06 <danbri__> C:|Braindump on plugging in Google (and other web services) behind RDF APIs and query languages

12:49:06 <dc_rdfig> titled item C

12:49:10 <libby> hm, danbri gets better results than dan brickley on images

12:49:40 <bijan> danbri, have you read edd's anti-google/soap api thing on his o'r blog?

12:49:40 <danbri__> C:Chumped so I can find it again when I get back to coding. And cos I'm wondering if it makes sense to anyone else...

12:49:41 <dc_rdfig> added comment C1

12:49:50 <danbri__> no! remind me url...

12:50:24 <AaronSw> oreillynet.com/cs/weblog/view/wlg/1303

12:50:32 <AaronSw> http://www.oreillynet.com/cs/weblog/view/wlg/1303

12:50:44 <bijan> As my man to give...oh! there he is!

12:50:57 <bijan> Have you laid out my waistecote as well, aaron?

12:50:58 <bijan> :)

12:51:11 <AaronSw> Heh.

12:51:26 <AaronSw> I laid it out in the ocean for you.

12:52:09 <bijan> Hmm. Seems like I need to get out my flogging stick.

12:54:17 * danbri__ reads edd's page -- what a great rant!

12:54:22 <danbri__> ...and cute hack too

12:55:16 <bijan> I think it will turn into a taglines

12:56:28 <danbri__> I made a backlinks scraper in ~4 lines of perl; not sure what SOAP buys me. Bit more reliability, perhaps... (though google ui pretty stable).

12:57:07 <bijan> Bad counterrant: http://davidwatson.org:8086/2002/04/18.html#a79

12:59:59 <AaronSw> he can't even spell Edd

13:00:17 <AaronSw> Oh, Edd is paid six figures?

13:00:40 <AaronSw> I don't remember his example using Oracle...

13:01:10 <AaronSw> wow, that rant was hideous

13:01:33 <bijan> yep.

13:01:54 <bijan> Edd should add a Rebol one liner :)

13:02:02 <bijan> I mean, there *is* a point her.

13:02:14 <bijan> Once you write SOAP clients for all the enviroments you ever need.

13:02:23 <bijan> Which means getting a decent XML setup etc.

13:02:29 <bijan> And testing for interop.

13:02:39 <bijan> Figuring out the marshalling to your native types.

13:02:47 <jang> i don't think the soap bubble is about to burst

13:02:48 <bijan> you prolly need an XML Schema thingy.

13:02:54 <AaronSw> I bet you that using GET would be easier

13:03:03 <bijan> *Then*, it's trivial and so easy your dog could do it.

13:03:07 <AaronSw> GET is a prereq for SOAP

13:03:20 <bijan> Er...no it isnt'.

13:03:50 <AaronSw> Er, POST is...

13:04:25 <bijan> It is?

13:04:35 <bijan> SMTP binding wouldn't count?

13:04:50 <AaronSw> What SOAP services have an SMTP binding?

13:04:54 <AaronSw> Real ones, I mean

13:04:56 <bijan> I thought that was the one real *advantage* of SOAP.

13:04:58 <bijan> Ah.

13:05:14 <bijan> But htat's differnet than saying that POST is a prereq for *SOAP*>

13:05:22 <bijan> I mean, it's not for the protocol *per se*.

13:05:41 <AaronSw> Yeah, I suppose. It'd be interesting to write an HTTP->SMTP mapping, actually.

13:06:01 <AaronSw> To: REQUEST-URI

13:06:01 <AaronSw> Subject: METHOD

13:06:01 <AaronSw> Header: Value

13:06:04 <bijan> Well, kendall and i sketched out a system where sync calls would be HTTP and asych calls SMTP.

13:06:14 <AaronSw> Mime-RPC?

13:06:15 <bijan> I think Ed Dumbill finally realized that Google gave away the keys to the safe and now mere mortals have access to the safe and that means that the I can connect these people to that data in 1 line of shell script crowd is nervous. It's like the TV show Name That Tune.

13:06:29 <bijan> I have trouble parsing that, much less making it make sense.

13:06:37 <bijan> Oh yeah, it's *exactlY* like Name that Tune! :)

13:06:58 <AaronSw> chuckle.

13:07:10 <AaronSw> I do not see that analogy at all...

13:11:02 <bijan> That's because you don't understand the Value of Ubiquitous Simplicity.

13:11:05 <bijan> Which I should trademark.

13:11:38 <bijan> "Bijan Parsia: The Value of Ubiquitous Simplicity(tm) in Action"

13:12:51 <bijan> "Bijan Parsia: Bringing the Value of Ubiquitous Simplicity(tm) to You (now available in seven different spellings)"

13:13:38 <AaronSw> Are you available as a consultant?

13:13:45 <sbp> "Bijan Parsia: Bringing the Value of Ubiquitous Simplicity(tm) to You (now part of the XML canon)"

13:13:47 <AaronSw> The enterprise need you to ubiquify their simplicity.

13:14:06 <bijan> I'm *all* about enterprise ubiquity and simplicity..

13:14:36 <AaronSw> [aside] how the f... whoa sbp is here...is SOAP simple?

13:15:01 <bijan> Well, compared to Corba.

13:15:01 <jang> doesn't the "S" in SOAP stand for "Simple"?

13:15:05 <jang> like in "SNMP"

13:15:11 <bijan> It's simple in leaving out, oh, distributed GC.

13:15:15 <jang> and remind me what the "L" in "LDAP" is for...?

13:15:23 <AaronSw> I was gonna take that one

13:16:28 <MarkB> The best thing the W3C has done for SOAP is retire the acronym. It's not simple, not for object access, and almost not a protocol

13:17:13 <sbp> MarkB: heh, heh

13:17:23 <sbp> I loved your rant on www-tag today, BTW

13:17:32 <MarkB> heh, thanks

13:17:43 <MarkB> the bubble is bursting. it's fun watching.

13:17:53 <AaronSw> It is?

13:19:02 <MarkB> the backlash is picking up steam anyhow. bursting should follow by the fall.

13:19:31 <bijan> Hmm. I was feeling that way about the SemWeb. I see some signs of a SW winter looming.

13:20:05 <MarkB> i expect the SOAP bubble burst to trigger interest in SW. I've had some good behind-the-scenes talks with bigcos about this.

13:20:12 <AaronSw> Hm. I think it's time to remarket SW as SOAP for dummies.

13:20:23 <bijan> yes, a soap winter couldhelp advert a SW winter.

13:20:41 <bijan> But at the risk of making the deferred SW winter worse.

13:20:43 <AaronSw> But could cause a nuclear wint...I mean...

13:20:59 <MarkB> yup. at least 2 bigcos are making SemWeb plans as we speak.

13:21:22 <bijan> Well, as long as they don't bet the farm on it being something that it's not.

13:21:27 <bijan> That's where you end up with trouble.

13:21:47 <MarkB> For sure.

13:21:47 <AaronSw> But the SemWeb is *everything*, right?

13:22:07 <bijan> Aaron, don't forget that I'm a Semantic Smacker(tm)

13:22:24 <AaronSw> I'm as Semantic Swhacker

13:22:35 <bijan> And when I smack, it *means* something!

13:22:40 <MarkB> i'm waiting for a tools company to take on the SW. It's pretty inaccessible to novices right now.

13:22:43 <sbp> a Semantic Semantic Web Hacker?

13:22:56 <AaronSw> Maybe Semantich Whacker...

13:23:10 <sbp> you sound drunk

13:23:23 <AaronSw> nope...hic

13:23:26 <sbp> :-)

13:23:28 <bijan> Hmm. In british, that'd be a Semantec Wanker.

13:23:39 <bijan> Which I'm happy to call you, aaron, is you like.

13:23:45 * MarkB feeds the kid

13:23:53 <AaronSw> is?

13:24:00 <sbp> careful: Aaron is your outer child

13:24:30 <bijan> not mine thank god!

13:25:04 <bijan> I'm my own outer child.

13:25:18 <bijan> If Aaron were my outer child, we'd *always* be bickering.

13:25:24 <bijan> No we wouldn't!

13:25:27 <bijan> Yes, we would.

13:25:27 <bijan> would not!

13:25:30 <bijan> would to!

13:25:57 <bijan> Hmm. This does feel like a conversation with various people with nicks beginning with "s"...

13:27:39 <sbp> no it doesn't

13:28:01 <bijan> You're right!

13:28:08 <bijan> It only feels like a conversation with *you* :)

13:28:16 <sbp> :-)

13:29:08 <sbp> I just love it when someone starts an argument by assertion with absolutely no evidence or logical progression of ideas to back it up... I just can't resist

13:30:04 <bijan> Er...now *who* would POSSIBLY do *that*?

13:30:57 <bijan> Safety is important because when a program processes hypertext (in the

13:30:57 <bijan> general sense of nodes with arcs), it may follow arcs over and over

13:30:57 <bijan> again, whenever it feels like, and may pass node references to other

13:30:57 <bijan> processes which will also follow arcs over and over again, whenever THEY

13:30:57 <bijan> feel like it.

13:30:59 <AaronSw> <MarkB> i expect the SOAP bubble burst to trigger interest in SW. I've had some good behind-the-scenes talks with bigcos about this.

