00:00:06 sbp has quit (Ping timeout for sbp[ch8as12-76-167-72.cw-visp.com]) 00:00:11 K:research by Belinfante, Art Quaife 00:00:11 commented item K 00:00:45 oh... I guess since Pat Hayes is taking pot-shots at RDF, I can say that it was he who made the bogus claim that induction isn't 1st order. 00:00:46 danbri has quit (Ping timeout for danbri[modem-101.allards-clownfish.dialup.pol.co.uk]) 00:01:54 DanC: I have an honest question i.e. this is simply a question I'm not trying to make a point 00:02:32 * DanC turns into a dad... 00:02:40 I can understand the utility of representing rules and statements as 1st order predicate calc 00:03:02 can you represent simple data? 00:03:14 Induction is second order. 00:03:19 Usually. 00:03:31 Unless, I suppose you use schemata 00:04:41 AaronSw has quit () 00:04:44 how would schemata help? 00:05:48 AaronSw has joined #rdfig 00:06:53 They may not. I just seemt o recall reading something to that effect but I can't recall the details and I'm not finding the book :) 00:07:10 AaronSw has quit () 00:07:37 AaronSw has joined #rdfig 00:07:40 what is the relationship between data and induction? 00:08:12 AaronSw has quit () 00:08:22 I don't understand your question. 00:08:32 AaronSw has joined #rdfig 00:08:34 You can use induction to reason about enumerable sets. 00:09:04 (mathematical induction, I presume we're talking about) 00:09:30 i am wondering that if i have a graph which represents some data ... just some data ... how would that be 'expressed' in 1st order predicate calculus 00:09:47 Depends. 00:09:56 Node(A) 00:10:07 Node(B) 00:10:13 Arc(A,B) 00:10:44 back to triples 00:10:52 No. 00:11:11 Those are predicates. 00:12:00 from a very simplistic view I have three symbols to store 00:12:36 I see 5, at least. 00:12:40 i don't understand the fundamental difference or the utility of making such a distinction 00:12:54 Between? 00:13:34 Arc(A,B) implies Node(A) and Node(B) hence is simply a different way to write it, no? 00:14:09 Not if it just implies it. 00:14:33 "just implies"? 00:14:47 I presume by "simply a differnt way" you mean equivalent. 00:14:58 Mere implicaiton doesn't sufficent for equivalence. 00:15:30 You need bi-implicaiton. 00:15:41 And Node(A), Node(B) doesn't imply Arc(A,B) 00:16:17 And you'd have to say an awful lot to make the implication clear. 00:16:26 i'm lost. 00:16:37 E.g., if Arc(A,B) then (Node(A) and Node(B)) 00:17:20 Actually, ForAllX,Y (if Arc(X,Y) then (Node(X) and Node(Y))) 00:17:42 couldn't a system which wants to interpret predicates understand a table that has as Arc(A,B) Node(A) Node(B)? 00:17:45 If you add that statment then you can derive Node(A) and Node(B) form Arc(A,B) 00:18:12 Why? 00:18:24 Even if so, you've got more than just the triple in the system. 00:18:26 ok, so then aren't they two ways of saying the same thing, representing the same data 00:19:15 WEll, Arc(A,B) will never be equivalent to Node(A) and Node(B) 00:19:23 Since the latter don't have to be connected. 00:19:37 ok. did I ask for that equivalence? 00:19:53 I read "saying the same thing" as "being equivalent" 00:20:05 i was careful. 00:20:35 how about "All symbols which are present in the second and third columns of the table are Nodes" 00:21:12 Well the symbols aren't, are they? 00:21:56 And you still have "Node" in there, yes? 00:22:02 "identify Nodes"? 00:22:08 I suspect i have *no* idea what you're after :) 00:22:19 Arc(A,B) implies Node(A) and Node(B) hence is simply a different way to write it, no? 00:22:29 exactly. 00:22:35 Was what you origianlly said. 00:22:39 But: 00:23:02 1) Arc(A,B) doesn't imply Node(a) and Node(B) without supplementary premises. 00:23:34 2) Even adding those premises does make them equivalent, since A & B being nodes doesn't entail that there is an arc between them. 00:24:12 what i am wondering from my knowledge of parse trees etc, is whether there are _any_ expressions that can be expressed in 1st order predicate calculus that cannot also be expressed as a collection of triple / or a graph given certain assumptions. 00:24:32 3) In a system with the requise additional premise, you can recover the facts that Node(A) and NOde(B) by deducing them from the fact Arc(A,B) + premises, which means you don't need to explicitly "enter those facts int he system" 00:25:37 Well, with the right interpretation, a bunch of triples can encode any 1st order prediacte wff. 00:25:48 Is this interesting? 00:25:52 It'll be rather verbose. 00:25:55 Typically. 00:25:59 so what could be problem be with RDF then? 00:26:20 Well, encoding isn't quite the same as expressing, yes? 00:26:36 It's sorta like saying "C is turnign complete thusit has backtracking" 00:26:46 Just because you can write a non-deterministic engine in C. 00:27:20 If you give a specific interpretationt o the triples. 00:27:35 "Subject predicate object" then there are lots of things you can't express. 00:27:57 For all x, if x is short, then x eats banannas 00:28:20 RDF has no variables, nor quantifiers, nor negation, nor implication, nor conjunction/disjunction, afaik. 00:28:24 basically turing complete is the issue i suppose. Can someone write a system using 1st order predicate calculus that can do something that cannot be done by a program which operates on a graph 00:28:57 Er.. you don't "write systems" using the 1st order predicate caluclus, I think. 00:29:20 If you're question is whether the first order predicate calc is turing ocomplete, I'll say, dunno off hand. 00:29:48 It's an *odd* thing to say since the fopc isn't about functions. 00:29:57 Ok, I have to run. TTYL 00:30:02 Gems and minerals beckon. 00:30:11 adios 00:30:53 jonb has left channel 01:06:21 sbp has joined #rdfig 01:10:34 sbp has quit (Ping timeout for sbp[pBBs09a06.client.global.net.uk]) 01:13:48 sbp has joined #rdfig 01:20:39 rigo has joined #rdfig 01:21:48 rigo has quit (Leaving) 01:22:08 rigo has joined #rdfig 01:22:44 sbp has quit (Ping timeout for sbp[pD8s13a07.client.global.net.uk]) 01:55:10 rigo has quit (Leaving) 01:59:11 rigo has joined #rdfig 02:00:37 rigo has quit (Leaving) 03:23:26 DanC_lap has quit (Ping timeout for DanC_lap[adsl-208-190-202-227.dsl.kscymo.swbell.net]) 03:49:36 * DanC studies mathematica notebooks of this goedel stuff... http://www.math.gatech.edu/~belinfan/research/autoreas/goedel/nb/ 03:53:57 rigo has joined #rdfig 03:54:18 rigo has left channel 04:20:29 jonb has joined #rdfig 04:25:47 jonas has quit (Ping timeout for jonas[jonas.rit.se]) 04:47:11 anybody wanna buy me a copy of this thing by godel? http://www.rarevols.co.uk/pages/00001178.htm 04:47:18 GÖDEL, Kurt. 04:47:18 The Consistency of the Axiom of Choice and of the 04:47:18 Generalized Continuum-Hypothesis With the 04:47:18 Axioms of Set Theory. 04:47:31 No. 04:47:34 Sorry. 04:47:45 Springer is have a big ole computer book sale though. 04:47:54 Lots of good stuff ripe for the picking. 04:47:55 btw... did you say that induction isn't 1st order? 04:48:08 math induction is second order, yes. 04:48:18 Standardly. 04:48:47 (A real good pick is "Logic for Applications" by Nerode and Shore, I've seen it for 20 bucks off) 04:48:48 I'm getting conflicting reports. Boyer showed me a bunch of work based on a finite axiomatization, in 1st order logic, of set theory, which is bigger than number theory. 04:49:41 i.e. work that shows that induction is 1st order; based on Goedel's work circa 1937. 04:49:45 Well, I do remember, perhaps wrongly, about some manevears involving schemata. 04:49:51 Yeah, but I'm sceptical :) 04:50:07 I wouldn't beat up on someone for saying it's 2nd order ;) 04:50:23 he said the myth that induction isn't 1st order is, regrettably, widespread to the point of orthodoxy in the AI community. 04:50:36 And the logic community. 04:51:03 I'm still skeptical. 04:51:16 well, I count at least 4 logicians, today, that say/know that induction is 1st-order. 04:51:57 A finite axiomatization in first order logic, of set theory, which is bigger than number theory, woudl entail that number there is complete, yes? 04:52:03 does this stuff address your skepticism? http://www.math.gatech.edu/~belinfan/research/autoreas/ 04:52:38 er... the axiomatization isn't complete. no consistent formal system is. 04:53:00 1st order prop and predicated calculi are. 04:53:08 Ie.., both consistent and complete. 04:53:18 *many* formal systems are. 04:53:28 No formal system as strong as number theory is. 04:53:39 sorry... no system including number theory is complete and consistent and recursively enumerable. 04:53:50 ... is what I meant to say. 04:53:54 Ah! 04:53:57 I was wondering :) 04:54:24 No, your url doesn't address my skepticism as far as I can tell. 04:54:29 Is there some bit I should be looking at? 04:54:43 anyway... I haven't found the proof of induction in the set theory cited above yet, but I have good reason to believe it's there (or doable). 04:54:44 "Belinfante's interest in computer assisted theorem proving has mainly concentrated on obtaining proofs of theorems in set theory using the first-order logic resolution-style theorem prover Otter" 04:54:56 That doesn't show that induction is first oder. 04:55:14 That jsut shows you can prove theorems of set theory using first order logic. 04:55:19 Which isn't the same, yes? 04:55:34 well, if induction is a theorem in set theory (which it is), then Q.E.D., no? 04:55:41 No. 04:55:44 ? 04:55:54 SInce they don't say they can prove *all* theorems of set theory. 04:56:08 For example, there are lots of theorems that I can prove without the axiom of choice. 04:56:29 That doesnt' make the axiom of choice superfluous. 04:56:37 er... of course they can prove all theorems in their set theory; that's what "theorem" means, no? something provable. 04:56:56 Re read it. 04:57:07 They are concentrating on obtaining proofs of theorems... 04:57:22 right, the axiom of choice is independent of a lot of stuff. But induction is entailed by well-foundedness in traditional set theory (which these guys have captured using a finite number of axioms). 04:57:23 They don't have a proof that their system is strong enough to prove *all* theorems. 04:57:38 That's the claim I don't see on that page. 04:57:42 Where are you reading that. 04:58:04 (I see 'otter" proofs which I don't know what they are.) 04:58:42 I guess the question is how strong this finite axiomization of Goedal's is. 04:58:55 hang on... there's a difference between: a theorem is provable (which is true by defintion) and: otter can find the proof. They make no claim that otter can find all proofs. But they do claim that all set theory theorems (including induction on numbers) are theorems in their finite axiomitzation. 04:59:12 That's the claim I haven't see. 04:59:21 how strong Goedel's: exactly; hence my desire for a copy of the 60UKP book. 05:00:27 I take that to be the interesting point. I confess ignorance with a skeptical bias. 05:00:46 a safe position ;-) 05:01:26 I think I'll call it a night. ciao. 05:02:02 See ya. 05:10:29 Aha! 05:10:35 http://www.edge.org/3rd_culture/huberdyson/huberdyson_p4.html 05:10:35 L: http://www.edge.org/3rd_culture/huberdyson/huberdyson_p4.html from bijan 05:11:02 L:|Something concrete to argue about wrt mathematical induction as second order. 05:11:03 titled item L 05:11:19 L:"PEANO ARITHMETIC, PA, is the first order approximation to second order arithmetic obtained by replacing P with the following schema of infinitely many axioms..." 05:11:20 commented item L 05:11:51 * DanC was drawn back by nagging curiosity... 05:11:55 L:Ok, that's what I remembered. 05:11:55 commented item L 05:12:07 yes, it's routine to do induction with infintely many axioms. 05:12:08 PA isn't what you mean, of coruse. 05:12:27 (Yeah, I mean that for jon.) 05:13:07 http://citeseer.nj.nec.com/context/333730/0 05:13:07 M: http://citeseer.nj.nec.com/context/333730/0 from DanC 05:13:13 What I want is a statement of induction that doesn't quantify over properties. 05:13:34 M:|Set Theory in First-Order Logic: Clauses for G¨odel's Axioms. 05:13:34 titled item M 05:13:57 quantifying over properties isn't necessarily outside of 1st order. (another bit of orthodoxy/myth) 05:14:24 Er..what do you mean by first order? 