Semantic Web Interest Group IRC Chat Logs for 2005-01-15

03:25:23 <kao_home_> kao_home_ is now known as kao_home

07:25:24 <sbp>http://slashdot.org/articles/05/01/14/2147259.shtml?tid=95&tid=217

07:25:24 <dc_swig> A: http://slashdot.org/articles/05/01/14/2147259.shtml?tid=95&tid=217 from sbp

07:25:40 <sbp> A:|On Finding Semantic Web Documents (/.)

07:25:41 <dc_swig> Titled item A.

07:25:55 <sbp> A:[http://swoogle.umbc.edu/|Swoogle] gets slashdotted.

07:25:55 <dc_swig> Added comment A1.

07:27:30 <sbp> hmm, 46 megatriples

08:30:02 <sh1mmer> moin

09:36:07 <bblfish> Hi I was wondering if anyone had any extra feedback on the thoughts I expressed in <http://lists.w3.org/Archives/Public/www-rdf-interest/2005Jan/0090.html>

09:36:07 <bblfish> (like "that was discovered 4 years ago", or I something I completely forgot to take into consideration...

09:36:39 <md-afk> md-afk is now known as PhUrl

09:38:09 <PhUrl> crschmidt: you awake

09:38:31 <PhUrl> good morning all

09:38:42 <bblfish> good morning

09:38:47 <PhUrl> <crschmidt> i think the telephone codes are what pushed it over

09:39:03 <PhUrl> I have been working with telefon codes for years

09:39:27 <PhUrl> in order to work with them effectivly, you need to build a tree structure in memory

09:39:34 <PhUrl> a radix tree

09:40:50 <PhUrl> databases just dont cut it for telefon numbers and codes

09:47:17 <PhUrl> comment about A: """"Plus a lot of the rah-rah booster club that's grown up around it sound a whole lot like the Royal Society folks in Quicksilver who keep trying to catalog everything in the world into a 'natural' organization.""

09:55:08 <sbp> bblfish: yeah, I was going to ramble on about how you can get an infoset representation in RDF just fine but that it doesn't mean anything other than being an infoset representation#

09:55:10 <sbp> s/#//

09:55:55 <sbp> and that the whole idea of getting a generic semantic from XML is silly, and fundamentally flawed since XML is there as a *generic* syntax upon which to build applications, and those applications can express their semantics in any way that they choose in the serialisation

09:56:24 <sbp> you can't just retroactively constrain everybody who uses XML to a single semantic interpretation just because it suits your needs on one or two applications

09:56:57 <sbp> and yet you're inferring that you can do exactly that just by deriving some prima facie bizzare argument based on the fact that most XML files you've looked at seem to have some containment semantic

09:57:14 <sbp> it's like saying that all trees are orange trees just because the only trees you've ever seen are orange trees

09:57:23 <sbp> so frankly I think you should give up

09:58:03 <sbp> and learn Notation3 instead of using the all-bandwith-consuming ASCII art triple diagrams, which are actually quite difficult to read in their redundancy

09:58:10 <bblfish> no, I was woring on what Roy Fielding said...

09:58:10 <bblfish> s/woring/working/

09:58:16 <sbp> yeah yeah

09:58:24 <PhUrl> i think that it is possible to interpret xml files into a set of facts about them given specific knowledge about the xml itself.

09:58:30 <bblfish> yes, I am trying to speak to people who may potentially not be deeply into rdf

09:58:49 <sbp> "almost all XML data formats"

09:58:52 <PhUrl> just because it is not a standard rules, but freeform does not mean it does not have semantics

09:59:03 <sbp> he didn't say that all of them follow that particular arrangement

09:59:15 <sbp> yet you went ahead and misrepresented him anyway

09:59:54 <bblfish> (btw. I do notation 3)

10:01:05 <bblfish> I was following a hunch... I found that it oddly gets one very close to rdf anyway. I was just going to look at the relation.

