00:06:53 mnot has left #rdfig 00:06:55 mnot has joined #rdfig 00:52:43 sbp has joined #rdfig 01:13:31 xbill has quit 01:16:03 timbl2 has joined #rdfig 01:16:41 tim-lurk has quit 01:18:08 * look is away: out of here 01:19:00 I'm thinking the right unixy way to handle authoritative language-definitions in non-XML files is the emacs local-variables hack, where you put in the file a magic string like: -*- formal-language-URI: "http://www.w3.org/2001/20/24/bnf2"; -*- 01:19:59 - /2001/20/? 01:20:06 typo 01:20:12 :-) 01:20:23 that's when I'll have this working. :-) 01:21:19 blindfold is a really neat idea... 01:21:39 http://www-106.ibm.com/developerworks/xml/library/x-matters14.html?open&l=118,t=grx,p=m14 01:21:39 A: http://www-106.ibm.com/developerworks/xml/library/x-matters14.html?open&l=118,t=grx,p=m14 from bijan 01:21:56 A:|developerWorks article on HaXML 01:21:56 titled item A 01:22:09 A:Er...nothing to do with RDF. 01:22:10 commented item A 01:22:16 Thanks. I hope I can get it working well enough. Or maybe someone else can....! 01:22:46 A:Except that it mentions my [Functional Programming and XML|http://www.xml.com/lpt/a/2001/02/14/functional.html] article, and I do a bit of RDF. 01:22:46 commented item A 01:23:24 A:Oh and Edd writes for developerWorks...I think. And I found out about it from the xmlhack list and xmlhack hosts the scratchpad. 01:23:24 commented item A 01:23:40 A:Er...I must be feeling a wee bit insecure or something. 01:23:41 commented item A 01:24:09 * bijan still doesn't know what blindfold is. 01:24:29 Some grammar thing. 01:24:35 Only not with grammars, or something. 01:26:18 A:Hey! He's a philosopher too! 01:26:19 commented item A 01:26:59 A: From his CV: Title of Dissertation: The Speculum and The Scalpel: The Politics of Impotent Representation and Nonrepresentational Terrorism 01:26:59 commented item A 01:27:42 A:Er..that's not what *my* dissertation is about, in case anyone is wondering. 01:27:42 commented item A 01:28:40 Yeah, I've nicely overloaded the term blindfold. It's whatever I'm working on at the moment. :-) But seriously, what sbp is talking about, I think, is the notion I'm pushing that one can/should use any formal language for the semantic web, and they can all play together if you just make formal language definitions (annotated BNF basically) for each one. 01:28:55 A:Writes a lot about Python. A *lot*. 01:28:56 commented item A 01:29:50 "formal language definitions"? 01:29:55 BNF annotated with...semantics? 01:30:01 yep. 01:30:45 As usual, I know how to code it, but I don't know all research in the area. 01:31:16 Hmm. well, ok, if limited to formal languages, it should be doable. 01:35:39 I have no intention of trying to have it parse things that can't already be parsed with computer programs. It's just a slightly prettied-up lex/yacc. 01:35:52 MikeM has joined #rdfig 01:35:56 Well, it wasn't the parsing side of it :) 01:36:59 Ah, the semantics side of it is either an RDF-Graph or an FOL Sentence, but we've had this discussion before. 01:37:12 We have? 01:37:38 * sbp waves to Mike 01:37:59 * bijan prolly doesn't understand the scope of the "any" in "any formal language". 01:38:49 any == mostly any that your're likely to want I think 01:39:24 Ok, then I don't understand "play together" :) 01:39:41 * sandro looks baffled. 01:40:39 ok, I have a first order propositional language, a second order monandic predicate language, and a propositional model language... 01:40:47 Er.. "modal" 01:41:14 What does it mean for all these to "play together"? 01:42:51 That's kind of not what I was talking about, but I'd love to talk about that, so I'll take a stab at an answer. 01:43:16 (That's shows I didn't understand what you meant :)) 01:43:17 hiho..... 01:48:42 em has quit 01:48:56 sbp has quit 01:49:06 My target for blindfold is things more like file formats than languages -- things you parse with perl -- and getting up to simple recursive languages like XML and n3. Pretty much all of these can only express conjunctions of predicates of ground terms. But a few of them get a bit more expressive. I don't really understand logics other than FOL and clear subsets of FOL (like Horn). 01:49:52 Ah! Now I understadn the scope of the "any". That seems reasonable. 01:50:57 I'm curious to what extent all monotonic logics can just be seen as ways of compressing an RDF Graph. 01:50:59 Er... 01:51:27 I can't quite fit negation into that view. 01:51:31 * bijan , at the moment, sees very little extent :) 01:52:31 Well, forall x RDF(a,b,x) is a very compact way to say (a,b,1), (a,b,2), (a,b,AaronSw), (a,b,TheColorBlue), etc. 01:52:56 * AaronSw is so the b of a 01:53:05 er...with the very *crucial* "etc" :) 01:53:22 the etc meaning EVERYTHING, yes. 01:54:02 not that , strickly speaking forall x RDF(a,b,x) doesn't express *any* RDF graph...nor any *possible* RDF graph. 01:54:09 Since RDF graphs are, I take it, all finite 01:54:30 RDF graphs are currently finite, but they probably should not be. 01:54:39 No, I really think they should. 01:54:57 sbp has joined #rdfig 01:54:57 If you make them infinite, you lose the possibiliyt of equating them with FOL sentences. 01:55:02 And a lot of other bad thing. 01:55:27 tim has joined #rdfig 01:55:44 (Note that the equivalence of the universal quantifier and infinite conjunction, and the existential quantifier with infinite disjunction, is well known) 01:56:00 Ah yes, we're having the reification level breaking problem here. Are RDF graphs the FOL sentence, or what is described by the FOL sentence? 01:56:19 That's actually how I thought you were going to go. 01:56:24 I.e., use the graph as a model. 01:56:40 But I couldn't quite see the use or the interest. 01:58:07 timbl-lap has joined #rdfig 01:58:07 tim has quit 01:58:25 I think maybe that is where I'm going (with the logic-for-compression) argument. Some RDF graphs can be described in FOL much more compactly than they can be enumerated (or described in a less-expressive logic). 