13:31:04 <bijan> (Prescod)

13:31:28 <bijan> Hmm. This makes me wonder about using, e.g., a linear logic type system with a serious semantic net.

13:32:02 <bijan> (er.. not ll *types*., but a ll *kind* of system)

13:32:10 <bijan> You cannot do that if the act of following an arc may have

13:32:10 <bijan> an implicit constraint like: "you should only follow it once, or only

13:32:10 <bijan> after following that other arc in the last five minutes."

13:32:26 <bijan> Which is exactly what a ll would allow us to express.

13:32:39 <bijan> Hm.m I can imagine arcs melting away as I traverse them...

13:32:40 <AaronSw> ll = logic language?

13:32:54 <bijan> linear logic

13:33:20 <bijan> Linear logics (and petri net like things) make statements resource bound.

13:33:33 <bijan> In standard logic, if you have P as a premise you can use it as many times as you like.

13:33:37 <bijan> In a LL, you can't.

13:33:51 <bijan> So it works very well for modeling, e.g., production processes.

13:34:15 <sbp> Error 410: Arc already traversed. - Sorry, but this resource is no longer available since it melted into the sea

13:34:29 <bijan> I point again to the Loom papers, for some backround in a relevant context.

13:34:51 <bijan> Hmm. Is it the loom papers for this?

13:34:55 <bijan> It may be another one.

13:35:20 <bijan> It's about comparing three kbs each with a different logic and how to merge them.

13:35:28 <sbp> .google "linear logic"

13:35:29 <xena> "linear logic": http://www.cs.cmu.edu/~carsten/linearbib/linearbib.html

13:38:18 <bijan>http://citeseer.nj.nec.com/hodas91logic.html

13:38:19 <dc_rdfig> D: http://citeseer.nj.nec.com/hodas91logic.html from bijan

13:38:34 <bijan> D:|"Logic Programming in a Fragment of Intuitionistic Linear Logic"

13:38:35 <dc_rdfig> titled item D

13:39:01 <bijan> D:How's *that* for an off-panting title? :)

13:39:01 <dc_rdfig> added comment D1

13:39:20 <bijan> D: And it's only the extended abstract!

13:39:20 <dc_rdfig> added comment D2

13:40:47 <MarkB> if a server cares whether an arc is followed more than once, then maintaining that knowledge is a significant side effect

13:41:26 <MarkB> => don't do that

13:42:11 <bijan> Hmm. I wonder...

13:43:21 <MarkB> there was another example about "1 millionth visitor". What if Google was the 1 millionth visitor? 8-)

13:44:01 <sbp> I wonder what percentage of websites get visited more than 1000000 times?

13:44:33 <JibberJim> If you count IIS worms - all of them not on virtual hostings ??

13:44:42 <sbp> heh, heh

13:45:54 <jang> <sbp> Error 410: Arc already traversed

13:46:02 <jang> isn't error 410: "payment required"?

13:46:04 <jang> in http?

13:46:14 <jang> which is almost as good.

13:46:36 <jang> and another reason to not wildly go firing off resource-intensive queries when a more careful plan might do better

13:46:40 <sbp> Hmm... no, "Gone"

13:46:50 <jang> sigh.

13:46:52 <AaronSw> [music: dun dun dun]

13:47:06 <jang> that doesn't give it away, aaron.

13:47:15 <jang> jaws?

13:47:27 <AaronSw> you know, the sort of dramatic chord thing

13:47:43 <sbp> 402 is Payment Required

13:47:45 <AaronSw> more dun dun daaaaaaaah

13:47:48 <jang> that's probably [dun dun daah!]

13:47:50 <jang> snap

13:47:54 <JibberJim> surely sures is [duuuhhh dun duuhh dun duh dun]

13:48:08 <bijan> Hmm. evenif there are examples where "link dissolving", or rather, resource contrained links (where resources may be measures as traversals") that doesn't mean that the whole idea is nuts.

13:48:08 <JibberJim> s/sures/jaws/

13:48:31 <jang> the IRC equivalent of tablature has just been (re) invented, it would seem

13:48:57 <jang> dwal

13:49:09 <jang> (that was me coughing up a lung, sorry).

13:49:26 <bijan> I'm not saying I in fact have a logic and a set of test cases which *do* work. I'm just not ruling them out *a priori*. espeically not for fear that google will win a lot of shopping sprees.. :)

13:49:53 <bijan> Hmm. Actually, somethign like this might work well for robot.txt like functionality.

13:50:40 <jang> the idea of traversing links as few times as possible, where traversing a link = calling up some remote, computationally-intensive (and possibly expensive) service

13:50:47 <jang> seems to make a lot of sense

14:05:18 <bijan>http://citeseer.nj.nec.com/376919.html

14:05:18 <dc_rdfig> E: http://citeseer.nj.nec.com/376919.html from bijan

14:05:35 <bijan> E:|"Intuitionistic Implication and Resolution"

14:05:35 <dc_rdfig> titled item E

14:06:02 <bijan> E:Should be read, I'll bet, by all the intuitionist fetishists working with N3.

14:06:02 <dc_rdfig> added comment E1

14:07:10 <bijan> E:"Some recent works rely on the intuitionistic theory of Hereditary Harrop formulas to provide extensions for Horn clauses while preserving the logic paradigm." Always a good thing, I think. I lik my Hereditary Harrop formulas *shaken*, not *stirred*.

14:07:10 <dc_rdfig> added comment E2

14:08:42 <bijan> E:"In particular, an intuitionistic implication is used to provide some form of scoping. However the lack of appropriate evaluation strategies, notably a strategy relying on a forward-chaining search, has limited its application. " Hmm. Sounds familar.

14:08:42 <dc_rdfig> added comment E3

14:09:03 <bijan> E:"This paper investigates a resolution schema for propositional Hereditary Harrop formulas, extending the resolution for Horn Clauses. A particular restriction of this resolution schema provides a specification of bottom-up search in presence of embedded intuitionistic implications." Yay! Academe saves the day!

14:09:03 <dc_rdfig> added comment E4

14:14:00 <bijan> We present a new framework for amalgamating two successful programming paradigms: logic programming and object-oriented programming. From the former, we keep the declarative reading of programs. From the latter, we select two crucial notions: (i) the ability for objects to dynamically change their internal state during the computation; (ii) the structured representation of knowledge, generally obtained via inheritance graphs among classes of objects.

14:14:09 <bijan>http://citeseer.nj.nec.com/476095.html

14:14:09 <dc_rdfig> F: http://citeseer.nj.nec.com/476095.html from bijan

14:14:24 <bijan> F:|"Linear Objects: logical processes with built-in inheritance "

14:14:24 <dc_rdfig> titled item F

14:14:33 <bijan> F:"We present a new framework for amalgamating two successful programming paradigms: logic programming and object-oriented programming. From the former, we keep the declarative reading of programs. From the latter, we select two crucial notions: (i) the ability for objects to dynamically change their internal state during the computation; (ii) the structured representation of knowledge, generally obtained via inheritance graphs among classes of objects

14:14:33 <dc_rdfig> added comment F1

14:14:52 <bijan> F:Still looking for the paper I can't find. But look at all the stuff I'm finding along the way! :)

14:14:52 <dc_rdfig> added comment F2

14:15:49 <bijan> We introduce an abstract form of interobject communication for object-oriented concurrent programming based on the proof theory of Linear Logic, a logic introduced to provide a theoretical basis for the study of concurrency. Such a form of communication, which we call forum-based communication, can be seen as a refinement of blackboard-based communication in terms of a more local notion of resource consumption.

15:29:21 <Seth>http://www.alpiri.com/

15:29:21 <dc_rdfig> G: http://www.alpiri.com/ from Seth

15:30:14 <Seth> G:|Tap Building the Machine to Machine Web

15:30:14 <dc_rdfig> titled item G

15:30:57 <Seth> G: R. V. Guha scores again !

15:30:57 <dc_rdfig> added comment G1

15:36:42 <JimJibber> JimJibber is now known as JibberJim

16:03:49 <jang> interesting end-of-telecon.

16:04:14 <jang> yet another piece of nonmonotonicity to add to the end of "RDF lets you say anything about anything" :-)

16:04:36 <jang> actually, it's one of the same old ones, but just cropping up again

16:05:12 <sandro> which one?

16:05:44 <jang> RDF lets you say anyhthing about anything except (in this case, to put a concrete example on it)

16:06:22 <jang> whenever you have a -foo-> b and a is of type bar then b is of type baz (ie, constrained containers)

16:06:48 <jang> because it uses what are (I s'pose) meta-syntactic constructs - well, just variables, and constraints over them

16:06:53 <jang> anyhoo

16:07:11 <bijan> it doesn't let you say, by means of individual statments, of each real number that it is in fact a real number.

16:07:36 <jang> well, not unless you can give them all URIs. Hrm, there must be a way to do that...

16:07:39 * jang is kidding

16:07:39 <sandro> You're being silly, Bijan.

16:07:39 <bijan> Though that's sorta pulling out the big gun.

16:07:42 <bijan> No no no!

16:07:56 <bijan> jang, there isn't a way to give them all URIs. that's the point. There aren't enough.

16:08:06 <jang> I am, I believe the expression is, pulling your chain

16:08:06 <bijan> Well, saying that rdf can say anything about anything is very silly.

