Semantic Web Interest Group IRC Chat Logs for 2001-12-13

01:32:00 <wendyout> wendyout is now known as wendy

02:16:05 * AaronSw wonders if mnot is around

02:16:16 <AaronSw> i want to chat about that sparta stuff

02:16:45 <AaronSw> Heh heh heh:

02:16:46 <AaronSw> [[[

02:16:48 <AaronSw> def _genName(self):

02:16:48 <AaronSw> ### we can do better than this...

02:16:48 <AaronSw> return "foo_boo"

02:16:50 <AaronSw> ]]]

02:16:56 <AaronSw> - http://test.mnot.net/sparta/Sparta.py

02:19:45 <sbp> heh, heh, heh

02:22:41 <AaronSw> that's even better than ruids

14:44:23 <danbri>http://lists.w3.org/Archives/Public/www-archive/2001Dec/0099.html

14:44:24 <dc_rdfig> A: http://lists.w3.org/Archives/Public/www-archive/2001Dec/0099.html from danbri

14:45:15 <danbri> A:|EARL 1.0: trying it with hypertext instead of reification

14:45:15 <dc_rdfig> titled item A

14:46:04 <danbri> A:EARL uses reification heavily. In N3 we have contexts syntax to make this prettier. In RDF (and RDF/XML) it can seem verbose. So I tried to recast EARL in a more hypertext style, where it makes claims about other documents that make (un-reified) claims.

14:46:04 <dc_rdfig> commented item A

14:46:27 <danbri> A:Comments etc from Sean et al welcome. Am I understanding EARL correctly?

14:46:28 <dc_rdfig> commented item A

14:49:35 <JibberJim> first comment- testMode is definately a property of the test not assertor...

14:55:55 <danbri> ah, it wasn't clear from parsing the sample RDF/XML in my head... looked like a typo.

14:56:34 <JibberJim> The same person could be expected to use multiple tools/test methods to evaluate a document.

14:57:03 <JibberJim> There's a typo in your example aswell (rdfs:Resouce)

14:59:14 <danbri> typo - doh

14:59:16 <danbri> these were real quick notes...

14:59:20 <danbri> does the basic idea make sense?

15:00:21 <JibberJim> I need to think more about that, I was just working on an EARL tool, and I'm plugging in your version to see what happens.

15:00:48 <danbri> ah, it was you that was asking about the old Javascript RDF query thing on www-rdf-interest?

15:01:01 <danbri> sorry didn't reply...

15:01:22 <JibberJim> Don't worry about it, I've built my own now...

15:01:31 <danbri> that stuff is pretty old. Uses early version of jang's js-prolog.

15:01:37 <danbri> ooh, url?

15:02:25 <danbri> we did all the rdf/xml to triples stuff server side. I experimented with a simple js XML triple syntax using a javaascript xml parser. JanG also has a basic rdf parser in js working somewhere

15:02:48 <JibberJim> http://www.e-media.co.uk/earl/rdfquery4.html is as far along as it is (that has your earl suggested format, which it seems to understand straight off - just hit the query button.

15:02:48 <dc_rdfig> B: http://www.e-media.co.uk/earl/rdfquery4.html from JibberJim

15:02:57 <JibberJim> IE/Mozilla only.

15:05:59 * danbri fires up mozilla

15:06:02 <JibberJim> IE and Mozilla have XML parsers so I use those rather than an XML implementation, means I can get the triples out myself. http://www.e-media.co.uk/earl/rdftoys.html demos that - matches the Validator for what I've tested so far, but definately doesn't understand everything.

15:07:06 <danbri> ooh, pretty

15:07:13 <JibberJim> Yes, but is it useful?

15:08:01 <danbri> I'm still trying to understand the way EARL works! But yes, clientside tools for this stuff definitely useful.

15:08:24 <JibberJim> The filters don't seem to work on yours, I wonder why that is. if you look at query3 it only returns certain parts.

