Semantic Web Interest Group IRC Chat Logs for 2002-05-02

00:45:17 <testing> testing is now known as testingrdfig

00:45:27 <testingrdfig> testingrdfig is now known as dot_owch4

00:46:26 <dot_owch4> dot_owch4 is now known as dot_owch7

00:46:40 <dot_owch7> dot_owch7 is now known as dot_owch8

00:47:38 <dot_owch8> dot_owch8 is now known as test1

01:09:59 * danbri waves, heads offline

01:10:04 <danbri> see some of you in hawaii...

01:10:24 <AaronSw> c'ya

01:10:35 <danbri> night!

01:13:18 <test1> test1 is now known as dot_owch9

01:13:24 <dot_owch9> dot_owch9 is now known as test1

01:13:41 <AaronSw> Neat, I didn't know about M-GET. TimBL should add that to his Mandatory Extensions Designissue

01:20:26 <mnot> 2774 is neat; the namespacing stuff for headers is way cool, esp. in the light of SW. Unfortunately, the infrastructure doesn't support M-* methods, really.

01:20:59 <mnot> Also, it's quite verbose; carrying around a URI for each namespace in each message doesn't scale well. It's great for ad-hoc extensions, or private ones, just not standardized ones.

05:53:19 <mnot> anybody awake enough for a quick n3 syntax question?

06:03:38 <mnot> hmm... http://openprojects.net/policy.shtml :

06:05:20 <mnot> I just caught them doing a portscan against my machine. This is nasty.

06:24:42 <bijan> maybe.

06:25:59 <bijan> But my reactions are slow

06:31:06 <mnot> That's OK; I found it.

06:31:07 <mnot> thx

06:33:17 <bijan> Heh.

06:33:18 <bijan> My pleasure.

12:23:44 <JibberJim> A: Query that outputs RDF rather than just HTML wanted!

12:23:44 <dc_rdfig> added comment A7

13:35:31 <rreck> i think i would like the topic to be more like "semantic web Backing"

13:55:50 * JibberJim hopes you don't also want noodling changed to canoodling...

14:08:51 <DanCon> hmm... it seems that most of www-tag has left for hawaii or something.

14:09:02 <bijan> Be grateful!

14:09:02 <DanCon> or maybe they're not up yet.

14:09:20 <bijan> Now instead of flaming you to hell in email...they can jump you in a back alley!

14:09:23 <bijan> Er...

14:10:04 <bijan> You could wear armor made of paul prescod email printouts...

14:11:39 <bijan> OTOH, maybe the Burning time is past...pat hayes and patrick stickler had a very cordial exchange on rdfcore!

14:27:05 <dfg_olin> Hi, I am trying to learn how to work with cwm and am looking for some simple examples of queries against the store.

14:27:32 <bijan> Are you trying to do those queries programmticaly, from python code, or using N3 code and filter?

14:27:39 <bijan> Er..from the command line.

14:27:57 <dfg_olin> N3 and filters. Some examples against the gedcom data would be nice.

14:28:30 <bijan> Er..I don't have any handy. The simples sort of filter is a simple rule.

14:29:11 <bijan> E.g., {:someKnownSubject :someKnownPropert: ?x} log:implies {?x a :Find}.

14:29:28 <bijan> (Pace syntax for universal vars.)

14:30:00 <dfg_olin> Pace syntax?

14:30:24 <bijan> Er..assuming the ?x works rather than declaring this log:forAll :x.

14:30:27 <bijan> having to declare.

14:31:15 <bijan> {?x a :Result} might be better as the consquent.

14:31:38 <dfg_olin> I'll go off and experiment. Thanks

14:32:14 <bijan> The trick is to make sure you have something useful in the consquent.

14:32:31 <bijan> i.e., the way you return results is by asserting stuff in the target context.

14:32:36 <bijan> (A way)

15:34:09 <DanCon> hm... I just discovered something about sms URIs in my maybe-spam inbox...

15:34:26 <DanCon> I'd really like to update http://www.w3.org/Addressing/schemes ...

15:34:44 <DanCon> maybe we could have a little IRC party, and invite swBot, and tell it about all the schemes we've discovered.

15:38:10 <AaronSw> sounds like fun

15:40:14 <DanCon> gerald? you up for that? I might have time today

15:42:23 <DanCon> of course, I'd have to get swBot running again.

15:42:44 * DanCon offers a 20 point bounty for anybody who updates swBot w.r.t. recent cwm/llyn stuff

15:43:00 <AaronSw> i think i still have a working copy...

15:46:30 <DanCon> is that using an old cwm?

15:46:52 <AaronSw> i just cvs uped cwm, so it should be the latest

15:47:02 <DanCon> you fixed swBot?

15:47:08 <AaronSw> swBot, colors are gray

15:47:11 <AaronSw> swBot, colors are gray .

15:47:12 <swBot> huh? Line 1: Bad syntax (expected directive or statement) at ^ in: "^colors are gray ."

15:47:18 <AaronSw> swBot, :colors :are :gray .