16:08:12 <jang> I'm familiar with cantor's argument

16:08:16 <bijan> Ouch. It was so pulled :)

16:08:43 <jang> maybe there would be enough if we had more leters in our alphabet.

16:08:56 <bijan> I'm ignoring your trolling now jang! :)

16:08:57 <jang> in the words of homer simpson, "anyhoo". have a nice weekend.

16:08:59 <sandro> RDF lets you say almost nothing at all unless you define some vocabulary, extending the language. If you do that it's easy to do what you ask. Without doing that, you can't say much at all. There are a few tiny things in rdf:, but not many!

16:09:15 <sandro> You too, Jan.

16:09:21 <bijan> sandro, rdf can never do what i just asked.

16:09:33 <bijan> And yes, RDF lets you say almost nothing about most things :)

16:09:45 <bijan> Which is why the anything about anything is *especially* silly.

16:10:31 <bijan> It's funniest, of course, when it follows quickly upon "RDF trades expressivity for simplicity" :)

16:10:42 <sandro> Of course it can, bijan. Here's the N-Triples: _:a sillyLog:factWithEnglishSemantics "each real number is a real number". or whatever.

16:10:58 <bijan> That's not what I asked.

16:11:17 <bijan> Sorry, it was a little unclear.

16:11:28 <bijan> it doesn't let you say, by means of individual statments (i.e., one for each), of each real number that it is in fact a real number.

16:11:29 <sandro> Oh, I see what you were doing.

16:11:50 <sandro> Yep, it still can. Watch this:

16:11:58 <bijan> Yeah, i was trying to bar quantification.

16:12:28 <sandro> _:a sillyLog:factWithEnglishSemantics "You may now infer an RDF statement about each real number saying that it is a real number."

16:12:47 <bijan> I don't see that that did that.

16:12:58 <bijan> did what I asked.

16:13:09 <bijan> Even putting aside that that's not rdf saying it.

16:13:46 <bijan> After all, that's saying something about a bit of english.

16:13:52 <bijan> And nothing at all about real numbers.

16:14:08 <bijan> Actually, it's saying something about a string.

16:16:00 <sandro> Well, I still don't exactly understand what you want, although I think you're doing a Cantor thing. Yeah -- my point is the bit about "RDF saying it". RDF itself says almost nothing. It's a big mistake that rdf: is even considered part of RDF. Take that out. Now you REALLY can't say anything, as it should be. Just like you can't say anything in XML. Like XML, RDF is a way of structuring languages, much more than it's a language

16:16:00 <sandro> itself.

16:16:25 <sandro> BTW, did you read my answer about what exactly CWM logic is, yesterday on www-rdf-logic?

16:16:54 <bijan> I don't know if i'm doing a "cantor thing", since by "cantor thing" I think diagonal args. I'm just using the fact thta tthere aren't enough uris to name the reals.

16:17:42 <sandro> But there are enough bNodes, so that's not a problem.

16:17:48 <bijan> And I *said* that I was pulling out the heavy nukes. or did you miss the point that the "anything about anything" is often found near "RDf is simple and tractable?"

16:17:59 <bijan> Heheh. How do you *know* there's enough bnodes?

16:18:09 <bijan> I *strongly* doubt that there are.

16:18:35 <bijan> Indeed, I'll bet that there are only a denumerable number of bnodes.

16:19:08 <bijan> Even if there weren't, I'm very sketpical that it'd help you, if you looked at the details.

16:19:13 <sandro> I don't think that's the case. I think any formalism you can make which can express how many bNodes it wants will find them.

16:19:50 <bijan> give me such a formalism that is still RDF.

16:19:56 <sandro> URIs are a red herring (cf uname). Some objects in the universe happen to have unambiguous names, so we use them sometimes.

16:20:39 <bijan> There arent' enough RDF sentences to do the task.

16:20:49 <bijan> There are only a denumerable number of RDF sentences.

16:20:55 <bijan> And my example required more.

16:20:57 <bijan> Bang.

16:21:04 <sandro> "is" RDF is kind of meaningless. Can be structured or described in RDF....? RDF is obviously at least as structurally expressive as a sequence, and most logics involve sentences which are sequences of symbols.

16:21:27 <sandro> Why are there only a denumerable number of RDF sentences?

16:21:36 <bijan> Er..I don't know what "structurally expressive sequence"

16:21:52 <bijan> Because they are finite sequences.

16:22:07 <bijan> Same reason that there are only a denumerable number of uris.

16:22:23 <bijan> And there are only a denumerable number of FOPC wffs.

16:23:20 <sandro> Never mind. You could be right -- give me a use-case of RDF needing more that Aleph-0 sentences and I'

16:23:31 <sandro> ... and I'll take up this argument again,.

16:23:44 <sandro> 'later

16:23:49 <bijan> First, I am right. I don't see why you doubt me.

16:23:50 * sandro goes to run an errand.

16:24:04 <bijan> second, it wasn't a usecase for RDF, but a refutation of the anything about anything line.

16:24:10 <bijan> Third, ta.

16:25:52 <sandro> maybe it should be rephrased as :RDF can say anything that anyone has any real reason to want to say. :-)

16:26:46 <bijan> But why say that?

16:26:56 <bijan> I mean, what exactly *is* the point.

16:27:20 <bijan> Aside from it probably not being true, RDF is no more expressive than any other first order finatary formulism.

16:27:37 <bijan> It's worse if its' just syntax.

16:29:54 <bijan> :RDF can say anything that anyone has any real reason to want to say if you add a lot of stuff to it, but maybe not *all* you have reason to want to say at the same time

17:07:46 <DanC_> hm... what ever became of "ACTION: 2002-02-15#4 PatH: Send a few paragraphs to the list to address this" -- http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2002Feb/0476

17:07:56 <DanC_> that action isn't mentioned in the next week's minutes.

18:08:20 <rreck> slowly but surely

18:10:06 <AaronSw`> AaronSw` is now known as auto_aaronsw

18:10:24 <auto_aaronsw> auto_aaronsw is now known as AaronSw`

18:17:32 <DanC_> ImportError: No module named Crypto.Util.randpool

18:17:53 <AaronSw`> edit cwm_crypto.py and set whatsitcalled = 0

18:18:04 <DanC_> hmm... we really need command-line opts for what modules to import

18:18:43 <DanC_> existentials.remove(pair[0]) # Can't match anything anymore, need exact match

18:18:43 <DanC_> ValueError: list.remove(x): x not in list

18:21:46 <DanC_> if len(gl) != l: raise RuntimeError("@@Length is %i instead of %i" %(len(gl), l))

18:21:47 <DanC_> RuntimeError: @@Length is 33 instead of 8

18:22:00 * DanC_ is losing miserably

18:22:19 <AaronSw`> what're you trying to do today?

18:22:36 <DanC_> just parsing two n3 files. cwm A.n3 B.n3 works, but not cwm B.n3 A.n3

18:22:45 <bijan> Where do you want CWM to go today? You can have anyplace as long as it's weedy.

18:22:54 <AaronSw`> heh heh

18:23:01 <DanC_> I'm trying to write a test case ala sameGuy.n3 that uses a property that takes literal/datatype values.

18:23:24 <DanC_> i.e. I'm trying to raise the 'why L disjoint R?' issue with an illustritive test case.

18:43:04 * DanC_ sends bug report

19:37:12 <bijan> Sandro, if you meant: http://lists.w3.org/Archives/Public/www-rdf-logic/2002Apr/0045.html

19:37:38 <sandro> I did, Bijan. (helllo.....)

19:37:38 <bijan> I take miles' point add that this paragraph seems confused:

19:37:39 <bijan> While no logic can completely express the semantics of a more-expressive

19:37:39 <bijan> logic, any Turing-equivalent logic (eg Horn clauses or Java) CAN

19:37:39 <bijan> express any effective inference procedures. Such procedures guide

19:37:39 <bijan> a computer to do all it ever could do with the (inexpressible)

19:37:39 <bijan> semantics of more-expressive logics.

19:38:14 <bijan> I don't they *can* express them. they can implement them for sure. Where these things meet and matter is a little unclear.

19:39:24 <bijan> I'm a little confused, further, by the thought that effective inference procedures "guide" what we'd want to or could do with the other logics.

19:39:48 <bijan> A logic might, I imagine, be undecidable, but yet all proofs in it be decideably checkable.

19:39:57 <bijan> To pull on of your and tim favorite tropes :)

19:40:18 <bijan> But note, you *still need the more expressive logic*.

19:40:36 <bijan> Because otherwise you can't express your proof to-be-checked.

19:41:12 <bijan> (similarly for heuristic methods of proof search)

19:41:19 <sandro> Sure you can -- make up some vocabulary for it, no problem at all. Then express (in your Turing logic) your proof-checking proceedure.

19:42:30 <bijan> Well, if you say so. Is that a promise to do that for any one reasonable example?

19:43:06 <bijan> hehe. "just make up some vocabulary"

19:43:17 <sandro> I'm not sure what you mean. Do you mean you know a case where it's impossible?

19:43:22 <bijan> If the vocabulary hs the right semantics you *ipso facto* have the extra logic.