05:14:50 er... fair question; the answer doesn't leap out at me. 05:15:06 I think the defintion is in terms of model theory. 05:15:16 Ok, I'd like to see a statemetn of mathematical induction that doesn't involve predicate place variables, with attendant quantifiers, and finite axioms. 05:15:44 yes, well, that's what M is, I think, if I could get a danged copy of it in front of me. 05:16:17 the otter stuff is an encoding of the M article, but I don't grok the language it's written in; I'm after a prose description. 05:17:23 Note that I've still not seen a claim about the strength of M on the otter page. 05:17:31 But it's late :) 05:17:48 Erhm...how is that article you chumped useful? 05:18:01 it's not, I guess. 05:18:28 Yeah, thus far, all I'm seeing is fragments. 05:18:51 I.e., how to translate some bit of higher order into first order to make them tractable for theorem provers. 05:19:57 well... the claim is that Goedel's algorithm covers the general case of converting set theory to 1st order. 05:20:32 But where's it made? 05:21:57 I mean, if it's a myth, there should be a nice proof to the contrary. 05:22:27 * bijan notes that it's a SW myth that second order means non-decidable. 05:22:47 Or that non-decidability is (necessarily) a huge negative. 05:26:14 "Every mathematical entity...can be construed inone way or another to be a set. Therefore only one kind of variable is necessay, for a set of sets is itself a set. Not only can set theory be formultaed as a first order theory, but as a finitely axiomatized 05:26:19 theory." 05:26:27 Ok, my skepticism is somewhat shaken :) 05:27:43 where did you find the "Every ..." excerpt above? 05:28:03 Margaris, *First order mathematical logic* 05:28:14 this paper that cites M covers a lot of the relevant ground: 05:28:16 http://citeseer.nj.nec.com/kerber94translation.html 05:28:17 N: http://citeseer.nj.nec.com/kerber94translation.html from DanC 05:28:41 N:|On the Translation of Higher-Order Problems into First-Order Logic (1994) 05:28:42 titled item N 05:29:42 N: cites Boyer et. al. "Set Theory in First Order ..." and covers a lot of surrounding ground. 05:29:43 commented item N 05:31:27 M: Boyer, Lusk, McCune et al. - 1986 05:31:27 commented item M 05:31:48 I love using chump as an annotated bibliography. 05:32:27 Yep. 05:32:32 I'm glad you like it. 05:32:34 I hope chump will start grokking more structured comments, ala: M: dc:author "Boyer", "Lusk", "McCune"; dc:date "1986" 05:32:40 Augmented chump coming mond... 05:32:47 Heh. 05:32:59 You're behind the curve! :) 05:33:09 you hack chump? 05:33:12 Your hopes are being fulfilled before you hope them. 05:33:25 have we met, by the way? I'm Dan Connolly. http://www.w3.org/People/Connolly/ 05:33:28 I wrote the original and am working the the py version right now. 05:33:37 No, not other than on irc. 05:33:38 I'm hector. 05:33:52 ah yes... Raleigh NC, right? 05:33:59 Durham, yes :) 05:34:12 I wasn't sure about the spelling anyway. 05:34:38 "Hector" was a joke on edd and when people started probing it made sense to switch to my real name ;) 05:34:48 so... wanna tell me the story of how you come to be connected to #rdfig, xmlhack, and chump? 05:35:04 I write for Monkeyfist.com 05:35:11 We have an irc channel. 05:35:29 I was tired of us losing good stuff and too lazy to bother with logs. 05:35:59 I was writing a chapter on Squeak networking for an anthology, and my co-author submitted a little bullatin board bot. 05:36:08 I hacked it to start saving urls. 05:36:20 Then kept evolving it. 05:36:27 Edd hangs out on #mf. 05:36:42 He and his pals at usefulinc wrote a clone in python (the chump) 05:36:54 Edd set up one for rdfig. 05:37:05 Rael started backing down from rdf in RSS1.0 05:37:08 I came over to argue with him. 05:37:11 The rest is history ;) 05:37:33 (I also write for XML.com...see my FP & XML and a forthcoming series on SWI-prolog and rdf) 05:37:43 And I'ma phil grad student at UNC-CH. 05:37:44 eikeon has joined #rdfig 05:37:49 DId I leave anything out? :) 05:38:34 I'll never know. ;-) 05:44:07  05:44:11 eikeon has quit (Leaving) 05:45:15 eikeon has joined #rdfig 06:12:25 A: 06:12:25 http://www.w3.org/2001/03swell/ 06:12:26 Logic Meets Communication Protcols: HTTP, MIME, XML, and RDF specified in RDF/n3 07:28:47 rigo has joined #rdfig 07:29:24 rigo has left channel 07:56:18 rael has joined #rdfig 07:56:46 * rael waves 08:02:11 [GlobalNotice] Morning all. The end of the week has arrived; please check the news and information on http://openprojects.net/ .... have a good morning and a great weekend! :) 08:05:37 eikeon has quit (Leaving) 08:38:10 rael has quit (Leaving) 09:17:38 danbri has joined #rdfig 09:26:50 * danbri catches up; is intrigued to hear about bijan's SWI-Prolog writing plans 09:30:01 dajobe has joined #rdfig 09:39:05 * dajobe has caught up 09:39:07 http://www.ia.com/ia/product/inbrief/BRF_RDF_UDDI_DRB.htm 09:39:07 O: http://www.ia.com/ia/product/inbrief/BRF_RDF_UDDI_DRB.htm from dajobe 09:39:23 O:|IN BRIEF: UDDI and RDF - from Information Architects 09:39:24 titled item O 09:39:44 O:slightly product focussd, but some thoughts on relationships 09:39:44 commented item O 09:39:51 -- from my inbox 09:47:09 bwm has joined #rdfig 09:47:25 http://www.ia.com/ia/sc_rdf.htm "SmartCode, the 1st and only RDF server", sigh. "builds applicatiosn with the power of metadata nodes", giggle... 09:47:25 P: http://www.ia.com/ia/sc_rdf.htm from danbri 09:48:08 yeah I saw that few months ago.... 09:49:21 ... because they linked to my RDF guide 09:49:43 I knew about the work; hadn't seen the PR bumf until now. 09:49:50 Just spotted that. 09:49:53 * dajobe danbri knows all 09:54:58 O:| IA and RDF other article [http://softwaredev.earthweb.com/mobi/article/0,,12067_630701,00.html|Java, RDF, and Wireless Computing] by their CTO Leon Shklar 09:54:58 titled item O 09:55:04 O: IA and RDF other article [http://softwaredev.earthweb.com/mobi/article/0,,12067_630701,00.html|Java, RDF, and Wireless Computing] by their CTO Leon Shklar 09:55:04 commented item O 09:55:09 O:|IN BRIEF: UDDI and RDF - from Information Architects 09:55:09 titled item O 09:58:37 * danbri stumbles across a post by David McCusker; is reminded that there's an ILRT mirror of the IronDoc site somewhere 09:58:39 * danbri rummages 10:00:09 Old old old... http://ilrt.org/rdu/mirrors/irondoc/ -> http://ilrt.org/rdu/mirrors/irondoc/irondoc/irqanda.htm#q-xml-rdf-fe-engine "Q: Might IronDoc be used as a storage engine to support RDF semantics?" 10:01:28 This was nice work; based on Bento, the structured file format from OpenDoc. I used to like David's posts on the opendoc usenet group, and later mozilla-rdf. Wonder where he's gone... 10:02:04 according to http://www.best.com/~mccusker/ - vaciation then startup stealth mode 10:02:27 So that's funny; I search for 'mccusker irondoc' on Google; "Hack the Planet Prime: I miss David McCusker's Posts!" is the first hit. 10:07:37 ta. Hmm, he did leave Netscape, then. But IronDoc still seems to exist in some form: " 10:07:37 The fractal blob code I'm writing now is the IronDoc blob system." http://www.best.com/~mccusker/map/ti/new.htm 10:31:46 Still can't connect to Broekstra's Sesame site http://sesame.aidministrator.nl/ 10:43:38 bwm has quit (Ping timeout for bwm[phobos.hpl.hp.com]) 10:54:31 danbri has quit (Ping timeout for danbri[modem-210.hoary-redpoll.dialup.pol.co.uk]) 11:11:06 danbri has joined #rdfig 11:20:46 bwm_ has joined #rdfig 11:22:01 edd has joined #rdfig 11:27:53 * danbri wonders what shared bookmark feeds in RSS should look like - http://ilrt.org/discovery/demos/RSSWeb/newitems-daniel.brickley@bristol.ac.uk/ ? 11:29:06 bijan has some thoughts on this. 11:29:16 he'll be awake in a few hours ;) 11:29:35 * edd is going to plant a rose in the garden... 11:29:46 Is that a euphemism? 11:29:53 * danbri grins, ducks 11:30:04 nope, it's actually what i'm off to do. 11:30:27 * edd desperately tries to Reclaim Life. later... 11:30:40 see you! 11:31:56 can anyone access www.w3.org at the moment? Something between me and it is failing. 11:32:17 It's you 11:34:06 I can access other sites outside the HP firewall 11:34:24 Could be sunspots - http://news.bbc.co.uk/hi/english/sci/tech/newsid_1252000/1252651.stm 11:36:04 works for me. Depending on your IP address I tihnk you'll be getting a different server from dns lookup though. 11:36:38 I get www.w3.org -nslookup-> 18.29.1.34, 18.29.1.35 11:37:38 18.29.1.35, 18.176.0.26, 18.29.1.34 11:38:09 works now - 11:38:48 from w3c homepage, [[30 March 2001: The XML Schema Working Group has updated the XML Schema Proposed Recommendation, 11:38:48 restoring the name 'decimal' to one datatype]] 11:39:09 news to me; wonder what happened there, whether daml+oil needs updating to match... 11:39:12 can't they add a '#' to their namespace? 11:42:04 "To recognize whether a node is an XML 11:42:04 Schema datatype definition it is sufficient to determine whether it is within an XML Schema context." (http://www.daml.org/2001/03/model-theoretic-semantics.html) 11:42:25 I'm trying to understand what "an XML Schema context" amounts to. 11:44:09 * edd adds several layers of indirection, just for good measure 11:47:11 to the rose? 11:47:17 go away, shoo! ;-) 11:47:51 * edd done. roses don't take so long to plant. 11:48:06 * edd was just thinking that to get all w3 specs to interoperte we might need several layers of indirection. 11:49:02 Oh, well in that case, how about you help me understand how to explain http://www.daml.org/2001/03/axiomatic-semantics.html and http://www.daml.org/2001/03/model-theoretic-semantics.html to good ole-fashioned web developers...? 11:49:17 * edd rotfl 11:49:45 * edd suggests you don't even try 11:49:56 Just skimming http://www.w3.org/TR/2000/WD-CCPP-struct-20000721/ note XML mistakes and odd/inconsistent RDF - must make a comment about this 11:50:30 My hunch is that the axiomatic doc will be the more readily digestible (perhaps as it has prose for each rule) 11:51:28 hmm, but to play the DA, why would anyone bother. 11:51:43 after a week or two of sitting here i don't see what i wanted to see yet 11:51:49 RDFCoreWG needs one of these - CC/PP Terminology and Abbreviations - http://www.w3.org/TR/2000/WD-CCPP-ta-20000721/ 11:51:57 consider: .... if in IR(?P) then x in IC(?C) 11:52:25 versus: 11:52:28 Ax48. (<=> (PropertyValue domain ?p ?d) 11:52:28 (and (Type ?p Property) 11:52:28 (Type ?d rdfs:Class) 11:52:28 (forall (?x ?y) 11:52:28 (=> (PropertyValue ?p ?x ?y) (Type ?x ?d))))) 11:54:26 RDFS should have had this meaning for rdfs:domain, but for historical reasons the WG left it weak. I like the notion of writing bits of RDFS more formally, but I can't imagine having (say) both these docs as appendices would endear us to WWW developers. 11:54:52 So we need to understand the role of these different formalisms... 11:56:18 edd: "after a week or two of sitting here i don't see what i wanted to see yet" 11:56:24 what do you mean? 11:57:47 well, a bit overdramatic, mebbe. i've seen some useful work done, but a lot of odd noise. a lot of folk just seem so far way from wanting to reach tangible needs. 11:58:00 maybe i'm feeling a bit tired or something. 11:58:51 unfair to say anything i guess as the channel's only just been announced. :) 11:59:23 Well, it has taken some of the chat off the mailing list, which I think useful in and of itself. Also we've got some neat tools assembled, which took some attention (both building, testing and their novelty value!). 11:59:50 The thing to do is think of some concrete projects... 12:00:12 i suggest we look at some existing open source stuff, web browsers, mail clients. 12:00:20 and see how we can hack SW-ness into them 12:00:53 for what we all lack in our experiments is decent sized globs of data 12:00:57 Bijan mentioned SWI-Prolog; the enabling enference hack might be a thing to revisit, for eg.? 12:01:05 RSS provides us with some, but we need more. 12:01:24 inference -- yes -- but *exposing data* needs to be considered as a first step! 12:01:25 You know my views on that. RDFWeb. 12:01:37 And you know my views on that, *prod*. 12:01:40 vapourware! 12:01:44 :) 12:02:28 So, I should try to reimpliment it over my latest perl hack. Today. 