10:03:31 <bblfish> There are many odd things. Atom is very close to rdf (just a few changes away). Atom interpreted just with an ontology gives us very close to what we want...

10:04:39 <PhUrl> xmlschema is similar to owl in some respects

10:05:02 <PhUrl> i made a xslt to turn a specfiic xsd into an owl

10:05:41 <PhUrl> it is really all about finding the meaning of a document and turning it into something usable

10:06:33 <bblfish> PhUrl: OWL has some nice OO and declarative elements.

10:06:42 <sbp> bblfish: trying to get XML people into RDF is good; fundamentally, the ethos of what you're doing is good. but you're misrepresenting XML, or at best misunderstanding it, and also portraying an image of RDF-folk that isn't all that savoury (not that we're not all guilty of it to some extent)

10:07:06 <bblfish> sbp: what is the image?

10:07:30 <bblfish> after all its all SemWeb stuff (OWL is)

10:07:42 <sbp> the image is the Mark Pilgrim effect, that of seeming to want to have an RDF semantic for all XML, all data in general

10:07:46 <PhUrl> bblfish: yea, OWL has more than xsd. But to begin with, you want to know the elements and containment relationships for example

10:08:44 * bblfish need to learn more about xsd (a lot more)

10:08:57 <bblfish> oh, all data has an mapping into tripple space :-)

10:09:27 <sbp> triples aren't the most efficient representation at the best of times though

10:09:36 <bblfish> (unless it is inconsistent)

10:09:42 <sbp> it's very often better to stick to native formats even for interchange

10:10:24 <bblfish> oh. I agree.

10:10:24 <bblfish> I want xml ers to use xml

10:10:34 <bblfish> (xml is not particularly efficient either btw)

10:10:37 <sbp> especially then the data is naturally >3 tuples...

10:10:44 <sbp> no, XML's evil :-)

10:11:02 <sbp> I like Bray's strawman for a simpler XML

10:11:27 <sbp> then you've got 1.1 pushing in the other direction due to i18n etc.

10:11:34 <sbp> can't complain though

10:11:38 <bblfish> but the next thing I want to show, is how having an ontology for your xml can help you design good xml.

10:12:15 <sbp> that I agree on too. the only thing I'm really grumbling at is having a universal ontology for all XML documents

10:12:40 <sbp> and the ASCII art triples :-)

10:12:50 <bblfish> no universal ontology. just a simple mapping...

10:13:24 <bblfish> (yes I am giving up on the ascii art. It is a lot of work, and on many mail clients it never gets through right, which is a real pitty)

10:13:47 <sbp> but you can't construe anything from any arbitrary XML document without a specification for the language that it uses

10:13:54 <sbp> (awesome)

10:14:37 <sbp> that is to say, anything other than the syntax--anything that's there in the infoset

10:14:47 <bblfish> true, you could get junk xml

10:14:50 <sbp> but you can't get from the infoset to the intended semantics very easily

10:14:52 <bblfish> that is why I need an ontology.

10:15:12 <sbp> well, take for example RDF/XML. how would your "simple mapping" do with RDF/XML? what triples would result?

10:15:29 <bblfish> xml+ontology = rdf mapping

10:15:31 <sbp> I'll bet they wouldn't be the triples that you get from parsing an RDF/XML document according to the RDF/XML Syntax Specification

10:15:46 <bblfish> well I was just going to look at that. Perhaps it works too... :-)

10:16:26 <sbp> I don't think so--your proposal, from what I can tell, looks unstriped

10:16:42 <sbp> so it'd come out pretty weirdly if you had, as most RDF/XML documents do, typed nodes all over the shop

10:17:35 <sbp> that is to say, your proposal generally places blank nodes between each of the elements and then treats the elements themselves as RDF properties, right?

