Semantic Web Interest Group IRC Chat Logs for 2003-08-31

00:00:12 <dmiles>http://cvs.sourceforge.net/cgi-bin/viewcvs.cgi/openmodality/planners/plan_tailspin.pl?rev=1.1&content-type=text/vnd.viewcvs-markup

00:00:12 <dc_rdfig> A: http://cvs.sourceforge.net/cgi-bin/viewcvs.cgi/openmodality/planners/plan_tailspin.pl?rev=1.1&content-type=text/vnd.viewcvs-markup from dmiles

00:00:17 <dmiles> ack ;P

00:00:25 <dmiles> well hey its a perminant url

00:01:13 <dmiles> A: Event planner that uses inference

00:01:14 <dc_rdfig> Added comment A1.

00:01:21 <dmiles> A:! Event planner that uses inference

00:01:21 <dc_rdfig> Added comment A2.

00:01:27 <dmiles> A!: Event planner that uses inference

00:01:42 <dmiles> a1:

00:01:42 <danbri_dna> cool, thanks.

00:01:48 <dmiles> A2:

00:01:48 <dc_rdfig> (dmiles) ! Event planner that uses inference

00:02:08 * danbri_dna stashes the url and heads off to bed

00:02:16 <danbri_dna> cheers dmiles! g'nite...

00:02:18 <dmiles> cool.. see you soon

00:02:24 <danbri_dna> &u

00:03:06 <dmiles> A1: this planner is writting in prolog that uses it planning abilities to prove causes that normally are tested in infernce engines

00:03:06 <dc_rdfig> Replaced comment A1.

00:03:27 <dmiles> A1: this planner is written in prolog that uses it planning abilities to prove causes that normally are tested from infernce engines

00:03:28 <dc_rdfig> Replaced comment A1.

00:03:43 <mdupont> dmiles: !

00:03:57 <dmiles> A2: ""

00:03:59 <dc_rdfig> Deleted comment A2.

00:04:04 <dmiles> hi

00:04:27 <mdupont> hows things going?

00:04:37 <dmiles> A:# Event planner that DOES common sense inference

00:04:37 <dc_rdfig> Added comment A2.

00:04:42 <dmiles> A2: ""

00:04:42 <dc_rdfig> Deleted comment A2.

00:04:47 <dmiles> A#: Event planner that DOES common sense inference

00:04:57 <mdupont> I am looking for exchange partners. what about a rdf interface to this planner?

00:05:19 <dmiles> RDF source matterial with soap query?

00:05:40 <mdupont> ok

00:05:45 <mdupont> that sounds good

00:05:45 <dmiles> the initual database for could actually use RDF for the data

00:05:53 <mdupont> with a perl soap::lite interface

00:06:05 <mdupont> initual?

00:06:37 <dmiles> the data it begins with

00:07:00 <dmiles> like right now the data is in big prolog files

00:08:04 <dmiles> once RDF tags added it may go from 65 megs to 150 megs

00:08:16 <dmiles> but still thats unavoiable

00:08:32 <dmiles> the used memory of course would not be any begger

00:08:39 <dmiles> any bigger

00:09:23 <dmiles> so.. now asking it to write ytou out a plan/proof could be done from a soap request

00:09:41 <dmiles> right now i live in a hospital..

00:09:52 <dmiles> so its only 3 days a mopnt

00:09:59 <dmiles> that i have internet access

00:10:10 <dmiles> but they are getting a perminant connection set up

00:10:23 <dmiles> it may only be a http proxy :(\

00:10:34 <dmiles> but i am sure i could get some irc interface

00:10:46 <dmiles> that cheats away through the proxy

00:11:12 <dmiles> 3 does a month i can leave the hospitol i meant

00:11:53 <dmiles> still i am interested in completing this

00:12:33 <dmiles> i had restarted the C# version of the prolog/inference engine

00:12:49 <dmiles> you had suggested just ussing the java version

00:13:25 <dmiles> what i am doing is turning the objects into "union" feilds

00:13:45 <dmiles> so that all the data exists in this datatype

00:14:00 <mdupont> 150 mbs?

00:14:10 <dmiles> and the object version of the union has the union loaded into it

00:14:19 <mdupont> hospital! OMG!

00:14:31 <dmiles> so eventually the goal is to remove the objects and only use unions

00:15:27 <dmiles> but starting out and maintaining things like "JohnFrost" into a StringObject with string methods

00:16:08 <mdupont> ko

00:16:19 <dmiles> but really internally its a union with the string field of the union initalized to "JohnFrost"

00:16:47 <dmiles> basiclly the union has an byte typeof

00:17:18 <dmiles> so compares compatre they are o1.typeunum=o2.typeenum

00:18:08 <dmiles> then the based on that (like STRINGTYPE) then the hashcodes are copared

00:18:39 <dmiles> or the hashcodes can be comapred first actually

00:18:52 <dmiles> then the type number

00:19:36 <dmiles> anyways this version of prolog/infernce engine is not nessisary

00:20:11 <dmiles> but its a good break sometimes.. and there is no language that is really optimized 100% towards infernce

00:20:28 <dmiles> not even logical axioms!

00:21:10 <dmiles> i belive the prolog engine is the easiest language to build this upon

00:21:38 <dmiles> the "inference code"

00:22:05 <dmiles> i was hired as an infernce programmer once and my imployer thought all i had to do was convert KIF to prolog

00:22:19 <dmiles> then run the prolog and he would have an engine

00:22:22 <mdupont> ok

00:22:32 <mdupont> i have been converting a lisp engine into rdf

00:22:34 <dmiles> but it was hard to explain its not that easy

00:22:58 <dmiles> if prolog had X/Y/Z improvments.. then it would be a dirrect transation

00:23:10 <dmiles> dirrect translation

00:23:21 <mdupont> i have the implementation of the c version of slate using the ecl

00:23:46 <dmiles> i wrote a lisp engine in prolog (ig axioms) http://cvs.sourceforge.net/cgi-bin/viewcvs.cgi/openmodality/lisp/

00:23:56 <mdupont> ok

00:24:07 <mdupont> i think i might have that

00:24:53 <dmiles> well its bout 1/2 done

00:25:04 <dmiles> on my computer at the hostitol its 90% done

00:25:33 <dmiles> this file: http://cvs.sourceforge.net/cgi-bin/viewcvs.cgi/openmodality/lisp/lisp_hashtables.pl?rev=1.1&content-type=text/vnd.viewcvs-markup for example is comp[lete

00:25:55 <dmiles> probly one of the easiest to see how it works

00:26:59 <dmiles> evalFunction/3 and evalSpecial/3 are differnt in respect to whether to eval arguments first

00:27:42 <dmiles> evalAccessor/3 uses lisp memory refernces

00:27:52 <dmiles> its a cell object

00:28:38 <dmiles> things like this is what inspires to rewrite the normal prolog

00:28:58 <dmiles> so its not limited the atoms/numbers/structions

00:29:58 <dmiles> euler is optimized towards infernce but looks internally like aprolog interpretor

00:30:14 <dmiles> so this is the same thing

00:31:00 <dmiles> but you cant write entire programs in Euler

00:31:14 <dmiles> you have to write applications arround it

00:31:53 <dmiles> so the prolog/inference is much like eular but maintaining the entire prolog language so full apps can be written in it

00:32:47 <dmiles> so.. you see i am devided between writing the code and writting interpretor

00:33:33 <dmiles> actually all infernce engine projects are devided like this

00:33:52 <dmiles> Knowelege Engineer/Engine Programmer

00:35:21 <dmiles> i just keep flipping between work on both

00:36:02 <dmiles> its like i could see you (mdupont) turning c programs into RDF

00:36:09 <dmiles> gcc output

00:36:21 <mdupont> you want to see some examplecode?

00:36:41 <dmiles> then having to head towards the soap feeder of it

00:36:48 <dmiles> like perl::soap

00:36:54 <mdupont> soap feeder of the gcc code?

00:36:57 <mdupont> wow

00:36:58 <dmiles> yes

00:37:07 <mdupont> what an idea!

00:37:17 <mdupont> i have have worked on rdf feeeder webserivces

00:37:21 <mdupont> using xmlrpc

00:37:23 <mdupont> and rdf

00:37:27 <dmiles> nod

00:37:28 <mdupont> i got that all worked outr

00:37:35 <mdupont> but the c# is limited

00:38:25 <dmiles> c# doesnt have soapish classes?