12:02:29 bijan's doing some nifty work making the chump understand N3 12:02:44 i think what could eventually happen is that the chump becomes some kind of knowledge acquisition engine 12:02:52 oh, cool. Maybe could also work on making me understand N3? 12:02:54 we already have command-line and HTTP-based interfaces to it as well. 12:03:10 That's the route all IRC bots seem to take... 12:03:11 heh, and me too while he's there. i'm not through with anglebrackets yet :) 12:03:28 N3 = XML2.0? 12:03:34 danbri: yes, but i'd like, frinstance, Mozilla to be able to throw stuff directly at a local chump. 12:03:55 that's why we did the HTTP engine. create a "bookmarklet" (which we have), and whoompf. 12:04:20 we actually did it for the purpose of feeding a palm pilot -- so you could browser around in Moz and click on stuf you wanted sent to your Palm. 12:04:38 in some parallel universe. Revisiting our shared bookmarks thing (above) I definitely see converged goal with Chump. My particular interest is w.r.t. discovery of those feeds, eg. I want to be able to find new items on logic bookmarked by RDFIG members this week. 12:04:49 have you seen "bookie"? 12:05:05 discovery's nice. i'd like to posit an RDF based alternative to UDDI :) 12:05:06 Libby hacked just such a thing; if you know the URL you can beam docs into her Palm Pilot :) 12:05:30 heh, i should tell mattb that 12:05:47 Well, since Mozilla's got built-in sherlock / screenscraper support, it's crying out for something like SWIPE. 12:06:04 I think it is time to do the SWIPE to Sherlock.src converter. Back shortly. 12:06:18 (re UDDI: exactly...) 12:07:48 BTW, w.r.t. effectiveness of this channel: http://lists.w3.org/Archives/Public/www-rdf-interest/2001Mar/0229.html (Jon/Sean + my old wordnet hack); the /Addressing URI schemes thing; the RSS feed from Chump... All stemmed from collaborations spawned here. 12:08:18 edd-win has joined #rdfig 12:10:31 * edd-win switches to windoze. 12:12:43 * baud hopes openoffice will become viable for that sort of thing by v6 12:13:38 * edd-win discoverd galeon last night 12:13:48 i highly recommend it to any GNOME user. stunningly fast UI. 12:15:25 I found an article (magazine) on creating GNOME apps via the Perl wrapper; looked really easy (and much prettier than the Perl/Tk things you see). I might have a go at that sometime. 12:16:29 danbri: Simon Cozens has some good stuff on perl.com about that too. 12:17:05 I would have a go at an RDF Gnome/KDE app. if I had a clear goal - *what it did* 12:17:10 edd-win: [off] My problem with galeon was it never worked with the version of mozilla I had installed. 12:17:48 thanks for the tip; hmmm perl.com and www.perl.com are different. 12:17:57 http://www.perl.com/pub/2000/10/gnome.html 12:17:58 Q: http://www.perl.com/pub/2000/10/gnome.html from danbri 12:18:04 danbri: there's a whole *very amusing* story about that :) 12:18:51 Q:"GNOME is the Unix desktop. It's a framework for writing graphical applications with Unix, providing drag-and-drop, interapplication 12:18:51 commented item Q 12:18:51 communication, CORBA components (what's called ``OLE'' in the Windows world) a standard, good-looking interface, and all the other 12:18:51 features that you'd expect from modern graphical applications." 12:19:17 Q:|Programming GNOME Applications with Perl, Part One 12:19:18 titled item Q 12:19:34 odd, that page/article isn't signed. 12:25:50 * danbri is installing Perl Gnome module in the background; fingers crossed I'll have something to show for this shortly. 12:29:37 Yes! It worked out of the box; I type 'install Gnome' wait a while, next thing a test GTk/Perl window pops up! 12:29:48 computers rule! I take back everything I've said about them. 12:39:12 computers rule *you*. i'll agree with that. 12:42:05 CPAN had the wrong version; gah. Computers suck. http://www.perl.com/pub/2000/11/gnome2.html -> http://projects.prosa.it/gtkperl/ for info if anyone's following same path. 12:42:43 * edd follows the apt-get path 12:43:34 bwm_ has quit (Ping timeout for bwm_[deimos.hpl.hp.com]) 12:50:40 * dajobe has wondered about putting Redland in perl CPAN 12:51:06 * danbri would like to see that 12:51:32 * edd-win would like to see some kind of RSS or RDF API find its way into many open source apps 12:51:50 dajobe: exampole. the RSS support inside Evolution is based on a pretty naive C processor. 12:51:57 edd: why do you think redland installs to a librdf.a then? 12:52:00 if C-based parsing could slip into more projects... 12:52:24 maybe just need to do a patch for one or two apps to use librdf... 12:52:33 wonder if we could rewrite the rss support in evolution using it, for example. 12:52:39 sure, investigating evoluation source was a todo 12:52:48 the hurdle seems to be getting to grips with the bonobo framework. 12:52:51 seems much more dooable than delving into mozilla 12:53:03 mozilla: agreed! 12:53:09 which I have tried 12:53:22 me too. it's a sorry jumbled experience. 12:53:54 danbri_ has joined #rdfig 12:54:07 danbri has quit (Ping timeout for danbri[modem-170.white-faced-ibis.dialup.pol.co.uk]) 13:43:17 jonas has joined #rdfig 13:44:11 * edd forgot to say earlier, w.r.t galeon -- it has "full screen" mode. this is the thing i miss when doing presentations with mozilla, it has no fullscreen. 13:44:30 galeon is mozilla based, right? 13:44:37 * danbri_ waves at Jonas 13:44:55 it uses the layout engine, yes. so for the moment you need moz 0.8 installed in order to run galeon 13:45:37 i believe work will eventually happen to unbundle the relevant libararies. 14:16:46 Just noticed Uche (4thought) is attending W3C web services workshop http://www.w3.org/2001/03/wsws-program - wonder who is there from SW activity? 14:17:08 ericp is going. 14:17:09 TimBL I think. 14:17:30 I was thinking of emiller since whole 1st afternoon is down for breakouts and SW is one of them 14:18:02 emiller can't be there i think 14:18:12 I know May was bad for him :-) 14:18:42 something to do with kids or somesuch :) 14:37:53 dajobe is now known as dajobe-out 14:44:37 danbri_ has quit (Ping timeout for danbri_[modem-185.achilles-tang.dialup.pol.co.uk]) 14:46:50 sbp has joined #rdfig 14:56:24 sbp has quit (Bye, bye!) 15:03:58 bitsko has quit (Leaving) 15:06:48 edd-win has quit (Ping timeout for edd-win[gravling.demon.co.uk]) 15:13:28 danbri has joined #rdfig 15:14:37 danbri has quit (Read error to danbri[modem-98.rottweiler.dialup.pol.co.uk]: EOF from client) 15:59:05 * jonb waves 15:59:38 * jonb is doing anything possible to avoid writing up my Extreme Markup abstract 16:02:28 * edd loads up his palm with plenty of GNOME app development documentation 16:02:53 * jonb thinks edd must have a big palm 16:03:04 8MB :) 16:04:59 for my talk: is there _something_ that necessitates RDF in RSS 1.0 i.e. that cannot be done in 0.91? 16:05:06 i.e. a killer feature 16:05:26 well, from my p.o.v.: 16:05:32 right :-) 16:05:33 * a clear path to extensibility 16:06:41 * a known structure for i18n 16:07:13 * no need to concoct custom object model, just use the RDF model. 16:07:23 object model? 16:07:49 the latter point is really w.r.t. using RDF tools to process it, you get more for free, and have to write less code. 16:07:59 basically RDF as a better XML 16:08:03 Right. 16:08:16 At the end of the day there's no *Killer* reason, it's just a sensible step. 16:08:30 It didn't have to be done. We could have used an XHTML m18n approach :-) 16:08:40 right. i am always looking for take home points. 16:08:44 mn, i'm sorry. 16:08:56 12 :-) 16:09:43 RSS and my healthcare stuff are the two things I can talk about 16:10:48 tools are the most important reason for someone to use something 16:11:30 and I really think this chump thingie is on to something 16:11:32 Indeed. The most important step forward RSS made, imho, was to break out of the weblog mindset. 16:11:54 The potential for extensibilty caused the scales to fall from eyes, as it were 16:12:44 * jonb thinks its unfortunate that XML interoperability isn't a higher priority for most RDF people 16:13:28 else I'd have much more to talk about 16:14:03 hmm, mail on mozilla-rdf list. someone moved bookmarks into LDAP, via DSML. 16:14:08 intriguing. 16:14:29 uses dublin core too. 16:15:10 right DC along the lines of: someday there will be a better search engine 16:15:34 the problem is that the XML community pitched that to the web community and so far... 16:16:23 you understand the level the xml devcon talks are at. 16:16:41 step 1. 16:16:43 step 2. 16:16:44 step 3. 16:16:55 repeat, lather, rinse, buy my product. 16:17:15 * edd is unfair for a few sparkling seconds 16:17:24 RSS is an easy sell in that context 16:17:45 RDF *should* be. 16:17:46 "Ever wonder how those newsfeed thingies work?" 16:18:06 "I always did until I learned ... ***RDF***!" 16:18:20 (heads blindly nod) 16:18:21 * edd considers that part of the new Semantic Web Advanced DEvelopment stuff should be the compilation of an RDF Resource Kit. 16:18:30 RPMs, .debs, windows installers plus docs. 16:18:31 and examples. 16:19:12 Actually the groups usually do a _great_ job at understanding RDF 16:19:27 "the groups" ? 16:19:55 what the devcon audience wants to know is how they can use RDF on a daily basis in their own projects 16:20:03 indeed. 16:20:16 i.e. go here and download this and ... 16:20:32 touch and feel stuff 16:21:05 or "how can i integrate RDF into my XML database?" 16:21:32 or "why would i ..>" :) 16:21:40 right. 16:22:00 not: "how can I rearchitect my entire project around RDF" 16:22:31 "cause i'd get fired with people laughing me out the door" 16:22:31 but you could have "If starting a new project with characteristics X Y and Z, could I use Redfoot?" (or Redland, or whatver) 16:22:41 of course :-) 16:22:51 James Tauber has a great thing with Redfoot, it's not public enough yet. 16:23:15 i've been meaning to give him a call, he's only 40 minutes from me 16:23:23 you should do. 16:25:06 i'm trying to resist the strong temptation to learn Py -- simply because i'd start rewriting my old stuff and get even less real work done. 16:25:21 lol 16:26:06 do you go to Extreme? 16:26:48 I'll be there this year. 16:26:57 or know if the audience has any real SW interest? 16:26:58 It'll be my first. If it's good engouh, it won't be my last. 16:27:26 I imagine it'd be a good topic. After all, TMs are part of the SW, right? ;-p 16:27:31 submitting a RDDL talk but deciding which way to fill it out 16:27:36 I'd love to see some SW there. 16:28:27 i think some of the TMs have a real possibility to be brought on board 16:28:48 the biggest hurdle is actually all the questions around URIs and resources 16:30:59 I'll be using SW as anything on the web that can be sensibly processed my machines 16:31:20 rather than human attached browsers 16:32:56 * jonb is going back to writing 17:22:51 SethR has joined #rdfig 17:30:36 danja has joined #rdfig 17:38:17 bijan has quit (Leaving) 17:38:49 Aaron - I nearly forgot : Sweeties kindaLike Candy 17:42:44 hey after i read the rss feed of chump into my sembrowser i dawned on me that 17:43:03 RSS **is** the low laying fruit of shared bookmarks 17:43:42 if i read the RSS feed in every day, i'll have those bookmarks, and so does 17:43:47 everyon else that does the same 17:44:19 bijan has joined #rdfig 17:54:08 eikeon has joined #rdfig 17:55:14 hey eikeon! 17:55:22 Hi there! 17:55:26 congrats on redfoot progress. 17:55:40 i'd like to write a bit on xmlhack for it but i couldn't find quite enough detail in your latest note. 17:55:46 Thanks... slowly but surely 17:56:41 hey edd ... don't you think tha we need a way to classify our chumps into different contexts ? 17:56:55 SethR: umm, waht? 17:57:23 there is no way to say a particular entry should go into a particular context 17:57:35 There is no concept of context. 17:58:04 * edd doesn't quite understand what the request is. 17:58:09 well there in the RSS taxo module 17:58:21 aha, you want to attach topics to an entry. 17:58:29 yep 17:58:40 yes, we've talked about that. your best bet is to send feature request to chump@heddley.com 17:58:45 so all the developers see it. 17:58:53 so that i can use it to grow my bookmarks 17:59:24 it was a real rush when i was able to read the rss feed into my sembrowser and enhance my bookmarks automatically :)) 17:59:44 i hear rael's been working on a similar app. 17:59:45 but i need to be able to assign them into topics for future reference 18:00:01 hmm, everyone's classification could be differnet. 