10:17:48 <sbp> whereas in RDF/XML you're using elements as classes quite a bit

10:17:56 <sbp> so you're going to end up using classes as properties

10:18:00 <bblfish> elements and attributes are rdf properties

10:18:03 <sbp> which is inconsistent since they're disjoint

10:18:23 <bblfish> ah, no

10:18:23 <sbp> well there you go then. you immediately fail when you apply the mapping to RDF/XML

10:18:26 <bblfish> hehe

10:18:45 <bblfish> mmmhh

10:19:08 <bblfish> yes. I see.

10:19:31 <sbp> I think it's not a tricky problem, it's an impossible one, sadly

10:19:48 <sbp> because you can never tell which are properties, which are classes, which are instances

10:20:48 <sbp> but that shouldn't stop people from coming up with such mapping that users of XML can then say that they conform to. so in other words, you're trying to find a mapping for all XML documents, but that won't work, so instead why not try to come up with a series of mappings that people can then choose between and overtly *say* work for their particular documents/languages?

10:27:41 <PhUrl> msg NickServ IDENTIFY mdupont

10:27:45 * PhUrl changes his password now

10:29:39 <PhUrl> sbp: bblfish i must have timed out. Did i miss anything? :)

10:30:06 <sbp> PhUrl: http://ilrt.org/discovery/chatlogs/swig/2005-01-15.html#T10-08-57

10:30:21 <sbp> too much to paste

10:32:06 <sbp> bblfish_: and http://ilrt.org/discovery/chatlogs/swig/2005-01-15.html#T10-19-31 for you :-)

10:33:03 <PhUrl> ok, it looks like all my comments got lost

10:33:09 * PhUrl summarizes

10:34:27 <PhUrl> in the end, the types of objects and the relationships between them can be expressed in unlimited ways. when you start looking at the possibilities of encoding the same information in c, for example : you can write struct point { int x,y,z;}; for an explicit struct. but also int point_array [3]; for just the data. and if you are hardcore, you can write int * point_p; and allocate the data off the heap. or even pack the entire point into a "unsig

10:34:27 <PhUrl> ned long long point_long;". and there are thousands of other equivalant representations.that is before you get into relative points and polar coords.so, i think that xml is more like c in the you have millions of possible interpretations, each reprsentation and interpretation schema will have be transformed into some formal semantics by someone who understands the message,not by a machine. unless of course you can extract the semantics from the s

10:34:32 <PhUrl> oftware the processes the xml itself, from the source code.

10:35:04 <sbp> yeah. good summary

10:40:24 <bblfish__> sbp: you could make your point above on the "Re: xml without rdf, but with an ontology [0]" thread. I think that is a good point.

10:41:23 <sbp> fair enough; will do

10:42:08 <bblfish__> oh, yes you could cc the atom list too on that one, since I included them in it.

10:43:37 <bblfish__> what we do have though is URIs: these are universal, and xml is just a tree of uris.

10:44:03 <bblfish__> bblfish__ is now known as bblfish

10:46:25 <bblfish> oh well, this is too complicated to spend too much more time on...

10:53:11 <sbp> nice: http://www.fakeroot.net/sw/rdf-formats-20040717/#features_overview (via dajobe)

10:53:23 <PhUrl> PhUrl is now known as md-afk

10:54:00 <sbp> N3 and RDF/XML have the most features on those criteria

10:54:26 <sbp> the genaeology is cool too. shame he misses the AEF

10:54:56 <sbp> (and there hundreds of mini-proposals that didn't go anywhere, the HTML WG's latest RDF-in-XHTML stuff, etc.)

10:55:02 <sbp> s/there/the/

10:56:12 <sbp> TriG's a bit odd; all this modularity of N3 syntax is leading odd places

10:57:27 <sbp> N-Triples/Turtle/N3 I can see the logic behind, but we're going to end up with one of those colour books of N3-variants at this rate

10:57:56 <sbp> "oh, I'll go for Eggshell White N3 for my application. Wheatfield N3 is a bit too over-developed"

11:05:20 <bblfish__> bblfish__ is now known as bfish

12:14:36 <md-afk> md-afk is now known as PhUrl

13:17:43 <PhUrl> PhUrl is now known as md-afk

14:01:52 <babilen> hi all....