00:38:35 <mdupont> it does

00:38:39 <dmiles> oh xmlrpc

00:38:42 <mdupont> but dotgnu has xmlrpc

00:38:45 <dmiles> i see

00:38:48 <mdupont> i can use perl soap lite

00:38:55 <dmiles> to talk to it?

00:38:58 <mdupont> and i am also looking into gsoap

00:39:38 <dmiles> in a few more weeks i am getting released from the hospitol

00:39:43 <mdupont> good luck

00:40:02 <dmiles> once the SSDI (govt disablity) kicks in

00:40:50 <dmiles> so its a place that supports me until then

00:41:03 <dmiles> but yes we have to work on this

00:43:06 <dmiles> what made you gret interested in lisp?

00:43:50 <mdupont> tunes

00:44:13 <mdupont> i have a dump in rdf of the slate engine written in lisp and converted to c.

00:44:23 <mdupont> it is a set of numbered functions that do things

00:44:29 <mdupont> and a set of identifiers.

00:44:44 <mdupont> this is introspected into rdf

00:44:50 <mdupont> so now i will publish that data

00:44:51 <dmiles> with me its simular existing usefull code is often writting in lisp enough to want to be able to use it inline to other things we do

00:45:23 <dmiles> lisp code itself is like knowledge

00:45:37 <dmiles> (same with prolog)

00:46:08 <dmiles> so its like having a system that gains knowledge ig code

00:47:57 <mdupont> i want to have a good example of lisp that does a lot expressed in the introspector

00:48:09 <mdupont> it will be interesting to write an interpreter of that code

00:49:02 <mdupont>http://introspector.sourceforge.net/2003/08/slate/Global-userdefined.n3

00:49:02 <dc_rdfig> B: http://introspector.sourceforge.net/2003/08/slate/Global-userdefined.n3 from mdupont

00:49:40 <mdupont> B:An introspector report extracted from the gcc and processed by cwm into an condenced form.

00:49:40 <dc_rdfig> Added comment B1.

00:50:19 <mdupont> B:It its the source code to a lisp engine that has a program loaded to process a smalltalk language.

00:50:20 <dc_rdfig> Added comment B2.

00:53:27 <nephrael> nephrael is now known as dmiles

00:53:49 <dmiles> computer didnt like the size of thaT file :)

00:54:03 <dmiles> my own computer wouldnt have minded

00:55:15 <mdupont> bitsko: [http://introspector.sourceforge.net/2003/08/slate/_L29.n3|L29] is a large 222k function that is written in lisp. this rdf representation should have some interest to lisp and semantic web interested people.

00:55:20 <mdupont> B:[http://introspector.sourceforge.net/2003/08/slate/_L29.n3|L29] is a large 222k function that is written in lisp. this rdf representation should have some interest to lisp and semantic web interested people.

00:55:20 <dc_rdfig> Added comment B3.

00:56:34 <dmiles> how do you title a chump?

00:57:31 <mdupont> B:[http://introspector.sourceforge.net/2003/08/slate/project.n3|project ] is an instance new owl project.owl vocabulary. Extracted from the .i file of the compiler, each include file is parsed and a tree is built. all stored in rdf. this is great for getting a handle on your source code.

00:57:31 <dc_rdfig> Added comment B4.

00:58:26 <dmiles> A:# Planner the doubles as an Inference Engine

00:58:26 <dc_rdfig> Added comment A2.

00:58:27 <mdupont> B:|SLATE ELC (Embedded common lisp) gcc introspector

00:58:28 <dc_rdfig> Titled item B.

00:58:35 <dmiles> A:| Planner the doubles as an Inference Engine

00:58:35 <dc_rdfig> Titled item A.

00:58:50 <dmiles> A:

00:58:50 <dc_rdfig>http://cvs.sourceforge.net/cgi-bin/viewcvs.cgi/openmodality/planners/plan_tailspin.pl?rev=1.1&content-type=text/vnd.viewcvs-markup

00:58:51 <dc_rdfig> Planner the doubles as an Inference Engine

00:58:52 <dc_rdfig> (1:dmiles) this planner is written in prolog that uses it planning abilities to prove causes that normally are tested from infernce engines

00:58:53 <dc_rdfig> (2:dmiles) # Planner the doubles as an Inference Engine

00:59:15 <dmiles> d2: ""

00:59:20 <dmiles> A2: ""

00:59:20 <dc_rdfig> Deleted comment A2.

00:59:25 <dmiles> thank you

00:59:35 <mdupont> /msg dc_rdfig help?

00:59:46 <dmiles> ok

00:59:49 <mdupont> that makes a quiet connection

01:00:43 <dmiles> i see

01:01:40 <dmiles> i feel i need to finish the inference engine to at least work on the KIF test problems i have

01:01:54 <mdupont> ok

01:01:58 <dmiles> make it again 100% sound/complete

01:02:01 <mdupont> so a gcc to kif converter

01:02:13 <dmiles> well all the infernce tests i have are in kif

01:02:26 <mdupont> that intereprets ELC data structures using cwm or kif

01:02:35 <mdupont> kif => n3

01:03:04 <dmiles> that woould be usefull

01:04:42 <mdupont> do you have Free kif tools?

01:07:07 <dmiles> hrrm

01:07:27 <dmiles> well there is somethin called KE-Text

01:07:42 <dmiles> it makes your kif source shorter

01:08:20 <dmiles> OpenCyc uses it to load CycL

01:08:40 <dmiles> CycL is just well managed form of KIF

01:09:57 <mdupont> ok

01:10:13 <mdupont> cycl, i could compile that with java

01:10:13 <dmiles> im looking for some for you

01:10:24 * mdupont needs to sleep

01:10:34 <mdupont> good luck dmiles!

01:10:38 <dmiles> i will be arround in 8hrs i think

01:10:42 <dmiles> talk toyou soon

01:10:43 <mdupont> ok

01:10:46 <mdupont> bye!

01:10:59 <mdupont> mdupont is now known as mdupont-ZZzz

01:12:56 <WillWare> anybody awake here?

01:34:46 <sandro> nope

04:44:15 * kham can't sleep here

04:47:33 <dmiles> who wants to go into bussiness text messaging spam?

04:48:09 <dmiles> perhaps a "talk dirty to me" texting bot

04:48:53 <kham> hmm.. a guy taslks dirty and the bot reply as a "hooker"

04:48:59 <kham> has possibilities

04:50:29 <kham> xcept I have very few programming skill

04:53:41 <dmiles> i could do it.. just need someone to help with capitol starting it

04:54:10 <dmiles> oh and i have some reasonable sex bot

04:54:27 * kham would like to but as a student the banks want what I owe them first :(

04:56:29 <jql> you just need the initial capital to begin your nefarious plot?

04:57:33 <dmiles> :)

04:58:15 <dmiles> well between a nympho eliza and selling mass texting it could be lucrative

05:08:07 <jql> does it involve windows messenger service spam as well? I love those

05:09:07 <dmiles> not sure if there is too much compitition .. could be fun

05:09:29 <dmiles> i dont no yeet if spammers figured it out (about texting)

05:09:51 <dmiles> but i have been ofline a while

05:32:45 <sbp> ...

05:33:01 <sbp> I have this thing where I think about my future employers and such before I start talking on a logged channel :-)

05:34:39 <sbp> man... sex text-spambot ideas on #rdfig...

05:38:08 * sbp hacks on a new RDF graph isomorphism utility, but can't make it pass the ntc test cases... hmph

05:39:21 <sbp> it's just the recursive ones, but still. I'd hoped that this approach was going to be better than my previous one

05:43:59 <WillWare> I have a dull-witted question about using directed graphs to do logic. I've been tinkering with some forward-chaining in Python (not sudying cwm as I should, just bumbling on my own).

05:45:09 <WillWare> I have been able to do AND clauses like this: (A implies (B implies C)) equivalent to ((A and B) implies C), so I can fit AND into the RDF-ish triple paradigm. OR is also easy, I just have multiple "implies" statements. But I can't figure out how to do NOT.

05:46:32 <WillWare> I was thinking maybe ((A implies B) implies C), equivalent to (((not A) or B) implies C), and make B always false.