15:08:34 <danbri> Does your parser assume the RDF is written in a certain way? I guess I'm suprised it worked since I changed the parseType=Resource bit, made the structure slightly different (at the XML not the RDF layer).

15:08:48 <danbri> I didn't know what to type in the filter box...

15:08:53 <JibberJim> I was surprised it worked to

15:09:22 <JibberJim> Filter box should match either predicates or objects and only return arcs with those on, it doesn't appear to work with yours.

15:09:50 <danbri> So it operates over the RDF abstraction, not the raw XML encoding? cool

15:09:54 * danbri views src

15:10:15 <JibberJim> It tries to work at the RDF, but only a subset of RDF used in EARL so far.

15:10:50 <JibberJim> looking at http://www.e-media.co.uk/earl/rdftoys.html is probably better that's the XML-triple conversion

15:11:29 <JibberJim> the querying is more EARL specific, and pretty much a hack, I still haven't got a concept of querying clear in my head yet.

15:13:16 <danbri> Is that RDF parser clientside? (xml-triple conversion...)

15:13:36 <JibberJim> Of course, there's no serverside involved.

15:14:01 * JibberJim stuff at the client unless there's a very good reason not to

15:14:16 <danbri> did you see: http://ioctl.org/rdf/jsrdf/

15:14:17 * JibberJim does ...

15:16:52 <JibberJim> I hadn't seen it no - doesn't work in IE...

15:17:11 <danbri> works in mozilla

15:17:39 <JibberJim> IE's script engine is more compliant than Mozillas...

15:18:06 <danbri> em: see also [rdf toys|http://www.e-media.co.uk/earl/rdftoys.html] for xml-to-triples mechanism.

15:18:41 <DanC> frap; forgot what I wanted to say in the process of joining.

15:18:47 <DanC> I really need IRC URIs on my desktop.

15:19:38 * danbri tries Jim's EARL data in Jan's RDF/.js parser. It barfs on rdf:parseType="Resource" I think

15:21:38 <DanC> does the RDFCore test suite include test for parseType="Resource" support? that's a feature that implementors might not realize the importance of.

15:21:54 <DanC> do you discuss parseType="Resource" in your striping thingy, danbri?

15:23:52 <JibberJim> It definately barfs on something, I think trying to author a genuine XML-RDF parser is a much bigger job than just using the XML parser available to you and parsing that for RDF, limits you to IE/Mozilla though, but I'm trying to lobby others...

15:25:38 <danbri> [[There are many other corner cases in the spec. RDF's rdf:parseType attribute, for example, complicates the simplistic striping model described here

15:25:42 <danbri> ]] -- http://www.w3.org/2001/10/stripes/

15:26:28 <danbri> dave picked up the striping thread for latest rdf-syntax draft, not sure what he's written to explain corner case

15:32:57 <DanC> I'd rather that parseType="Resource" weren't relegated to the role of "corner case"

15:33:21 <DanC> most days I think it should have been the default

15:33:38 <DanC> that alone might have make RDF more acceptable to the XML community.

15:41:29 <danbri> interesting... I can see that p.o.v.; I just got used to thinking in terms of stripes cos 90% of syntax is comprehensible that way. But I'm not arguing that stripes are good.

15:45:59 <danbri> A:[live version|http://www.w3.org/2001/12/hyperearl/earl-hypertext.rdf] (still has bugs)

15:45:59 <dc_rdfig> commented item A

15:50:32 <DanC> BLURB: re-thinking cwm's string output stuff

15:50:32 <dc_rdfig> C: re-thinking cwm's string output stuff from DanC

15:50:55 <DanC> C:rather than just strings, I'd rather see it support (a) tab-delimited I/O, and

15:50:55 <sbp> Uh oh... :-)

15:50:55 <dc_rdfig> commented item C

15:51:06 <DanC> C:... and (b) structured XML

15:51:07 <dc_rdfig> commented item C

15:51:19 <DanC> we really need an infoset vocabulary that cwm groks

15:52:01 <DanC> in thinking about rdf:parseType="Literal" like lisp (quote (a b c)), I'm thinking that a SAX-based vocabulary might be less awkward than an infoset-based vocabulary.