19:43:52 <bijan> THe situation is "having extra vocabulary with the right semantics to express the proof" and *not* having the more expressive logic.

19:44:18 <sandro> No, it's not.

19:45:15 <sandro> I'm not sure what you mean by "having" a more expressive logic. It's obviously trivial to "have" the syntax, and I'm granting that for many logics it will be impossible to formally express the semantics in our base language.

19:46:06 <bijan> formally express the semantics...

19:46:13 <bijan> There's a lot of meanings to this.

19:46:25 <bijan> One's metalogic needed have the same style semantics as the object level logic.

19:46:29 <bijan> That's one point.

19:46:31 <sandro> But there is very, very important middle ground, which is doing things, like, oh, maybe inference on the logic. How you want to do THAT can be expressed in our base language.

19:47:09 <bijan> On the other hand, if you want to express a statement in a certain logic, it needs to be capable enough to *say* that.

19:47:47 <sandro> Right. And it's not going to be. So we don't. And that's fine, because this is computers.

19:48:02 <bijan> Oh, ok, if it's computers.

19:48:13 <sandro> Computers only handle Turing-equivalent logic.

19:48:57 <bijan> you need to be precise about "handle"

19:49:13 <bijan> Because that's just false on the readings I keep generating.

19:49:21 <sandro> Not when I'm spouting bromides like THAT I don't. :-)

19:49:42 <bijan> Oh, ok. Then I stop.

19:50:11 <AaronSw`> Busy Bijan of N: How many bromides can a bijan of N states handle before it halts.

19:50:40 <sandro> The point is, cwm is just a programming-language interpreter, of a very similar nature to a tabled-prolog system.

19:51:12 <bijan> It astonishes me that Pat Hayes missed that.

19:52:50 <sandro> Are you agreeing with me, or disagreeing indirectly by saying that if I were right, Pat would have said it already?

19:53:15 <bijan> I'm saying that, at best, there is a lacuna in your argument.

19:54:07 <AaronSw`> .wn lacuna

19:54:08 <xena> lacuna defined as:

19:54:09 <xena> - n 1: a blank space or missing part [syn: {blank}]

19:54:10 <xena> - 2: an ornamental sunken panel in a ceiling or dome [syn: {coffer}, {caisson}]

19:54:21 <bijan> I meant, of course, sense 2 :)

19:54:29 <sandro> Which would be closed if I finished my cwm's-vocabulary-rdf to tabled-prolog translator, and it passed the cwm test suite?

19:54:37 <AaronSw`> Sandro's big into ornamental arguments, I guess.

19:54:52 <bijan> not at all, as far as I can tell.

19:55:03 <bijan> Indeed, I don't yet even see the structure of the argument.

19:55:21 <sandro> Okay, I'll structure it for you.

19:55:38 <bijan> I mean, I wrote most of a CWM interpreter in prolog.

19:55:48 <bijan> which part exactly did I miss?

19:55:57 <bijan> Or will tabled prolog add something?

19:56:23 <sandro> You wrote an interpreter, not a translator, to my great surprise.

19:56:37 <bijan> Technically, I wrote a compiler.

19:56:54 <bijan> With a metacircular interpreter for applying the rules using forward chaining.

19:57:00 <sandro> But it doesn't compile to prolog.

19:57:09 <bijan> yes it certainly does.

19:57:44 <bijan> the code and fairly reasonable discussionof it is available on the web.

19:57:53 <sandro> It turns the n3 file " <a> <b> <c>. " into "rdf(a, b, c)." ?

19:58:26 <bijan> Of course, since I built it ontop of rdf_db

19:58:39 <bijan> The rules then get compiled into a different form, of course.

19:58:46 <bijan> Still prolog though.

19:59:31 <sandro> And it turns "this log:forAll <x>. { <a> <b> <x> } log:implies { <a> <c> <x>}." into "rdf(a, c, X) :- rdf(a, b, X)." ?

19:59:45 <bijan> Not at all.

19:59:48 <bijan> Why would I do that.

19:59:55 <bijan> It turns it into something like: http://www.unc.edu/~bparsia/sw/cwmclone/music_brainz_rule.P

19:59:58 <sandro> because it has identical semantics!!!!!!!!!!

20:00:24 <bijan> Well, that's never been all that clear to me.

20:00:36 <bijan> That it has the idenitical semantics.

20:01:16 <sandro> I understand that it has not been clear to you. So I suggested trying it and noticing that (lo!) it passes the test suite. That would be some evidence that it has identical semantics!

20:01:31 * AaronSw hits the "TYPO - electricute bijan!" button

20:01:42 <bijan> Heh.

20:02:36 <sandro> It will only pass the test suite if the prolog is tabled, of course. Untabled prolog has different semantics.

20:02:51 <bijan> To be precise, I don't know that :- and log:implies have the same semantics.

20:03:29 <bijan> That's consistent there being many expressions mapping between them with the same semantics.

20:03:32 <bijan> But be that as it may.

20:03:38 <sandro> You also need to create skolem functions for log:forSome variables in the consequent of log:implies, of course.

20:04:53 <bijan> Sure. i certainly acknowledge that n3 is some varient of horn logic.

20:04:57 <bijan> but you knew that too.

20:06:37 <bijan> I'm afraid I haven't seen your n3 to tabled-prolog translater. So if you want me to examine it, i'll need a pointer.

20:06:42 <sandro> You just think it might be different from tabled prolog in some suble way.....?

20:07:00 <sandro> subtle

20:07:05 <sandro> eek!

20:07:09 * sandro brb

20:07:14 <bijan> I suspect that some constructs might not map easily or exactly.

20:12:29 <sandro> Such as?

20:12:57 <bijan> log:forSome. [] in rules.

20:13:20 <bijan> Builtins

20:13:39 <bijan> Not a construct, but some aspects of the forward chainer.

20:13:54 <bijan> Variables in general.

20:14:05 <sandro> Yeah, as I said, you have to make log:forSome (and thus []) into Skolem functions.

20:14:14 <bijan> Variable scope.

20:14:22 <bijan> Nested contexts.

20:14:47 <bijan> I'm not saying that all of these are problematic, but I certainly haven't mapped out all the details of all of them.

20:14:58 <sandro> Builtins are an interesting case. They probably map very nicely to prolog rules or builtins, but I'd kind of rather not do that.

20:15:00 <bijan> There is no document which completely describes them, much less formally.

20:15:38 <bijan> And, even if one understands llyn.py, it certainly has some bugs, so Appeals To Tim become necessary.

20:15:46 <sandro> Variable scope and nested contexts should be dealt with using the standard algorithm for putting a FOL formula into implicative normal form.

20:16:07 <bijan> Sandro, note that I'm not asking you to explain all these.

20:16:22 <bijan> you asked me the bits i don't quite get in terms of clean mapping.

20:16:25 <bijan> These are some.

20:16:45 <bijan> You're asserting the contrary won't fix my doubts, because my doubts lie in the exact details.

20:16:53 <bijan> Your

20:17:21 <bijan> Of course, your translater might help, assumign we feel confident about the test suite.

20:17:24 <bijan> Which I don't, btw.

20:17:27 <sandro> True. But I'm trying to convince you that the mapping is mostly very clean, and implementing the mapping is VASTLY more useful to the semantic web effort than just re-implementing cwm.

20:18:01 <bijan> Well, cwm and euler.

20:18:24 <sandro> Yeah....

20:18:35 <bijan> And I don't have confidence in my understanding N3 absent my reimplementation.

20:18:49 <bijan> THough I'll happily read any detailed formal specs you care to produce.

20:18:58 <sandro> While I'm at it, it's also very important that the clone *not* read n3, but read standard RDF/XML. But that's another issue.

20:19:00 <bijan> Especially if all the relevant user communities sign off on them.

20:19:12 <bijan> It does read standard RDF/XMl.

20:19:26 <bijan> But i don't see why that would be important.

20:19:46 <bijan> Other than, I'd like that and it was trivially easy building ontop of a prolog with extensive RDF/XML support.

20:19:48 <sandro> It reads log:implies with its formulas and scoped variables from RDF/XML files?

20:19:56 <bijan> Of course not.

20:20:07 <bijan> Well, it depends on what you mean.

20:20:11 <sandro> Right -- I didn't think so -- but it should.

20:20:23 <bijan> Since RDF/XMl doesn't have either, there's nothing of that sort to read from them.

20:20:37 <bijan> Scoped variables actually, maybe.

20:20:37 <sandro> As far as I know, there is no program today that properly turns n3 into RDF/XML without losing important information.

20:20:56 <bijan> I don't interpret log:forAll statements until they're in the triplestore.

20:21:00 <bijan> So really, I'm usin gdark triples

20:21:05 <bijan> Directly int eh top level context.

20:21:38 <bijan> Well, I do plan doing a mcdermott approach.

20:21:53 <bijan> Which is, afaict, the only way to get formulae into strict RDF1.0 RDF/XML

20:21:59 <sandro> Ugh. I'm wasting my time. I should be coding, instead of trying to convince you to code.

20:22:10 <bijan> Were you?

20:22:15 <sandro> Yep.

20:22:16 <bijan> I'm happy to take requests, natch.

20:22:38 <bijan> But "CWNclone should..." doesn't sound like a request :)

20:22:55 <bijan> "For the sake of the semantic web effort..."