05:46:49 <WillWare> Haven't actually tried that in my code yet.

05:47:15 <dmiles> sbp:)

05:47:40 <dmiles> ((A and B) implies C),

05:47:55 <dmiles> from (A implies (B implies C))

05:48:43 <dmiles> wait

05:49:04 <dmiles> well starting out.. A sets up intial typeof domains

05:49:40 <dmiles> WillWare: are you familar with Modus Tollens?

05:50:06 <WillWare> Nope, but I'll go Google it.

05:50:14 <dmiles> the contrapositive of the implication

05:50:16 <dmiles> like

05:50:40 <dmiles> (implies A B)

05:50:51 <dmiles> also maens not B means not A

05:50:56 <dmiles> also means

05:51:19 <dmiles> so for each implication there is reverse negation

05:51:27 <dmiles> so looking at...

05:51:31 <dmiles> (A implies (B implies C))

05:51:51 <dmiles> ( not (B implies C) ) implies (not A))

05:52:54 <WillWare> There's a problem with ((not B) implies (not A)). It assumes you've already invented the directed-graph inverter, because you've already got (not B). If you can't do a logical inversion in directed graph logic, you can't obtain (not B).

05:53:33 <dmiles> oh actully...

05:53:36 <WillWare> But I think ((A implies false) implies B) might do it.

05:53:54 <dmiles> i am writting a dirrected graph inferenece engine...

05:54:01 <dmiles> (in prolog)

05:54:12 <dmiles> what i had to do is make two nodes for each litteral

05:54:18 <dmiles> actaully two graphs

05:54:39 <dmiles> sorry 2 subgraphs of one graph

05:54:54 <dmiles> lets say A is (loves john mary)

05:55:12 <dmiles> and we wanted not A

05:55:19 <dmiles> we inventr a new litteral

05:55:26 <WillWare> I thought about doing two graphs, one representing known-true statements and one representing known-false statements. It seemed reasonable but I was too lazy to think through all the ways information could pass back and forth between them.

05:55:36 <dmiles> called "~love"

05:55:51 <dmiles> WillWare: wow you will succeed

05:56:21 <dmiles> yes there is as you know success/fail of both known-true and known-false

05:56:38 <dmiles> so there are 4 possible results of any proposition

05:56:58 <dmiles> but there are six tests

05:57:18 <dmiles> lets say first we have a bitmask

05:57:32 <dmiles> known-true | known-false

05:57:36 <dmiles> ---------------

05:57:42 <WillWare> Just for yucks, here's an example of one of my AND statements: ((X, sibling, Y), implies, ((X, species, Z), implies, (Y, species, Z)))

05:57:43 <dmiles> T X

05:57:46 <dmiles> T F

05:58:00 <dmiles> ok yeah

05:58:18 <dmiles> that creats 5 rules

05:58:30 <dmiles> polycanonicalization

05:58:55 <dmiles> not sibling for example fails the entire proposition

05:59:07 <dmiles> thats what cuts out 3 of the 8 to make 5

06:00:02 <WillWare> not-sibling just means that this rule doesn't fire. Another one might fire that derives same-species some other way.

06:00:28 <dmiles> yes

06:00:36 <dmiles> thats what got you interested in graph

06:00:50 <dmiles> you see the heuristics based on best path first

06:01:10 <dmiles> once best-path is none the expensive logic tests are good

06:01:18 <dmiles> best-path done

06:01:51 <dmiles> you are completely on the right track

06:02:30 <dmiles> have you heard of clausal norml form?

06:02:31 <WillWare> I'm not sure where the 3/5/8 come from, and I don't know what "polycanonicalization" is, but I can see that I want to prune possibilities as early as possible.

06:02:39 <dmiles> yup

06:02:43 <WillWare> Nope, don't know "clausal normal form" either.

06:03:03 <dmiles> this is a form that uses entailment instead of implication

06:03:24 <dmiles> you convert your logical axioms to entailments

06:03:35 <dmiles> that use variables that are existentaul\

06:03:47 <dmiles> these entailments are graph nodes

06:04:08 <dmiles> the plycanonicalization is the conversion

06:04:41 <dmiles> however you have to chose what a node realy needs

06:05:06 <dmiles> i found i need litteral connections based on fully bound and unbound axioms

06:05:24 <dmiles> meaning ~A unbound~A

06:05:30 <dmiles> A and unbound A

06:05:47 <dmiles> which is forAll x A and exists x A

06:06:15 <dmiles> meaning (forall x (not Ax))

06:06:21 <dmiles> meaning (exists x (not Ax))

06:06:28 <dmiles> meaning (forall x (Ax))

06:06:36 <dmiles> meaning (exists x (Ax))

06:07:08 <dmiles> are nodes

06:07:12 <dmiles> rukles are edges

06:07:23 <dmiles> err i meant vertexes not nodes

06:07:41 <dmiles> but knowing what edges first you get the rules you want

06:08:09 <dmiles> then all those rules have litterals

06:08:22 <dmiles> those will show you the subgraph

06:08:47 <dmiles> then you can run those optimizers like transitive closures

06:08:55 <dmiles> (the warshall)

06:10:32 <dmiles> tell me if i lost you

06:11:22 <WillWare> Yeah, I'm afraid you did.

06:12:33 <dmiles> well the four versions of A .. you see what i mean?

06:13:02 <dmiles> those are known-false

06:13:16 <dmiles> those are not-know-false

06:14:04 <dmiles> well actually do you accept know-false, know-true , known-true-false not-know-true-flase

06:15:06 <WillWare> I know what "known true" and "known false" mean. I think I know what "not known true false" means, it's something about which I acknowledge my ignorance. I'm not sure what you mean by "known true false".

06:15:34 <dmiles> known true-false is when you can in one context prove A true.. and another context false.. but now you need to compbine both contexts toi gain access to another rule

06:16:15 <dmiles> you are dealing with a provable litteral that basiclly is inconsistent

06:16:47 <dmiles> before allow the fact into the KB though you should have queried the negation first

06:17:01 <dmiles> and if it was found nopt allow the new fact

06:17:20 <dmiles> but we cant always control how we get facts

06:17:31 <dmiles> so there are acceptable cases

06:18:04 <dmiles> well lets say you cant prove "known true"

06:18:09 <dmiles> what woul;d that mean?

06:19:43 <WillWare> If I can't prove A is true, that probably doesn't mean anything at all. Later I might gain more information that allowed me to prove it true or prove it false. So I'm gonna guess "no significance whatsoever".

06:20:18 <dmiles> right

06:20:28 <dmiles> same with "known false"

06:21:07 <dmiles> so for each proposition or fact there are 4 possible results to a query right?

06:21:28 <WillWare> I can see three: true, false, don't know.

06:22:13 <dmiles> ok that works .. i cant really prove the 4th to be usefull

06:22:32 <dmiles> 4 being inconsistent_kb

06:22:59 <dmiles> but ok lets use those there

06:23:03 <WillWare> So the 4th is where you have two different chains of reasoning, one resulting in true and another in false?

06:23:04 <dmiles> err three

06:23:08 <dmiles> yes

06:23:56 <dmiles> ok so here is where i am going...

06:24:20 <dmiles> actually how long you going to be here?

06:25:06 <dmiles> i really want to talk to you about this .. 2 months ago i was wondering if anyone used graph for infernce engine

06:25:12 <WillWare> Not long. I'm on the East Coast and it's 2:30 in the morning. I think I'll just plow ahead with my known-true and known-false nets and see where I get.

06:25:26 <dmiles> wiat 2 nets?

06:25:39 <WillWare> sorry? "wiat"?

06:25:42 <dmiles> wait

06:25:45 <dmiles> 2 nets?

06:26:07 <WillWare> sorry, directed graphs

06:26:15 <dmiles> 2 dirrect graphs?

06:26:19 <dmiles> er directed

06:26:36 <WillWare> right, one for known-true propositions, the other for known-false propositions

06:27:25 <dmiles> i think you could and should combine them but create a known-false-sibling / known-true-subling

06:27:40 <dmiles> known-true-sibling

06:28:07 <dmiles> becasue often the comp[iling to the graph will mix them

06:28:27 <dmiles> i wish there was times to give you examples

06:28:54 <dmiles> i am not at my own computer today

06:29:07 <WillWare> Well the problem I was originally trying to solve was this: how in a directed graph do you represent a negation, like (A and not B) implies C. And the thing we haven't established is how to perform a negation in a directed graph, though I think I'm onto something with the ((A implies false) implies B) idea.