18:00:04 buy don't yo need to put that into our chump syntax 18:00:19 xgavin has joined #rdfig 18:00:22 New chumpbot on the way. 18:00:25 doesnt matter 18:00:28 Uses a subset of N3 18:00:51 Let's you say :A rss:whatever "blah blah" 18:01:50 cool! real kewl .... are you gonna make a working example ? 18:02:00 * timbl excited about the idea of an n3-subset chumpbot. 18:02:07 What's a working examle? 18:02:23 chump something in n3 18:02:57 Welll...I'm hacking the bot so that it understands a subset of N3. 18:03:04 It's "on the way" 18:03:13 Ergo, I'm making a working example? 18:03:14 sorry i was too anxious 18:03:25 dajobe-out is now known as dajobe 18:03:30 Yes, don't pester and it'll be here by monday. 18:03:32 it'll keep a set of prefixes - and start with some defaults? 18:03:36 Yes. 18:03:47 dc:, rss:, rdf: built in. 18:04:03 then who knows if we do n3, is Semenglish too far off ? 18:04:26 Yes, it's too far off ;) 18:04:47 i guess you get scared just like Aaron .. huh? 18:05:18 Since I don't know how arron gets scared, I can't confirm. 18:05:29 danbri has joined #rdfig 18:05:33 My next project is AnnoteaBot. 18:05:46 * danbri arrives just in time :) 18:05:52 what language is the bot written in? 18:05:58 Which version? 18:06:03 latest 18:06:13 Origianlly in Smalltalk. The chump is a clone in python. 18:06:38 so if i provide the parser for semenglish, could you plunk it in? 18:06:40 ahem, with a few neat bits :) 18:06:50 And a few missing bits! :) 18:07:07 well, the HAIKU directive isn't needed in this context :) tho it's pretty neat, i'll grant. 18:07:11 SethR depends on your budget :) 18:07:15 And the limerick??!! 18:07:20 What about wordnet! 18:07:34 you can reintroduce wordnet with danbri's fancy rdf wrappings 18:07:47 Or, for that matter, with wordnet.py. 18:08:10 So, aside from foo, what's extra? 18:08:10 wrapping the wordnet executable more likely to prove fulfilling; my wrapper only does the nouns; lots of the fancy searches are more useful. 18:08:31 danbri: there's an excellent python module for dealing with wordnet. 18:12:52 s/on/only/ 18:13:11 Ok, a teaser: 18:13:12 >>> x = dswlb.DailyChump('C:\\PROGRAMMING\\PYTHON\\rdfchump\\chump-1.01\\src') 18:13:12 >>> x.process_input('Bijan', 'http://www.xml.com/') 18:13:13 ':A dc:relation ; dc:creator "Bijan"' 18:13:27 >>> x.process_input('edd', ':A My pride and joy') 18:13:28 'commented item A' 18:13:33 ^^^^^needs fixing ;) 18:13:39 >>> x.process_input('edd', ':A') 18:13:40 ':A dc:relation \012{:A :comment """My pride and joy"""} dc:creator "edd".\012' 18:14:02 Sorry, the idle repl doesn't print nicely :) 18:14:07 But y'all get the idea. 18:14:17 * danbri is officially intrigued 18:14:19 * edd impressed. 18:14:54 What's the best environment for Perl IRC bot hacking? Infobot? 18:15:10 Something will be working by tomorrow or monday. I hope ;) 18:15:19 can we have a db of prefixes and just add them at will? 18:15:36 Design is frozen. 18:15:44 But luckily for you, that's in the design. 18:15:54 kewl !! 18:16:20 do we add them in line or go somewhere else and add them? 18:16:50 OK, here's an issue: how can we enable our RDF/IRC bots to have some notion of authentication, ie. accept assertions from irc user 'danbri' and know that it's me? 18:17:00 inline would be {sru uriPrefix http://robustai.net/ns#} 18:17:25 No, I use @preix dc: "uri" syntax 18:17:38 So, "in irc line" but not "in statement line" 18:17:40 ok same thing 18:18:04 danbri: beeeeg issue. 18:18:21 There's no, afaik, good, truely secure way. 18:18:36 no, medium sized issue! as long as [oh right, agree] you don't want it reeeely secure. 18:18:59 Ah, well the simpliest is to have a nick-->user registry. 18:19:12 That's how I do it with DWLB. 18:19:25 I claim the Web is one of those. We just need the vocabulary to name channel / nick pairs. 18:19:29 you should chump that and it would end up in my bookmarks automatically :) 18:19:40 Chump what? 18:19:51 ie. I claim I'm danbri in irc://irc.opennetworks.net/rdfig but someone else might be danbri elsewhere. 18:20:08 Yes. 18:20:53 SethR: [off] means it doesn't get logged. bijan didn't want it circulating. but now its in the logs... 18:20:53 ("Not for general circulation" probably includes worldwide syndication via RSS ;-) 18:20:55 Seth: What part of "not for general circulation" and [off] don't you understand. 18:21:28 Does the logger (or any irc clients) understand [off] ? Is it a standard convention in IRCland? 18:21:40 dunno, I just invented it 18:21:48 * bijan now wonders how to get that *out* of the logs. 18:21:53 ask me nicely 18:22:02 but I don't like doing that 18:22:10 dajobe: Please undo seth's faux pas? 18:22:30 ok, keep quite for a bit 18:22:32 Logging turned off 18:22:58 Logging turned on 18:22:58 logger, hello 18:23:50 sorry i missed that 18:24:14 Well, no harm ultimately done :) 18:24:24 so anyhoo, I can ship logger's perl code for general abuse if people care enough 18:26:24 thunderous silence - obviously not, ok closed source it remains 18:26:25 * edd places N3 docs onto his Palm for later use, too. 18:27:11 I'd be interested in having a look at the basic wrapper, though really want a bot that allows arbitrary plugins to be independently authored (hence infobot appeals) 18:28:22 danbri: Yes. I've been considering various bot framework designs. The Chump code is pretty good (kudos to edd & co), though not yet totally modular. 18:28:41 I.e., it's still easier to modify it than drop in plugins (for me, anyway) 18:31:46 I'm thinking the same with my RDF stuff; I want it to load RDF rendering snippets that can provide views of certain namespaces (in text, html, ...). 18:32:00 If that gets plugged into a bot, becomes the same headache instead of two different ones. 18:32:11 yes, i've been wondering about those kind of modular architectures too 18:32:24 if there is enought interest, I could do loggerV2 now I know set of requirements it is likely to have 18:32:32 usual prototype, throw away game 18:32:46 in GNOME there's a central registry and a daemon. apps query the daemon for object servers that support particular interfaces. 18:33:10 i was wondering about some kind of rendering registry framework, where a module registers itself as being able to render certain things. 18:33:18 or whatever.... 18:34:09 a RDDL document? 18:34:54 more than that. needs something to handle registrations. 18:35:30 and a RDDL document describes a namespace. what about the inverse problem where a software module wants to say what namespaces (and actions thereon) it can grok? 18:38:34 edd: In my ears, it sounds like what Wraf does. :) 18:39:16 who would a software module register itself with, a central repository or another piece of code? 18:39:16 jonas: wraf is just perl, right? 18:39:31 GNOME: that's the old CORBA trader stuff done locally, right? 18:39:31 Label GNOME not found. 18:39:36 re GNOME: that's the old CORBA trader stuff done locally, right? 18:39:46 jonb: depends on the context. i've been thinking about how i'd integrate RDF throughout the desktop 18:40:26 create something like the windoes registry in RDF 18:40:27 danbri: right. been reading some of the docs. design seems quite clean. 18:41:39 So if you like that, edd, what's your beef with Web Services? Same difference with a bit of network lag in the middle...? 18:42:36 re 'rendering registry framework', heaven save us from another framework, but I think I know what you mean. 18:43:04 * edd has to run now, will save danbri's qs for later. 18:43:09 Currently in my perl I can register a function with an RDF node such that calling view( ) on the node subsequently gets a custom view of that node. 18:43:33 edd: Wraf is just Perl, yes. But it will interface with other services 18:43:35 somethings pinging a lot... 18:43:37 Question is just the query / modelling question of how best to characterise the classes of resource that functiion could also usefully render. 18:43:38 or an XML browser that automagically loads code to render some elements based on their namespace 18:44:07 e.g. SVG, MathML etc. 18:44:28 or RSS embedded in XHTML for that matter... 18:44:28 jonb: been done & demonstrated with HTML browsers and java, some WWW paper 18:44:46 * jonb knows this idea is not new :-) 18:44:50 * danbri wonders whether to dig out his old OpenDoc docs... (I guess MS COM attempted this too) 18:45:24 i'm interested in providing RDDL infrastructure for those tasks 18:45:30 e.g. RDDLClassLoader 18:47:04 for example the 'community' XSLT extension framework that Jeni Tennison and Uche Ogbiju are working on 18:47:18 So ('Byte Guide to OpenDoc' in hand), I claim there's a major difference between mixed namespace document rendering (SVG + MathML + CML + HTML...) and mixed namespace data rendering (RDF etc). OpenDoc was a brave stab at the former, where you have reasonable sized islands of specialised information for display... 18:47:24 s/Ogbiju/Ogbuji 18:47:29 shellac has joined #rdfig 18:48:11 i agree the two are different 18:48:44 the second is generally a transform into XHTML 18:48:47 danbri: My OD is super rusty, but I thought that the data model was orthoganally chunky. 18:48:58 * bijan is proud that he doesn't understand what he just wrote. 18:50:55 I think I know what you mean... big chunks of data that can be radically different, just making use of shared services (storage), UI widgets etc... 18:51:16 XML is a lot more integrated; different XML apps have more in common (and can be mixed more closely) 18:51:44 * bijan thinks of saying, "Bento"! "Structured storage" "something!" 18:52:00 I'm logging. Sorry, nothing found for 'bento' 18:52:00 logger, grep bento 18:52:05 I'm logging. I found 2 answers for 'Bento' 18:52:06 0) 2001-03-31 18:51:44 * bijan thinks of saying, "Bento"! "Structured storage" "something!" 18:52:07 1) 2001-03-31 10:01:28 This was nice work; based on Bento, the structured file format from OpenDoc. I used to like David's posts on the opendoc usenet group, and later mozilla-rdf. Wonder where he's gone... 18:52:07 logger, grep Bento 18:52:20 Ah, IronDoc. 18:52:34 Or FeDoc, or whatever it was last. 18:53:22 Indeed. Seems to be still hacking on it... 18:53:28 His efforts seem to have collapsed sometime in the past year, but perhaps they've revived. 18:53:49 * bijan no longer surfs in those circles. 18:55:19 danja has quit () 18:55:45 which circles? (I guess I surf the same circles but don't know what kind of circles they are) 18:56:15 The circles that include mccusker :) 18:57:03 Hmm. Going to http://www.treedragon.com/ it seems that "Biscut" is the thing du jour. 18:57:07 ns/opendoc/etc and dist-obj list and various Dave Winer sites... 18:57:26 Yes. I've been Winer free for over a year now ;) 18:58:00 danbri: re: querying -- the GNOME daemon seems to have evolved some kind of capabiltiies query lang for the purpsoe. 18:58:42 Edd: oh, now that's interesting (re web services); all the CORBA trader stuff had the crudest query matching capability last I looked. 18:59:14 remind me to find a URL at some point. 18:59:24 ...which was 2+ yrs ago, admittedly. It might be that most Web services expose over generic APIs and the structures of intrest are queries and responces 18:59:33 ASK/TELL again... 18:59:41 aka HTTP GET, HTTP POST 19:00:52 * bijan gives up on figuring out what McCusker is doing. 19:17:58 shellac has quit ([x]chat) 19:19:24 danbri has quit (Read error to danbri[modem-406.scissor-tailed-flycatcher.dialup.pol.co.uk]: Connection reset by peer) 19:22:56 SethR has left channel 19:24:42 SethR has joined #rdfig 19:24:58 SethR has quit () 19:36:09 bitsko has joined #rdfig 19:39:31 danbri has joined #rdfig 19:46:03 Gee... 900 rows debugging info just for getting one item from a container. :) 19:47:06 danbri: looked at infobot, not suitable for logger development (if it remains just a logging tool) 19:47:36 didn't work well as a daemon last I used it; had to keep a terminal window open. 19:48:08 http://freshmeat.net/projects/xmltv/ XMLTV is a set of programs to process TV listings and manage your TV viewing. 