01:59:03 Well, what do you mean by "compact"? 01:59:21 can be transmitted in fewer bits. 01:59:22 I mean, if you give up uris and use single letters with subscripts, then *definitely* :) 01:59:38 But I mean losslessly. 02:00:42 How is that lossy? 02:00:44 The RDF graph is topology and labeling. Giving up the URIs would be giving up the labels, which is lossy. 02:00:51 Well, changing labels. 02:01:09 I'm presuming you can recover the uris with your translator. 02:01:51 I'm not concerned with that. I'm talking about compression that work by having the de-compressor do valid inference. 02:01:57 er, sound 02:03:22 Sumtin wrong with valid inferences? :) 02:03:57 I dunno what valid inference means. 02:04:01 (but it sounds good) 02:04:40 Standardly, a valid inference is one where if all the premises are true, so to is the conclusion. 02:04:55 Er...inference/argument 02:05:09 For sound, you add that the premises are all true. 02:05:43 "all the premises are true" == " the premises are all true" ? 02:05:53 ie, what's the difference? 02:06:02 Nothing. 02:06:17 Sound just discharges the antecedant. 02:07:36 What does that mean? 02:08:03 I.e., validity says If P then Q. Soundness adds, oh, and P. 02:10:23 GabeW has quit 02:11:56 I don't understand. We have the sentence F ^ (F==>G) and the inference method modus ponens. Is that valid or sound or...? 02:12:17 modus ponens is a sound inference rule. 02:12:45 Which means, roughly, that it's truth preserving. 02:13:15 But it's not going to tell us F. We already knew F. 02:13:33 What would a valid but unsound inference method be? 02:14:07 hmm. one generally doesn't call the *rules* valid. 02:14:29 An inference made *according to the rules* would be a valid inference, generally. 02:14:43 The problem is valid and sound are way overloaded :) 02:14:59 okay, let's pass on that. 02:15:39 What do you think of using the emacs hack for blindfold? :-) [ popping 7 levels off the stack there ] 02:15:54 When referring to arguments, a valid one is one where the conclusion follows from the premise, generally taken to mean that if the premise are true, the conclusion is true. 02:16:08 A sound one is valid, plus the premises are true. 02:16:10 No idea ;) 02:16:18 Okay..... 02:16:28 Full continuations aren't implemented in my cortex yet. 02:18:18 The problem is this: you want to be able to see a data file (probably as the contents returned by an HTTP GET) and be able to get the RDF triples / FOL sentence out of it. I can do that if I know the language, but in non-XML text, how would I find out the language? 02:19:19 Ok. 02:21:00 MIME Type! 02:21:09 There are clues in the filename/URI text, and in the media-type returned by the web server, but neither of those gives you a URI from which you can fetch the language definition. So my best hack is to look for text like -*- formal-language-URI: "http://www.w3.org/2001/20/24/bnf2"; -*- in the contents of the file, which is a trick emacs does. 02:21:40 Ah. Sorta a doctype. 02:22:03 I can't see the mime type (media type) working in practice. There are too many formats that are sent as text/plain or application/octet-stream. 02:22:13 Yeah, very much like a doctype. 02:22:56 a magic string... 02:23:16 a magic string. scary. 02:23:28 Well, it doesn't really bother me, I guess. 02:23:33 of course, then you still have the sub problem of finding out whether or not a format uses the "magic string" thing too 02:23:42 Why not just an HTTP header? 02:24:09 using HTTP-EXT? 02:24:26 Yeah -- interesting idea. I should talk to Prud'hommeaux about that. 02:24:54 I'd like it to be in e-mail too, but I guess that has X-headers, too. 02:25:10 I don't understand the "sub problem" sbp. 02:25:35 In general, headers seem to be the right thing :) 02:27:02 X-Formal-Language-URI: http://www.w3.org/2001/10/x 02:27:16 Or something -- I don't know how HTTP-EXT works. 02:30:27 xbill has joined #rdfig 02:49:35 what do you want from http-ext? 02:50:23 an extra field: here's a new typing deeley, and hey, it's a URI 02:53:20 no need to use http-ext, just define a header. You don't have to make it an X-header either; that was only for MIME, and is frowned upon now in any case. 02:54:05 So it's a header free-for-all? 02:54:16 All HTTP proxies and clients will just pass along any header they don't recognize? 02:54:18 woo, everybody grab a header! 02:54:19 tho interestingly, http-ext does define a namespace-like mechanism; however, it's only scoped within the headers, iirc 02:54:21 yes 02:54:41 there is a header free-for-all. There's been noise made about establishing a iana registry for headers, but no motion. 02:54:43 ftp://ftp.isi.edu/in-notes/rfc2774.txt 02:54:45 mnot, but don't we want that? To create a new namespaced headers? 02:54:53 Is that true for mail transfer agents as well? Or i X-... important? 02:54:53 isn't that what HTTP-ext is for? 02:55:34 for MIME (e.g., smtp), yes, X- is defined, but IETF doesn't like that any more, because it leads to abuse too easily. 02:56:01 So what good is HTTP-ext? 02:56:03 So what should one use for SMTP headers? 02:56:10 AaronSw: yes, we want to, but the namespace is only valid for headers themselves. 02:56:18 So? 02:56:20 anythin you want, AFAIK ;) 02:56:21 HTTP-EXT was made for exacly this... 02:56:37 (sort of thing) 02:56:55 HTTP-EXT was designed to avoid header collisions, and to allow extentions to be mandatory (e.g., you MUST understand this to process the message) 02:57:50 if you just want to introduce a HTTP header, you can do one of three things: 02:57:50 Can I quote you on that? Mark Nottingham tells me that the X- approach is neither necessary nor recommended for HTTP (or even SMTP, AFAHK). 