06:29:43 <dmiles> i dont see it

06:30:13 <dmiles> (A and not B) implies C creates 6 edges

06:30:55 <WillWare> (A implies B) is logically equivalent to (not A or B). For instance "raining implies I'll get wet" is true if it's not raining, or if I am getting wet. Now make B false all the time. Then (not A or B) reduces to (not A). Voila, inversion.

06:30:59 <dmiles> known-true A and known-false B => known-true C

06:31:11 <dmiles> is one

06:32:04 <dmiles> if known-flase C and known-true A then not known-true B

06:32:22 <dmiles> wait

06:32:29 <dmiles> if known-flase C and known-true A then not known-false B

06:32:57 <WillWare> I think we must be making different assumptions about this. I'm going by what I've seen in some (admittedly weak) tutorial material about RDF and the whole build-a-graph-with-triples idea.

06:33:16 <dmiles> ok

06:33:48 <dmiles> i am sure they mean a closed world though

06:34:17 <dmiles> and your advanced into open world axioms

06:34:38 <WillWare> "Open world axioms"? How do you mean?

06:35:05 <dmiles> meaning known-true known-false unkown .. is open world

06:35:23 <dmiles> closed world is: known-ture / unkown-true

06:35:38 <dmiles> closed world is: known-true / unknown-true

06:36:30 <WillWare> Oh, THAT's what close world is. I just assumed "closed world" meant that you're only dealing with a limited problem domain, like my toy problem with three siblings and a pet dog.

06:36:30 <dmiles> well if they had any success then its all good

06:37:05 <dmiles> closed world means that the world is complete

06:37:30 <dmiles> well you know everything

06:37:41 <dmiles> so if you dont know it .. it MUST be fasle

06:37:57 <WillWare> OK, so with "closed world" I never need to assume I'm ignorant about anything.

06:38:10 <dmiles> RDF is claiming closed world constantly

06:38:27 <dmiles> correct

06:39:38 <WillWare> Gee, if RDF is supposed to make its way out into the big scarey Internet, isn't the "closed world" assumption going to run into trouble? You're trying to merge knowledge from umpteen different websites, it'd seem you'd need to be ready to plead ignorance on some things.

06:39:56 <dmiles> my gripe exactly ;P

06:40:42 <dmiles> those is why negation needs to be explicit

06:41:36 <jql> I don't know that GWB is the president, therefore I know that he's not? Fun.

06:41:50 * jql hides in his corner of the US of A

06:43:10 <dmiles> after "compiling" RDF into entailments (or edges) you have to become aware of the 3 results

06:44:00 <WillWare> When you say compiling RDF into edges, I assume you mean parsing RDF/XML to produce an in-memory directed graph?

06:44:03 <dmiles> i'll look into this thing WillWare (the current theory your on)

06:44:12 <dmiles> yes

06:44:45 <dmiles> compiling RDF into edges, I mean parsing RDF/XML to produce an in-memory directed graph

06:45:43 <dmiles> A implies B defauklts to (forall x (Ax implies Bx))

06:46:01 <dmiles> all varible args are universally quantified right>?

06:46:42 <WillWare> OK, this is where I was running into trouble understanding your notation earlier. "Ax" is a statement involving a variable x?

06:46:56 <dmiles> your sibling vatrables are forAll X Y Z

06:47:02 <dmiles> correct

06:47:24 <dmiles> "Ax" is a statement involving a variable x

06:48:27 <dmiles> that means the contrapositive (negated form) is (not (exists Bx)) then (not (exists x (Bx)))

06:48:33 <dmiles> erm

06:48:39 <dmiles> (not (exists Bx)) then (not (exists x (Ax))

06:49:22 <dmiles> so not only is it dealing with know-false and known-true its dealing with exists/forall

06:50:02 <dmiles> i am tired as well so i am not very clear

06:50:19 <WillWare> The introduction of "forAll" and "x" seem like an unnecessary complication. What are they buying you?

06:51:00 <dmiles> kalish and monash theorms that speed up inference

06:51:12 <WillWare> OK, that's worthwhile.

06:51:30 <dmiles> and also 100% sound/complete infernce

06:53:14 <WillWare> You're sleepy, I'm sleepy, we should both crash. You want a copy of the Python script I'm hacking this stuff with? I can mail it to your sourceforge account, it's the only email address I can conveniently find for you.

06:54:15 <dmiles> mea_culpa@comcast.net will work to

06:54:23 <dmiles> actualy better

06:55:17 <dmiles> my polycanonicalizer http://cvs.sourceforge.net/cgi-bin/viewcvs.cgi/logicmoo/openmodaliy/iso-prolog/Attic/logicmoo_cmp_sentence.pl?rev=1.2&content-type=text/vnd.viewcvs-markup

06:56:09 <dmiles> stuff like..

06:56:10 <dmiles> Fml = not forall(X,F) -> Fml1 = exists(X,not(F));

06:56:10 <dmiles> Fml = not exists(X,F) -> Fml1 = forall(X,not(F));

06:56:32 <dmiles> Fml = (A <=> B) -> Fml1 = (A and B) or (not A and not(B));

06:57:32 <dmiles> i mixed nessisay/possiblity forall/exists and temporal logic in a single compiled form

06:58:45 <WillWare> OK, thanks, I'm going to hit the hay now. Got various work to do tomorrow and I don't want to be in a complete stupor. Adios.

06:58:58 <dmiles> goodnight

08:32:24 <GNUWookie> GNUWookie is now known as mdupont

08:32:36 * mdupont waves

08:36:53 <mdupont> A: [http://introspector.sourceforge.net/2003/08/slate/_init_OBJECT.n3|Init Object] this is the ini function that is used by the ECLS to initialize all the elements of the data structures needed for the other functions in the object.c file.

08:36:53 <dc_rdfig> Added comment A2.

08:36:57 <mdupont> A:

08:36:57 <dc_rdfig>http://cvs.sourceforge.net/cgi-bin/viewcvs.cgi/openmodality/planners/plan_tailspin.pl?rev=1.1&content-type=text/vnd.viewcvs-markup

08:36:58 <dc_rdfig> Planner the doubles as an Inference Engine

08:36:59 <dc_rdfig> (1:dmiles) this planner is written in prolog that uses it planning abilities to prove causes that normally are tested from infernce engines

08:37:00 <dc_rdfig> (2:mdupont) [http://introspector.sourceforge.net/2003/08/slate/_init_OBJECT.n3|Init Object] this is the ini function that is used by the ECLS to initialize all the elements of the data structures needed for the other functions in the object.c file.

08:37:08 <mdupont> A2:""

08:37:08 <dc_rdfig> Deleted comment A2.

08:37:12 <mdupont> bitsko: [http://introspector.sourceforge.net/2003/08/slate/_init_OBJECT.n3|Init Object] this is the ini function that is used by the ECLS to initialize all the elements of the data structures needed for the other functions in the object.c file.

08:37:17 <mdupont> B:[http://introspector.sourceforge.net/2003/08/slate/_init_OBJECT.n3|Init Object] this is the ini function that is used by the ECLS to initialize all the elements of the data structures needed for the other functions in the object.c file.

08:37:17 <dc_rdfig> Added comment B5.

08:38:33 <mdupont> B:[http://introspector.sourceforge.net/2003/08/slate/Global.n3|Global] an 1.4MB n3 file containing all the global data from the compilier while compiling the object.c file

08:38:33 <dc_rdfig> Added comment B6.

08:39:30 <mdupont> B:[http://introspector.sourceforge.net/2003/08/slate/project-cwm.n3|Project CWM] a project file that describes all the files and directories involved. Used the great CWM tool to clean it up.

08:39:30 <dc_rdfig> Added comment B7.

08:40:23 <mdupont> B:[http://introspector.sourceforge.net/2003/08/slate/project-select.n3|Project Selector] the beginnings of the data for select function to filter out all delcarations in only these interestings directories.

08:40:23 <dc_rdfig> Added comment B8.

11:43:11 <ChanServ> [#rdfig] This channel is logged and blogged: http://logicerror.com/rdfIRCWelcome

12:14:18 <mdupont> anyone use XUL?