19:48:08 R: http://freshmeat.net/projects/xmltv/ from danbri 19:51:04 R: From a [thread|http://www.xent.com/feb01/1575.html] on [fork|http://www.xent.com/] 19:51:05 commented item R 19:51:29 that's been around for a while 19:52:50 It gets Listings for UK from Ananova; scraped presumably. 19:53:22 Since I got the Tivo, I had a renenewed interest in such things! 19:53:49 you don't need machine readable TV listings, since you have a machine that does that for you 19:54:45 eh? I surely do. Tivo's too clunky, and has a vague notion of series etc. 19:55:25 wonder how http://www.doc.ic.ac.uk/~epa98/work/apps/xmltv/latest-devel/xmltv.dtd compares to the IMDB model... 19:55:29 this is so [off], but anyway, you are thus still interested in esentially making a tivo from bits of hardware + software 19:56:24 Highly on topic; Tivo is one of the most interesting mass-market metadata applications out there. (and 'yes' to your question!) 19:56:42 We could talk about higher order logics and skolem functions if you prefer! ;-) 19:56:49 the metadata is hidden, so not very interesting 19:57:50 the metadata is hidden, so should be un-hidden in a way that motivates networked services to do likewise 19:58:27 the metadata (subscription) is also part of the $$$ model 19:58:50 Part of Tivo's business model, sure. But but but... 19:59:15 Imagine someone came along with another box that could talk to their server (but say it also did home management of MP3 and photographs). 19:59:52 sure 19:59:56 It would be in the business interests of Tivo-the-service-provider to allow that box to use their metadata feed, and to make use of their personalisation services etc... 20:00:44 Which means they'd need to open up the interface between box and networked service. Once that's done, it becomes a marketplace where consumers could switch to a better provider. 20:00:46 i feel they would sue you for "reverse engineering" something or change protocol, play AIM games which now have got very funky - passing MD5-hashes of AOL's executable as part of the protocol 20:00:53 And we get the portal stickyness battles all over again. 20:01:31 The won't change protocol if they want lots of boxes to be engineered to use their service. 20:01:57 It's like the CDDB situation; or, from a different angle, portal's after stickiness, locking in users. 20:02:16 yeah I was thinking of cddb 20:02:31 In the US there's already Replay TV as well; more doubtless to come. 20:02:46 I thought Replay was collapsing? 20:03:03 hadn't heard that. 20:03:59 hacking the hardware/protocol to break someones business model is happening a lot - DVDs, AIM, CueCat, CDDB, Furby 20:04:07 maybe tivo is next or http://www.xboxlinux.com/ 20:04:43 It could go the other way; opening the protocol could get them more users (if the collaborative filtering service is good) 20:05:15 I suspect they'd have made the box more open, if not for fear of it turning into the multimedia Napster. 20:06:03 Again this is very on-topic for RDF/SW, as it relates to the question of: Why are there no (visible, public) PICS Label Bureaus in widespread use? 20:06:08 the relationship between these hardware vendors and media companies is getting rather strained or close, I'm not sure which. See silly hard drive issues, DVDs again, maybe tivo had restrictions imposed by them too 20:26:18 danbri: is it worth fixing your wordnet thingie to generate well-formed XML when the word isn;t in the database? 20:26:42 it now does an 20:26:45 No. 20:27:12 It's a really unpleasant Perl hack. Better to generate a tree of .xml files from the source data files, or use one of the Wordnet APIs. 20:27:34 I do want to do that; but can't justify any more time fiddling with the old perl. 20:27:42 Unless you've got a good use for it, I guess. 20:28:37 it seems like the SW thesaurus is a good LHF app 20:29:13 yes, very true. 20:29:37 Is it? 20:29:37 sbp has joined #rdfig 20:29:39 i want to do this the very same way as UMLS ought be done 20:29:42 There was another WordNet RDF announcement; after mine, before the stanford one. Clever stuff; lost the URL though. 20:30:20 * bijan doesn't understand what a SW thesaurus is (as opposed to a non-SW thesaurus) 20:30:41 URI based thesaurus 20:30:43 a broad umbrella framework with sub-vocabularies cross-linked. 20:30:57 Er... 20:31:08 The UMLS works well within the medical domain; it assigns CUIs to medical concepts... 20:31:19 What's all that when it's in non-sw-savvy english? ;) 20:31:22 CUI? 20:31:35 concept unique id? 20:31:42 Some UMLS acronym; concept unique identifier. 20:31:59 yup. We use these in the catalogue records for the images in http://www.brisbio.ac.uk/ 20:31:59 i want to create URIs for those 20:33:05 for which the concepts are to be linked with the actual medical records defined in the ASTM/XChart format 20:33:31 bijan: the model they've used for UMLS takes a lot of independently developed thesaurus systems and maps them into a loose whole. 20:34:06 * danbri wonders if JonB knows the Galen folk 20:34:11 danbri: my ASTM stuff has a DICOM <-> XML mapping 20:34:23 * jonb has heard of it 20:34:35 danbri: Ah. So the advantage is a bigger thesaurus? 20:35:35 * jonb has looked through Biomed: cool! 20:35:58 that's the sort of stuff we are doing as well 20:36:05 jonb: You've out acronym'd me. bijan: no, the advantage is that medical info systems that want to use a thesaurus can use a common information model, regardless of dataset. 20:36:09 UMLS subsets well. 20:36:11 * bijan points out that LHF may not be worth plucking if it doesn't further goals (e.g., making SW tech popular) 20:36:23 http://www.opengalen.org/ OpenGalen ("Making the impossible very difficult") 20:36:24 S: http://www.opengalen.org/ from danbri 20:36:39 defn's ASTM: American Society for Testing and Materials 20:37:08 a standards org that issues medical specs including jonb's committee 20:37:30 defn DICOM: Digital Image Communication for Medicine 20:37:55 an ACR/NEMA standard for medical image transmission supported by imaging vendors 20:38:02 a tagged binary format/protocol 20:38:17 bijan: The goal would be the long delayed use of RDF for classifying things, in cases where the inbuilt RDF machinery (class hierarchies) doesn't match yr descriptive vocab. 20:38:36 Lots of people have taxonomies expressed in thesarus-like structures, broadterm/narrowerterm. 20:39:06 Ah, that's sounding more interesting. 20:39:21 Jon: re Biomed, I could probably expose all the records in text (maybe even XML!) for SW testbed. The images have constraints on their use; though perhaps not the thumbnails. 20:39:40 Though without tools for using rdf based classifications, it may be a touch moot. 20:40:13 i am writing/have written standard (as in ANSI/ISO) terminologies for healthcare 20:40:42 which incorporate other terminologies (e.g. ICD, SNOMED, HL7, CPT) 20:41:06 and snuck in RDF compatibility 20:41:20 have a nose around http://www.brisbio.ac.uk/ROADS/cgi-bin/thesaurus.pl (unpublished prototype) 20:41:40 search on heart 20:42:23 jonb: anything with a url to look at? 20:42:25 what needs to be done to bring this all together is assign URIs 20:42:42 . http://www.openhealth.org/ASTM/ 20:43:25 contains the ballot standards (DTDs) also some experimental/example RDFS 20:43:53 thanks. The Biomed thesaurus URL I just gave is the UMLS subset corresponding to terms that have associations to records in our image archive. 20:44:23 what would be really really cool is to expose these terms in the same fashion as wordnet 20:44:54 UMLS has some constraints on use, though they used to expose a CORBA service if I remember right 20:45:26 a bunch of my apps are behind firewalls 20:45:34 BTW TimBL's claim that all namespace URIs should have a # at the end concerns me, for large namespaces it makes it difficult to partion a la wordnet 20:45:39 yes there are constraints but only for certain vocabs 20:45:58 sandro_ has joined #rdfig 20:46:00 hmmm... 20:46:05 What about Mesh? 20:46:09 * danbri waves at sandro 20:46:30 dajobe has quit (barnes.openprojects.net adams.openprojects.net) 20:46:30 edd has quit (barnes.openprojects.net adams.openprojects.net) 20:46:30 sandro has quit (barnes.openprojects.net adams.openprojects.net) 20:46:30 bijan has quit (barnes.openprojects.net adams.openprojects.net) 20:46:30 timbl-mit has quit (barnes.openprojects.net adams.openprojects.net) 20:46:30 AaronSw has quit (barnes.openprojects.net adams.openprojects.net) 20:46:32 bijan has joined #rdfig 20:46:34 i think we agree that namespaces may have '/' its just the ones that end in alphanumerics that cause problems 20:46:42 danbri, according to the specs that is the only thing which is consistent. 20:47:22 consistent with what? This is your claim that things things in namespaces must be disjoint with HTTPResource? 20:47:27 timbl-mit has joined #rdfig 20:47:35 * timbl wonders at the giant exodus 20:47:47 Server trouble. 20:47:58 It was a network split between IRC servers, I think. 20:48:14 dajobe has joined #rdfig 20:48:21 ah, the dreaded netsplit 20:48:31 The HTTP implicitly says that http; URIs idenifiy works - abstract generic documents. 20:49:12 "implicitly says" is another way of saying "doesn't say". 20:49:38 * bijan moans at the pain of Python over Smalltalk. 20:49:38 No, it isn't 20:49:40 "works" can include all the works of the mind. We already all URIs in http:* to identify abstractions, and services. 20:49:55 dan: "doesn't explicitly say" :-) 20:50:07 Does the HTTP/1.1 spec detail the kinds of things that an HTTP server can be asked about? 20:50:16 What the HTTP protocoll allows you to do is to ask for a rendition of a work. 20:50:21 i think the namespace designer can use good judgement to partition terms 20:50:39 The spec says what the HTTP server can reply. "200 here is your document" 20:50:46 Well, I tried to partion the WordNet space in a way TimBL thinks should be illegal... 20:50:56 Here is a rendering of the work you asked about. 20:51:08 no. it is just partitioned into multiple works. 20:51:11 Note that PURL redirects such as the DC namespace can resolve to a URL with a # on the end 20:51:32 I think of works as the result of creative deeds; on that basis a physical object or abstract concept could be a work. 20:51:46 rigo has joined #rdfig 20:51:59 from a practical POV people frequently partition works into multiple HTML documents 20:51:59 Here is the problem. For example: When I pick up an HTTP document, I want to store the http header info whcih tells me about the response. Its subject is the document I tried to get/got. 20:52:01 Actually I don't care what the HTTP spec says, I'm looking for the easiest solution to some practical problems. 20:52:13 Agreed. 20:52:36 If dublic core maintains that is is that URI identifies an abstract idea of "title", I have a problem. 20:52:38 Oh, the subject is the work, the rendering is a manifestation that you got sent over the network. 20:52:57 This isn't about DC, really. 20:53:00 Right. 20:53:02 Why the problem? 20:53:03 Well, Dublin Core can then point that abstract idea at their current "server truth" 20:53:09 No this is not about DC, it is about anything. 20:53:16 No, but the problem that applies to DC applies to many 20:53:22 exactly 20:53:55 Anything which uses the same URI for a document about Dn and for Dan himself. 20:54:12 You can't use an HTTP URI for Dan because HTTp can't return you Dan. 20:54:21 in the exact same way that UMLS contains multiple terminologies, any terminology can be partitioned into multiple namespaces 20:54:28 that's in HTTP 1.2 20:54:31 :-) 20:54:41 But URI strings are opaque: you can use them to identify Dan within a certain context 20:54:43 Now, if we had a "277 Here is some stuff ABOUT what you wanted" then we could have abstarct things. 20:54:45 So, there is a subset of all URI name-able things that you're happy giving http://* names to. 20:55:06 277, neat idea... 20:55:12 sbp: Strings are opaque, but the URI spec implies that they represent generic documents. 20:55:30 Aside from that, there is the architectural quirk that HTTP content negotiation, lang neg etc allow the same URI to be associated (in complex ways) with 'concrette' documents at various stages of their lifecycle. 