02:57:51 1. define it, use it, and screw anybody who happens to use the same one 02:57:51 just point them to the RDF IG logs :-) 02:58:14 2. define it, have it published in a RFC (informational or standards track) to legitimise it (this is the current approach of P3P, SOAP, etc.) 02:58:44 3. Use the HTTP Extension Framework, and incur the overhead of an extra header with a URI on it for every extension in every message you send. 02:59:17 re: quote - sure; that's to the best of my knowledge, of course. Hold on, I'll do some digging for some more reliable references... 03:01:08 Somehow I'm guess my employers will frown on option 1. :-) (and my buddy Pru' who just implemented HTTP-Ext in Apache, libwww, and mozilla will prefer that solution. :-) 03:01:43 heh, heh 03:02:23 Eric and I have talked about the extension framework... He finally did it, eh? 03:02:36 Yep. 03:02:39 cool 03:03:00 I'm not sure he's gotten it accepted into the apache and mozilla trees yet, though. 03:03:05 IMHO it's unworkable for just adding headers, because of the overhead. The mandatory stuff is intriguing, but noone seems to be using that. 03:03:08 And the rest of use will curse your name. 03:03:17 ...not that that would be a *change*, mind. 03:03:50 * sandro looks like he's about to cry. :-( 03:04:37 Oops. 03:04:54 C'mon "sandro" just sounds like *such* a stud muffin name. 03:05:25 nile has joined #rdfig 03:05:27 * bijan is just kidding of course. 03:05:34 nile has left #rdfig 03:07:33 hmm, can't find anything specific... do remember that it was discussed at the plenary meeting in London, and frowned upon. 03:09:17 So, the extension framework would be fine for experimentation, but IIRC you'd have to add something like this to every message: 03:09:35 Opt: "http://www.example.com/ns"; ns=10 03:09:44 10-MyHeader: foobarboo 03:10:03 Note that the namespace in 'Opt' is the namespace for the HTTP headers, not anything else. 03:10:13 Tho it would be interesting to see how they mix up 03:10:25 Yeah. Interesting. 03:10:41 Yes, HTTP-ext is like namespaces for XML. It binds new headers to a URI - the header itself is irrelevant. I agtree with mnot that IETF frowns on X- -- it didn't help. 03:11:19 But if the useit to give a *real*N mime type, then why not start using the content-type properly?! 03:11:52 "As people have stopped using real-content-type properly we ar eintroducing real-real-coontent type". 03:12:09 Suggestion: add it to the content type as a param 03:12:41 There was a lot of discussion about this in SOAP regarding SOAPAction and content-types. 03:12:48 Content-type: application/octet-stream; bfg=http://whatever/foo.bfg 03:13:33 except you can't really declare new parameters for old content types, can you? 03:13:49 Not sure. 03:14:02 I mean, anyway, for Sandro's purposes, people are going to be starting from scratch anyway, so it's irrelevant 03:14:09 s/anyway// 03:14:12 Not sure whether there is an "ignore unkown params" rule. 03:14:16 also, doesn't that change the interpretation of the resource? Think that's frowned upon with media type parameters 03:14:31 * sandro wants to be able to (easily) chump a part of this log -- to title a section of the conversaion.... 03:14:47 you can do a BLURB: thing, sandro 03:14:52 Well, a blindfold grammar does rather specify the semantic interpretation of hte resource. 03:15:05 or link to the URI-Ref for this part of the logs 03:15:33 The header I'm imagining is Formal-Language-Definition: 03:15:54 FLD: 03:15:54 Label FLD not found. 03:16:11 I see no reason to abbreviate it. 03:16:14 Why not boostrap RDF metadat with just 03:16:20 is it going to be something that's explicitly retrivable? 03:16:40 RDF-prop: foo.bfg 03:16:56 I'm thinking in terms of retreivability at the moment, but there's probably a place for 3rd-party information, too. 03:16:57 HTTP Headers are problematic; there aren't any good UIs in Web servers for associating metadata with resources, and often the authors don't have administrative control 03:17:13 (wonder what happens when th econtent-type and fld don't match - security hole?) 03:17:17 ooh, RDF in the headers. Just like CC/PP 03:18:06 N3: ; . 03:18:25 just replace HTTP headers with N3 full-stop! 03:18:53 That's the second time I've heard that suggestion in a week... 03:18:55 this a :Response; :code "200 O.K." . # etc. 03:19:03 Bin there done that patentented it^H^H^H^H^H^H^H^H^H^H got the tshirt 03:19:12 AHA! 03:19:13 heh, heh, heh 03:19:45 An RDF mapping of HTTP and SMTP headers is well overdue. DanC has of course written a bunch of larch about it [n] 03:19:48 Actually this ties in with what I'm doing for work now; modeling HTTP messages as data structures 03:20:19 well, some other people are working on it too, porting lots of different Internet protocols into RDF Triples 03:20:28 mnot, You mean not for fun?! ;-) 03:20:34 unfortunately not ;) 03:21:09 It's hard to sell RDF, tho; I've since taken references to RDF and RDFS out, and my boss stifles me every time I say 'model' 03:21:18 The next step is to write the conditoin for a response to be a vlaid response to a given request in trems of N3 rules. 03:21:27 YES 03:21:43 and cache coherence, and so forth for HTTP features 03:21:43 yeah, good idea. Then you can implement it in CWM, eh? 03:21:46 solution: don't say "model". It isn't a proper use of hte word anyway. 03:22:00 mnot, have you seen DanC's stuff? 03:22:07 Don't think so 03:22:37 http://www.w3.org/XML/9711theory/ 03:22:37 B: http://www.w3.org/XML/9711theory/ from timbl-lap 03:22:44 Sandro says, "the semantics" of the graph... Tim, don't you have a similar property in log:? the range of log:parsesTo or something 03:22:52 B| Specifying Web Architecture with Larch 03:23:09 s/property/Class/ 03:23:11 you need a colon like in B:| 03:23:25 B: dc:creator [ con:mailbox ]. 03:23:25 commented item B 03:23:34 OK, interesting 03:23:35 B:| Specifying Web Architecture with Larch 03:23:36 titled item B 03:24:16 I was going a bit more into struture; e.g., model cache control as a bag holding statements about max-age, etc. 03:24:44 based on some work I did a while back in Python (old, nasty code): http://www.mnot.net/python/httpheadertypes.py 03:25:05 Somewher Dan defines what an HTTP transactoin guarantees you, give sthe rules for proxies, and then shows how a goo dproxy will preseve HTTP validity of transactions. 