12:19:42 <mdupont> is seems to me that alot of this XUL is not valid rdf

12:54:17 <mdupont> I am getting invalid RDF from mozillas own XUL when fed to rapper

12:54:18 <mdupont> rapper: Error - URI file:contentholder.rdf:109 - property element menupopup has multiple object node elements, skipping.

12:54:38 <mdupont> that is used to create multiple submenus in XUL/RDF

12:55:11 <mdupont> and here is another on, invalid uri : rapper: Error - URI file:help-index1.rdf:31 - Illegal rdf:ID value 'bookmarks:boomarking_search_results'

12:55:42 <mdupont> as in <rdf:Description ID="bookmarks:boomarking_search_results" ...

12:55:48 <mdupont> any ideas ?

12:57:43 <mdupont> it produces also some nice segfaults in rapper dajobe

12:59:12 <dajobe> I wouldn't expect XUL's rdf to be valid

12:59:26 <dajobe> I cannot confirm that crash with your example

13:00:18 <mdupont> ok, it is not suprising to me either

13:00:24 <mdupont> i will make a test case for you

13:00:39 <dajobe> check you are using raptor 0.9.12

13:00:49 <mdupont> ahh

13:01:02 <mdupont> i think i am using the new debs

13:02:59 <mdupont> hmm rapper does not have a version command..

13:03:18 <dajobe> just type 'rapper'

13:05:56 <mdupont> Usage: rapper [OPTIONS] <source file: URI> [base URI]

13:05:56 <mdupont> Parse the given file as RDF/XML or NTriples using Raptor

13:06:15 <dajobe> in that case, you are using an old one

13:07:50 <mdupont> this crashed my rapper http://unstable.elemental.com/mozilla/build/latest/mozilla/extensions/wallet/editor/WalletAddress.xul

13:08:06 <dajobe> ack, yes it does here

13:08:11 * mdupont upgrades raptor

13:08:26 <mdupont> do you want me to file a bug report?

13:08:31 <dajobe> no need

13:08:34 <mdupont> ok

13:08:50 * mdupont wants to get a grip on XUL

13:09:06 <mdupont> it would be nice to be able to somehow clean up this XUL

13:09:10 <mdupont> and make it proper rdf

13:09:15 <mdupont> so that we can reason about it.

13:11:14 <mdupont> is there a way to skip this error message : rapper: Error - URI file:XMLPrettyPrint.xml:40 - Document element rdf:RDF missing.

13:11:24 <dajobe> try --assume

13:11:43 <dajobe> Additional options:

13:11:43 <dajobe> -a, --assume Assume document is rdf/xml (rdf:RDF optional)

13:11:57 <mdupont> great

13:12:02 <mdupont> that is working fine

13:12:13 <mdupont> now one thing that is bothernig me is this message

13:12:16 <mdupont> rapper: Warning - URI file:XMLPrettyPrint.xml:42 - element binding has non-namespaced parts, skipping.

13:12:47 <mdupont> <bindings xmlns="http://www.mozilla.org/xbl" xmlns:html="http://www.w3.org/1999/xhtml"><binding id="prettyprint">

13:12:58 <mdupont> the default ns is set

13:13:07 <mdupont> how come it complains about binding then?

13:13:11 <dajobe> <binding id="prettyprint">

13:13:16 <mdupont> yes

13:13:23 <dajobe> id has no namespace

13:13:35 <mdupont> ok, and the default namespace does not apply?

13:13:40 <dajobe> that's right

13:13:46 <dajobe> this is XML Namespaces FAQ #1

13:14:05 <mdupont> ok, so the doc is invalid? I will get an xml cleaner

13:14:14 <mdupont> maybe html tidy ...

13:16:38 <dajobe> it's not html

13:16:54 <mdupont> tiday has limited support for xml sez the manpage

13:25:56 <mdupont> tidy does not do it

13:26:14 <mdupont> is there a way to get all the redland libs at once from debian apt-get?

13:26:40 <dajobe> "all"?

13:26:57 <dajobe> you mean both

13:27:12 <dajobe> get librdf0 and it'll get libraptor0

13:27:17 <dajobe> but rapper is not a library

13:28:09 <deltab> dajobe: I looked it up: it's FAQ #11.10 :-)

13:28:28 <dajobe> around here, #1

13:29:11 <mdupont> well, In any case, it is quite annoying... :(

13:34:25 <mdupont> ok

13:34:27 * danbri wonders if mdupont has thought about setting up a blog for the introspector stuff...

13:34:38 <mdupont> i am hacking xmlprinter now

13:34:53 <danbri> ...there's lots of detail on it in the rdfig.xmlhack.com chump archives, but it's kinda smeared across many pages with no index etc., hard to get the big picture

13:34:53 <mdupont> danbri: you mean instead of spamming the rdfig?

13:35:00 <danbri> not spamming! :)

13:35:05 <mdupont> ;)

13:35:13 <mdupont> i have a blog, but it sucks

13:35:16 <danbri> but there's a lot of detail and its hard to see the big picture...

13:35:31 <danbri> sucks: howso?

13:35:40 <mdupont> yes, check out the introspector.sf.net that links to the blog, the wiki and the mailing list

13:35:51 <danbri> hmm maybe better chump search might help...

13:35:54 <mdupont> The wiki has someone the big picture now

13:36:12 <mdupont> i need to pull this all together, It is such a *mess*

13:36:36 <mdupont> s/someone/somewhat/

13:36:36 <danbri> oh cool, hadn't seen http://gccintrospector.blogspot.com/

13:36:47 <mdupont> yeah, that is kinda out of dat

13:36:54 <mdupont> i need to write in it

13:37:03 <mdupont> but the chumpbot is nicer to use

13:37:40 <danbri> yup, chumpbot++

13:37:49 <mdupont> :)

13:38:05 <mdupont> i wonder if there is a way to define a ns for the whole document?

13:38:06 * danbri wonders if there's an irc interface to any of the weblog APIs...

13:38:11 <mdupont> without the element

13:43:22 * mdupont was working on an xmlrpc api into a rdf blogger/wiki thingie

13:43:39 <mdupont> but the dotgnu platform is *not* ready to use

13:44:01 <mdupont> at least for that type of application

13:50:03 <mdupont> so i have a cleaner now dajobe

13:51:09 <mdupont>http://introspector.sourceforge.net/2003/08/xmlclean.pl

13:51:09 <dc_rdfig> C: http://introspector.sourceforge.net/2003/08/xmlclean.pl from mdupont

13:51:34 <mdupont> C:|XML cleaner hack based on xmlpretty

13:51:34 <dc_rdfig> Titled item C.

13:52:05 <mdupont> C:This inserts bogus namespace attributes into your XUL xml so that rapper can rip it

13:52:06 <dc_rdfig> Added comment C1.

13:52:37 <mdupont> C:Needs the hacked [http://introspector.sourceforge.net/2003/08/CleanWriter.pm|CleanWriter] module

13:52:37 <dc_rdfig> Added comment C2.

13:53:21 <mdupont> danbri: you have a point there, I need to get this stuff into a better form

13:53:35 <mdupont> i like redlands webpage, it is so clean and easy to read

13:53:54 <mdupont> my pages tend to be very messy, like my self. you should see my computer room...

14:04:43 * danbri grins, glances around guiltily at piles of papers and cables...

14:21:26 <mdupont> hehehe

14:21:47 <mdupont> danbri: we have a problem with mozillas rdf

14:21:57 <mdupont> it is not making rapper very happy

14:22:11 <mdupont> i thought they are oh so standards compliant ... :)

14:22:53 <mdupont> property element binding has multiple object node elements, skipping.

14:23:10 <mdupont> basically all the of GUI elements of mozilla are using an implicit context object

14:23:19 <mdupont> or an implicit list of subobjects

14:23:32 <mdupont> this is easier to program but not stadard...

14:23:36 <mdupont> standard

14:23:39 <deltab> parseType="collection"?

14:25:12 <mdupont> is that possible in xml?

14:25:20 <mdupont> here is a good example of my problems

14:26:45 <mdupont> C:example output is [http://introspector.sourceforge.net/2003/08/button.rdf|here]

14:26:45 <dc_rdfig> Added comment C3.