15:47:18 <swBot> huh? Line 1: Bad syntax (Prefix not bound) at ^ in: "^:colors :are :gray ."

15:47:22 <DanCon> swBot, @prefix : <#>.

15:47:23 <swBot> <mid:rdfn3chat-499@irc.openprojects.net> OK. KB size: 0

15:47:31 <AaronSw> swBot, :dogs :have :fleas .

15:47:32 <swBot> <mid:rdfn3chat-500@irc.openprojects.net> OK. KB size: 1

15:47:36 <AaronSw> i guess i did.

15:47:49 <DanCon> swBot, help?

15:47:50 <AaronSw> hm, have no way to get stiff out of it, tho

15:47:50 <swBot> more docs, source at http://dev.w3.org/cvsweb/2000/scribe-bot/

15:47:52 <swBot> $Id: rdfn3chat.py,v 1.13 2001/08/30 18:46:48 connolly Exp $

15:47:52 <swBot> Line 1: Bad syntax (expected directive or statement) at ^ in: "^help."

15:48:35 <DanCon> swBot, @prefix log: <http://www.w3.org/2000/10/swap/log#>.

15:48:36 <swBot> <mid:rdfn3chat-502@irc.openprojects.net> OK. KB size: 1

15:48:56 <DanCon> swBot, this log:forAll :what. { :what :have :fleas } log:implies { :what a :Answer }?

15:48:56 <swBot> :dogs a :Answer .

15:48:56 <swBot> query done.

15:49:15 <DanCon> get stuff out: oh; you mean getting it to save its state?

15:49:58 <AaronSw> yeah

15:50:07 <AaronSw> i guess i could just do ?x ?y ?z

15:53:01 * AaronSw hacks

15:53:29 <DanCon> it has a save feature

15:53:37 <AaronSw> yeah, but it uses HTTP put...

15:53:44 <DanCon> and?

15:54:19 <AaronSw> i don't think my server's set up for PUT.

15:54:43 <DanCon> ugh. you don't maintain any web servers that grok PUT?

15:54:49 <DanCon> do you have write access to any of W3C's web space?

15:55:12 <DanCon> I could probably chacl someplace for you to PUT addressing stuff

15:55:13 <AaronSw> hm, maybe enabling mod_PUT will make it work...

15:56:26 <AaronSw> i think i have write access to the rdf core test cases dir

15:57:54 <AaronSw> feh, it'll be easier to save it locally and scp it up

15:58:52 * DanCon -> webont

15:59:08 <AaronSw> c'ya

15:59:09 <DanCon> yeah, you can change swBot's save code to save locally

16:01:40 <AaronSw> swBot, @prefix : <#> .

16:01:40 <swBot> <mid:rdfn3chat-28493@irc.openprojects.net> OK. KB size: 0

16:01:45 <AaronSw> swBot, :Dogs :have :fleas .

16:01:45 <swBot> <mid:rdfn3chat-28494@irc.openprojects.net> OK. KB size: 1

16:01:49 <AaronSw> swBot, save!

16:02:03 <JibberJim> Don't think he liked that :-)

16:02:08 <AaronSw> :-)

16:04:27 <AaronSw> swBot, save!

16:05:00 <AaronSw> swBot, @prefix : <#> .

16:05:00 <swBot> <mid:rdfn3chat-46603@irc.openprojects.net> OK. KB size: 0

16:05:03 <AaronSw> swBot, save!

16:05:17 <AaronSw> swBot, :Dogs :have :fleas .

16:05:18 <swBot> <mid:rdfn3chat-46604@irc.openprojects.net> OK. KB size: 1

16:05:18 <AaronSw> swBot, save!

16:05:31 <AaronSw> oh, heh, it did work

16:05:41 <JibberJim> He didn't tell us though...

16:07:31 <AaronSw> swBot, save!

16:08:50 <AaronSw> swBot, save!

16:08:50 <swBot> saved!

16:09:01 <AaronSw> swBot, save!

16:10:22 <AaronSw> swBot, save!

16:11:03 <AaronSw> swBot, save!

16:11:03 <swBot> saved!

16:11:14 <AaronSw> woo, it works

16:11:22 <AaronSw> swBot, @prefix : <#> .

16:11:22 <swBot> <mid:rdfn3chat-58196@localhost> OK. KB size: 0

16:11:28 <AaronSw> i should put n3s in front of it...

16:14:16 <AaronSw> swBot, :x dc:title "Something or other" .

16:14:16 <swBot> <mid:rdfn3chat-2869@localhost> OK. KB size: 1

16:14:20 <AaronSw> cool!

16:14:24 <AaronSw> swBot, save!

16:14:24 <swBot> saved!

16:14:40 <AaronSw> thanks, sbp!

16:16:04 <JibberJim> Does it have a load?

16:16:42 <AaronSw> nope

16:17:16 <AaronSw> swBot, { ?x dc:title ?y } => { ?x dc:title ?y}?

16:17:16 <swBot> Line 1: Bad syntax (object_list expected) at ^ in: "^> { ?x dc:title ?y}."

16:17:31 <AaronSw> swBot, { ?x dc:title ?y } log:implies { ?x dc:title ?y}?