15:52:15 <DanC> sean, interested in implementing that? do you grok XSLT? interested in learning?

15:52:37 <sbp> I can do XSLT to some extent, and have been doing XML RDF + XSLT => XHTML

15:52:40 <DanC> logger, pointer?

15:52:40 <DanC> See http://ilrt.org/discovery/chatlogs/rdfig/2001-12-13#T15-52-40

15:52:51 <DanC> C:cf [discussion|http://ilrt.org/discovery/chatlogs/rdfig/2001-12-13#T15-52-40]

15:52:51 <dc_rdfig> commented item C

15:53:00 <sbp> But I had hoped that the XSLT days would be other now with --strings :-)

15:53:20 <DanC> well, the XSLT would be parallel to an implementation in cwm, I guess.

15:53:35 <sbp> it would be nice to have XSLT available from CWM

15:53:52 <DanC> it's just that I have the infoset vocabulary stuff implemented in XSLT, and I'd like to migrate my existing test cases before/contemporary-with implementing in swap

15:54:13 * sbp looks up the infoset stuff

15:55:02 <sbp> neat result graph: http://www.w3.org/2000/10/swap/infoset/infoset-diagram.svg

15:55:13 <DanC> thx

15:56:17 <DanC> C:rdf-mathlit-noLit stuff in [RDF syntax experiments Makefile|http://www.w3.org/2001/04rs22/Makefile] is my current test case for rdf:parseType="Literal"

15:56:18 <dc_rdfig> commented item C

15:56:43 <DanC> C:I think maybe a SAX vocabulary would be less awkward than the current infoset-based vocab

15:56:44 <dc_rdfig> commented item C

15:58:19 * sbp has to go take a call... (only one available phone line; ugh)

15:58:22 <DanC> timbl did some sketching a while ago of what HTML should look like in n3: [ :p ("this is a paragraph with" [ :em ("emphasized")] "text in it")]. or something like that.

15:58:42 <DanC> go cablemodem and use IP telephony, sbp

15:58:53 <sbp> Hmm... that would be neat - I'm currently just piecing HTML fragments together, which is not the best approach

15:58:55 <DanC> $40/month for the whole she-bang

15:59:26 <sbp> can't get broadband down here...

15:59:44 * sbp really must go

15:59:46 <sbp> c'ya

15:59:49 <DanC> ciao

16:14:01 <JibberJim> Sean, did you see DanBri's post about EARL?

16:14:28 <sbp> yes, I did; just going through it now

16:15:03 <DanC> anybody know how to read an ICAL calendar into evolution?

16:15:18 <JibberJim> My RDF query interface queried his example ok...

16:19:34 <DanC> hmm... evolution/local/Calendar/calendar.ics

16:21:20 <sbp> so, it queried both files? How did it know about the other, or did you just read both?

16:21:37 <JibberJim> The first example where they're not seperate

16:21:43 <sbp> ah

16:21:57 <sbp> could it be made so that it would grok the second, with the pointer to doc2.rdf?

16:22:14 <JibberJim> Of course it could!

16:23:08 <JibberJim> It can be made to do anything with enough Tuits (round ones of course.)

16:23:25 <sbp> Note that I made this suggestion before somewhere... but it got flamed down for some reason. I made it pretty recently in fact

16:23:41 <sbp> DanBri's suggestion about TestMode is interesting

16:24:08 <JibberJim> s/interesting/wrong :-)

16:24:43 <sbp> "sbp: could do: everything in this doc asserted by... in rdf, implicit defn of what's going on in doc" - http://www.w3.org/WAI/ER/2001/12/03-minutes

16:25:09 <JibberJim> TestMode has to be related to the test, a single assertor could assert multiple things with different testmodes.