14:27:10 <mdupont> C:Strange rapper error rapper: Error - URI file:button.rdf:115 - Literal property element field has property attributes

14:27:11 <dc_rdfig> Added comment C4.

14:27:30 <mdupont> C:Typical error :Error - URI file:button.xml.clean:143 - property element binding has multiple object node elements, skipping.

14:27:31 <dc_rdfig> Added comment C5.

14:27:39 <mdupont> C5:Typical error :Error - URI file:button.rdf:143 - property element binding has multiple object node elements, skipping.

14:27:39 <dc_rdfig> Replaced comment C5.

14:31:43 <Wack> Literal property element field has property attributes <- should use rdf:value there right?

15:02:44 <mdupont> Wack: yes?

15:02:56 <mdupont> but that line number is strange

15:14:40 <mdupont> Wack: here is an example

15:16:07 <mdupont> <RDF:RDF xmlns="http://fo#" xmlns:XUL="http://fo#" xmlns:RDF="http://www.w3.org/1999/02/22-rdf-syntax-ns#" xmlns:NC="http://home.netscape.com/NC-rdf#"><bindings><binding><handlers><handler XUL:event="command"><![CDATA[good]]></handler></handlers></binding></bindings></RDF:RDF>

15:16:24 <mdupont> with the error : rapper: Error - URI file:///home/mdupont/XUL/button-test.rdf:2 - Literal property element handler has property attributes

15:34:34 <dajobe> mdupont: that crash seems to be caused by an libxml2 bug, but I can work around it

16:08:07 <mdupont> dajobe: cool

16:27:49 <mdupont> dajobe: can you comment on this error :

16:27:52 <mdupont> <RDF:RDF xmlns="http://fo#" xmlns:XUL="http://fo#" xmlns:RDF="http://www.w3.org/1999/02/22-rdf-syntax-ns#" xmlns:NC="http://home.netscape.com/NC-rdf#"><bindings><binding><handlers><handler XUL:event="command"><![CDATA[good]]></handler></handlers></binding></bindings></RDF:RDF>

16:27:56 <mdupont> with the error : rapper: Error - URI file:///home/mdupont/XUL/button-test.rdf:2 - Literal property element handler has property attributes

16:28:01 <mdupont> any ideas what that means?

16:28:13 <dajobe> seems pretty clear to me

16:28:15 <mdupont> i now have rapper 9.13 :)

16:28:20 <dajobe> "Literal property element handler has property attributes"

16:28:26 <dajobe> a literal property element

16:28:27 <dajobe> handler

16:28:30 <dajobe> which has property attributes

16:28:34 <dajobe> that's forbidden.

16:28:36 <mdupont> what is a literal property?

16:28:44 <dajobe> a property with a literal

16:28:46 <mdupont> and why is it forbidden?

16:28:52 <dajobe> I can rearrange these words around

16:28:52 <mdupont> ok

16:29:00 <dajobe> literals cannot have properties

16:29:02 <mdupont> because this invalidates most of XUL

16:29:08 * dajobe shrugs

16:29:12 <dajobe> I hope they know what it means

16:29:16 <mdupont> ok, i get the picture

16:29:24 <mdupont> thanks dajobe,

16:29:28 <dajobe> having said that....

16:29:35 <mdupont> sorry to ask such silly questions

16:29:39 <dajobe> redland allows you to use literals as nodes

16:29:41 <dajobe> but anyway

16:29:58 <mdupont> i think it is strange...

16:30:10 <dajobe> the rdf/xml syntax has never allowed it

16:30:31 <mdupont> i am pretty dissappointed with this XUL stuff

16:30:48 <mdupont> they say it is rdf, but is just a jumble of xml

16:31:03 <deltab> who said it was rdf?

16:31:20 * mdupont looks

16:31:42 <dajobe> it doesn't look like rdf to me

16:31:45 <deltab> there are places in XUL where RDF can be used as iinput

16:31:51 <dajobe> yes, that's different

16:35:32 <mdupont> http://www.mozilla.org/xpfe/xulrdf.htm

16:35:36 <mdupont> yes your right

16:35:43 <mdupont> it is *not* rdf

16:35:46 <mdupont> DOH!

16:36:58 <mdupont> now i am dissapointed, and confused.

16:37:11 <mdupont> deltab: i feel kinda silly now

16:51:39 <mdupont> dajobe: i found another bug!

16:53:57 <mdupont> C:BUG rapper: Error - URI file:///home/mdupont/XUL/button.rdf:29 - property element property has multiple object node elements, skipping. this line number is wrong. It is not counting the line numbers of the CDATA Section.

16:53:58 <dc_rdfig> Added comment C6.

16:54:11 <mdupont> dajobe: there you go, C6 is the bug

16:54:19 <dajobe> that one I'm not worryig about

16:54:31 <mdupont> based on the button.rdf file.

16:54:41 <mdupont> ok, well I know what the problem is now :)

16:57:19 <dajobe> it's an xml parser bug, I don't count lines

16:57:47 * dajobe -> out

17:15:44 <mdupont> ok, bye dajobe

17:15:45 <mdupont> tahnks

17:19:06 <jql> howdy

17:21:57 <mdupont> hi

17:23:39 * jql awakens to an... interesting email on intro-devel

17:23:52 <mdupont> which one?

17:23:58 <jql> brian

17:24:06 <jql> your response

17:24:12 <mdupont> oh yeah, my good friend brian

17:24:23 <mdupont> he is so cuddly

17:24:28 <jql> indeed

17:24:42 <jql> makes me wanna just join in your little conversation

17:24:44 * jql hides

17:25:13 <mdupont> one of the first people who is going to be lined up against the wall, when the revolution comes

17:26:27 <mdupont> please feel free to get into the fight with him

17:26:47 <mdupont> his ignorance of the semantic web is amazing

17:27:06 <jql> naw, I have no ammunition against... that. might as well argue for west-bank settlements with a palestinian

17:29:56 * danbri encourages mdupont and jql to lay off the militaristic metaphors...

17:30:36 <jql> yeah, I was about to start a comparison with pat buchanan, which was gonna lead to hitler. it's all so clear to me now

17:30:54 * jql goes back to coding

17:31:01 <mdupont> danbri: did you see any of these statements from mr brian t. rice about the semantic web?

17:31:20 <danbri> no

17:31:53 <mdupont>http://cliki.tunes.org/TUNES%20vs%20the%20WWW

17:31:54 <dc_rdfig> D: http://cliki.tunes.org/TUNES%20vs%20the%20WWW from mdupont

17:32:12 <mdupont> D:|Tunes page dissing the semantic web and w3c big time

17:32:12 <dc_rdfig> Titled item D.

17:32:15 <danbri> but I see a lot of criticisms of SW, RDF, W3C, FOAF etc... ...but I don't use #rdfig to encourage folks to pick fights. better to just let it go and work harder

17:32:49 <danbri> or answer specific factual criticisms, errors and ignore the fud

17:32:58 <mdupont> ok

17:33:30 <mdupont> fine, I will stop the hate terms, I will take that back with the "against the wall" statement

17:36:07 <danbri> thanks

17:36:28 <mdupont> dje: here is another mention of the semantic web [http://lists.tunes.org/archives/tunes/2001-April/003115.html|MentionMail]

17:36:31 <mdupont> D:here is another mention of the semantic web [http://lists.tunes.org/archives/tunes/2001-April/003115.html|MentionMail]

17:36:31 <dc_rdfig> Added comment D1.

17:36:51 <mdupont> danbri: do you mind if I collect and chump the relevant documents under D:?

17:37:25 <danbri> sure

17:37:28 <danbri> go for it, i mean

17:37:40 * danbri is skimming and unimpressed by http://firstmonday.org/issues/issue7_7/bates/

17:39:33 <danbri> it's a fair point that "ontologists" (in the recent KR sense) undervalue what's been done in the library world. But to be fair, many library classification schemas are inconsistently strucutred becuase they have no formalised structure... Soergel, who s/he cites against the ontologists, is plenty aware of this I believe.

17:40:58 <mdupont> B:The idea of TUNES [http://cliki.tunes.org/Metatext|Metatext] seems to be what the semantic web is, but they say it is much much better.

17:40:58 <dc_rdfig> Added comment B9.

17:41:29 <jql> is the comparison really fair? library indexes are designed to have no duplicated, so their system is designed to have each book under its One True Index.