03:25:38 Tim's log:semantics matches my usage, I think, except that he's assuming the document is RDF/XML (I think). If I can get this language labeling down, we can widen that to "anything with properly declared semantics" 03:25:43 cool; will have to look at it more deeply tomorrow. Thx 03:26:05 as in log:semantics rdfs:subPropertyOf sandro:semantics . 03:27:01 Actually log/semantics doesn't assume rdf/xml, it uses first the mime type and then given no clues a quick peek. 03:27:10 I don't actually know what subPropertyOf means. :-) sandro:semantics has a larger domain. 03:27:15 I agree to widening totally generally 03:27:22 in principle.... just do it 03:27:36 It uses mime-type with a hard-coded table, right? 03:27:59 Yes - it only has two parsers, and choses what looks lik ethe most likely. 03:28:46 It doens't have th etechnology to dynamically craete a parser -- that is blindfold. But dynamically doingt sounds good to me. Furthers the idea of grounding the semnatics of a document. 03:29:19 Alos not ethat it uses Python's meachanism - no idea what - which actually gives mime types for files. 03:29:24 heh, grounding the serialization in URI space as well as model? 03:29:29 Also note that it uses Python's meachanism - no idea what - which actually gives mime types for files. 03:30:05 * timbl-lap is wandering off but glad I tuned in! 03:30:08 timbl-lap is now known as timzzzz 03:31:23 kutgw 03:31:39 :-) 03:31:52 Where's blindfold when you need it! 03:31:57 k= not 03:32:05 u= -> 03:32:22 Oops. 03:32:42 k= -> 03:33:01 u= & 03:33:10 tgw are all sentences. 03:33:23 ->&tgw (reverse polish) 03:33:30 t&g -> w. 03:41:50 Can an http header be repeated meaningfully? Can we have multiple RDF-Property: headers? 03:42:24 I don't think so 03:42:39 as I found out to my cost when my server spewed out duplicate Content-Type headers 03:42:47 It was a nice idea. 03:43:01 I think a client is obliged to take the former value 03:43:13 * sbp should check 03:43:43 nope, it's the latter 03:45:23 (if at all - I'm still not sure that multiple values are disallowed) 03:58:53 Hmm, someone else today was proposing repeated headers... I hope they're allowed. 04:06:35 sbp has quit 04:33:11 http://www.w3.org/2001/06/blindfold/langIdent.html 04:33:11 C: http://www.w3.org/2001/06/blindfold/langIdent.html from sandro 04:33:52 chihchun: trying to write down some of what we've been talking about 04:33:58 C:trying to write down some of what we've been talking about 04:33:58 commented item C 04:36:11 sandro, I don't see anything in the spec against repeated headers. 04:41:12 timzzzz has quit 05:07:57 the semantics of repeat headers depends on the individual header 05:08:10 a header can be repeated, if the header is a list 05:08:16 e.g., 05:08:37 Cache-Control: max-age=30 05:08:45 Cache-Control: must-revalidate 05:08:49 is equivalent to 05:08:58 Cache-Control: max-age=30, must-revalidate 05:09:09 ordering is *not* significant 05:09:48 behaviours of repeated headers like Content-Type is undefined, IIRC 05:10:07 The only header that doesn't follow this convention is Set-Cookie, because !@#$% Netscrape jumped the gun 05:10:18 * mnot goes back to watching Farscape 05:17:58 Heh heh heh. 05:25:24 Well, 200 email messages down, 927 to go. I better go to bed... 05:25:32 c'ya 05:26:22 Ouch. 05:27:01 Yeah... :-( 05:27:06 What's that for? 05:27:07 Just mail? 05:27:18 As opposed to what? 05:27:22 * bijan is supposed to be summarizing a months worth of squeak mailing messages. 05:27:24 >1400 05:27:35 Eek. 05:28:15 Tis a bummer. 07:48:43 chihchun_ has joined #rdfig 07:49:29 chihchun has quit 08:45:40 larsbot has joined #rdfig 09:37:44 chihchun_ has quit 09:38:33 chihchun_ has joined #rdfig 11:31:16 larsbot has quit 11:40:22 tim has joined #rdfig 11:57:12 em has joined #rdfig 12:10:42 tim has quit 13:06:56 timbl has joined #rdfig 13:15:58 Re last night. You *can* meaningfully repeat RFC822 headers in general! 13:16:11 Obviously, for some headers it does not make sense. 13:16:16 The content-type shoul dbe unique. 13:16:29 But think of "Received:" headers in email. 13:17:07 This is comment on http://www.w3.org/2001/06/blindfold/langIdent.html 13:19:00 2. If the document is XML, then you MUST use the namespace I think. That defines the meaning of the document. IMHO it leads to confusion and possible security failure if you have two places to define what it means, an FDL and an NS. I would prefer that one make a simple RDF assertion of what the FDL is for the namespace. 13:21:27 Youre mthod 3 with RDDL is therefore better IMO. You can also use embedded RDF. 13:22:11 Documents "in the wold" by definition have no surrounding protocol, so there is no formal way you can hold somebofy to their meaning, unless you start with natural language. 13:23:35 An interesting question is how to introduce say N3 as having semantci sin an otherwise english langue message, like an IRC lof or an email. 13:25:06 If I just say, "In n3, { <> a com:check }" can we assume that I can be held to that, assuming the document was in english? 13:26:17 s/wold/wild/ 13:26:24 no reference to Stow on the Wold. 13:27:24 C: Reviewed by timbl 13:06 BST 13:27:24 commented item C 13:41:06 lasDesk has quit 13:42:20 lasDesk has joined #rdfig 13:46:12 sbp has joined #rdfig 13:51:59 ambient has joined #rdfig 13:52:33 In RDFS, how can I restrict the range of a property to bags with members of a certain class? 13:52:56 sbp has quit 13:57:26 * timbl doesn't know of a way 13:58:17 I have just realized that the chump information is a mapping from bits of comment to documents. An annotation system needs a mapping from documents to annotations. 