16:17:31 <swBot> Line 1: Bad syntax (Prefix log not bound) at ^ in: "^ log:implies { ?x dc:title ?y}...."

16:17:42 <AaronSw> swBot, { ?x dc:title ?y } log:implies { ?x dc:title ?y} .

16:17:42 <swBot> <mid:rdfn3chat-2872@localhost> OK. KB size: 12

16:17:49 <AaronSw> oh, it's the query thing...

16:18:33 <AaronSw> swBot, :x dc:title "Something or other" .

16:18:33 <swBot> <mid:rdfn3chat-97923@localhost> OK. KB size: 1

16:18:38 <AaronSw> swBot, { ?x dc:title ?y } => { ?x dc:title ?y}?

16:18:38 <swBot> :x <http://purl.org/dc/elements/1.1/title> "Something or other" .

16:18:38 <swBot> query done.

16:18:44 <AaronSw> yes!

16:22:19 <sbp> ooh, cool stuff, Aaron

16:22:41 <AaronSw> i just did formula = n3s.process(formula) -- is that right?

16:24:08 <sbp> yep, that's fine

16:24:56 <sbp> if you want it to be able to read in some AN3 from the Web, I added a bit of code locally to stip out the comments correctly

16:25:03 <sbp> but it's not really necessary

16:25:51 <AaronSw> hi redmo... bye redmonk

16:25:58 <sbp> heh, heh

16:28:12 <DanCon> hmm... => and ?foo mean that for many cases, you don't need to declare the log: prefix

16:28:51 <AaronSw> n3s declares it for you if you need it, tho

16:28:59 <AaronSw> along with a number of other popular prefixes

16:59:02 <sat> a history question....who were the chairs of RDF WG, before the current folks

17:00:07 <AaronSw> I think it was Bob someone... at IBM

17:00:23 * AaronSw waves from first telecon of the day

17:00:26 <AaronSw> I always wanted to say that

17:01:10 <larsbot> :)

17:01:14 <AaronSw> "David Singer of IBM was the chair of the original RDF Schema group"

17:01:51 <sat> thanks...its interesting these guys are no longer members of RDF WG ?

17:02:39 <em> sat, myself and Bob Schloss were the chairs of the RDF M&S Working Group; David Singer (IBM) was chair of the RDF Schema working group

17:03:04 <em> Bob Schloss was also from IBM, I was at OCLC at the time

17:03:23 <bijan> Why is it interesting?

17:03:31 <bijan> Chairing a working group has to be hell.

17:03:35 <bijan> No fun whatsoever.

17:03:46 <AaronSw> oh?

17:03:47 <em> there are (actually i find surprising) still several people on RDF working groups that were around in early days

17:03:55 <JibberJim> em standing it for 5 years must make him some kind of saint then...

17:04:03 <bijan> Or masochist.

17:04:11 <sat> I wasn't sure if membership is rotated or people just drop out due to lack of interest

17:04:16 <JibberJim> ah a more likely explanation indeed...

17:04:28 <bijan> Well, it really does depend, but the chair has to be adminstratively minded, which, for me, sucks :)

17:04:47 * em prefers 'dedicated', but masochist may also be correct...

17:04:47 <bijan> sat, there's lots of possible explanation.

17:04:55 <bijan> other commitments for example.

17:05:06 <em> hmm... i think some sort of history page would be useful

17:05:14 <bijan> Being on a working group, or any standards body, is typically a lot of work.

17:05:22 <bijan> And comparatively little "inherent" reward.

17:05:26 <em> sat, why (out of curiosity) do you ask?

17:05:26 <bijan> I.e., it's not like you're doing research.

17:05:33 <bijan> Or building a product. Etc.

17:05:38 <bijan> It's all politics :)

17:05:54 <sat> just I heard someone say that original folks are no longer interested...I was trying to figure out

17:06:07 <bijan> So, I would be careful about inferring from changing membership to anything about the spec.

17:06:12 <bijan> Or the tech

17:06:22 <bijan> Or even their feelings about such :)

17:06:29 <em> right... most have gone on to build products... many (if not all) arent' neccessary interested in standards work, but rather this as a means towards a shared ends

17:07:05 <bijan> Plus, the *corp* that was sponsering them 1) may have lost interest or 2) needed them for other things or 3) had to give someone else (with different interests) a shot.

17:07:06 <bijan> Etc.

17:07:10 <sat> please excuse my lack of knowledge in standards work

17:07:41 <bijan> FWIW, I'd prolly burn out in a day :)

17:08:31 <bijan> Plus, it's not always a useful career move, e.g., depending on your institutioin.

17:08:37 <em> sat, no excuses neccessary...

17:08:44 <em> its a fine question to ask

17:08:54 <bijan> I.e., I don't know how CS depts rank contributing to a w3c wg in tenure decisions, for example.

17:08:56 <bijan> Yep.

17:09:03 <bijan> (agreeing with em)

17:09:39 * em remembers the reason why he started poking around ...

17:09:54 <bijan> Since being an active WG member requires a fair bit of time, if you're also needing to publish in refereed journals...