20:22:56 <DanC_> take them and laugh at them, huh, bijan?

20:23:03 <bijan> Heh :0

20:23:06 <bijan> Not at all.

20:24:04 <bijan> I'm not quite sure why you'd think that. To date, I've received *no* request for features in CWMClone.

20:24:09 <bijan> So I just putter on my own.

20:24:21 <sandro> Are you pointing out the difference between saying "please turn left" and "there's a shortcut that will save is 3 months, at the next left.".?

20:24:35 <bijan> I work on the assumption that the closer I get to a drop in replacement for cwm, the more likely people will be to use it.

20:24:38 * DanC_ takes back the poke at bijan

20:25:03 <bijan> Thank you, DanC, I really didn't see that it was warrented so I appreciate the retraction.

20:25:21 <bijan> sandro, I have trouble distinguishing between your theoretical arguments and your requests for features.

20:25:30 <bijan> If you say, "CWM should be a certain way" that's one thing.

20:25:34 <sandro> I can understand that. :-)

20:25:38 <bijan> But CWMClone is a clone of, well, cwm :)

20:25:53 <bijan> If you'd like me to try to add a different inference engine, I have no inherent objection.

20:26:04 <bijan> Indeed, I've pointed out that I intend to add an euler one.

20:26:59 <sandro> Bijan! I would make me sooooo happy if you would try to make (1) something which encoded n3 in RDF/XML [in a way Tim liked], and (2) something which did cwm-think and cwm-filter on that by turning it into simple and clean tabled prolog and running that.

20:27:10 <sandro> Bijan! *IT* would make me sooooo happy if you would try to make (1) something which encoded n3 in RDF/XML [in a way Tim liked], and (2) something which did cwm-think and cwm-filter on that by turning it into simple and clean tabled prolog and running that.

20:27:30 <bijan> Well, 2 is unlikely, since I don't have a tabled prolog handy.

20:28:08 <sandro> They're not hard to find, are they?

20:28:21 <bijan> Well, XSB doesn't have much rdf support.

20:28:26 <bijan> I suppose I cross compile.

20:28:36 <bijan> no one I know likes working with XSB though.

20:28:52 <bijan> And 1 is on the list. can't promise tim likeatude.

20:29:33 <sandro> I wouldn't want you to use any special features of XSB. I'd like the generated code to work in at least two tabled prologs (I'd chose XSB and B-Prolog, but I haven't done much research in the area).

20:30:06 <bijan> B-Prolog has some nice features, doug tells me, but it's also very rough ins pot.s

20:30:44 <bijan> But i'll keep it in mind.

20:30:54 <AaronSw> Hm, surely there's someone who hasn't implemented daml:colelction

20:31:12 <bijan> I think that that's behind a lot of other features, *but*...

20:31:23 <bijan> If you give me a mapping, I prolly can modify my compiler fairly easily.

20:31:37 <sandro> Sorry, Aaron, how is that relevant?

20:31:52 <AaronSw> sorry, was being semi-random

20:33:52 <AaronSw> Hm, looks like Sandro and me disconnected somewhere. Sandro wants people to deal with a) RDF/XML and b) N3 logic hacks.

20:33:55 <AaronSw> bleech.

20:35:12 <bijan> A mcdermottian encoding via reification shouldn't be that bad.

20:35:17 <bijan> And reasonably roundtripable.

20:35:47 <bijan> I mean, structurally, it should be just like his "if".

20:36:49 <AaronSw> I want to see RDF/XML stabbed dead, dead, dead!

20:38:47 <bijan> The advantage is that you don't have to write an n3 parser.

20:39:19 <AaronSw> N-Triples is better for that, IMO.

20:40:36 <bijan> same diff, in this case.

20:41:26 <bijan> i.e., the key bit is how do you deal with, e.g., formulae.

20:44:06 * DanC_ wonders why AaronSw wants RDF/xml dead

20:44:13 <DanC_> I mean: what's wrong with KIF, then?

20:44:20 <DanC_> rather...

20:44:25 <DanC_> nothing's wrong with KIF, but...

20:44:28 <sandro> Aaron, I'm fine with TODL or whatever. I'm still hooked on arbitrary syntaxes. I just find RDF/XML a convenient syntax for some conversations and communities.

20:44:29 <AaronSw> I like KIF.

20:44:36 * sandro has to run. chat later.

20:44:42 <DanC_> if you're not interested in RDF/XML, why do you bother with RDF at all? why not just use KIF/prolog/whatever?

20:45:03 <AaronSw> Are you serious?

20:45:07 <DanC_> yes

20:45:11 <AaronSw> Because it's hard to quote URIs in KIF.

20:45:20 <DanC_> really?

20:45:36 <AaronSw> Well, just judging from the output of your software.

20:45:38 <DanC_> I think |http://...| works.

20:45:39 <bijan> easy in prolog.

20:45:43 <AaronSw> Oh, neat.

20:45:46 <bijan> I would think so.

20:45:58 <AaronSw> Well, I think we should all switch to KIF then.

20:46:00 <DanC_> well, |...| is common-lisp syntax. And most KIF readers are actually common lisp readers.

20:46:03 <bijan> Plus, if you use a common lisp, it's trivial to add useful reader macros.

20:46:15 <bijan> And other macros.

20:46:21 <AaronSw> I guess N-Triples is still easier to parse than KIF.

20:46:40 <bijan> Indeed, writing useful syntaxes for KIF (given common lisp) is about as good as it gets.

20:46:43 <bijan> Better, even :)

20:46:58 <AaronSw> DanC, you seriously think the only thing RDF has over KIF is RDF/XML?

20:47:20 <DanC_> there are lots of good reasons to switch to KIF, but there are lots of good reasons to use betamax rather than VHS too. deployment is what's interesting, to me.

20:47:50 <AaronSw> RDF/XML deployment is negligible, IMO.

20:47:52 <DanC_> the only thing RDF has over KIF is that web folks are willing to look at and try to use RDF.

20:48:08 <AaronSw> And why is that?

20:48:13 <DanC_> dunno

20:48:34 <DanC_> well, stuff like the RDF primer helps.

20:48:40 <DanC_> tim bray's xml.com article

20:48:42 <bijan> W3C impriture helps.

20:48:47 <bijan> More lit on it.

20:48:50 <bijan> More tools.

20:48:57 <bijan> Used in other specs.

20:49:07 <bijan> (E.g., I predict SVG will bring folks in)

20:49:25 <bijan> RSS1.0 can't hurt (ow!)

20:49:31 <AaronSw> Wow, that is incredibly disheartening.

20:49:41 <AaronSw> So why did W3C create RDF?

20:49:41 <bijan> Brutal, ain't it?

20:49:54 <bijan> Aaron, you must become a Common Lisper.

20:50:00 <bijan> Then all your despairs become one!

20:50:02 <AaronSw> yeah, yeah, after i learn C

20:50:23 <DanC_> er... W3C created RDF to do generic metdata stuff. you know PICS was spelled with ()'s, don't you?

20:50:29 <AaronSw> Yep.

20:50:40 <DanC_> folks complained about the interactions between () and <> when putting PICS in HTML.

20:50:41 <AaronSw> Wouldn't KIF have worked for that?

20:50:57 <AaronSw> Heh.

20:51:01 <DanC_> KIF, i.e. s-expressions, did work for PICS, as far as that goes.

20:51:12 <DanC_> i.e. there's no problem, technically.

20:51:12 * AaronSw remembers that dying FAQ question

20:51:13 <bijan> Hm. if I may put on my embittered lisper hat (though I only channel them): Why XML instead of sexpr! That woudl solve *everything*!

20:51:56 * bijan imagines trying to convince folks that we should move from XML generally to sexpre *so that* we could replace RDF with KIF...

20:52:05 <DanC_> <> won over () because <> got integrated with URIs before ()s did

20:52:14 * bijan imagines getting pointy bracketed to death.

20:52:27 <AaronSw> I thought PICS had URIs...

20:52:39 <bijan> The second imagining is RDFMT entailed by ther first (not really)

20:52:41 <DanC_> yes, PICS had URIs; several years after HTML had them, though.

20:53:11 * DanC_ wonders why AaronSw finds this disheartening? what did you think was going on?

20:54:08 <AaronSw> So the W3C intentionally wrote a worse spec that existing stuff in a failed attempt to integrate metadata with HTML? Do I have to explain why that is disheartening?

20:54:21 <DanC_> hmm.... <> also won over () because, in fact, most stuff that people write is unstructured/informal. in <> languages, data characters dominate over markup. in (), you have to work harder to do data.

20:54:26 <AaronSw> And now we've spent three years cleaning up the mess

20:54:38 <AaronSw> And we're still worse off than we could have been...

20:54:47 <bijan> But that's not what happened. i mean, especially the "intentionally" bit.

20:54:49 <DanC_> worse spec: worse in what way?

20:55:09 <DanC_> technically, <> and () are equally just fine.

20:55:47 <bijan> Hmm. elhwBijan says, "No way! McCarthy is God. Eat your veggies!"

20:55:49 <AaronSw> Yeah, but RDF M&S is totally confusing, has no formal grounding, informally specifies RDF/XML and even if it didn't RDF/XML is torture for humans to right and even worse for machines to parse.