14:02:23 <babilen> i am a bit confused on how to use rss 1.0...

14:02:42 <crschmidt> babilen: By "use", do you mean to create or consume?

14:02:57 <babilen> i mean create....

14:03:28 <crschmidt> babilen: Alright, go ahead

14:03:39 <babilen> i want to syndictate content i have on my website, (html 4.01) as an rss feed but want to add as much metadata as i can....

14:03:54 <babilen> but i reallly do not get where this metadata shall come from....

14:04:31 <babilen> and when i use an automated script to generate rss feeds from my html files, i can't see a way to add this metadata...

14:04:58 <crschmidt> babilen: well, what kind of metadata do you want to add?

14:05:09 <eaon> usually rss is created by a content management system that creates both html and rss/rdf

14:05:59 <babilen> ok, i get the idea of an cms, but in which format will the data of this cms be stoerd, in order to be able to create html files and rss feeds?

14:06:12 <sbp> usually database backed

14:06:18 <crschmidt> babilen: typically, systems use a database to store the data and then create the data from there.

14:06:35 <crschmidt> Some (Wordpress, Drupal) do it on the fly, others (Movable Type) regenerate from time to time.

14:07:18 <babilen> ok, but what does this mean. i edit my thml files by hand right now, and want to settle on an format which gives me the possibility to create meta-data rich web sites and rss feeds...

14:07:35 <babilen> all these systems are meant for blogs, aren't they?

14:08:00 <crschmidt> Those are, yes. Others aren't: Plone, I think, is a general content management system

14:08:13 <crschmidt> babilen: can you give an example of a page you're trying to add RSS to?

14:09:38 <babilen> it's a page of my university department, i wanted to create an rss feeds, which gives information on what has changed on the website.

14:10:30 <crschmidt> babilen: In general, updating HTML + RSS by hand gets annoying after a while. If you want to do it though, I'm sure we can walk you through an example?

14:12:07 <babilen> ok, we have an announcement page, but the content of this page reaches from mp3 files of lectures, to announcement of the books we bought, to intresting discussions on our wiki....

14:12:41 <babilen> i would like to add meta-data to all of this....

14:13:39 <babilen> so that an, possible non-existend, rss reader would display them properly..... like gives all the information it has on the artist of an mp3 file, or the author of an book....

14:14:58 <crschmidt> babilen: Okay. How far have you gotten? What are you having problems with?

14:16:08 <md-afk> md-afk is now known as Phurl

14:16:23 <Phurl> back, hi crschmidt

14:16:35 <Phurl> <crschmidt> i think the telephone codes are what pushed it over

14:16:36 <babilen> i can create rss feeds with an script, but all i have in there is the text i have on the web page, but no information what's behind it.....

14:16:59 <Phurl> I have been working with telefon codes for years. in order to work with them effectivly, you need to build a tree structure in memory. a radix tree. databases just dont cut it for telefon numbers and codes

14:17:38 <Phurl> for american codes, it is possible to map them sometimes onto xy coords and then use a geometric index

14:17:55 <babilen> and i am not sure, where i should add this information. you mentioned cms systems, as i have never used one i am not sure what possibilities i would have, if i would use one.....

14:18:07 <crschmidt> babilen: okay, don't worry about that.

14:18:14 <Phurl> babilen: sorry to interrupt

14:18:15 <crschmidt> babilen: Do you know what RDF is or how it works?

14:18:44 <babilen> yeah, i read the rdf primer and try to figure out how that fits into the whole pictures......

14:19:19 <crschmidt> babilen: Have you created an RSS 1.0 feed yet?