17:10:02 <bijan> Well, you may have to make decisions :)

17:10:38 <em> i'm interested in hearing peoples 'top 10 sw projects' list...

17:10:52 <em> err... how do i chump a blurb again?

17:11:01 <bijan> ..BLURB: blah blah blah

17:11:07 <bijan> Without the leading ..s

17:11:33 <em> BLURB: List your favorite Semantic Web project here!

17:11:33 <dc_rdfig> A: List your favorite Semantic Web project here! from em

17:11:49 <JibberJim> Well that's got to be Google surely...

17:11:54 <bijan> Hehehe.

17:12:10 * em dosen't suppose rdfig chump has a notion of a sticky item

17:12:48 * em agrees with JibberJim & bijan on this... google that mine graphs are up on my top

17:12:59 <bijan> em, see: http://monkeyfist.com/articles/815

17:13:38 <em> already have bijan ... glad to see this was the catalyst for you :)

17:13:46 <em> very nicely written btw

17:14:24 * larsbot is away: I'm busy

17:14:27 * larsbot is back (gone 00:00:04)

17:15:36 * em will be back shortly...

17:16:01 <bijan> thanks.

18:08:21 <larsbot> what's the difference between a type and a class?

18:09:28 <larsbot> is there a difference at all?

18:09:42 <bijan> In what context.

18:10:03 <larsbot> I'm thinking in general

18:10:06 <bijan> In rdf/s, not really.

18:10:14 <bijan> There is no in general :)

18:10:19 <larsbot> I expected that :)

18:10:29 <bijan> Sorry.

18:10:32 <larsbot> say in knowledge representation, then

18:10:39 <bijan> Depends.

18:10:43 <bijan> Sorry.

18:10:47 <larsbot> ok, logic, then

18:11:08 <bijan> Sigh.

18:11:15 <bijan> Picking another fairly general area doens't help :)

18:11:35 <larsbot> let's turn it around, then: can you think of a context where there is a difference?

18:11:48 <golbeck> in programming languages there's a differenc e:)

18:11:55 <bijan> For certain understandings of the terms, yes.

18:11:55 <larsbot> only in misdesigned ones :)

18:12:06 <larsbot> could you give an example?

18:12:21 <golbeck> me or bijan

18:12:36 <larsbot> bijan

18:12:40 <bijan> I might want classes to have a type, but not for that to be merely being membership in a class.

18:12:43 <larsbot> I know about programming languages :)

18:12:49 <golbeck> bah

18:13:10 <larsbot> uh, bijan, what's your context now?

18:13:12 <bijan> "class" is tricky because it's easy to think of them as sets.

18:13:26 <larsbot> so they aren't just sets?

18:13:45 <bijan> Well, in some terminologys, sets and classes are disjoint.

18:13:53 <bijan> E.g., classes can contain themselves.

18:14:06 <bijan> The "collectoin of all sets" is a class.

18:14:16 <bijan> There's *no winning* here unless you pin down your terms.

18:14:55 <larsbot> and you're saying they're not pinned down in logic, only within specific contexts within logic?

18:15:15 <bijan> There are a lot of logics.

18:15:37 <larsbot> right, and they handle this differently?

18:15:40 <bijan> FOPL doesn't, in its object theory, typically have a distinguished notion of class *or* type.

18:15:58 <bijan> You can explicate FOPL in partially in terms of sets/classes (e.g., model theory)

18:16:08 <bijan> You can add types to your quantifiers, etc.

18:16:48 <larsbot> so model theory distinguishes between the two?

18:17:01 <bijan> There is no "two".

18:17:12 <larsbot> well, there are two terms: "type" and "class"

18:17:31 <bijan> But they are wildly equivical terms.

18:17:35 <bijan> So they are legion :)

18:18:00 <larsbot> I can understand why I am confused, at least :)

18:18:10 <larsbot> does model theory have a notion of "type"?

18:18:18 <bijan> Then, as Socrates would, I'm leaving things there :0

18:18:30 <bijan> depnds on the model theroy.

18:18:34 <larsbot> ah, I see

18:18:38 <bijan> And with that, I have to stop (deadlines)

18:18:47 <larsbot> fair enough; good luck and thanks

18:19:27 <bijan> Your welcome.

18:19:28 <bijan> THanks.

18:31:56 <rallye> What is the advantage of description logics over frame-based?

18:34:33 <DanCon> hm... W3C specs need quick-ref pages

18:34:46 <DanCon> I just want to know the URI of the XML Schema datatypes namespace

18:36:02 <JibberJim> .google xml schema datatypes namespace

18:36:03 <xena> xml schema datatypes namespace: http://www.xml.com/pub/a/2000/11/29/schemas/dataref.html

18:41:33 <bijan> rallye, I'm not sure you can classify things that way.

18:41:47 <bijan> A DL is, in general, a highly tractable subset of FOL.

18:41:53 <bijan> (computationally tractable0

18:42:09 <bijan> That is it trades off expressivity for computational tractability.

18:45:43 <rallye> In a KR system isn't expressivity far more important? I mean, how intractable can this become?