20:56:04 <bijan> It's not worse for machines.

20:56:16 <AaronSw> Well, it is for the people who teach the machines...

20:57:01 <DanC_> confusing: guilty. no formal: I think there always was a reasonable formal grouding; it just got muddled in the 1st REC.

20:57:21 <bijan> Hmm. I take the MT to be proof of exactly that.

20:57:32 <AaronSw> of what?

20:57:45 <bijan> that there was always a reasonable formal basis.

20:58:12 <AaronSw> then why did we throw out rdf:value, containers and reification?

20:58:13 <bijan> There was no good *explication* of that basis.

20:58:20 <bijan> Hmm.

20:58:37 <DanC_> human torture: umm... guilty again. (due to a misguided attempt to get RDF to fit inside HTML, and due to being an early XML adopter)

20:59:03 <bijan> Arron, you mean, aside from them sucking?

20:59:07 <DanC_> throw out rdf:value?

20:59:10 <bijan> Doh.

20:59:17 <DanC_> containers and reification were botched. guilty.

20:59:19 * AaronSw wonders why Yahoo thinks Kansas City, MO is 16 miles from his house

20:59:37 <AaronSw> well the WG never decided what rdf:value meant

20:59:42 <AaronSw> and doesn't seem to be going to

21:00:11 <DanC_> you seem to have overly high expectations about rdf:value. It's just like rdfs:label; it's a handy property.

21:00:21 * bijan wonders how he ended up on the "defendng rdf" side.

21:00:35 <AaronSw> I can tell you what rdfs:label means, or where to use it. I can't do the same for rdf:value.

21:00:40 <bijan> Ah, it's the AntiAaron side :)

21:00:54 <DanC_> do tell what rdfs:label means, in formal terms, AaronSw.

21:01:03 <AaronSw> I didn't say in formal terms

21:01:16 <DanC_> ok, then in any terms

21:01:26 <AaronSw> it associates an object with a name

21:01:41 <DanC_> ok, so rdf:value associates an object with a value.

21:01:48 <bijan> Heh.

21:01:49 <AaronSw> what do you mean by value?

21:01:54 <DanC_> what do you mean by name?

21:02:11 <bijan> D'oh! my "/me waits for it" was too slow!

21:02:26 <AaronSw> a name is something like "Fred" or "N-Triples" or RDF. My friend's name is Fred, the format's name is N-Triples

21:02:36 <AaronSw> What's is N-Triple's value?

21:02:43 <DanC_> I usually use <http://x> rdf:value "abc" analagously to <a href="http://x">abc</a>.

21:02:44 <AaronSw> Give me an example of usage here.

21:02:58 <bijan> ooo. You're friend's name better not be Fred. Fred would be unhappy.

21:03:00 <AaronSw> That's totally contrary to the current specs

21:03:00 <bijan> yoru

21:03:04 <bijan> (damn)

21:03:22 <DanC_> really? the current specs say "don't use rdf:value for stuff like link text"?

21:03:33 <AaronSw> pretty close

21:03:38 <AaronSw> we went over this once

21:03:39 <DanC_> recall: I stipulated to the M&S spec being confusing.

21:03:44 <AaronSw> they use it for what you use currying for

21:04:10 <DanC_> er... when you say "current specs" do you include M&S?

21:04:15 <AaronSw> yeah

21:04:30 <AaronSw> I mean RECs

21:04:40 * DanC_ re-adjusts conversational vocab

21:04:41 <AaronSw> and Proposed RECs or whatever RDF Schema is

21:04:55 <bijan> Candidate.

21:05:31 <AaronSw> I call RDF Core's specs the new specs.

21:06:11 * AaronSw gives up trying to find a phone number for "195 Skokie Valley", just calls United

21:06:11 * DanC_ re-reads recent conversation with adjusted interpretation of "current"

21:08:11 <DanC_> ok, if by "why did we drop rdf:value [along with containers and reification]" you're talking about currying stuff, then guilty again. The M&S spec waved its hands around n-ary stuff and botched it.

21:08:50 <DanC_> containers and reification were in there to reproduce PICS functionality.

21:09:04 <AaronSw> not to mention aboutEach*

21:09:06 <DanC_> hmm... well, containers were motivated by other metadata apps too, I suppose.

21:10:15 <DanC_> I'm still looking for the disheartening part. Yes, web technologies suck. Shall we have an "HTML sucks" party as well? or an HTTP sucks party?

21:10:17 <AaronSw> So if we're guilty on all these things, and had a negligible user base (compared to say SOAP and XML-RPC) who's idea was it to have a WG go and fix the spec instead of writing a new one?

21:10:32 <bijan> Ooo, I'd go to an HTML sucks party!

21:10:36 <DanC_> whos idea: I'll give you one guess.

21:10:36 <AaronSw> The HTML sucks party only happens when Ted Nelson is around.

21:10:43 <AaronSw> Hm. Yours?

21:10:48 <bijan> Aaron, you're kidding right?

21:10:48 <DanC_> no

21:10:57 <AaronSw> aw, there goes my guess

21:11:00 <AaronSw> bijan, on which?

21:11:15 <bijan> That only ted nelson is the HTML sucks partier.

21:11:19 <AaronSw> DanC, so who's idea was it?

21:11:26 <AaronSw> bijan, try again, i didn't say that

21:11:49 <bijan> Sorry, better, that ted nelson's presence is necessary to an HTML sucks party

21:12:15 <DanC_> Eric Miller is pretty invested in RDF. He's got quite a few promise cards played.

21:12:15 <AaronSw> Well, it certainly makes it much more fun.

21:12:33 <bijan> That I won't dispute!

21:12:44 <bijan> "HTML is a CRIME against all that is DECENT and GOOD in HUMANITY!"

21:12:53 <AaronSw> "Hamburger meat! With legs!"

21:13:04 <AaronSw> DanC_, are you fingering EricM?

21:13:24 <DanC_> fingering in what way? EricM's on record as leading the RDF charge.

21:13:42 <DanC_> I'm a pretty happy soldier.

21:13:44 <AaronSw> OK, he's gonna get it...

21:14:05 * xover has a filthy imagination... :-)

21:14:23 <AaronSw> I suppose it's an American idiom.

21:14:45 <DanC_> of course, when the "fix RDF" plan was hatched, it was supposed to take 6 monhts.

21:14:46 <bijan> Do. Not. Go. There.

21:15:14 <AaronSw> So when's Plan B gonna start?

21:15:20 <DanC_> we were supposed to be well into RDF v3 by now ;-)

21:15:44 <DanC_> Plan B? well, plan B was last call by WWW2002

21:15:44 <AaronSw> I'm sick of working in serial. Parrelelize!

21:16:00 <AaronSw> Er, I was thinking about the Plan B that involved tearing M&S into shreds

21:16:36 <bijan> Parr... Para...p...I can't *spell it* much less say it! :0

21:16:37 <DanC_> M&S is in shreds, no? 5 of them: syntax/model-theory/primer/schema

21:16:40 <DanC_> +datatypes

21:17:05 <AaronSw> That's not exactly the sense I meant.

21:17:19 <DanC_> you meant replacing it with KIF?

21:17:27 <AaronSw> Yeah, or *something*.

21:17:46 <DanC_> what sorta *something*?

21:17:57 <DanC_> I mean: to what end?

21:18:14 <AaronSw> To have a foundation for the Semantic Web that isn't made of cobwebs.

21:18:38 <DanC_> every time I look at alternative designs, they all have something very much like RDF at the core. We could play marketing games and call it "XML Object Serialization" or something, but...

21:18:50 <AaronSw> I'm fine with say, N-Triples.

21:18:58 <AaronSw> Or TODL.

21:19:10 <DanC_> well, you have n-triples. so?

21:19:23 <DanC_> .google TODL

21:19:24 <xena> TODL: http://www.todl.com

21:19:24 <dc_rdfig> Label TODL not found.

21:19:29 <AaronSw> .google sandro todl

21:19:30 <xena> no results found.

21:19:43 <AaronSw> hm. it's sandro's ntriples+literals as subjects and stuff

21:20:02 <DanC_> DanC_ is now known as Geenie

21:20:03 <bijan> Personally, i would have thought that if you're happy with ntriples, what care you RDF/XML.

21:20:13 <Geenie> prong! ok, TODL is a REC.

21:20:18 <Geenie> what do you put in the press release?

21:20:22 <bijan> And if what you wanted was KIF, what point ntriples?

21:20:25 <Geenie> Geenie is now known as DanC_

21:20:38 <AaronSw> Press Release: "W3C releases innovative new spec for Web services deployement"

21:20:54 <DanC_> innovative? gimme a break!

21:21:04 <AaronSw> Hey, you asked for a press release!

21:21:05 <bijan> What's probably needed is an ntriples/XML.

21:21:17 <bijan> So you can embed it in things like SVG easily.

21:21:24 <DanC_> well, W3C has this wierd way of doing press releases... we actually write stuff we believe in them.

21:21:33 <bijan> Heh.

21:21:41 <AaronSw> weird.