17:41:38 <jql> duplicates, that is

17:42:00 <mdupont> I think that we can safely say that there is not the problem of many ontologies

17:42:15 <mdupont> just look at the type of ontologies we are collecting for the introspector project

17:42:27 <mdupont> each one is very different

17:42:38 <mdupont> there is not one true representation

17:43:01 <mdupont> later we might be able to map some onto another or onto a synthesis of them all

17:43:18 <mdupont> but i dont see a problem with individuality

17:43:48 <jql> libraries don't index their books by Hugo Award Winning, but I think that's a perfectly reasonable index to give a book. Hence, the value of the semantic web

17:43:57 <jql> someone just needs to define a hugo award

17:43:58 <jql> :)

17:44:38 * danbri finds http://www.w3.org/People/Berners-Lee/FAQ.html#have

17:44:53 <danbri> handy reply, though I suspect timbl meant 'centralised database' not 'decentralised' in last sentence

17:46:12 <mdupont> danbri: the best thing is that I dont think brian has seen the advanced logic that is possible in CWM

17:46:48 <mdupont> as far as I can tell, we can do most of the things that tunes is promising with the semantic web + a large source of rdf files representing the programs themselves

17:48:34 * danbri closes the browser window; http://cliki.tunes.org/TUNES%20vs%20the%20WWW has links to a bunch of interesting stuff, but just gives impression of holding a grudge and scrabbling around for evidence to support it.

17:48:49 * danbri doesn't mind him doing that, isn't particularly interested in digging further

17:51:33 <mdupont> dje: more hate [http://tunes.org/~water/tunesDraft.html|Tunes] """Finally, you may wonder where do these technologies or similarly-aspiring ones like XML or Tim Berners-Lee's Semantic Webs fail (aside from the relatively verbose nature of the syntax). The answer lies in the limitations of the expressiveness of the format, and how the notion of environment can implicitly get in the way. The history of Lisp can testify to this where often the n

17:51:33 <mdupont> re of the level of flexibility in the environment greatly affected the usefulness of whatever language features were there. Also, the substructural logics field explicitly addresses the limitations imposed when relying on predicate logic and other very similar logics, as the Semantic Web idea does. """

17:52:08 <mdupont> D:more hate [http://tunes.org/~water/tunesDraft.html|Tunes] """Finally, you may wonder where do these technologies or similarly-aspiring ones like XML or Tim Berners-Lee's Semantic Webs fail (aside from the relatively verbose nature of the syntax). The answer lies in the limitations of the expressiveness of the format, and how the notion of environment can implicitly get in the way. """

17:52:08 <dc_rdfig> Added comment D2.

17:52:22 <mdupont> D:Cont:"""The history of Lisp can testify to this where often the nature of the level of flexibility in the environment greatly affected the usefulness of whatever language features were there. Also, the substructural logics field explicitly addresses the limitations imposed when relying on predicate logic and other very similar logics, as the Semantic Web idea does. """

17:52:22 <dc_rdfig> Added comment D3.

17:54:30 <mdupont> D:See also the tune definition of [http://cliki.tunes.org/Ontology|Ontology]

17:54:30 <dc_rdfig> Added comment D4.

17:56:44 <mdupont> dje: My current mailing with tunes [http://lists.tunes.org/archives/tunes/2003-August/003797.html|Here]

17:56:47 <mdupont> D:My current mailing with tunes [http://lists.tunes.org/archives/tunes/2003-August/003797.html|Here]

17:56:47 <dc_rdfig> Added comment D5.

17:58:07 <mdupont> dje: Quotes from Brian's response for review : """Arrow's use of the term "ontology" relates to the philosophy of Martin Heidegger, and is more poetic than technical. The Semantic Web's turn on it is related to "formal ontologies" whose research started in the 1980's on knowledge representation and is basically a really limited, fairly useless perversion of the idea."""

17:58:10 <mdupont> D:Quotes from Brian's response for review : """Arrow's use of the term "ontology" relates to the philosophy of Martin Heidegger, and is more poetic than technical. The Semantic Web's turn on it is related to "formal ontologies" whose research started in the 1980's on knowledge representation and is basically a really limited, fairly useless perversion of the idea."""

17:58:11 <dc_rdfig> Added comment D6.

17:58:49 <mdupont> D:Put an emphasis on <emp>fairly useless perversion of the idea.</emp>

17:58:49 <dc_rdfig> Added comment D7.

18:00:09 <mdupont> D:more quotes """The Semantic Web is one possible reification, within an inefficient (less-expressive in a few ways) XML format (or syntactic equivalent), and is only one standard, even if it is a meta-standard. Yes, it's useful, but it's not of the same scale or scope."""

18:00:09 <dc_rdfig> Added comment D8.

18:01:06 <mdupont> Here is a more concise and silly misrepresentation of the www [http://tunes.org/Interfaces/tunesvswww.html|TunesVsCGI]

18:01:11 <mdupont> D:Here is a more concise and silly misrepresentation of the www [http://tunes.org/Interfaces/tunesvswww.html|TunesVsCGI]

18:01:11 <dc_rdfig> Added comment D9.

18:01:43 <mdupont> D:Quote """CGI scripts do allow dynamic generation of documents, but are a crippled low-level way to do that, and do NOT allow browsing of meta-document. """

18:01:43 <dc_rdfig> Added comment D10.

18:02:23 <mdupont> D:More """The Web is but yet another kind of coarse-grained client/server crap ...."""

18:02:23 <dc_rdfig> Added comment D11.

18:03:28 <mdupont> D:more gems of enlightenment """Also, using URLs is like having to work with pointers, not objects, in an environment with an unreliable buggy moving GC. It just can't work, so that a reliable meta- browsing server on the WWW cannot include meta-browsable external references."""

18:03:28 <dc_rdfig> Added comment D12.

18:03:43 * mdupont washes his hands of tunes.

18:03:48 <mdupont> what garbage

18:04:17 <danbri> why care so much about this? the Web's full of websites claiming to have basis for some new wondertech will obsolete Internet/Web, Turing machines, data formats etc...

18:04:23 <mdupont> sorry for all this BS, people. I just wanted to get this stuff out in the open, so you have a chance to respond.

18:04:42 <mdupont> danbri: I guess I should just drop it. your right, who cares

18:05:20 <danbri> thanks. i don't feel any need to respond though... looks like his/their worldview is immune to seeing RDFish folks as collaborators

18:05:35 * mdupont gets back to some real work.

18:06:14 <mdupont> danbri: that is the problem, the tunes worldview excludes any form of collboration, AFAICT

18:06:57 <mdupont> but it does not matter really, if they ever do produce anything...

18:07:02 <mdupont> then it might.

18:11:08 <danbri> imho best polish your own demos/docs, so they can better see where you're coming from...

18:11:20 <mdupont> yes

18:11:38 <mdupont> I am slowly getting down to something interesting with all this

18:11:56 <mdupont> but i need some more months before you can really use my cwm code

18:12:15 <mdupont> btw: something completly different, and more interesting

18:13:27 <mdupont>http://psyco.sourceforge.net/

18:13:27 <dc_rdfig> E: http://psyco.sourceforge.net/ from mdupont

18:13:50 <mdupont> E:|Psyco a Python accelerator

18:13:50 <dc_rdfig> Titled item E.

18:14:14 <mdupont> E:This might be interesting for the CWM crew to squeeze out some more performance.

18:14:14 <dc_rdfig> Added comment E1.

18:14:42 <mdupont> E:QUOTE """In short: run your existing Python software much faster, with no change in your source.Think of Psyco as a kind of just-in-time (JIT) compiler, a little bit like Java's, that emit machine code on the fly instead of interpreting your Python program step by step. The result is that your unmodified Python programs run faster."""

18:14:42 <dc_rdfig> Added comment E2.

18:18:48 <danbri> E:Anyone tried it?

18:18:49 <dc_rdfig> Added comment E3.

18:20:13 <danbri> E:"...where Psyco shines is when running algorithmical code --- these are the first pieces of code that you would consider rewriting in C for performance. If you are in this situation, consider using Psyco instead! You might get 10x to 100x speed-ups"

18:20:13 <dc_rdfig> Added comment E4.