13:58:50 Has anyone thought of indexing all chumped documents and providing an annotation server? 13:59:15 i'll be doing something like that sometime 13:59:37 Are you maintaining chump? 13:59:40 no 14:00:18 You were thinking of sucking up the chump output? 14:00:26 yes 14:00:33 Why not/ 14:00:54 I think it would be a great way of feeding an annotation server with real stuff. 14:01:37 I suppose you could generate a set of annotatoin documents and then actually contact a running annotation server to make the annotations as an annotating client -- or you coul drun the server. 14:01:51 Well anyway, consider yourself encouraged ;-) 14:02:39 (I had been revieing Snadro's doc and realized he might never know). 14:03:44 which doc, sorry? 14:06:12 davb has joined #rdfig 14:15:13 * AaronSw disconnects 14:17:28 Tim: SemantiChimp uses some annotea stuff. I intend to hook up to an annotation server. Just FYI. 14:18:06 (re doc: C as it happens - a doc sandro chumped early this am BST. the point is he could use a ist of where people have said things about it) 14:20:28 bijan, where is SemantiChimp running? 14:20:44 No where at the moment :) 14:21:06 I have a mostly feature complete version, but other things kept me from setting it up on my normal channel. 14:21:20 * AaronSw reconnects 14:21:22 And CVS frustration kept me from pubbing to to the sourceforge site. 14:21:47 But I can hook it up to a test channel if you'd like. 14:22:42 * AaronSw would really like that. 14:23:04 * bijan tries to remember how to run it... 14:25:44 Well, I fired it up, and it looks like it's running on monkeyfist.com, #testBot 14:28:20 irc://monkeyfist.com/testbot 14:28:51 afs has joined #rdfig 14:29:10 Hi Andy! 14:29:34 ambient, that's quite a FAQ and the answer is there's really no way to do it. 14:30:44 darn 14:31:32 can you do it in DAML? 14:33:42 Hmm, I really don't know... 14:37:20 * look is back (gone 13:19:15) 14:44:05 ambient has quit 14:51:29 sbp has joined #rdfig 14:51:46 Hi there. 14:52:02 Hi 14:52:09 I just updated the NTriples parser 14:52:53 * AaronSw disconnects 14:53:03 Aaron missed the coolest bot command: 14:53:05 > die! 14:53:05 *** Signoff: testbot (bijan HATES me!!) 14:53:07 * AaronSw reconnects 14:54:03 Oops. 14:54:13 Wonder how I got disconnected like that. 14:54:15 Heheh that's a great line. 14:55:37 sbp has quit 14:56:03 davb has left #rdfig 15:45:14 timbl, the problem with using just the namespace is that XHTML documents then can't have semantics. Maybe mixing namespaces will solve this, but I don't know how to mix semantics. 15:45:46 (Or they can't have semantics more than "this is heading text" stuff.) 15:46:34 oierw` has quit 16:14:29 oierw` has joined #rdfig 16:34:28 afs has quit 16:35:28 look has left #rdfig 16:41:46 GabeW has joined #rdfig 16:41:56 danb-lap has joined #rdfig 16:44:21 sbp has joined #rdfig 16:51:14 tav has quit 16:51:14 tav`` has quit 16:51:14 chihchun_ has quit 16:51:14 danb-lap has quit 16:51:14 dc_rdfig has quit 16:51:14 sandro has quit 16:51:14 kifbot has quit 16:51:14 oierw` has quit 16:51:14 GabeW has quit 16:51:14 bijan has quit 16:51:14 Iron_SpermWhale has quit 16:51:15 sbp has quit 16:51:15 jang has quit 16:51:15 xena has quit 16:51:15 timbl2 has quit 16:51:15 MikeM has quit 16:51:15 xbill has quit 16:51:15 lasDesk has quit 16:51:15 em has quit 16:51:15 AaronSw has quit 16:51:15 deltab has quit 16:51:15 mnot has quit 16:51:15 Disconnected from irc.openprojects.net (Connection reset by peer) 16:52:03 logger has joined #rdfig 16:52:03 Users on #rdfig: logger timbl grove dajobe sbp danb-lap GabeW lasDesk em chihchun_ xbill MikeM timbl2 tav`` tav @AaronSw jang bijan kifbot sandro Iron_SpermWhale dc_rdfig deltab xena 16:52:03 This channel is logged and blogged: http://logicerror.com/rdfIRCWelcome 17:26:54 SethR has joined #rdfig 17:29:27 * sandro is trying to work out the semantics of mixing namespaces. 17:30:05 * sbp is trying to work out what to do with SWAG 17:30:30 make a dictionary 17:31:04 We've made a dictionary. Now what? 17:31:58 imho we dont's have a dictionary ... and also we need the ability for people to upload there schema automatically 17:32:40 and then we need trust annotation and the ability to filter retrievals by trust ... all the stuff i scoped in my email 17:34:24 let's face it ... swag died because it didnt provide what was useful 17:34:27 sbp has quit 17:34:40 sbp has joined #rdfig 17:35:25 yeah, I agree 17:36:02 but those things are really difficult to build, and even when you do, getting them open, interoperable, and usable is very difficult 17:36:48 very true ... i dont think it is something that we can just do so very casually .. would take a lot of work .. and perhaps that is the problem ... we would need funding 17:36:59 yes 17:37:02 SethR, can you reference you email? I'm interested in learning more about what you thought the problems with SWAD were 17:37:14 s/SWAD/SWAG/ 17:37:24 err... yeah, that too :) 17:37:26 :-) 17:37:50 it's ongoing discussion on SWAG-Dev: the archives are a mess, but you're free to go through them! 17:38:10 yeah... i'll get right on this :) 17:38:14 my email addressed the oppurtunity, not the problems with the existing swag ... that has not been really voiced except here 17:38:19 the problems with swag is that no one is coding 17:38:31 that's one problem 17:38:36 i gree 17:38:48 what else is there? 17:38:50 anyway em ill chump the email to which i refered 17:38:51 I should send my half finished Python implementation to the list 17:39:00 thanks SethR 17:39:06 the other things are that we lost all sense of direction and organization 17:39:20 we were actually going to *create* vocabularies as well 17:39:23 that was our initial goal 17:39:27 No it wasn't. 17:39:32 then we decided that we needed a dictionary to house them 17:39:33 You're terribly confused. 17:39:40 Yes, it was! go through the logs 17:40:04 no, I'm pretty sure. Well, the original message was: can we agree to share a schema? 