20:55:31 "resources", yes? 20:55:55 difference between resource and entity 20:56:07 I would not call that an architectural quirk. 20:56:14 jon: well, network entity then... :-) 20:56:24 the document is the entity, no? 20:56:28 It is the only way you can acutally reply to a message with real bits. 20:56:30 "200" can't mean here _it_ is. "it" is a view, rendering, snapshot, serialisation "of" the thing named. 20:56:52 Yes, a rendering of the document. 20:56:53 What kind of thing is http://www.w3.org/Icons/WWW/w3c_home 20:57:12 it's the concept of the W3C logo, which is time-variant 20:57:14 That is the icon of the W3C. 20:57:15 A work; something someone made? On which date? (the gif, the png...) 20:57:18 is the relationship between an entity and the resource it represents formally defined? 20:57:26 IT used to be green, now it is blue... not the same thing; the same concept 20:57:30 The W3C logo. 20:57:30 An abstraction, not directly representable in bytes. 20:57:36 yes 20:57:57 An sbtarction, but represntable in bytes, knowing that most represnetations lose something. 20:58:08 "rendition" tends to give a btter feel for that loss. 20:58:08 So for us, right here right now: dc:title "The W3C Logo" . 20:58:21 yup 20:58:25 There are two or three different bags-of-bytes, each of which represents manifests some hidden work we're not concerned with here. 20:58:37 (eg. someone had to do something to make the SVG one...) 20:58:56 And languge negotiation, changes over time... Regardless, that URI names the abstraction. 20:59:00 (Yes, CharlesMcCathie Neville I think) 20:59:15 what can we say about namespace URI ending in '/' vs '#" ? 20:59:46 I think a namespace URI should be allowed to end in any character permitted by the URI specification 20:59:47 I'm claiming that the relation to the bytestream rendering (which might be called --http1.1:200-->, perhaps) shows that we're happy with HTTP naming abstractions. 21:00:29 We could model just how happy we are with a resource returned w.r.t. the concept 21:00:32 And if so, it's a slippery slope between things that can be 'rendered' into a bytestream, versus those that can only be 'described'. A distinction I believe we can't do more than handwave around. 21:00:46 ok: lets assume for the sake of discussion that we are 21:00:59 More evidence: Web Services. What does http://example.com/cgi-bin/foo?blah=blee _name_ ? 21:01:27 the reason to end in '/' vs '#' is described as practical not theoretical: i.e. the returned document will be massive 21:01:48 assuming infinite network bandwidth and disc space there is no issue 21:01:50 Again, we're happy with abstractions. It's stretchy to claim there's a single document being rendered; better to talk about messages being sent to a service. 21:02:10 jonb: exactly. Also assume no IPR concerns about exposing a complete dump of the data. 21:02:23 Yeah, a query string passed to a CGI script 21:02:23 yada yada 21:02:36 DAnbri: re handwaving: No, the difference is fundamental. 21:03:16 The idea of a document, a piece of ifnromation, is very core to the web architecture. 21:03:19 if fundamental then _insist_ on an implicit '#' 21:03:22 How do you characterise web services, then? 21:03:54 When you do a web servcie with a POST then that is quite different. We are talking GET here. 21:04:10 If you do a cgi script with a GET then you have a virtual document. 21:04:15 Jon: How can you insist on an implicit # when you could also refer to a document without it? 21:04:17 you say that http://www.w3.org/Icons/WWW/w3c_home is a "piece of information"? (rather than a work that has renderings as such pieces?) 21:04:44 millions of web services are exposed over GET 21:04:58 "poece of information" is not a very precise term. 21:05:07 We agree :) 21:05:08 But it is certainly not a person. 21:05:20 we agree :-) 21:05:24 For all intents and purposes, we can say that in the context of some piece of machinery, the W3C Logo URI references a network entity which is an image. That is the *current* result of the concept of that URI, the W3C logo 21:05:35 Can we name resources with http: URIs that don't expose GET interfaces? (only HEAD, for example?) 21:05:46 I can use RDF to talk about documents. 21:06:00 If I do, and I dereference the URI used, i should get htat document. 21:06:27 You can say that a particular URI references a particular resource... just invent a predicate for it 21:06:29 what about a word? e.g. http://xmlns.com/wordnet/1.6/Person 21:06:32 only HEAD: The HEAD cammands gets just headersabout the document. 21:06:52 are you saying that it must be http://xmlns.com/wordnet/1.6#Person 21:06:52 Jon: That references a piece of RDF about that word 21:06:58 What do you mean, 'get that document'? I prefer it when you talk about messages between machines. You get one machine to send _something_ to another... 21:07:12 The concept is just "About the word "Person"" 21:07:59 The messages are part of a protocol - that defines their meaning. 21:08:12 HTTP is a protocol for getting renditions of documents. 21:08:32 You're blurring HTTP with HTTP's GET method 21:08:51 It's a shame about the name of HTTP really... an image isn't hypertext 21:09:00 Although HTTP GET has a special standing in the URI space, since we don't (do we?) have URI views of POST, PUT etc... 21:09:04 But I suppose it can be used or referred to in hypertext 21:10:30 sbp, Jon: http://xmlns.com/wordnet/1.6#Person in RDF references NOT a bit of RDF, but the abstract RDF Thing which is DESCRIBED by that document 21:10:45 I can see this worldview, but it forces us to get really alice-in-wonderland about '#' semantics. 21:11:01 timbl: right. 21:11:09 O.K. 21:11:13 I had to clarify - that it wasn't for transferring hyeprtexct, but for transferring things in a way (fast etc) suitable for hypertext systems. 21:11:24 Tim, tell me how you know that? That's a URL in my namespace. Do you understand what the #Person bit means on the basis of the media type, by talking to me, or what? 21:11:45 FragID's are defined by media type aren't they? 21:12:01 Tim: aha 21:12:04 By alice-in-wonderland, I mean: who gets to say how #Person is interpreted in the URI Ref ' http://xmlns.com/wordnet/1.6#Person' 21:12:22 So HTTP is a protocol for any media suitable for a hyprtext system... neat 21:12:32 media: tim seems to say "only when you're retrieving a resource rendering" does that bit of the URI spec kick in. 21:12:42 RDDL being XHTML can deal with either HTML or XML ID frag id semantics 21:12:48 ie. when you're not in a retrieval contenxt, _something_ else determines what #foo means. 21:12:49 danbri: 1) if i trust any document which talks about ...#Person I will know thinsg about it and 2) if I do a HTTp GET then I go through a process which incldues the media type and I end up with more (rather definitive in this cae) infromation about ...#Person. 21:13:16 hence #Person refers to something that has both human readable prose and machine readable URIs 21:13:28 ? 21:13:33 Jon: What is the default content type for RDDL? text/html I presume. Then it's xhtml:@id 21:13:50 heh, sure. You can't have it both ways. Either http://example.com/foo#foo has a meaning independent of some passing HTTP conversatiion, or not. 21:14:17 Using RDDL to describe a namespace (URI) implies the namespace is described by a document 21:14:20 To do this properly, RDF needs its won content type. because text/xml fragids refer to parts of a document, not to an abstract object described by th document. 21:14:22 I run xmlns.com; do I get to say what those names mean? Or is it set by the contenxt of the HTTPResource they're named within? 21:14:53 Yes, and this should be explicitly stated in RDF, or in the MIME (e.g. application/rdf+xml) 21:15:05 You get to say what they mean and putting something at the HTTPResource is a classic and standard way of saying it. 21:15:08 Tim: that's the problem, yes. 21:15:09 that is, the precise purpose of RDF FragID's 21:15:36 How would I get to say what http://www.w3.org/Icons/WWW/w3c_home#foobar means? 21:15:37 (Well, in N3 the #Person refs to the abstarct thing anyway!) 21:15:54 danbri, you wouldn't. 21:15:56 Ah, N3 :-) 21:16:05 W3C have put several things there; each with a different semantics for #foobar 21:16:22 How would I set about determining what http://www.w3.org/Icons/WWW/w3c_home#foobar means? 21:16:29 No, I don't think any of the images types ahve a defined fragid semantcis at all. 21:16:31 i am suggesting a -convention- that the URI reference dereferences to something that describes the resource 21:16:38 (N.B. IE5.5/6 doesn't properly render the GIF image at that URI without the extension anyway...) 21:16:47 (image frags: I was being optimistic) 21:16:59 In SVG I'll bet you could! 21:17:05 W3C has not allocated the name http://www.w3.org/Icons/WWW/w3c_home#foobar. (--editor in chief) 21:17:10 the convention is that the byte stream entity describe the abstract resource 21:17:13 SVG: true. 21:17:14 Label SVG not found. 21:17:31 quiet, bot. 21:17:35 :-) 21:17:47 Hmm, I suppose more accurate to say: W3C has not *yet* allocated... 21:18:21 Why would it in the future? 21:18:28 So, say there was a frag-semantics doc for PNG (it could be arranged), would the PNG consortium (or whatever) (and foobar meant, er, filesize) have auto-allocated that name for us? 21:18:37 Say when the day comes that you change it over to SVG 21:18:52 #foobar could reference the "3" in "W3C" :-) 21:18:52 * timbl wonders whether there is an CVG in fact 21:18:57 Yes, there is 21:18:59 s/C/S/ 21:19:03 * sbp gets the URI 21:19:10 * timbl wonders whether the logger takes sed commands 21:19:23 Tim: http://www.w3.org/TR/SVG-access 21:19:30 It's about half way down the page 21:19:36 Can the semantic web (as opposed to RDF) work in a way that the entity not describe the resource (e.g. RDF(S), DAML+ONT) 21:19:46 Ah, the SVG logo itslef should have a URI too... hang on... 21:20:16 s/ONT/OIL 21:20:19 But anyway, you know that if something is an SVG image then any fragment of it will be an SVG imagebit of soem sort. 21:20:22 W3C SVG Logo: http://www.w3.org/TR/SVG-access/w3csub.svg 21:20:34 yep 21:20:38 SVG is an XML format 21:20:47 scalable vector graphics 21:21:15 Now, when the document is XML, then you have to look to the namespace to find out what the thing means. 21:21:18 part of the logo is:- W3C 21:21:32 So I suppose http://www.w3.org/TR/SVG-access/w3csub.svg#logotitle references that 21:21:42 right. so how 'look to the namespace'? 21:21:43 But our 'something' in this case is the abstraction '/images/logo' not a specific rendering; only the renderings have media-type based fragid semantics. 21:21:50 yes, and it must have type svg title. 21:21:59 That is a well-known bug. 21:22:28 I wish the DC specializers were in wider use. 21:22:41 because o this limitation, publishers cannot allocate URIs with fragment identifiers which could be construed difefrently for documents for whcih they support content negotiation. 21:22:43 Quite. That's why I'm averse to using #, and prefer to attach the HTTPResource space. 21:23:02 Yes, that's is a big problem 21:23:11 bijan -- want to join the DC Architecture WG and help get them into wider use? 21:23:19 No. 21:23:26 stop complainin' them! ;-) 21:23:35 One way to be safe though is to use just an id as a gragid because most new lamngauges will be encouraged to supprort that as a fragid syntax 21:23:36 I think it is up to the document creator to be sure that multiple media types support the same fragment ids 21:23:39 gotta go.. 21:24:05 I'm part of the grass roots that the DCArchWG needs to nurture for the specializers to flourish. 21:24:05 Yes, the document creator should be careful 21:24:19 * danbri scratches his head, wonders how this'll ever get fixed. 21:24:23 Ya gotta have peons too! 21:24:28 :-) 21:25:01 * jonb thinks that the answer should be practical 21:25:35 There should be a theoretical and pragmatic answer to this 21:26:10 there appear to be multiple theoretical answers depending on who you talk to 21:26:30 i just want code that works 21:26:39 Code that someone else wrote. 