18:21:16 * mdupont is building it now

18:25:40 <mdupont> wow, a c program

18:31:08 <mdupont> gawd, it takes a long time to compile

18:33:38 * danbri tries http://search.cpan.org/author/ZOOLEIKA/RDF-Simple-0.03/

18:34:46 * mortenf fights coredumps in hacked c for redland/mysql...

18:37:23 * steez avoids coredumps with Jena

18:37:47 <mdupont> danbri: have you seen jqls Rdf::Laces?

18:38:19 <mortenf> nothing wrong with redland, it's me that's the problem :/

18:39:24 <mdupont> mortenf: i told you not to derefence void pointers... <g>

18:39:31 <mortenf> heh

18:45:24 <danbri> cat simpletest.pl

18:45:24 <danbri> #!/usr/bin/perl

18:45:24 <danbri> use RDF::Simple;

18:45:24 <danbri> use RDF::Simple::Parser;

18:45:25 <danbri> use LWP::Simple;

18:45:29 <danbri> my $uri = 'http://www.zooleika.org.uk/bio/foaf.rdf';

18:45:31 <danbri> my $rdf = LWP::Simple->get($uri);

18:45:37 <danbri> my $parser = RDF::Simple::Parser->new(base => $uri);

18:45:39 <danbri> my @triples = $parser->parse_rdf($rdf);

18:45:41 <danbri> danbri@fireball:~$ perl simpletest.pl

18:45:43 <danbri> No _parse_* routine defined on this driver (if it a filter, remember to set the Parent property) [XML::SAX::PurePerl=HASH(0x83e4930)] at /usr/local/share/perl/5.8.0/XML/SAX/Base.pm line 2308.

18:45:46 <danbri> any ideas?

18:45:48 * danbri thinks zool's offline...

18:45:50 <danbri> I installed a bunch of CPAN dependencies without much thought, maybe broke something

18:45:55 <danbri> mostly went w/ defaults

20:28:11 <mdupont> mdupont is now known as mdupont-zzz

20:39:06 <dajobe> zool's stuff looks good

20:39:14 <dajobe> the parser is a translation of sean's 10k python one

20:39:18 <dajobe> however...

20:39:30 <dajobe> in the goal of simplicity, urls=literals=bnodes

20:39:42 <dajobe> and regexes and heuristics used to distingusih them

21:23:53 <sbp> only the same amount of "regexes" and "heuristics" used to dintinguish between terms in NTriples

21:24:23 <sbp> though in an offline version, I make the unescaped literal values and the type available as properties of a sub class of literal

21:26:06 <dajobe> I was describing zool's not yours

21:26:17 <sbp> ah. my apologies

21:26:46 <dajobe> yours was aiming for a different thing - 1 file does it all

21:27:13 <dajobe> seems pretty complete to me, apart from Unicode NFC checking (would need another file) and XML canonicalisation (another file too)

21:28:01 <sbp> actually, I think NFC checking is fairly easy... I just didn't look into it. XML canon would be a pain, but there's an excellent implementation of it somewhere that I could "borrow"; use it if it's available, ignore if not. hmm

21:28:25 * sbp looks at RDF-Simple-0.03

21:28:25 <dajobe> rdflib just used that - it's easy if it edxists

21:29:49 * sbp can feel a 15KB version coming on

21:30:35 <sbp> I wanted to make it as small as possible just to prove that it's not really that complex. now I've done that, I can make it more comprehensive, I suppose

21:30:48 <dajobe> I only looked at your code, maybe you could try testing it

21:30:56 <dajobe> ... I didn't test it either

21:31:30 <dajobe> I meant against some test cases, maybe pyunit it etc.

21:32:28 <sbp> yes, absolutely. I only tested it against all the ones in the syntax specification, and a few "out in the wild" ones. but I used a isomorphism utility

21:32:59 <dajobe> I'm mostly at the ruggedness stage - making sure it doesn't collapse when the XML's crap

21:33:06 <dajobe> or should I say "XML"

21:33:13 <sbp> sigh. yeah

21:33:59 <sbp> it's sad how high a percentage of "RDF" files are not well formed in RSS file surveys etc.

22:26:47 <sandro> what kinds of errors, sbp?

22:27:07 <sandro> it hurts that there are no good/popular RDF browsers

22:30:36 <niq> RDF browsers: would you wish to see it integrated in a Web browser?

22:31:06 <niq> and if so, what would it look like?

22:32:37 <ChrisDodo> ok, i've just created a new weather rss 1.0 service - please kick the tyres, tell me if anything's wrong: http://www.undergroundlondon.com/weather/

22:33:37 * niq smells irony of weather "underground"

22:34:03 <ChrisDodo> :)

22:34:15 <mortenf> it returns an invalid content type

22:34:28 <ChrisDodo> what should it be?

22:34:40 * JibberJim met a guy at svgopen who I unfortunately completely failed to contact info off, who was a weather company, and was looking to RDF descs of his weather info.

22:35:35 <mortenf> basically it's fine, but: Content-Type: application/xml+rss\n\n<?xml version="1.0" encoding="utf-8"?>

22:35:58 <mortenf> seems the newlines aren't interpreted properly

22:36:11 <ChrisDodo> ah.

22:36:14 * ChrisDodo hmms.

22:36:31 <niq> it's \r\n on the Web

22:36:31 <mortenf> in php switch the ''s with "'s ...

22:36:47 <ChrisDodo> heh, it's Perl

22:36:59 <ChrisDodo> lemme have a look

22:37:14 <niq> things like CGI mandate the server to convert automatically

22:38:00 <mortenf> oh, and could you please skip the encoded html?

22:38:14 <mortenf> (there are modules for e.g. images)

22:38:44 <ChrisDodo> ok, i'll look into it. there's a plain feed available for now.

22:39:17 <mortenf> thx (should have read the docs...)

22:42:37 <mortenf> dajobe, any recent changes that would make the serializer output rdf:datatype="" for NULL datatype uris (it seems wrong)?

22:43:08 <dajobe> seems wrong to me; no datatype, no argument at all should be written

22:43:25 <mortenf> hmm, i'll double-check my NULL-ness then...

22:43:43 * dajobe gets raptor reading rdf/xml from inside svg

22:44:26 <mortenf> nice - here's a test: http://xml.mfd-consult.dk/images/foaf-explorer.svg :)

22:44:51 <dajobe> 30 triples

22:45:01 <mortenf> sounds right

22:51:55 <mortenf> hmm, datatype uri problem was mine, apparently i stored an empty string instead of null...

22:52:05 <dajobe> ok

22:54:41 <mortenf> ah, much better; i've now got input *and* output...

22:54:50 <mortenf> btw, i like the new bnode names!

22:54:56 <dajobe> great

22:55:25 <dajobe> *NEW* and *IMPROVED* More Uniquer bnodes Than Before

22:55:30 <mortenf> :)

22:55:47 <mortenf> are you still planning on storing a sequence number hidden away?

22:55:57 <dajobe> persistently?

22:56:03 <dajobe> maybe sometime later

22:56:06 <mortenf> ok

22:56:12 <dajobe> you can alwasy do it in the application layer

22:56:23 <mortenf> yep, just remembered you talked about it

22:56:27 <dajobe> <rdf:Descrption rdf:about="your-uri" rdf:value="1234"/>

22:56:37 <dajobe> yeah, hidden triples

22:57:17 <dajobe> actually would be for describing the graph/store - such as kind of store, URI, creation date etc.

22:57:30 <mortenf> ah, ok

22:57:43 <mortenf> perhaps a special context?

22:57:49 <dajobe> m2=model.get_meta_model() ?

22:58:03 <dajobe> yeah, could be

22:58:04 <mortenf> ETOOMUCHMETA

22:58:18 <dajobe> however I want to hide the stuff from the graph content triples

22:58:36 <dajobe> but still thinking about it

22:58:38 <mortenf> yep

22:58:50 <dajobe> in SQL terms, it's the DESCRIBE ...

22:59:05 <dajobe> rather than the contents of the tables

22:59:08 <mortenf> i'll save modelID 0 for you :)

22:59:16 <dajobe> lol, ta

22:59:48 <mortenf> anyway, optimizing will have to wait, off to bed...

22:59:54 <dajobe> nn

23:54:46 <deltab> A:|A planner that doubles as an inference engine

23:54:46 <dc_rdfig> Titled item A.