17:40:28 things that came from that were that we need more "words" on the SW, and that we need a "dictionary" in which to house those words 17:40:29 And I explained that what he really wanted was a dictionary. 17:40:55 yeah, of couse. The dictionary was the next logical step. I don't disagree with that 17:41:15 http://groups.yahoo.com/group/swag-dev/message/987 17:41:15 D: http://groups.yahoo.com/group/swag-dev/message/987 from SethR 17:41:20 """ 17:41:22 I 17:41:22 dont't want to have my stuff not be easily aggregated with 17:41:22 others. Consequently I would like to share a vocabulary 17:41:22 with a group of people who are seemingly going in the same 17:41:22 I'm just saying that we were meant to be filling the dictionary with our own stuff as well as other people's, and we only just got started 17:41:23 direction visa vie the semantic web. I find that both you 17:41:23 Aaron and you Sean share an important attitude with me. 17:41:26 """ 17:41:28 that was the first message. 17:41:36 """ 17:41:38 1) Agree to import as many of the existing standards schemas 17:41:39 that we can agree upon .. may I suggest the following for a 17:41:39 start: 17:41:39 """ 17:41:46 well, Seth's here. Just ask him 17:42:00 It's clearly layed out in: http://groups.yahoo.com/group/swag-dev/message/5 17:42:17 if you'll recall, there was a slight revival when Seth and I started to work on some vocabularies 17:43:39 actually i think you are both right .. we wanted a dictionary AND we wanted to be able to add our own words to the dictionary 17:43:40 example of early vocabulary work: http://groups.yahoo.com/group/swag-dev/message/42 17:43:51 yes, I'm saying that we needed to do both 17:44:02 I'm also saying that we neglected the vocabulary side of things 17:44:32 well i think aaron is right ... we didnt write the code to do what we wanted done 17:44:49 we all just sat back and assumed that aaron was gonna do it 17:44:55 heh, heh :-) 17:44:58 :-) 17:45:14 yeah, that was rather sucky of us 17:45:16 at least i did .. cause im soooo slow at actually coding anything now 17:45:35 Well, we have the Python implentation work on 17:45:50 If we port it over to use the RDF API, then we can be compatable with that line of work 17:46:11 lurk_pef has joined #rdfig 17:46:50 I didn't really feel comfortable knowing that we were using Oracle, TCL etc. I mean, they're really good systems, but isn't Oracle rather expensive? 17:47:07 With Python, people can just drag it off the Web, contribute to it, install it, and so on 17:47:15 * em wonders if http://potlach.org:8000/ would help/hurt some of these discussions 17:48:05 Oracle is effectively interchangable with any other DB for this project 17:48:14 and there are many free RDBMSs 17:48:18 hmm... haven't added the searching capabilities on this interface yet. 17:48:27 cool, an RDFS viewer. I wrote one in XSLT once, and I think the SWAG-D has one... but yours seems quite advanced 17:48:44 but surely you'd prefer Python, Aaron? 17:48:51 nowadays 17:49:04 em, thats similar to the original vision i think .. have you seen the suo ontology navigator? 17:49:08 yeah, we need really decent querying abilities 17:49:11 em just borowed the Redfoot stuff ;) 17:49:17 yep :) 17:49:23 plus soe added code base 17:49:36 * sbp has to move SWIPT.query over to use RDF API too... ugh 17:49:37 SOE? 17:49:42 err.. some 17:50:32 and i think the annotation and filtering to annotation is really improtant .. it makes it work .. otherwise you will get chaos 17:51:21 that allows chaos to rule, but you can still retieve solid stuff .. if we had this today, ill bet we could convert the IEEE to use it ... really i think we could ... too bad we dont have it though 17:52:04 we have the beginnings of a dictionary 17:52:23 so what do we do need? Keep on keeping on? Port to Python? Something else? Tell me 17:55:00 oierw` has joined #rdfig 17:55:02 SethR, suo ont navigator ... no. pointers? 17:55:13 sbp has quit 17:55:15 if it were me, i would say use the python and interface it to something like Berkeley or even Oracle since we seem to have that available 17:55:39 just a second, em 17:56:21 http://ontology.teknowledge.com:8080/rsigma/index.html 17:56:21 E: http://ontology.teknowledge.com:8080/rsigma/index.html from SethR 17:57:28 E:|IEEE Standard Upper Ontology (suo) ontology browser 17:57:29 titled item E 17:58:54 GabeW has quit 17:59:15 E: first time you use it you need to select an ontology ... select Merge 17:59:16 commented item E 18:01:36 D: Suggestion for a democratic dictionary to SUO and to SWAG 18:01:36 commented item D 18:02:11 GabeW has joined #rdfig 18:05:24 bijan has quit 18:11:40 lurk_pef has quit 18:11:43 http://search.dataconsortium.org/ 18:11:43 F: http://search.dataconsortium.org/ from SethR 18:11:48 thanks SethR for the pointer... 18:12:05 f:|Data Consortium Dictionary 18:12:23 F:|Data Consortium Dictionary 18:12:23 titled item F 18:12:49 F: seems close to what i was proposing for SWAG 18:12:49 commented item F 18:13:28 SethR - I know that John McClure (of dataconsortium.org and legalxml.org) would love to work with RDF'ers 18:14:41 the SUO is availiable in daml... i've added this to the schema navigator along with the daml vocab... simple example http://potlach.org:8000/view?subject=http%3a//reliant.teknowledge.com/DAML/SUO.daml%23BinaryRelation 18:14:48 hmm... nice 18:15:07 * em notes to add search service component tonight 18:16:58 * em curses the fact that no one uses isDefinedBy 18:17:05 Heh. 18:17:06 your schema navigator is cool too .. first time i saw it 18:17:17 been a bit silent on this SethR... 18:17:44 i'm planning on porting this to a more stable machine shortly have provide codebase back to redfoot team 18:18:30 you should provide an jxta pipe into it 18:18:37 but its a continuation of the early work on the DCMI registry work.. just in different code base 18:18:55 the DCMI registry work included the jxta connections 18:19:51 who is potlach.org ? 