16:26:26 <sbp> yeah. Originally, TestMode was part of the TestCase! I think it's best off of the Assertion... but can you think of a use case where they'd be different?

16:26:30 * sbp waves to chaals

16:27:10 * chaals waves back to Sean

16:30:06 <JibberJim> Yeah I could see person doing tests some of which it would use MANUAL on and some HEURISTIC

16:30:27 <sbp> fair enough

16:30:41 <JibberJim> I don't know if that's realistic, I can just see it...

16:31:28 <JibberJim> http://www.e-media.co.uk/earl/rdfquery3.html now has some filters in so you can only look at or exclude certain things

16:31:28 <dc_rdfig> D: http://www.e-media.co.uk/earl/rdfquery3.html from JibberJim

16:32:56 <sbp> heh, cool names; and I like how it's displayed

16:33:13 <sbp> D:|EARL RDF Query in JavaScript

16:33:14 <dc_rdfig> titled item D

16:33:19 <sbp> D:By Jim Ley

16:33:19 <dc_rdfig> commented item D

16:33:50 * JibberJim uses characters from Moomin Troll books rather than foo bar etc.

16:34:31 <sbp> seems prety sensible to me. I don't like metasyntactic variables (except for "blargh", perhaps)

16:34:32 <JibberJim> rdfquery4.html shows it with DanBris different EARL, it still understands it which I was very pleased about...

16:35:42 <sbp> the cool thing about client-side is that it's basically instant

16:38:41 <JibberJim> yep, I much prefer client side for everything, HTTP delays are just too annoying.

16:42:12 <JibberJim> Where can I find some other RDF test cases?

16:44:31 <sbp> .google RDF test cases

16:44:31 <xena> RDF test cases: http://www.w3.org/TR/2001/WD-rdf-testcases-20010912

16:44:46 <sbp> tip courtesy of A. Swartz

16:45:31 * DanC gets ICAL/evolution help in irc://irc.gnome.org/gnome

16:47:15 <sbp> <AaronSw> also http://www.w3.org/2000/10/rdf-tests/

16:50:04 <JibberJim> Hmm... those tests lock up my query engine entirely!

16:54:24 <danbri> doh! just missed sean :(

16:58:19 <lasDesk> lasDesk is now known as las

17:06:58 <las> zakim, ??p5 is las

17:07:34 <JibberJim> Why's there some strange authentication header on those test cases?

17:08:15 <JibberJim> e.g. http://www.w3.org/2000/10/rdf-tests/RDF-Model-Syntax_1.0/ms_7.1_1.rdf

18:23:50 <danbri> danc?

18:24:11 * danbri looking for a quick url that describes rdf as subset of FOL ('binary / conjunctive' or whatever)

18:24:42 * chaals wonders how to get the logger to give a reference to this point of the logs ;-)

18:24:55 <dajobe> logger, here

18:24:55 <dajobe> See http://ilrt.org/discovery/chatlogs/rdfig/2001-12-13#T18-24-55

18:25:01 * chaals suspects danbri might want a better ustified argument that that ;-)

18:34:22 * bwm waves to dajobe

18:34:34 <dajobe> hi bwm

18:34:48 <danbri> hiya brian; sorry I couldn't make it earlier. sounded fun!

18:34:51 <dajobe> qnames for uris! honestly :)

18:34:54 <bwm> reviewing id and about test cases

18:35:25 <bwm> my xml ain't so hot: rdf:ID="D&#xFC;rst"

18:35:50 <bwm> the &#xFC - is that a single hex character?

18:36:14 <dajobe> it is an ü

18:36:17 <bwm> danbri: no problem - we didn't get to go to pub I was overruled by the chair :(

18:37:17 <bwm> dajobe: and the %C3%BC

18:37:20 <danbri> bummer!

18:37:34 * danbri writes about RDF and unicorns

18:37:35 <bwm> is two bytes which is the right encoding?