18:46:01 <bijan> Well, start with the fact that FOPL isn't decidable.

18:46:29 <bijan> Second that proof techniques for full FOPL can get very complicated.

18:46:52 <bijan> Often all you *need* is what a DL gets you.

18:47:07 <bijan> I.e., you aren't after superduper complex knowledge, just a lot of it.

18:47:17 <bijan> And you need guaruntees about termination and performance.

18:47:32 <bijan> After all, RDBMS are a highly tractable subset of FOL.

18:47:43 <bijan> An incredibly useful subset, in fact.

18:47:58 <bijan> *Huge* trade off between expressivity and tractability there.

18:48:28 <bijan> And RDBMS have had decades of implementation research & innovation.

18:48:58 <rallye> So if this is true, is it better to use DL's instead of frame-based unless one has more complex ideas to express?

18:49:18 <bijan> It's not DL vs. Framebased.

18:49:24 <bijan> Frame is more a notational style.

18:49:35 <bijan> I.e., webont will have a framey layer.

18:49:56 <bijan> IOW, you can have expressively limited frame languages.

18:50:17 <bijan> And, overall, there's a buncha criteria you need to evaluate.

18:50:26 <bijan> E.g., what specific reasoning techniques you want to use.

18:50:28 <bijan> Storage.

18:50:39 <bijan> Complexity of the KR langauge for end users.

18:50:43 <bijan> Match with the domain.

18:50:44 <bijan> Etc.

18:51:54 <rallye> I must have confused myself then. I have seen many sites comparing the two (granted, not in any clear way).

18:52:43 <rallye> Sorry.

18:52:52 <bijan> No, it's ok.

18:53:19 <bijan> The thing is that it's easy for an approach or notation to *feel* really different.

18:53:27 <bijan> But to turn out ot be a notational variant of FOL :)

18:53:49 <larsbot> you should have heard John Sowa at KT 2002 :)

18:53:55 <bijan> Heh.

18:53:57 <bijan> I'll bet.

18:54:06 <bijan> Note that not *everyone* agrees with me on this front :)

18:54:07 <larsbot> he had a long rant about how God was a higher standard authority than both the W3C and ISO :)

18:54:08 <AaronSw> what's Sowa's position?

18:54:13 <AaronSw> lol

18:54:14 <AaronSw> no notation without denotation!

18:54:35 <larsbot> basically he didn't want to see any more KR stuff that wasn't grounded in logic

18:54:52 <bijan> Ah, I saw an email msg about that.

18:55:02 <larsbot> wouldn't surprise me; it was highly amusing

18:55:04 <bijan> Yeah yeah, but that was a little unclear.

18:55:04 <bijan> Argh!

18:55:14 <bijan> Must work. Not get sucked in.

18:55:19 <larsbot> :-)

19:35:23 <sbp> heh @ <AaronSw> no notation without denotation!

19:35:36 <AaronSw> Drew McDermott wrote that paper

19:35:42 <AaronSw> (IIRC)

19:35:55 <larsbot> reminds me of Derrida: every decoding is another encoding

19:36:03 <AaronSw> it was the sequel to artificial intelligence meets natural stupidity, i think

19:36:08 * sbp tips some URIs into Boston harbour

19:36:12 <larsbot> do you have a URI?

19:36:19 <AaronSw> it's in Boston Harbor

19:36:25 <larsbot> too bad :)

19:36:29 <AaronSw> oh, my URI? it's http://www.aaronsw.com/

19:36:41 <larsbot> well, no, to that paper

19:36:47 <larsbot> I've seen your site several times :)

19:37:05 <larsbot> and quite like it, FWIW

19:37:10 <AaronSw> Thanks. :)

19:37:17 <AaronSw> here's the citeseer link... http://citeseer.nj.nec.com/context/70329/0

19:37:35 <larsbot> thanks :)

19:37:48 <AaronSw> and http://citeseer.nj.nec.com/context/25043/0

19:38:25 <AaronSw> those pages are all "Hayes, McDermott, Hayes, McDermott"

19:38:31 <larsbot> so I see

19:38:39 <AaronSw> it's like some alien conspiracy...

19:40:36 <larsbot> yeah, I'm still looking for a guide to alien language and psychology

19:40:46 <larsbot> tried to find a book on model theory at Foyles, but failed

19:40:59 <larsbot> well, I found one, but it was obviously written by aliens for aliens

19:41:42 <AaronSw> heh.

21:02:56 * DanCon tries to convince cwm that bob is not a Continent...

21:06:07 <DanCon> I got as far as

21:06:09 <DanCon> ex:Bob a [

21:06:09 <DanCon> ont:complementOf [

21:06:09 <DanCon> ont:oneOf (

21:06:09 <DanCon> ex:Australia

21:06:09 <DanCon> ex:Antarctica ) ] ],

21:06:25 <DanCon> really odd; it'll do setps 1 and 2, but not 3 thru 6.

21:07:10 <DanCon> wow... when I shorten the list of continents to 2, it works.

21:44:10 <DanCon> hi. So... what conclusions should we be able to draw from this cert thingy?