18:19:58 http://onionnetworks.com/caw/caw.txt 18:19:59 G: http://onionnetworks.com/caw/caw.txt from AaronSw 18:20:09 G:|HTTP Extensions for a Content-Addressable Web 18:20:09 titled item G 18:21:50 G:Some very interesting uses of P2P, hashes, URN, URN resolution and HTTP 18:21:50 commented item G 18:26:35 http://robustai.net/images/semCloud.jpg 18:26:36 H: http://robustai.net/images/semCloud.jpg from SethR 18:26:55 H:| Picture of the semantic cloud 18:26:55 titled item H 18:27:48 H: by my sister :) 18:27:48 commented item H 18:29:17 nice - is there a version without so noticeable JPEG artifacts? 18:30:36 well since she's my sister i could ask her to adopt it any way anyone would like 18:32:54 i was hoping she would animate it .. make the cup float into the cloud .. stuff like that 18:50:48 sandro, how's TAG coming? 18:50:57 Perhaps we should go ahead with PTS after all? 18:51:06 mnot has joined #rdfig 18:51:17 Hi mnot 18:53:43 the tag URN NID has been submitted to the IETF. just a couple days ago, I think. 18:54:51 Oh? To the URN-NID list for review or right to IANA? 19:03:57 hody 19:04:00 howdy, even 19:04:12 hey there hi there ho there 19:04:24 :nodes... 19:04:33 doh... wrong system... 19:04:35 * MikeM nodes 19:04:41 no comments yet either.... 19:04:57 ?? 19:05:37 MOO user, MikeM? 19:05:43 * MikeM nods 19:06:00 * AaronSw goes to see if the #rdfig is back up yet... 19:08:13 s/#rdfig/#rdfig MOO/ 19:10:24 * timbl Installing Active python wants to overwrite files being used by mIRC32, IE, Wordpad, Outlook, Word... ouch ... oh welll. 19:10:34 timbl has quit 19:12:21 hazmat has joined #rdfig 19:13:23 sbp has joined #rdfig 19:14:38 * em curses the fact that no one uses isDefinedBy 19:14:49 I'm partially guilty of that, I must admit... 19:15:01 one of my few RDFS bad habits 19:15:40 * sbp chases up TAG-URN 19:18:03 Hmm... seems that the urn lists seem to be split up 19:21:26 SethR has quit 19:21:45 AaronSW, I really don't know. I'm not paying that much attention. 19:22:32 * AaronSw disconnects 19:22:46 * AaronSw reconnects 19:23:54 hmmph 19:24:36 Is it pure coincidence that MikeM shows up moments after Aaron asks me about tags? Or is this a different MikeM? 19:24:48 Heh heh heh. 19:24:56 We're everywhere... 19:25:37 It was submitted to urn-nid@apps.ietf.org on 19 October. 19:26:31 er... that's the 01 draft, isn't it? 19:26:33 Yeah. http://www.ietf.org/internet-drafts/draft-kindberg-tag-uri-01.txt 19:26:40 seen that, read it, patented the T-Shirt 19:26:40 that, has never been seen. Use "seen ?" for help. 19:26:56 heh, heh: thanks, xena! 19:27:19 That draft's been up since about Sep 15 or so. 19:27:28 Yeah 19:28:46 the comment period for the urn-nid list is two weeks I think. If I can count correctly you should be able to ask the RFC-Editor to publish it at the end of next week.... 19:29:04 Of course now I'm deeply in the other camp, trying to make the semantic work entirely on fetching XML namespace URIs. 19:29:15 blargh 19:29:24 hmm.... looking at the list there was the request from some that tag and Larry's dated URI stuff be merged.... hmm..... gotta go re-read to see where that leaves both proposals.... 19:29:36 crud, I'd like to discuss this more, but I've gotta run 19:29:37 c'ya 19:29:40 sbp has quit 19:30:54 Hey, if we're gunna do that we might as well make it a WG (Sean, Timothy, Me, Larry, ....) and get this a standards-track URI. :-) 19:31:09 Heh heh. 19:31:29 We can just call this the WG... 19:36:05 timbl has joined #rdfig 19:38:36 Actually, I can't see Larry every agreeing to semantics-free identifiers. 19:38:49 danb-lap has quit 19:38:57 Which is why the two proposals can't be merged. 19:40:14 danb-lap has joined #rdfig 19:44:54 * AaronSw disconnects 19:47:48 danb-lap has quit 19:48:43 danb-lap has joined #rdfig 19:57:45 SethR has joined #rdfig 19:59:56 sbp has joined #rdfig 20:08:22 http://dailynews.yahoo.com/h/cn/20011025/tc/msn_com_shuts_out_non-microsoft_browsers_1.html 20:08:22 I: http://dailynews.yahoo.com/h/cn/20011025/tc/msn_com_shuts_out_non-microsoft_browsers_1.html from dajobe 20:08:28 I:|MSN.com shuts out non-Microsoft browsers 20:08:28 titled item I 20:10:54 I:Eek, it's true. I get a page with the heading "Attention: Web Browser Upgrade Required to View MSN.com 20:10:54 commented item I 20:11:07 I:"All of our development work for the new MSN.com is...W3C standard," ""For browsers that we know don't support those standards or that we can't insure will get a great experience for the customer, we do serve up a page that suggests that they upgrade to an IE browser that does support the" standard" 20:11:07 commented item I 20:11:32 that's absolutely appaling 20:11:43 I:apalling isn't it? 20:11:44 commented item I 20:12:39 * dajobe can't spell 20:12:54 I'm absolutely shocked... I can't believe that they could be so ignorant 20:13:08 * sbp gets ready to warn the WAI 20:16:55 SethR has quit 20:19:30 I:Heh, and they say that they're W3C standards complaint? Their page [http://validator.w3.org/check?uri=http://www.msn.com/|doesn't validate], and it doesn't follow WCAG 20:19:30 commented item I 20:19:51 I:s/complaint/compliant/ 20:19:51 commented item I 20:22:42 * danb-lap just tried msn.com in NS4.7, got through OK 20:22:59 MSN just got /.-ed 20:25:57 * sandro can't get through with Opera. 20:27:24 I was shut out on Mozilla 0.9.5 20:40:16 http://rdfweb.org/people/danbri/2001/10/dot/jdot/ 20:40:16 J: http://rdfweb.org/people/danbri/2001/10/dot/jdot/ from danb-lap 20:40:34 J:|Dot graph layout code, in Java (by Joel Crisp) 20:40:35 titled item J 20:41:14 J:Joel (ex-ILRT hacker) sent me this a couple years ago, I never really investigated. Something like a Java version of GraphViz? 