14:19:26 <babilen> i understand it as a way to express ontologies....

14:20:14 <babilen> i started to write a script in python to create one.....

14:20:52 <babilen> my problem is more a kind of chicken and egg problem......

14:21:25 <babilen> as i started to parse the html file, i thought that it might be quite cumbersome to add all the meta data....

14:21:57 <babilen> so i thought i better come here and ask, which format should be the basis for my content....

14:22:15 <sbp> .g GRDDL

14:22:23 <phenny> GRDDL: http://www.w3.org/2004/01/rdxh/spec

14:22:23 <dc_swig> Label GRDDL not found.

14:22:28 <sbp> try that

14:22:45 * crschmidt has to run, bbiab

16:14:50 <d2m> sbp: any chances you add your pyrple/afon N3 parser to rdflib ?

16:15:11 <sbp> afon is out of date now

16:15:15 <sbp> n3p has taken its place

16:15:21 <sbp> http://inamidst.com/n3p/

16:15:27 <sbp> and very possibly

16:15:56 <sbp> eikeon and I have discussed working on a joint venture, but it's moving ahead a bit slowly mainly because I have some reservations about the whole thing that I haven't discussed with him yet

16:15:59 <d2m> hmm, good news :)

16:16:48 <sbp> and what's wrong with pyrple?

16:17:28 <d2m> nothing wrong, works fine for me

16:17:38 <sbp> why do you need afon ported to rdflib then?

16:17:56 * sbp points the spotlight at d2m

16:18:03 <sbp> eh? eh?

16:18:20 <d2m> just a though while at the point

16:18:47 <sbp> fair enough... :-)

16:19:05 <sbp> seriously, as I say, it'd be nice to merge them both entirely

16:19:12 <sbp> then everybody can have the best of both

16:37:56 <balbinus_> balbinus_ is now known as balbinus

17:27:09 * bijan notes, belatedly, that rdflib is getting a turtle parser for sure (ply based)

17:27:45 <dajobe> is that public?

17:27:56 <bijan> Hmm.

17:28:00 <bijan> I want to say "yes"

17:28:04 <bijan> It's in our svn

17:28:06 <bijan> Which is public

17:28:11 <dajobe> that'll do

17:28:29 <bijan> Hey dave, have you filmtrusted yet?

17:28:41 * bijan in filmtrust pimpage mode

17:28:43 <dajobe> I didn't follow the link

17:28:59 <dajobe> is that http://svn.mindswap.org/rdflib/trunk/rdflib/syntax/parsers/

17:29:00 <bijan> I'll invite you

17:29:03 <dajobe> files named N3* ?

17:29:28 <bijan> That's n3

17:29:33 <bijan> there's specifically turtle, I believe

17:29:37 <dajobe> hmm

17:29:52 <bijan> But where, I have no idea

17:30:35 <bijan> I will buzz ron about it

17:30:36 <dajobe> I'm >< this close to adding """strs""" to turtle

17:30:40 <dajobe> but I ran out of time

17:30:55 <dajobe> need to check what's escaped rules in python/n3

17:31:23 <dajobe> found it in http://svn.mindswap.org/rdflib/trunk/test/

17:31:30 <bijan> Weird place for it :)

17:31:45 <dajobe> yeah, guess it'll svn mv to the right place

17:32:18 <bijan> Yep

17:32:42 * bijan hasn't touched it yet, himself

17:33:12 <bijan> Argh! FilmTrust hasn't grown ALL HOUR!!!

17:33:12 <bijan> :)

17:33:45 <dajobe> I was looking at my imdb rates (many years of them). No easy export

17:33:49 <dajobe> imdb ratings

17:34:14 <bijan> Ohhhh too bad

17:34:35 <d2m> bijan: there is a way to check this out ?