21:44:30 <DanCon> reading SPKI syntax... (* set read write) <- * is a regular symbol there, yes?

21:44:54 <reagle-tu> An agent can use it to permit the subject to do certain things

21:45:19 <DanCon> the subject... there's a key around there somewhere, yes?

21:45:24 <reagle-tu> Yes, ... a SPKI TAG has a simple syntax that defines string and some globs (per scheme: e.g., http: ftp:, etc.)

21:45:29 <DanCon> I'm not sure how to read (subject (hash sha1 |Ve1L/7MqiJcj+LSa/l10fl3tuTQ=|))

21:45:40 <reagle-tu> that defines the path over which the access (e.g., read write) apply

21:46:19 <reagle-tu> the subject of the cert

21:46:27 <reagle-tu> has a key iwth that fingerprint

21:46:39 <DanCon> ah... it might be clearer to write it that way...

21:46:50 <reagle-tu> that's how the sujbect is identified. the issuer is identifier with their fingerprint, there the person stating the cert

21:47:00 <DanCon> :cert :subject [ has :fingerprint "Ve1L/7MqiJcj+LSa/l10fl3tuTQ=

21:47:02 <DanCon> "]

21:47:23 <reagle-tu> my second example has:

21:47:23 <reagle-tu> :cert has :issuer [cr:SHA "TLCgPLFlGTzgUbcaYLW8kGTEnUk="];

21:47:23 <reagle-tu> has :subject [cr:SHA "Ve1L/7MqiJcj+LSa/l10fl3tuTQ="];

21:48:20 <DanCon> er... fingerprint... is that really just the SHA-1 of the key? that's not how PGP fingerprints work, is it?

21:48:27 <reagle-tu> so that was the little step i made today! :)

21:48:29 <DanCon> anyway... I think I grok now...

21:48:37 <reagle-tu> yep, and yep

21:48:46 <reagle-tu> (at least, to the bets of my knowledge)

21:49:44 <DanCon> so the rules are something like "if a request matches a tag, and it's signed by a key that's the subject of a certificate, then the issuer of the certificate authorizes the request"

21:50:00 <reagle-tu> yes.

21:50:18 <reagle-tu> (I'm not spki expert, so any answer is "to the best of my understanding")

21:51:35 <DanCon> { :cert :tag [ :matches ?req ]; :subject ?whoK; :issuer ?CK. ?req :signedBy ?whoK. } log:implies { ?CK :authorizes ?req }.

21:52:10 <DanCon> (where ?v is almost-n3.)

21:52:20 <AaronSw> i think the A stands for abbreviated

21:52:28 <reagle-tu> yes...

21:52:54 <AaronSw> swBot now speaks AN3, BTW

21:53:12 <DanCon> does it know sha-1?

21:53:34 <DanCon> swBot, :cert1 :issuer [ cr:SHA "TLCgPLFlGTzgUbcaYLW8kGTEnUk="].

21:53:43 <AaronSw> hmmm

21:53:54 <DanCon> sorry, swBot! didn't mean to blow your mind!

21:53:56 <AaronSw> Invalid N3: The prefix "cr" has been used but not declared

21:54:02 <reagle-tu> your rule is the evaluation rule ... (and spki specifies "intersection rules" by which mutliple certs can be collapsed and evaluatied.

21:54:37 <AaronSw> what do i except there?

21:54:38 <reagle-tu> that was part of the question... should the cert itself contain the eval rule?

21:55:29 <DanCon> well, in a very general web-of-trust system, yes, a certificate is just a file of facts and rules. But if you're trying to model SPKI, the rules should probably be factored out.

21:56:20 * DanCon plays around in emacs, trying to get cwm to come to a relevant conclusion...

21:58:15 <AaronSw> swBot, :cert1 :issuer [ cr:SHA "TLCgPLFlGTzgUbcaYLW8kGTEnUk="].

21:58:38 <DanCon> in my version of swBot, it reported parse errors, rather than blowing up

21:58:44 <AaronSw> that's what i'm trying to fix

21:58:45 <AaronSw> swBot, :cert1 :issuer [ cr:SHA "TLCgPLFlGTzgUbcaYLW8kGTEnUk="].

21:58:45 <swBot> huh? 'Invalid N3: The prefix "cr" has been used but not declared':None

21:58:58 <AaronSw> ok, that's a little better

21:59:40 <AaronSw> swBot, @prefix cr: <#> .

21:59:41 <swBot> huh? <class exceptions.IndexError at 0x3b91e0>:<exceptions.IndexError instance at 0x694880>

22:00:09 <AaronSw> huh?

22:03:39 <AaronSw> swBot, @prefix cr: <foo#> .

22:03:41 <swBot> huh? <class exceptions.IndexError at 0x3b91e0>:<exceptions.IndexError instance at 0x685b20>

22:03:50 <AaronSw> i think i borked it or something.

22:04:40 <reagle-tu> DanC, gotta go soon, but I'd be very happy to look at anything, I also need to look at your example in test more carefully.