20:41:14 commented item J 20:48:48 danb-lap has quit 20:55:31 larsbot has joined #rdfig 20:57:21 sbp has quit 21:03:10 http://www.w3.org/DesignIssues/Business.html 21:03:10 K: http://www.w3.org/DesignIssues/Business.html from timbl 21:03:26 K:|Business Model for the Semantic Web 21:03:26 titled item K 21:03:51 K:An attempt to explain why it important for you and your company. 21:03:51 commented item K 21:03:58 K:Draft 21:03:58 commented item K 21:04:49 JHendler has joined #rdfig 21:06:45 em, K needs those diagrams you were going to russle up! 21:07:00 timbl is now known as tim-gone 21:10:39 shellac has joined #rdfig 21:12:05 JHendler has quit 21:29:08 Bayta has joined #rdfig 21:31:31 shellac has quit 21:39:38 mnot has quit 22:05:51 sbp has joined #rdfig 22:07:22 sbp has quit 22:07:26 sbp has joined #rdfig 22:09:40 cool "Business Model" for the SW thing 22:10:59 a QOTW contender: "For data, we are still pre-Web." 22:11:33 K:Great quote: "For data, we are still pre-Web." 22:11:33 commented item K 22:12:12 Of course, RDF doesn' 22:12:21 doesn't make transferring data 100% easy 22:12:40 it makes it a lot easier if someone hs already made the connections, and carefully published the data 22:32:54 sbp has quit 22:36:33 .google 4suite 22:36:34 4suite: http://4suite.org 22:48:59 mnot has joined #rdfig 22:58:19 dajobe has quit 23:01:27 sat has joined #rdfig 23:05:16 sbp has joined #rdfig 23:07:33 * sbp doesn't like the constant allusion to "namespaces" in the RDFS draft 23:08:27 K: For EAI, exchanging RDF documents may not be very attractive, given performance concerns. Typically they convert XML to propritory format at adapters and then send it to broker. 23:08:28 commented item K 23:08:31 or the fact that rdfs:comment is still very "raw" 23:09:29 I am not sure if EAI is really good application for RDF..I have had such a hard time emitting events in XML format to adapters. People wanted me to use canonical format 23:09:35 * sbp should bundle up these concerns and send to www-rdf-comments... 23:12:08 Pfffff http://onionnetworks.com/caw/caw.txt not found 23:12:50 I like to think Semantic web makes plain data into knowledge for your computer 23:12:57 aw, and no Google cache either... 23:13:08 define "plain" data :-) 23:13:40 s/plain data/data 23:13:49 heh 23:14:38 I think that *one* of the aims of the Semantic Web is to make that data easily accessible, sure. But proprietary formats could be said to make "data into knowledge for your computer" 23:15:26 the twist of the SW (and it's a twist that depends on a great deal of previous work) is the interoperability gained by using a common model (RDF), and URIs to identify the terms. It supposedly make it esier to transfer information 23:15:57 and to evolve it, etc. In my experience, it doesn't make it any easier to maintain, and it does incur some overhead in that you have to learn about RDF first 23:16:15 but once you have a couple or few systems going, the benefits are there straight away 23:17:10 agree, but there isn't much economic benifit for a typical enterprise vendor who likes to sell you a broker to do this integration 23:17:54 the economic benefit is that we're going to do this anyway, and once more and more people start to get into it, if you don't hop onto the train, you're going to be left behind 23:18:13 (if it succeeds). So you either pioneer, or you trail :-) 23:18:33 yes, that is the hope. It has to be driven not by companies, but by developers. 23:18:52 yeah. Grassroots at first, and then the companies can pile in... 23:19:00 but its much harder than html stuff, so how many developers will join grassroots is a question 23:19:31 Yes, that is true. And of course, benefits from HTML were probably more obvious, because you could "see" (triple quote) the documentation 23:19:50 but there are quite a few developers devoted to making the SW work, and an awful lot moer out on the fringes, intregued 23:20:08 perhaps, one of the first applications could be this national ID program, folks are talking about 23:20:22 national ID program? Blargh 23:20:47 in US, they seem to be talking about need for a national ID program....we should develop a simple schema for it 23:21:25 :-) 23:23:27 Is RDF grammar a context free grammar?. 23:23:47 what do you mean by "context"? 23:24:16 I should clarify, as per the chomsky's hierarchy, type 0, type 1, type 2, type 3, I was wondering if its really type 2 23:25:07 RDF's XML encoding is not context-free, at very least because of namespaces. There are various other features which I also think are not, but I haven't really tried to see if there is a CFG approach to them. 23:25:45 On the other hand, RDF's real grammar is a graph grammar. I have no idea how to apply Chomsky to graph syntaxes. 23:26:41 ok..I should probably ask my professor too, he is teaching me formal languages 23:27:32 RDF's XML coding: also because it specifies that the front and end tags are optional, but in reality, you can't just have one? 23:27:38 Are you curious about RDF/XML, or RDF's graph syntax? 23:27:53 RDF/XML 23:28:23 sbp -- no, if the you read the paragraph in M&S before the BNF, it's not to be taken that way. 23:28:34 * sbp reads 23:28:42 Yeah -- namespaces blow CFG out of the water. 23:28:49 they're basically macros. 23:29:04 (as far as a parser is concerned) 23:29:20 ah, yeah 23:29:25 * sandro hopes he understood sbp and M&S right. :-) 23:30:25 yeah, it says "Syntactic features inherited from XML are not reproduced here. These include all well-formedness constraints, " so I guess that means that it's implied 23:32:03 mnot has quit 23:33:29 Yeah. Nowadays I would say: the document is assumed to have been first processed into canonical XML before this grammar is applied. 23:33:47 yeah, that's a good way of putting it 23:34:44 Although that still doesn't address the namespace issue. There should be a CXML with namespaces all expanded, looking sometihng like <{namespaceFullName}element {namespace}attribute="value"> or some such. 23:34:55 Then you'd have a stab at a CFG. 23:35:03 (not that it really matters.) 23:44:15 sat has quit