21:22:09 <AaronSw> What I really want is for RDF core to stop worrying about insane problems like RDF/XML syntax bugs and not being able to do anything beyond the holy writ of M&S (like add daml:collection) and start doing interesting things like logic

21:22:15 <DanC_> yes, it limits the scalability/power/dominance of W3C considerably. But, well, with a brit at the helm, that's sorta what you get.

21:22:29 <bijan> Brits are *always* to blame.

21:22:43 <DanC_> AaronSw, what you want is a new WG

21:22:49 <AaronSw> Yes, I do.

21:23:04 <AaronSw> I've already drafted a list of members.

21:23:06 <Seth> personally i dont see what XML adds to a meta-data krep language .. anyone care to tell me what?

21:23:24 <DanC_> technically: zero.

21:23:51 <bijan> As long as you've drafted the *list* and not anyone on it, ok :)

21:23:56 <DanC_> socially: familiarity (recall: <>s got integrated with URIs 1st) and text-friendliness.

21:23:56 <Seth> i mean you can embed any krep language inside of a xml document if you want

21:24:00 <AaronSw> see http://rdfig.xmlhack.com/2002/03/30/2002-03-30.html#1017460652.727047

21:24:05 <AaronSw> Hey, look, Bijan is on it!

21:24:21 * AaronSw ducks

21:24:23 <bijan> D'oh!

21:24:36 <Seth> text-friendliness ... to who?

21:24:37 <bijan> Sorry, must decline, health/job/family reasons.

21:24:49 <AaronSw> Heh.

21:25:03 <DanC_> ooh! let's do WG membership by election! (GAG! hork! phtphtp!)

21:25:22 <AaronSw> Oooh, even DanC is horking now.

21:25:42 <bijan> Hmm. I think this was more on the "Let's do WG membership by Aaron-fiat" model, DanC.

21:25:53 <AaronSw> We clearly need Fantasy Working Groups

21:26:08 <DanC_> yes, AaronSw, I nominate you as Benevolent Dictator of W4C.

21:26:19 <AaronSw> Is that the World Wide Wireless Web?

21:26:22 <bijan> I prefer my Fantasies leisurely and rather more solitary.

21:26:48 <AaronSw> Hm, I can't find sbp's XHTML deam team.

21:27:14 <bijan> I don't know why you want a WG for that. Just join a research team.

21:27:34 <AaronSw> it's hardly research

21:27:47 <bijan> Really?

21:28:00 <AaronSw> Oh, HTML or RDF?

21:28:10 <bijan> There's something remotely like even W3C member general understanding about what a "web logic" should be.

21:28:11 <bijan> Ah.

21:28:12 <bijan> RDF>

21:28:15 <DanC_> "hardly research" <- famous last words. (1st hand experience, here)

21:28:24 <bijan> that too.

21:28:39 <bijan> I follow kent pitman -- usually, a posteriori specs are the way to go.

21:28:56 <bijan> You need a wide body of experience and active deployment before you standardize.

21:29:02 <AaronSw> I don't particularly care what it's called

21:29:04 <bijan> Oops. s/spec/standard/

21:29:22 <bijan> It's not what it's called, it's what it does and what it's supposed to do.

21:29:42 <bijan> Standard building is *always* about fussing and politics.

21:30:09 <bijan> And finishing what you have, rather than starting over.

21:31:25 <AaronSw> So, DanC, what do you rate the chances of this ever happening?

21:32:00 <DanC_> chances of what? a new WG? I'd rate the odds of a "what next for SemWeb/RDF?" workshop this fall at ~80%

21:32:49 <dmiles> dmiles is now known as dmiles-wc

21:33:09 <AaronSw> I was thinking more along the lines of a group to work on / push S-Triples and maybe inference/query/rules on top of it.

21:33:51 <DanC_> umm... that sounds kinda like SWAD.

21:33:55 <sbp> <AaronSw> Hm, I can't find sbp's XHTML deam team.

21:33:58 <DanC_> i.e. us. here. now.

21:34:00 <sbp> Murray Altheim, MarkB, Tantek, DanC, Al Gilman, Roger Gimson, Masayasu, Chaals, Ian, Dave Raggett, AaronSw, Jason White

21:34:07 <AaronSw> Ah, thanks.

21:34:10 <sbp> np

21:34:43 <bijan> Is AaronSw the only person beside DanC who's in the intersection of these dream teams?

21:34:44 <DanC_> ew... that's the stuff of bad dreams, not fantasy. Good guys all, in that list, but talking about HTML with them? I'd rather get my teeth drilled.

21:34:53 <sbp> lol!

21:34:59 <bijan> Espeically that aaron character.

21:35:15 * DanC_ heads out for family stuff.

21:35:21 <AaronSw> I wonder why it'd be so painful.

21:35:33 <sbp> painful: I think it's the nature of HTML :-)

21:35:36 * AaronSw considers resigning from RDF Core

21:35:47 <bijan> If you have to ask, you are the source and not the recipient of the pain :)

21:35:58 <sbp> heh, bijan's nth rule

21:36:21 <bijan> Corollery 1: if you have to ask *me*, please go away.

21:36:36 <AaronSw> SWAD doesn't do ?x .

21:36:40 * DanC_ did my time in HTML-standards-purgatory already, ThankYouVeryMuch.

21:38:19 <AaronSw> Hm, Sandro should take out bNodes from TODL.

21:39:05 <Seth> aaron, what's S-Triples ?

21:39:12 <AaronSw> http://www.w3.org/2002/03/todl/

21:39:19 <sbp> S-Triples was the old name for them

21:39:33 <AaronSw> them is sbp's name for TODL

21:39:56 * AaronSw plans Semantic Web World Domination

21:40:07 <AaronSw> 1. Remove bNodes from TODL

21:40:10 <AaronSw> 2. Combine with REST

21:40:18 <AaronSw> 3. Hype

21:40:32 <AaronSw> 3a. Write O'Reilly Book

21:40:36 <AaronSw> 3b. Go on Book Tour

21:41:20 <sbp> it's not my name for it at all! sandro came up with both

21:41:30 <AaronSw> 3c. Do sweeps-week TV show: "Who wants to be a machine-to-machine interface?"

21:41:34 <Seth> i dont see not context in todl .. did i miss it?

21:41:36 * bijan decides not to put his retirement plan funds in SemanticAaron stock.

21:41:42 <Seth> give me a s-triple

21:42:06 <AaronSw> pick any n-triple

21:42:17 <Seth> seth is great

21:42:51 <Seth> :seth :is :great

21:42:58 <bijan> that isn't either.

21:43:19 <Seth> <:seth> <:is> <:great>.

21:43:20 <sbp> no QNames, no keywords, no nothing

21:43:26 <sbp> nope, that isn't either

21:43:26 <bijan> Still no.

21:43:35 <Seth> i give up

21:44:28 <Seth> <http://foo/seth> <http://foo/is> <http:/foo/great>.

21:44:38 <AaronSw> actually, i think you need a space before the .

21:44:53 <bijan> Oh dear, I hope not! :)

21:44:55 <Seth> so now give me the striple of it

21:45:02 <sbp> that is the striple of it

21:45:02 <AaronSw> That is the striple

21:45:08 * sbp ^5's Aaron

21:45:18 <Seth> so there aint no context in it ?

21:45:25 <AaronSw> Oops, no space necessary.

21:45:47 <bijan> Indeed. Let's not elevant bad style advice into required syntax, even accidently.

21:45:56 <bijan> elevate

21:46:13 <Seth> could somebody answer my simple question?

21:46:20 * AaronSw hits the TYPO button again

21:46:23 <AaronSw> Bzzt!

21:46:33 <sbp> Seth, there are no formulae constructs in TODL

21:47:06 <Seth> so why would aaron want to build the next semantic web language based on it, is he daft?

21:47:15 <bijan> What, don't mine cause fluxuations in the stock market.

21:47:37 <AaronSw> No, yours decide game show winners.

21:48:00 * Seth goes back to coding the next semantic web language in python

22:17:02 * jordan is back

22:17:15 <sbp> wb

22:18:10 <jordan> thanks

22:36:01 <dmiles-wc>http://logicmoo.sourceforge.net

22:36:01 <dc_rdfig> H: http://logicmoo.sourceforge.net from dmiles-wc

22:36:26 <dmiles-wc> H|: LogicMoo Tryout

22:36:35 <dmiles-wc> H:| LogicMoo Tryout

22:36:36 <dc_rdfig> titled item H

22:37:11 <dmiles-wc> H: Tryout basicly working.. been a while since chumped

22:37:12 <dc_rdfig> added comment H1

22:37:49 <dmiles-wc> H: Uses OpenCyc and Jamud to allow inference on Role-playing game

22:37:49 <dc_rdfig> added comment H2

22:37:53 <dmiles-wc> H:

22:37:54 <dc_rdfig>http://logicmoo.sourceforge.net

22:37:54 <dc_rdfig> LogicMoo Tryout

22:37:55 <dc_rdfig> (1:dmiles-wc) Tryout basicly working.. been a while since chumped

22:37:56 <dc_rdfig> (2:dmiles-wc) Uses OpenCyc and Jamud to allow inference on Role-playing game

22:38:30 <Seth> you should chump the telnet uri

22:38:49 <dmiles-wc> good idea

22:39:30 <dmiles-wc>telnet://12.255.207.235:8808

22:39:46 <dmiles-wc> :P well i'll add it on

22:39:47 <bijan> Chump don't recognize telnet uris

22:40:16 <dmiles-wc> H: Telnet interface[telnet://12.255.207.235:8808]

22:40:16 <dc_rdfig> added comment H3

22:40:42 <dmiles-wc> H: Web interface[http://12.255.207.235/logicmoo]

22:40:42 <dc_rdfig> added comment H4

22:41:07 <dmiles-wc> kinda right ?