22:05:14 <AaronSw> swBot, @prefix cr: <foo#> .

22:05:28 <AaronSw> um.

22:05:33 <AaronSw> swBot, :cert1 :issuer [ cr:SHA "TLCgPLFlGTzgUbcaYLW8kGTEnUk="].

22:05:33 <swBot> <mid:rdfn3chat-9824@localhost> OK. KB size: 3

22:05:40 <AaronSw> swBot, @prefix cr: <foo#> .

22:05:47 <AaronSw> swBot, save!

22:05:48 <swBot> saved!

22:06:04 <AaronSw> hm, seems to have worked

22:06:26 <DanCon> swBot, where did you save it to?

22:06:32 <DanCon> sorry, swBot!

22:06:41 <DanCon> [[[

22:06:42 <DanCon> :cert1

22:06:42 <DanCon> s:issuer [ s:sha1 "TLCgPLFlGTzgUbcaYLW8kGTEnUk="];

22:06:42 <DanCon> s:subject [ s:sha1 "Ve1L/7MqiJcj+LSa/l10fl3tuTQ="];

22:06:42 <DanCon> a s:propagate; #@@??

22:06:42 <DanCon> s:tag [ sT:http;

22:06:44 <DanCon> s:prefix "http://acme.com/company-private/personnel/"];

22:06:45 <DanCon> s:tag [ a sT:pkpfs;

22:06:47 <DanCon> sA:path "//ftp.clark.net/pub/cme/spki.*"

22:06:51 <DanCon> sA:operations (set read write) ];

22:06:53 <DanCon> s:not-before "1998-03-01_12:42:17";

22:06:55 <DanCon> s:not-after "2012-01-01_00:00:00".

22:06:57 <DanCon> ]]]

22:07:09 <reagle-tu> cool

22:08:16 <AaronSw> i don't think swBot likes DanCon today

22:08:25 <AaronSw> swBot, where'd you save it to?

22:08:26 <swBot> Line 1: Bad syntax (expected directive or statement) at ^ in: "^where'd you save it to."

22:09:07 <AaronSw> what he means, of course, is http://www.aaronsw.com/2002/swbotstate.n3

22:09:17 <reagle-tu> i presume your triple brackets are a quote convention, not part of the n3.

22:09:51 <reagle-tu> yes.

22:10:00 <DanCon> yes, [[[ ]]] is just IRC-gunk

22:10:09 <AaronSw> swBot, { "foo" crypto:sha ?x } => { ?x a :Answer } .

22:10:10 <swBot> <mid:rdfn3chat-29338@localhost> OK. KB size: 6

22:10:10 * DanCon fleshes out evaluation rule...

22:10:12 <AaronSw> swBot, { "foo" crypto:sha ?x } => { ?x a :Answer } ?

22:10:12 <swBot> query done.

22:10:15 <AaronSw> hm

22:10:40 <DanCon> swBot, @prefix : <#>.

22:10:47 <DanCon> swBot, :Dogs :get :Fleas.

22:10:47 <swBot> <mid:rdfn3chat-29341@localhost> OK. KB size: 13

22:11:03 <DanCon> swBot, { ?what :gets :Fleas } => { ?what a :Answer}?

22:11:03 <swBot> query done.

22:11:14 <AaronSw> odd

22:11:24 <DanCon> are you sure you got the scoping of ?v's right?

22:11:38 <AaronSw> they're syntactically converted to log:forAlls

22:11:47 <DanCon> but at what level of {}?

22:11:55 <reagle-tu> (typo in example? I presume "s:cert1")

22:12:06 <DanCon> swBot, this log:forAll :what. { :what :gets :Fleas } => { :what a :Answer}?

22:12:06 <swBot> query done.

22:12:25 <DanCon> no, :cert1 is a name I gave the certificate, not something from spki vocab

22:12:39 <AaronSw> your example gets converted to:

22:12:39 <AaronSw> this log:forAll :what .

22:12:39 <AaronSw> { :what :gets :Fleas } log:implies { :what rdf:type :Answer } .

22:12:44 <DanCon> I'm using @prefix s: <http://example/@@spki-vocab#>.

22:13:00 <DanCon> that looks right, Aaron.

22:13:10 <AaronSw> it worked earlier this morning...

22:13:19 <AaronSw> swBot, :x dc:title "something or other" .

22:13:19 <swBot> <mid:rdfn3chat-29344@localhost> OK. KB size: 26

22:13:32 <AaronSw> swBot, {?x dc:title ?y } => {?x dc:title ?y }?

22:13:32 <swBot> :x <http://purl.org/dc/elements/1.1/title> "something or other" .

22:13:32 <swBot> query done.

22:13:36 <AaronSw> hmm

22:13:41 * reagle-tu is still parsing syntax...

22:13:57 <AaronSw> swBot, { "foo" crypto:sha ?x } => { "foo" crypto:sha ?x}?

22:13:57 <swBot> query done.

22:14:04 * reagle-tu wonders why ther's a semi colon after sT:http

22:14:13 <AaronSw> swBot, { "foo" crypto:md5 ?x } => { "foo" crypto:md5 ?x } ?