18:37:49 <libby> soz I couldn't make it either brian: boring EU stuff all day intead.

18:38:11 <dajobe> for rdf ids see here: http://ilrt.org/discovery/2001/07/rdf-syntax-grammar/#rdf-id

18:38:21 <bwm> Bet you are having a ball; project manager; all that power to screw people over

18:39:04 <bwm> dajobe: was expecting one byte; just wanted to confirm two was right.

18:39:23 <dajobe> it is the %-escaped version of the utf-8 encoding of 0xFc

18:40:12 <bwm> ok - forgive my ignorance - thanks.

18:40:15 <dajobe> jjc told me not to worry about it just now, until IRIs get sorted out and he finishes some email from w3c i18n

18:40:49 <bwm> its good to make the point that encoding might be necessary - we can correct later if necessary

18:41:10 * bwm feels he can now send message saying he has reviewed test case

18:57:58 * bwm danbri chilly

19:00:13 <lilo> [Global Notice] Repeating the message for those lost in the confusion. We began experiencing severe network problems with one of our main rotation servers and we've removed the server from the rotation and shut it down. Apologies for the inconvenience. If you'd like to help and have a server with excellent connectivity, please take a look at http://openprojects.net/sponsoring_servers.shtml .... thanks.

20:14:00 <danbri>http://rdfweb.org/people/danbri/2001/12/puzzle/unicorny.html

20:14:01 <dc_rdfig> E: http://rdfweb.org/people/danbri/2001/12/puzzle/unicorny.html from danbri

20:14:22 <danbri> E:Practical RDF puzzles: photos, unicorns, web ontologies...

20:14:22 <dc_rdfig> commented item E

20:14:32 <danbri> E:|Practical RDF puzzles: photos, unicorns, web ontologies...

20:14:32 <dc_rdfig> titled item E

20:15:28 <danbri> E:Rough notes I've just made. It doesn't make any great claims, just tries to describe a problem...

20:15:28 <dc_rdfig> commented item E

20:16:05 * danbri expects it to be ripped to bits...

20:16:09 <JibberJim> Why is foaf:mbox a UniqueProperty there are lots of mailto:'s that map to more than one person?

20:16:31 * lasDesk browses it briefly, sees discussion of people who don't exist, looks forward to reading it in more detail.

20:16:59 <sbp> DanBri: it should be an UnambiguousProperty

20:17:19 <sbp> thanks for the EARL idea, BTW; I shall review that a bit more in detail...

20:17:25 <danbri> doh... I always get those back to front

20:17:30 <danbri> earl: cool!

20:18:00 <danbri> Why is foaf:mbox a *Property...? only some mailboxes can be truthfully claimed to be at the sharp end of a foaf:mbox arc.

20:18:44 <lasDesk> Actually, I don't think mbox is either unique or unambiguous. family@myhouse.com should mail to everyone in my family, and I have many mailboxes.

20:18:59 <sbp> well, you only use it on unambiguous boxes!

20:19:17 <sbp> the range will be :UnambiguousMailbox, or whatever

20:19:30 <danbri> for each of the mailboxes that you own and control (even if others have access), they can be used to pick out you.

20:19:37 <sbp> i.e. not all boxes will apply. It is a perfectly valid (and certainly useful) property to have

20:20:21 * danbri realises the foaf schema is mixed up

20:20:21 <danbri> <rdf:type rdf:resource="http://www.daml.org/2001/03/daml+oil#UnambiguousProperty"/>

20:20:23 <JibberJim> that means it's impossible to refer to Fred in family@myhouse.com as they don't have an individual one, but mail can get to them...

20:20:27 <danbri> This property is a 'unique property' in the DAML+OIL sense

20:20:49 <danbri> why do I find it so hard to remember that piece of terminology!

20:21:04 <danbri> maybe he has a homepage instead...

20:21:05 <sbp> no, unique property means that it has a cardfinality restriction of one. I'm pretty sure that I have more than one email box

20:21:20 <JibberJim> Some doubt in that Sean?