17:34:36 <bijan> Still, people ratings are probably overall more important

17:34:49 <dajobe> do what I did, svn co <url above>

17:34:52 <crschmidt> d2m: http://trust.mindswap.org/FilmTrust/

17:34:57 <dajobe> ah

17:34:59 <crschmidt> d2m: oh, unless you didn't mean filmtrust

17:35:00 <bijan> Yeah, the this was ambiguous for me

17:35:03 <bijan> So I stalled :)

17:35:05 <crschmidt> heh

17:35:13 <crschmidt> Well, now he has both answers!

17:35:14 <d2m> crschmidt: no i ment the rdflib

17:35:17 <bijan> Indeed

17:35:29 <bijan> d2m, it's public svn

17:35:32 * crschmidt is a filmtrust user!

17:35:35 <bijan> co away

17:35:38 <bijan> YAY CRSCHMIDT!

17:35:46 <bijan> you get one snark free answer of your choice from me

17:35:52 <d2m> bijan: but your server refuses my connection

17:35:56 <crschmidt> Wait, those exist?!

17:36:02 <bijan> Yes

17:36:03 <crschmidt> ;)

17:36:06 <bijan> And you just used yours up

17:36:07 <bijan> Too bad

17:36:10 <sbp> heh!

17:36:41 <bijan> d2m...no idea why

17:36:43 <bijan> Shouldn't

17:36:51 <bijan> can you get there in a browser?

17:36:56 <d2m> yes

17:37:22 <bijan> Well...er...it's not really different I dont' think, but svn and i have a hate thing going on

17:38:11 <sbp> firewall?

17:38:16 <d2m> yes

17:38:17 <sbp> svn uses port 3690

17:38:30 <bijan> Hmm.

17:38:32 <eaon> not the http version

17:38:59 <sbp> ah true, that's only for svn:// URIs

17:38:59 <bijan> I find that a firewall would say that the *server* refused the connection to be a bit confusing

17:39:13 <bijan> I mean, if it's blocking

17:39:16 <sbp> yeah. could've been a misreport though

17:39:29 <bijan> Should it say, "Yo dude, don't touch that!"

17:39:35 <bijan> shouldn't it

17:39:40 <Phurl> Phurl is now known as md-afk

17:39:59 <sbp> if it's a fir3wall, yeah

17:40:10 <bijan> Or maybe, "Really, I *know* it's porn; *you* know it's porn; now your coworkers know."

17:40:18 <sbp> heh, heh

17:41:07 <sbp> "Here's me, firewall with a brain the size of a planet, and you want me to give you access to *what* exactly?"

17:42:18 <bijan> "A piece of paper that's fallen to the floor!"

17:45:39 <crschmidt> ^add http://trust.mindswap.org/cgi-bin/FilmTrust/foaf.cgi?user=CloCkWeRX

17:45:40 <julie> Adding http://trust.mindswap.org/cgi-bin/FilmTrust/foaf.cgi?user=CloCkWeRX to my database...

17:46:58 <julie> Added 1118 statements from http://trust.mindswap.org/cgi-bin/FilmTrust/foaf.cgi?user=CloCkWeRX. Model size is 2113033.

17:57:42 <hex_> hex_ is now known as earle

17:57:49 <earle> earle is now known as hex_

19:23:58 <dajobe>http://lists.w3.org/Archives/Public/www-rdf-interest/2005Jan/0113.html

19:23:59 <dc_swig> B: http://lists.w3.org/Archives/Public/www-rdf-interest/2005Jan/0113.html from dajobe

19:24:04 <dajobe> B:|Raptor 1.4.4

19:24:04 <dc_swig> Titled item B.

19:24:14 <dajobe> B:a bunch of fixes in RSS parser/serializer. mostly reported by crschmidt

19:24:14 <dc_swig> Added comment B1.

19:24:18 <dajobe> B:thanks

19:24:18 <dc_swig> Added comment B2.

20:33:39 <aniasis3321> aniasis3321 is now known as aniasis