22:14:14 <swBot> query done.

22:15:55 <reagle-tu> are questions marks special, or just part of the identifer?

22:16:13 <AaronSw> ?x means that x is a free variable, i.e. something your searching for

22:16:15 <sbp> swBot, { ?what :get :Fleas } => { ?what a :Answer}?

22:16:15 <swBot> :Dogs a :Answer .

22:16:15 <swBot> query done.

22:16:31 <AaronSw> any idea what's wrong with my crypto example, sbp?

22:16:31 <sbp> s/gets/get/

22:16:49 <sbp> Hmm...

22:16:54 <AaronSw> hm, maybe md5 goes the other way

22:17:04 <reagle-tu> aaron, is that just a stylistic convention, or necessary to the processing?

22:17:08 <AaronSw> swBot, { "foo" is crypto:sha of ?x } => { "foo" is crypto:sha of ?x}?

22:17:08 <swBot> query done.

22:17:22 <AaronSw> it's necessary to the processing, otherwise it would thing it was just a normal URI.

22:17:31 <AaronSw> with a ? in front, cwm will search for all things that fit that pattern

22:17:36 <sbp> no, I think you had it the right way before

22:17:46 <reagle-tu> and a question mark after?

22:17:54 <reagle-tu> {...}?

22:18:00 <AaronSw> a questionmark at the end of the statement tells swBot to do a query

22:18:20 <reagle-tu> and => is short for log:implies ?

22:18:25 <AaronSw> yep

22:18:36 <reagle-tu> (when => is log:forAll implied then?)

22:18:45 <sbp> swBot, "blargh" a :TargetString .

22:18:46 <swBot> <mid:rdfn3chat-29350@localhost> OK. KB size: 60

22:19:04 <AaronSw> "?x" is short for "this log:forAll :x" .

22:19:10 <reagle-tu> ah, got it.

22:19:11 <sbp> swBot { ?x a :TargetString; crypro:sha ?y } log:implies { ?x :sha ?y }?

22:19:14 <sbp> ugh

22:19:16 <sbp> swBot, { ?x a :TargetString; crypro:sha ?y } log:implies { ?x :sha ?y }?

22:19:17 <swBot> Line 1: Bad syntax (Prefix crypro not bound) at ^ in: "^ crypro:sha ?y } log:implies { ?x :sha ?..."

22:19:24 <sbp> ugh!

22:19:25 * DanCon gets rules to parse, checks in, chacls...

22:19:29 <sbp> swBot, { ?x a :TargetString; crypto:sha ?y } log:implies { ?x :sha ?y }?

22:19:29 <swBot> query done.

22:19:43 <DanCon> the rules don't work yet, reagle, but...

22:19:57 <DanCon>http://www.w3.org/2000/10/swap/test/spki2may.n3

22:19:57 <dc_rdfig> B: http://www.w3.org/2000/10/swap/test/spki2may.n3 from DanCon

22:20:16 <DanCon> B:a SPKI example, converted to RDF/n3

22:20:16 <dc_rdfig> added comment B1

22:20:18 <reagle-tu> aaron, when you say swBot, is that the same as cwm, or does it have different conventions than the cwm that is in CVS?

22:20:20 <DanCon> B:in progress; not working yet

22:20:20 <dc_rdfig> added comment B2

22:20:42 <AaronSw> swBot is just an IRC interface to CWM. the swBot ?-at-end-of-sentence is equivalent to cwm's -filter option.

22:21:12 <DanCon> well, swBot is a bit ahead of cwm-in-cvs. cwm-in-cvs doesn't grok ?foo

22:21:32 <AaronSw> it does if you pipe it thru n3s, which is what swBot does.

22:21:36 <reagle-tu> thanks DanC, I will study it as I'm still trying to be conversant with the syntax.

22:21:46 * DanCon debugs...

22:24:02 <reagle-tu> hang loose till hawaii! :)

22:24:06 <AaronSw> swBot, {"blargh" crypto:md5 ?y } => { ?y a :Answer } ?

22:24:07 <swBot> query done.

22:24:25 <DanCon> you outa here, reagle? do come again and poke me about this

22:24:31 <reagle-tu> yep

22:24:37 <reagle-tu> i will

22:30:41 <DanCon> aha! got the rule to work.

22:30:52 <DanCon> :cert1 a s:propagate;

22:30:52 <DanCon> s:issuer [

22:30:52 <DanCon> sA:authorizes :req1;

22:33:37 * sbp runs an entailment test on eep3

22:38:38 <DanCon> B:rule works on the example data now (v1.2)

22:38:38 <dc_rdfig> added comment B3

22:38:47 <DanCon> B:|cwm does a little SPKI

22:38:48 <dc_rdfig> titled item B

23:39:18 <Ansi> Ansi is now known as Ansi_zzZZ

23:58:13 <sbp> Hmm... 8/11 N3 parsers support the "has prop" construct in N3. Does *yours*?

23:59:11 <sbp> rdfn3.g, n3spark.py, and Cwmclone.P being the ones that don't (I may be wrong about Cwmclone)