20:21:24 <sbp> I'm pretty sure that it's an unambiguous property

20:21:40 <sbp> nope, no doubt. Just hedging my bets :-)

20:21:56 * JibberJim thinks Unicorns do exist anyway...

20:22:20 * JibberJim knows Unicorns exist every bit as much as RDF...

20:22:20 <sbp> unambiguous means { :x :y :z . :p :y :z . :y a daml:UnambiguousProperty } log:implies { :x = :p } .

20:23:19 <sbp> :DanBri foaf:mbox <mailto:dan@[...]> . :DanielBrickley foaf:mbox <mailto:dan@[...]> . => :DanBri = :DanielBrickley .

20:25:29 <JibberJim> I thought unicorns did unmentionable things with virgins not ran about in fields, or have I got my mythology confused?

20:25:43 <sbp> heh! yes, I'm afraid so :-)

20:27:02 <JibberJim> "Though it would fight savagely when cornered, it could be tamed by a virgin's touch"

20:27:04 * danbri fixes up the U*Property terminology in unicorny and foaf docs

20:27:07 <danbri> does it make sense, anyway?

20:27:37 <danbri> sort of relates to the modeling style I'm thinking about for EARL... ie. use hypertext (multi docs) to capture multi viewpoints

20:27:43 <danbri> (instead of reification)

20:29:34 <sbp> the problem with that is: how do you assert things about the root context? I suppose you could have a predicate relationship from the document to it's root context: <doc2.rdf> :formulae [ :assertedBy :Sean ] .

20:31:38 <JibberJim> I don't fully see the problem with asserting that a unicorn exists, it clearly does exist, it's well defined well classified and any query would soon come up with someone asserting that it was mythological.

20:32:21 <sbp> the concept of a unicorn exists, certainly; whether or not a physical unicron exists is something up for debate

20:33:38 <JibberJim> Everything being expressed in RDF has a similar problem, I don't know any of you exist...

20:34:05 <JibberJim> A unicorn exists, but has no physical existence, it's simply a property of unicornness

20:34:08 <sbp> it's not a problem - you simply assume that everything in the root context is takes as true by the assertor of that context

20:34:30 <chaals> logger, grep unicorn

20:34:49 <logger> I'm logging. I found 66 answers for 'unicorn' (showing 0...4)

20:34:51 <logger> 0) 2001-12-13 20:34:30 <chaals> logger, grep unicorn

20:34:52 <logger> 1) 2001-12-13 20:34:05 <JibberJim> A unicorn exists, but has no physical existence, it's simply a property of unicornness

20:34:53 <logger> 2) 2001-12-13 20:32:21 <sbp> the concept of a unicorn exists, certainly; whether or not a physical unicron exists is something up for debate

20:34:54 <logger> 3) 2001-12-13 20:31:38 <JibberJim> I don't fully see the problem with asserting that a unicorn exists, it clearly does exist, it's well defined well classified and any query would soon come up with someone asserting that it was mythological.

20:34:55 <logger> 4) 2001-12-13 20:27:04 <danbri> * danbri fixes up the U*Property terminology in unicorny and foaf docs

20:34:55 <chaals> hmmm.

20:34:55 <chaals> grep unicorn

20:36:11 <AaronSw> hi everyone

20:36:22 <Shalrath> heh

20:36:26 <AaronSw> shalrath here is wondering how to add metadata to the website he's building

20:36:53 <AaronSw> and i have to disappear now.

20:37:08 <AaronSw> best of luck to you :)

20:37:11 <Shalrath> heh, catch ya later

20:37:59 <sbp> Uh oh

20:38:19 <Shalrath> hello

20:38:25 <Shalrath> would you care to hear some ideas?

20:38:35 <sbp> go for it

20:40:06 <Shalrath> im currently working on a site which would feature an engineering forum..