22:41:23 <dmiles-wc> dmiles-wc is now known as dmiles

22:41:39 <dmiles> wc was a polite way to say water closet

22:42:10 <danbri__> danbri@fireball:~ > telnet 12.255.207.235 8808

22:42:11 <danbri__> Trying 12.255.207.235...

22:42:14 <danbri__> [....]

22:42:18 <danbri__> hangs...

22:42:18 <dmiles> ?!?

22:42:24 <dmiles> hrrm let me check

22:42:40 <danbri__> --- 12.255.207.235 ping statistics ---

22:42:40 <danbri__> 4 packets transmitted, 0 packets received, 100% packet loss

22:42:50 <danbri__> ip address right?

22:42:58 <dmiles> yeah but i dont reply to pings

22:43:06 <dmiles> try to http there

22:43:31 <danbri__> Could not connect to the server

22:43:37 <dmiles> herrm i only close ports up to 7000

22:43:39 <danbri__> could be my net providor

22:43:41 <dmiles> i better check

22:43:48 <AaronSw> Shoot, danbri stole my TimBL reply!

22:44:22 <Seth> it says im not a human person ... now how'd it know that?

22:44:29 * danbri__ replaying a rather old argument, aired many times...

22:44:47 <AaronSw> oh wait, no you didn't

22:44:49 * danbri__ tries url from mit network too; still no joy

22:44:49 <AaronSw> you left out time

22:45:19 <danbri__> And language and content negotiation, HTTP-extensions, HTTP POST and all sorts of other issues

22:45:40 <dmiles> danbri__: who is youtr isp?

22:45:57 <danbri__> HTTP doesn't distinguish between service (value added by the HTTP protocol) and things more intrinsic to the resource...

22:46:10 * danbri__ being cryptic; will explain better another time

22:46:41 <danbri__> BTW I got positive feedback from posting to www-rdf-interest, I should remember to send stuff there more often... hundreds of people on that list...

22:47:22 <dmiles> hrrm i bet i am hozing my own client ports

22:47:49 <dmiles> by the way i set up the firewall

22:49:45 <sbp> Hmm... I was going for a different approach to Tim's D/C/H/V thing, but I guess I'll can it now

22:49:58 <dmiles> danbri: could you confirm again?

22:50:13 <dmiles> i changed part of the ffirewall

22:50:35 <danbri__> nioe

22:50:43 <danbri__> nope, i meant

22:50:50 <danbri__> hangs still

22:51:55 <sbp> And what of Tim's latter point? i.e. if w3.org denotes the homepage of the W3C, then <http://www.w3.org/> is the homepage, and [ :homepage <http://www.w3.org/> ] is the organization, and if w3.org denotes the W3C, then <http://www.w3.org/> is the W3C... but what denotes the homepage?

22:52:20 <AaronSw> um, [is :homepage of <http://www.w3.org/>

22:52:34 <AaronSw> probably http://www.w3.org/index or /index.html

22:52:37 <sbp> that seems a little silly

22:52:46 <sbp> "/Overview"

22:52:52 <AaronSw> That'd make more sense

22:53:01 <AaronSw> www.w3.org is sort of a bad example

22:53:08 <sbp> why?

22:54:01 <dmiles> hrrm danbri, i will more into it

22:54:17 <sbp> more to the point, what if somedomain.com/w3c was the homepagey-thing but denoted the organization, and there were no alternative HTTP URI for the homepage?

22:57:20 <sbp> going once...

22:57:49 <tomch> it's an awkward question, I suppose :-)

22:58:29 <sbp> right, and each side is quite convinced that they're right, but I feel that Miles Sabin is closer to an answer than anyone else

22:58:37 <sbp> (at the moment)

22:59:08 <sbp> because HTTP was designed for hyperlinks, and they have no formal semantics

22:59:27 <danbri__>http://rdfweb.org/2002/02/foafpath/simple-2002-04-02.jpg

22:59:28 <dc_rdfig> I: http://rdfweb.org/2002/02/foafpath/simple-2002-04-02.jpg from danbri__

22:59:41 <danbri__> I:|RDFWeb co-depiction graph

22:59:41 <dc_rdfig> titled item I

22:59:43 * AaronSw looks around for EricM

23:00:00 <AaronSw> sbp, that's what Content-Location is for

23:00:08 <AaronSw> it tells you the URI of the homepage, from a Resource URI

23:00:26 <bijan> I:Hm. May I recommend a find graph drawing algo?

23:00:26 <dc_rdfig> added comment I1

23:00:27 <danbri__> I:Edges are photos; this isn't _really_ RDF, though derrived from harvested RDF. From libby and damian's tools.

23:00:27 <dc_rdfig> added comment I2

23:00:36 <danbri__> I:Sure...

23:00:36 <dc_rdfig> added comment I3

23:01:03 <sbp> heh, "me@sranksinatra.com"

23:01:13 <bijan> I:At this point, anything other than what you've got :) At least eliminate overlapping nodes.

23:01:13 <dc_rdfig> added comment I4

23:01:18 <sbp> did I tell you that I managed to codepict myself with HJS?

23:01:46 <AaronSw> sbp, it's spelled right

23:01:49 <bijan> Actually, I can't see why those overlapped.

23:01:52 <AaronSw> the real HJS?

23:02:01 <bijan> It doesn't seem to be a general property of the picture.

23:02:06 <danbri__> hjs?

23:02:14 <AaronSw> I think danbri forgot to hit the "prettify graph" button

23:02:19 <AaronSw> Homer J. Simpson, I assume

23:02:35 <bijan> Or hit the uglify button :)

23:02:44 <danbri__> I:It's from RDFAuthor, damian has a button that jiggles the graph a bit, and generally prettifies it. I didn't press it many times.

23:02:44 <dc_rdfig> added comment I5

23:02:45 <bijan> I'm not sure I get the purple squares.

23:03:01 <danbri__> just edgelabels switched off

23:03:15 <bijan> Ah. Can one delete them entirely?

23:03:25 <danbri__> point isn't that its pretty, just that the network is getting bigger :)

23:03:31 <bijan> Hehe.

23:05:27 <bijan> Hmm. Any hyperbolic tree rdf graph browsers out there?

23:05:29 <AaronSw> ooh, the rael data is in there

23:05:46 <sbp> Aaron: try reading the definition of Content-Location

23:05:52 <AaronSw> Pfft.

23:06:00 <sbp> [[[

23:06:00 <sbp> The Content-Location entity-header field MAY be used to supply the

23:06:00 <sbp> resource location for the entity enclosed in the message when that

23:06:00 <sbp> entity is accessible from a location separate from the requested

23:06:00 <sbp> resource's URI.

23:06:01 <sbp> ]]]

23:06:04 <sbp> - RFC 2616

23:06:15 <AaronSw> Isn't that what I said?

23:06:28 <sbp> no, you wanted to point to a different resource with it

23:06:46 <AaronSw> "supply the *resource* location" [emph added]

23:07:17 <sbp> if I get a bit of crud back from dereferencing the W3C as a consortium, you can't very well going pointing to soemthing that is the homepage

23:07:30 <AaronSw> Why not?

23:07:47 <AaronSw> (as long as the crud is the homepage)

23:08:02 <sbp> because surely it means the resource location for the entity that you just got back from the representation?

23:08:12 <AaronSw> Right.

23:08:15 <sbp> s/representation/resource/

23:08:35 <sbp> so if that resource is the W3C-as-consortium...

23:09:16 <AaronSw> w3c-as-consortium :representation w3c-homepage ; :content-location w3c-homepage .

23:09:48 <sbp> ah, I guess I'm reading the words "resource location for the entity" differently

23:10:21 <sbp> i.e. wrongly. *ahem*

23:11:10 <sbp> Hmm... that means that the domain of your content-location property must be u:Fixed

23:11:27 <sbp> else it changes with the seasons

23:11:47 <sbp> oh, it's a property of the representation, not the resource as you had it...

23:12:47 <sbp> in that case, the range must be u:Fixed

23:13:19 <sbp> so w3.org as a consortium can't return Content-Type: /Overview

23:13:50 <AaronSw> it's not u-fixed

23:13:54 <AaronSw> Content-Type?!

23:14:02 <sbp> oops, Content-Location

23:14:11 <sbp> heh, Content-Type on my mind, on my mind...

23:14:11 <AaronSw> Why does it have to be fixed?

23:14:53 <sbp> if it's the resource location for the entity

23:15:13 <AaronSw> Hm. I think that's poorly worded.

23:16:13 <sbp> yep; it's all to easy for me to come along and say that the range is u:Fixed, and I agree that that's probably not as they intended

23:16:26 <AaronSw> ah: "it is only a statement of the location of the resource corresponding to this particular entity *at the time of the request*."

23:16:36 <sbp> aha. fair enough