21:26:39 Yeah, so choose the most practical 21:27:10 i write lots of code 21:27:13 How about, we use 'http://xmlns.com/wordnet/1.6/Person' but the RDF that comes back, makes assertions about the entity 'http://xmlns.com/wordnet/1.6/#Person' instead. 21:27:13 Someone *good*, and friendly like. Willing to accomodate my special needs! 21:27:17 No, that wouldn't work. 21:27:24 nah 21:27:37 did I take from the above thread, that when you have a uri#fragid, you *always* have to GET uri in order to determine mime type and hence what #fragid meant 21:27:52 What's the point? http://xmlns.com/wordnet/1.6/Person is good enough, why add a #? 21:28:10 lets assume implicit GET to simplify the discussion 21:28:47 sbp: because according to TimBL and a fair % of web architecture specs, it names a 'document' not a class of things (people). 21:29:56 SeanP has joined #rdfig 21:30:07 is there something fundamentally wrong with treating URIs that end in '/' differently than those that end in alphanumeric 21:30:14 * danbri considers investigating http://www.linuxcare.com.au/tridge/tivo-ethernet/ 21:30:30 Yes, if just talking about generic URIs... they're opaque 21:30:40 get over it 21:30:40 But in certian contexts I guess it's O.K., although a bit odd 21:31:06 sbp has quit (Killed (NickServ (GHOST command used by SeanP))) 21:31:09 SeanP is now known as sbp 21:31:17 * sbp is over it 21:31:18 who ever started the rumor that URIs are opaque, there plenty of fine BNF in the RFC? 21:31:41 The string is opaque... that's where the agreement comes in 21:31:53 THey're...uh...*structured* but opaquely so! 21:31:59 We all agree that ?xxx is a query string... doesn't have to be 21:32:17 biajn: in a way I guess 21:32:22 well anyways danbri i think it would be useful to do something with wordnet and UMLS in the same way 21:32:25 s/biajn/bijan 21:32:54 Oh, how about using ? in the namespace, does that work? 21:33:11 eg http://xmlns.com/wordnet/1.6?Person 21:33:25 I suppose it should... find the XMLNS spec 21:33:33 1.6 is the document, Person is the query... nope. It would be the name of the document. 21:33:34 a anmespace is any URI I bet 21:33:59 I think the URI is fine as it is. ANme me one thing wrong with it 21:34:05 apache will look for something at .../1.6 21:34:12 edd has joined #rdfig 21:35:33 . ? http://.../1.6/Person# vs .../Person -- what is the upshot on the difference? 21:37:20 what is the problem with .../1.6?word=Person ? 21:37:21 there wouldn't be the slightest difference if server software were perfect 21:38:28 the namespace would be http://.../1.6?word= 21:39:06 so the work is 'all the words' but what is returned is the result of filtering that by the query 21:39:59 You could even set up /Person to redirect to /1.6?word=Person 21:40:19 In fact, for backwards compatability, and Cool URIs Don't Change... 21:40:45 danbri has quit (Ping timeout for danbri[modem-123.brown-sailfin-tang.dialup.pol.co.uk]) 21:42:31 http://www-106.ibm.com/developerworks/xml/library/x-tipxrdf.html 21:42:31 T: http://www-106.ibm.com/developerworks/xml/library/x-tipxrdf.html from edd 21:42:39 T:|Documenting style sheets using RDF 21:42:39 titled item T 21:42:39 Comments as first-class resources 21:42:47 T:|Documenting style sheets using RDF Comments as first-class resources 21:42:47 titled item T 21:42:50 Hey, that was Aarons idea... 21:42:53 T:By Uche Ogbuji. 21:42:53 commented item T 21:43:13 T:Another nifty using-RDF-to-good-effect piece from Uche. 21:43:13 commented item T 21:44:09 OH, maybe not 21:50:09 Uche does some excellent work... I remember that XSLT sheet he made in the very early days of RDDL 21:50:32 jonb: yes, wrongto treat .../ differently fromn ...alpha. Client can't look into string except as spec says. 21:51:52 I'm still confused, did XML Schema chose a good namespace URI or not? From RDF's point of view 21:52:54 I don't think it did, although the XSD folk could put the appropriate resources at each concatenated URI 21:53:27 So http://(XSD NS)/XMLSchemaelement could reference some form of descriptino about that element in RDF 21:53:43 Bad, from RDF's PoV if you want to regard the things ID'd in th spec as RDF things. 21:53:45 I mentioned this a while back on the mailing list when the ugliness of the RDF-style concatentated concat(namespace,localname) URis became aparent 21:54:23 The concatenation isn't a problem, it's the people who design the namespaces w.r.t. purpose 21:54:36 ... when namespace URI didn't end in / or # etc. - ugly to humans, computers don't care 21:54:59 Well, the XHTML namespace has the same problem 21:55:04 IDs in spec - well, I assume RDF assumes XML IDs == RDF IDs 21:55:39 No, just the ID of whatever resource 21:56:18 (I presume, just guessing) 21:56:44 So anything that RDF points to has its FragIDs defined by the media type of the network entity 21:56:52 The reason that we can't use things without "#" to identify abstract obejcts is that we need to be able to use then to refer t the thing you get from the server. HTTP gives you the right to ask about this thing, and you put the URI of it as the 2nd word of the request. 21:57:35 spb: No, anything ANYTHING points to with a URI has its FragIDs defined by the media type of the network entity 21:57:47 IF and WHEN you use HTTP. 21:58:25 Ah, so it's all-pervasive 21:58:48 timbl can I rephrase that. You are asserting - We cannot use things without "#" to identify abstract objects. 21:59:36 So what does http://www.w3.org/1999/xhtmllink identify w.r.t. RDF? 22:00:17 I still don't see why that can't be used to point to the concept of a the XHTML element, even if it is more difficult to post something there on the server 22:02:41 Ah, I think I've got it:- 22:02:50 "A client which understands the http: protocol can immediately conclude that the fragementid-less URI is a generic document. This is true even if the publisher (owner of the DNS name) has decided not to run a server." - http://www.w3.org/DesignIssues/Fragment 22:03:33 So if you *don't* include a #, the disadvantage is that you cannot immediately identify it as a document namespace tht can contain "things" defined by its own MIME type 22:04:47 Suddenly, "bind" looks more attractive than it did 5 minutes ago... 22:04:51 So do we have to GET on every namespace URI to find out more - mime type and then use that to determine if we use RDF uri creating rules (concat) or some others (nsuri#XMLid). Won't this come up a *lot* for XML schemas. Hasn't daml thought about this? 22:05:15 Yes, but only if you want to find out more about what is being identified 22:06:05 sounds to me like we should implement more stuff and see how this works 22:06:37 Well, what we have already seems to work kinda well... excapt when it comes up to stuff like the XHTML/XSD namespace 22:06:45 which deson't matter anway, cause it's not in RDF 22:08:11 But then the arguement is that if they *did* have a # on the end, then at least you could identify them as a single document which has "concepts" identified inside it 22:08:51 well, since we have to put special namespace support for opaquish-namespace URIs, we probably still can do that, by other means 22:09:38 there is/will be an rdf schema for xml schemas I assume so that's another way 22:10:02 an rdfs for XSD? I hadn't thought about that... 22:10:35 It's neat that the XSD people are using RDDL now for the namespace 22:10:49 I had to write one for DC, since I couldn't find that the other day. And I'm writing another RDFS this week too... 22:10:49 How long before RDF processors grok RDDL? 22:10:51 yeah, kudos to henry 22:11:04 edd: absolutely 22:11:23 The RDFS for DC is actualyl at the namespace doc... at the top 22:11:36 my route to this is - add XSLT support to Redland, add Xlink mapping support, thus RDDL support 22:11:57 Great! 22:12:13 and also means it can grok future funky XML -> rdf mappings 22:12:27 * sbp suddenly finds he has to move his entire server stuff over... oh no! 22:15:22 * jonb is going to need to read over these comments 22:15:56 can a namespace identify an abstract object? 22:16:00 yeah, I've struggled to pull out some short, snappy thoughts, conclusions or statements 22:17:12 dajobe: You DON'T *have* to to a get on everything. 22:17:19 dajobe: You DON'T *have* to to a get on anything. 22:17:21 that's what I hoped 22:17:29 and expect 22:17:34 dajobe: The mime type thing tells you what happens if you do. 22:17:46 The owner of the name uses this to publish documents. 22:18:45 People I think get confused because HTTP is a space of idenitiers which has a lookup service. It isn't the lookup service, though. 22:18:58 I've been saying "RDF doesn't look into URIs" but the #frag thing and discussion above seems to say something else. danbri's W3C logo thing 22:19:00 The fact that HTTP specae contains documents is my fault. 22:19:04 I suppose we could all move over to Masinters HTCPCP and use the BREW method 22:19:22 RDF doesn't look into URIs. 22:19:44 well I do know this, my code has no methods to de-compose URIs 22:20:20 the problem is that a QName is two parts and a URI one. 22:20:31 But if I ask you: dc:title ?x what is the first thing you will do? Use http to get some information about the document in question. 22:20:54 I don't have that coed wither. But it is on the wishlist. And Dan ahs a lot of Larch to do it. 22:21:07 Tim: is the "?" groked by CWM? 22:21:07 not necessarily, if I don't have a matching statement - I return () empty list 22:21:24 ?x that is 22:21:37 { :x log:URIStartsWith "mailto:" } log:implies { :x rdf:type rdf822:mailbox } . 22:21:52 No, ? doesn't grok the ? 22:21:58 I have wonderd about it though. 22:22:00 oh no, not abouteachprefix again 22:22:04 You would say 22:23:11 And please, please, please fix the anonymous contexts problem for 1.32... [ :a { :b :c :d . }; :e { :f :g :h . } ] goes all odd when -rdf 22:23:21 timbl: above is taking URIs (x;) as conecpts in RDF, surely they aren't there but are used to identify rdf things 22:23:21 cwm foo.n3 -filter=bar.n3 where bar.n3 has { dc:title :x}log:implies{:x a :result }. <> log:forAll :x . 22:24:10 Yes, and { ?x a :unknownVariable . } 22:24:15 if using that syntax 22:24:28 anonymous contexts ... ooops I fixed it in N3, didn't check -rdf. 22:24:58 It puts out the second third ... contexts under a different anonymous node 22:25:28 I need email with tese bugs in. I need of course an email list for it. 22:25:50 Yeah, a CWM list would be great, I wolsn't have to keep cheacking for updates :-) 22:25:52 timbl: the practical problem is that a URI such as http://xmlns.com/wordnet/1.6 is intended to represent 'all words' 22:26:01 I have some test data online for this bug... 22:26:18 huri me the uri... 22:26:26 and http://xmlns.com/wordnet/1.6#Person to represent 'Person' 22:26:32 Tim: Here we are, http://uwimp.com/2001/03/cwmbug/in.n3 http://uwimp.com/2001/03/cwmbug/out1.n3 http://uwimp.com/2001/03/cwmbug/out2.n3 22:26:52 in.n3 is the N3 in, out1.rdf is what CWM gives, out2.rdf is the corrected RDF 22:27:21 supposing it is intended to return an RDF Schema for 'Person' when dereferencing, the problem is that a huge document returning all words would be returned from http://xmlns.com/wordnet/1.6 22:27:36 when all we need is the #Person fragment 22:27:45 how to solve this technical problem? 22:27:51 Believe it or not, I was making an rdf:seq list of my favourite Simpsons episodes in N3 when I spotted the problem :-) 22:28:35 jonb: you get back an rdf fragment with rdfs:isDefinedIn pointing to the full UR 22:29:02 but what URI to use in GET? 22:30:13 SeanP has joined #rdfig 22:30:19 jonb: this is fine http://xmlns.com/wordnet/1.6/Person 22:30:32 sbp has quit (Killed (NickServ (GHOST command used by SeanP))) 22:30:35 SeanP is now known as sbp 22:30:46 but this is what timbl is saying isn't correct? 22:31:43 I don't know, I've lost the thread 22:31:57 * timbl mails sbp's bug to himself 22:32:07 * timbl wonders when he will ahve a better bug tracker 22:32:17 * dajobe notes we've just logged the most data than any other day on #rdfig 22:32:18 rdfig could use a bugzilla install 22:32:37 dajobe: and on a saturday. what an indictment. 22:33:12 * jonb thought he had a little grip but lost it... 22:34:41 "Person" is a word with a length of 6 characters. http://xmlns.com/wordnet/1.6/Person has a length of 86344 bytes. 