20:40:16 <Shalrath> engineering, and the application thereof

20:40:56 <Shalrath> in order to help describe objects in the site (objects would be anything like forums, posts, users, etc), im thinking about implementing a metadata system

20:41:16 <Shalrath> something that would use several man categories, with a tree structure of sub categories beneath them

20:41:36 <Shalrath> users could use these tags to categorize their projects, posts, and profiles

20:42:01 <Shalrath> the interface would make heavy use of filtering/searching by these tags

20:42:27 <Shalrath> anyways.. would you like to hear more?

20:43:33 * sbp is just trying to figure out what to make of this

20:43:59 <sbp> I guess this is a bit like a BBS, but with intelligent threading?

20:44:06 <Shalrath> heh

20:44:09 <Shalrath> you could say that

20:44:16 <sbp> well, please do go on

20:44:26 <Shalrath> the tag system was originally concieved to handle another part of the site.. the 'projects' section

20:44:40 <Shalrath> applying it to the forums and users was an afterthought

20:46:05 <Shalrath> at the moment, im rethinking about how to construct this

20:46:27 <Shalrath> so i thought I would get some advice from people who actually do this kind of stuff.. heh

20:47:10 <sbp> well, it seems as if you have a pretty ephemeral vision there. I guess you'd be wanting every user of the system to have some profile, described in RDF< and then get them to categorize their posts

20:47:28 <sbp> you could then give a summary of the types of topics that they post on

20:48:08 <Shalrath> actually, the little things like post categorizations would be inherited from the forum they post in - but they could edit them if they want

20:48:32 <Shalrath> again, the tags would be more useful in the project section

20:48:39 <sbp> right, well that's good. People tend to hate categorizing their posts anyway :-)

20:49:29 <JibberJim> So what are the issues you have, it seems well thought out, and ready to implement?

20:51:11 * JibberJim thinks a Unicorn definately exists, just one of the properties of it is it's imaginary so there should be no problem saying there's a picture of a unicorn...

20:51:28 <sbp> who says it's imaginary?

20:52:16 <JibberJim> Someone who's defining what a unicorn is, there must be some resource which says what a unicorn is otherwise the whole unicornness is meaningless to all

21:31:23 <danja> like the unicorns piece danbri

21:32:58 <danja> strikes me that the WordNet style approach would be most useful

21:33:55 <danja> the unicorn exists as a concept in some ontology somewhere

21:34:59 <danja> I think photos are probably a bit confusing here though

21:35:55 <danja> a person in the photograph probably has as specific mbox at the time the photo was taken

21:36:40 <danja> unlikely for the unicorn

21:37:44 <danja> still virgins in Bristol then..?

21:47:08 <danbri> the wordnet approach doesn't work so good, except perhaps for unicorns, since not-existing seems to be a central aspect of unicorn-ness.

21:47:42 <danbri> it doesn't work cos you don't want to have to move a whole bunch of classes over from 'mammal' to 'stuff that's in the head' whenever you find out the classes had an empty extension.

21:47:50 * danbri -> offline for a bit

21:48:56 * DanC wonders if gerald's IRC client is rigged to beep at the mention of 'photo' in #rdfig

21:50:04 * danbri mutters something about nude photos of gerald

21:50:14 <danbri> nope. oh well.

21:53:01 <sbp> aw man, www-archive is *still* rejecting my mail!

21:53:35 <sbp> I'll have to send it from another ISP...

21:55:00 <danbri> bounce it me if you're in a hurry. danbri@w3.org

22:10:22 <danja> nah, what about 'real mammals' and 'mythological beasties' as subclasses of animalish concepts..?

22:11:04 <danja> maybe suo might be useful here?

22:13:46 <danja> i.e. the Thing with URI is a concept

22:14:39 <danja> or an easier approach might just be to have the property 'imaginary'

22:15:35 <danja> though as far as the machine's concerned, doesn't this apply to everything without a binary representation?

23:45:54 * sbp sends reply to DanBri