22:34:47 perhaps we can just say that .../1.6/Person references the RDF Schema describing the abstract concept "Person"? 22:35:30 * jonb notes that people are complicated 22:35:48 http://xmlns.com/wordnet/1.6/Person has language RDF, but "Person" has language English. 22:35:48 U: http://xmlns.com/wordnet/1.6/Person from timbl 22:36:21 U: A document in rdf-xml 22:36:22 commented item U 22:36:38 U: Not a class. Not a word. Not a person. 22:36:38 commented item U 22:37:03 * timbl suspects jonb is taking the rdfig list as a biased sample ;-) 22:37:51 * edd can confirm that his wife, though not being on the rdfig list, also fits the profile 22:38:25 I'm not on the list. 22:38:40 But I'm just sneaky, not complicated. 22:38:54 perhaps there could be an arc from a URI reference (i.e. with a #) to the URI referencing the RDF Schema 22:39:12 . ? foo:RDFSchemaFor 22:39:25 rdfs:isDefinedBy ??? 22:39:39 oh, hang on... 22:40:55 So /Person is a document. /1.6#Person is a class So link the class to the document 22:41:56 O.K. then, not all that hard { rdfs:isDefinedBy . } 22:42:13 Sounds good 22:42:14 Use the # for the namespace, and the / for the schema 22:42:37 a bit like using log.n3 for a namespace and logic.n3 for the schema, eh Tim? :-) 22:42:45 [or is that the other way around?] 22:42:45 But I will be rwriting code to look up schemas from the namespace. 22:42:56 neat! 22:43:16 * timbl assumes that he ahs gotten muddled soemewhere between log and logic - where 22:43:17 look up schemas from the namespace? 22:43:46 the CWM namespace uses log.n3, but the schema is at logic.n3 22:43:57 I think it's probably just an accident 22:44:18 s/CWM namespace/CWM uses log.n3 as the logic namespace 22:44:23 woops 22:44:33 where does the above statement by sbp live? 22:44:48 In the # document 22:45:01 I would guess would be the best place to put it in this instance 22:45:40 so the namepace doc is still a huge doc with one of those for each concept 22:46:03 SethR has joined #rdfig 22:46:10 Yes: but it can be generated... difficult to d/l though 22:47:06 No, I will fix it. I like to allow people to derfer the namespace and get something useful. 22:47:22 Tim: thanks 22:47:34 You could put some RDDL up there! 22:48:41 RDDL? i can't parse that with an RDF parser. 22:49:02 including a resource with nature RDFS 22:49:19 RDDL is trivially converted into RDF 22:49:27 So in summary ns uri would be http://xmlns.com/wordnet/1.6# and contain a zillion statements like { <#Person> rdfs:isDefinedBy . } 22:50:08 Tim: You could make a parser that dereferences some RDDL, extracts any links to RDFS, and follows them... complicated I know, but still... 22:50:28 which brings us to the mythical rdf processing model; where RDDL, xlink, xml schemas could be part of that 22:50:38 dajobe: or you could get your server software to redirect 22:50:46 the java code performs the indirection automatically 22:50:49 Yeah, it was a half-joke really 22:51:03 and a py implementation needs to be done -- ahem. 22:51:21 Why would you want anything other than N3 at an N3 specific namespace? Well, you could have an XML RDF version I suppose 22:51:40 . 22:51:46 suppose you want to put N3 at say the XML Schema namespace 22:51:49 the key bit is: rdf processors must follow rdfs:isDefinedIn in some circumstances, not yet determined 22:52:18 :Sean :dereferences . 22:52:36 XML Schema has been great about not bulldozing the namespace 22:52:59 You never *have* to foloow anything. But the link suggests it would get you something useful were you to. 22:53:24 sure, but if you went to look up a concept and only found a rdfs:isDefinedIn, you might be encouraged to do something 22:54:06 (oops wrong names change my rdfs:isDefinedIn to rdfs:isDefinedBy in above) 22:54:30 22:54:49 err href="Person.rdfs" 22:55:18 :HTTPResponse "403" . 22:55:34 * timbl goes to fix 22:56:02 :Sean :thanks :TimBL . 22:57:14 /me fixed 22:57:14 22:57:14 22:57:14 22:57:30 SeanP has joined #rdfig 22:57:43 sbp has quit (Killed (NickServ (GHOST command used by SeanP))) 22:57:50 SeanP is now known as sbp 22:58:43 * sbp kicks his modem 22:59:02 * dajobe yawns - that's enough for me today 22:59:14 Ah, hooray: { :HTTPResponse "200 O.K." . } 22:59:55 when you say log:N3Document a rdfs:Class -- don't you mean to use rdfs:subClassOf rather than rdf:type (i.e. isn't 'a' == rdf:type) 23:00:07 dajobe has left channel 23:00:47 Ihaven't read or written tht doc for a while 23:02:04 yes a rdf:type. And log:N3Document has type Class. It isn't a subclass of Class, or all N3 documents would be classes 23:03:41 * sbp ponders releasing more N3 examples onto the Web... 23:04:28 just trying to grok -- so :N3Document is one specific document rather than a class of documents 23:05:06 * timbl was looking to hack http://www.w3.org/2001/Talks/04-tbl/Overview.html into a general talk on sw tech... 23:05:21 :N3Docuemnt is a class, not a docuemnt. 23:05:22 { <> a :N3Document . } 23:05:31 yes. 23:05:39 acually, <> a IRC:session . 23:05:43 :-) 23:05:53 Heh heh 23:05:53 ... but they are getting closer 23:06:03 Well, this gets logged as XML RDF 23:06:16 yes, there is a danger in translating <> ! 23:06:26 so isn't :N3Document rdfs:subClassOf rdfs:Class ? 23:06:35 I had a rpoblem with that -- tranlated <> a n3:document into xml! 23:06:59 no, :N3Document a rdfs:Class. 23:07:06 :N3Document rdfs:subClassOf doc;Work 23:07:11 :N3Document rdfs:subClassOf doc:Work 23:07:19 ok 23:07:27 Eek, that is a problem... does CWM do that every time that it converts from N3 to RDF? Could you make it automatically convert that to a :XMLRDFDocument? 23:07:54 Oh, and what's this "$" identifier I see? 23:07:57 I suppose so... it could in fact put <> ...lots of metadata about how tehe doc was made. 23:08:03 $? 23:08:26 where? 23:08:26 From cwm.py:- 23:08:27 - Introduce this or $ to indicate the current context 23:08:27 - Introduce a difference between <> and $ in that <> log:parsesTo $ . 23:08:27 serialised subPropertyOf serialisedAs 23:08:36 - http://www.w3.org/2000/10/swap/cwm.py 23:08:39 Oh, I decided "this" was better than "$". 23:08:50 <> means "this document", "this" means 23:08:56 "this context" 23:08:58 . 23:09:18 So "this" is a keyword in N3 for the current document context? 23:09:29 scratch "document" 23:09:38 That is the idea... 23:09:51 the document hasn't quite cuaght up yet. 23:10:04 the documentation hasn't quite cuaght up yet. 23:10:21 aha, I get lost y'see :-) 23:11:15 Some of that stuff in the top blurb looks great, and I still wonder how many more things CWM can do that I'm not aware of... I really should learn Python 23:12:04 Then I could help Seth make a version of CWM with semEnglish support... that's be fun 23:12:08 ({ this log:forAll :x. :x wh:loves :y } log:implies { :y a psyc:crazy }. this log:forAll y} 23:12:13 s/that's/that'd 23:13:17 thsi context applies for all x . x loves y, which implies that y is crazy, where this context applies for all y 23:13:42 so to love x is crazy 23:14:37 hang on, I read that wrong 23:15:14 if everyone loves you, you uwill go crazy. Only a logical example, Ino experimental evidence. 23:15:37 aha, I see. :x is log:forAll 23:15:42 * edd :) 23:16:03 Yeah, there is evidence: movie starts and so on... 23:16:09 s/starts/stars 23:16:23 most of them are nuts 23:17:03 { :y rdfs:domain :movieStar . } 23:18:29 * sbp waits for "- Find all synonyms of synonym" 23:19:43 try http://xmlns.com/wordnet/1.6/Synonym :-) 23:20:07 What's a word for a word which is not an example of itself? nonautonym? 23:20:15 jon: no! You mean http://xmlns.com/wordnet/1.6#Synonym :-) 23:20:36 get danbri to fix his computer first 23:21:05 :synonym a :word . :word a :Langauge_unit . :Language_unit a [...] 23:21:06 * timbl has to go 23:21:15 see ya 23:21:22 timbl is now known as timbl-gone 23:22:49 Hmm... "a word for a word which is not an example of itself" - my head reels 23:23:04 is there a converter from N3 into RDFS so i can see what log.n3 would look like as an RDF Schema? 23:23:14 antiautonym maybe :-) 23:23:20 Yes: CWM! 23:23:27 proper set 23:23:31 URL? 23:23:43 - http://www.w3.org/2000/10/swap/cwm.py 23:23:44 "red" 23:23:53 There is a Web interface at SWAG, hang on a sec... 23:24:32 http://swag.semanticweb.org/n3tordf?uri=http%3A%2F%2Fwww.w3.org%2F2000%2F10%2Fswap%2Flog.n3 23:24:32 V: http://swag.semanticweb.org/n3tordf?uri=http%3A%2F%2Fwww.w3.org%2F2000%2F10%2Fswap%2Flog.n3 from sbp 23:24:36 aaargh 23:24:51 V:|The N3 Logic Namespace Schema in XML RDF 23:24:51 titled item V 23:25:06 V:Thanks to the SWAG interface for TimBLs CWM 23:25:06 commented item V 23:25:34 Hey, it gives an error! 23:26:21 oh, " rdfs:domain: log:context; 23:26:34 " should be rdfs:domain log:context . 23:26:52 eek, nasty 23:27:26 and there's a couple of other typos as well 23:27:43 what are you trying to do, sean ? 23:27:46 makes you wonder, doesn't it? 23:27:52 edd: he he he 23:27:53 whether it's all an elaborate prank or not. 23:28:08 SethR: trying to parse the N3 Logic NS document 23:28:33 has it been translated to rdf/xml ? 23:29:00 I tried, using the SWAG interface 23:29:05 but it has parsing errors 23:29:21 can you do it whth cwm? 23:29:38 yeah, that's what I'm using: the SWAG interface to CWM 23:30:11 why not try it locally? 23:30:28 I'm just about to 23:30:52 * sbp 1) fixes log.n3 23:30:59 let me know where the rdf/xml ends up .. id like to read it into SethBot 23:31:17 I'll run it through CWM and upload it to the Web 23:31:24 :) 23:31:56 * sbp starts Python 23:32:09 well i think n3 in the chump is going to be the best thing since sliced bread :)))) 23:32:35 .. and you know what it's gonna force me to do 23:32:51 he he he 23:34:58 hooray! It parsed at last 23:35:14 Does anyone want me to paste it in channel? :-) 23:35:22 yes 23:35:34 better for me if you put it on the web 23:35:39 Well, go to "#cwmpaste" then, he he he 23:36:23 can you put it into the scratchpad? 23:36:33 Just about to 23:36:49 * jonb is humorless 23:37:28 http://uwimp.com/2001/03/log/log.n3 23:37:29 W: http://uwimp.com/2001/03/log/log.n3 from sbp 23:37:39 W:|Corrected N3 For the Logic Namespace 23:37:39 titled item W 23:37:48 http://uwimp.com/2001/03/log/log.rdf 23:37:49 X: http://uwimp.com/2001/03/log/log.rdf from sbp 23:38:04 X:|XML RDF For the Logic Namespace 23:38:04 titled item X 23:38:30 hey thats n3 23:38:49 W is, X is RDF 23:38:57 s/RDF/XML RDF 23:39:01 ok 23:40:14 must have an error, it doesnt parse 23:40:31 What, http://uwimp.com/2001/03/log/log.rdf doesn't parse? 23:41:36 not through my implementation of jason diamonds parser 23:41:43 but it might be my bug 23:43:05 what's rdf:is ? 23:43:31 no idea 23:44:10 a problem is that the about="#N3Document" should be rdf:ID="N3Document" 23:45:39 good catch 23:47:53 * sbp writes up the bugs and mails them to TimBL 23:48:15 * sbp expects the reply "I have got a consortium to run you know..." 23:48:38 whats 23:50:12 it gives the parse type of the resource 23:50:39 "The parseType attribute changes the interpretation of the element content." - http://www.w3.org/TR/REC-rdf-syntax/ 23:50:51 CWM bungs it in automatically 23:51:24 is it legitimate rdf? 23:51:28 yes 23:51:34 It's in the RDF Rec. 23:51:56 There's lots of junk in there :-) 23:52:45 I suppose that the p(s,o) model is theoreticall "it", but that practically, we always ned to define other stuff to help us process it 23:55:11 Seth: any closer to a semEnglish package, or a (sem web wiki) dc:description yet? 23:56:33 Heh, one Simpsons blackboard gag is "Pork is not a verb". I suppose in N3 that's { { "pork" a rdf:predicate . } a log:falsehood . } . :-) 23:56:44 where is the element defined? 23:56:59 log:means 23:57:07 (I think) 23:57:24 but in the rdf it shows just as no ns prefix 23:58:31 baud has quit (Ping timeout for baud[194.70.26.133]) 23:58:41 im thinking when bijan gets n3 in here, this will be very close to a sem web wiki 23:58:43 SeanP has joined #rdfig 23:58:57 sbp has quit (Killed (NickServ (GHOST command used by SeanP))) 23:59:03 SeanP is now known as sbp 23:59:08 in fact... 23:59:18 baud has joined #rdfig 23:59:58 it depends on how well the n3 gets translated into the rss feed 23:59:59 { this a log:falsehood log:forAll :x . :x rdfs:label "pork"; a rdf:Predicate . }