Semantic Web Interest Group IRC Chat Logs for 2003-11-06

00:00:29 * Ian_ installs gnome-gv

00:00:52 <DanC> do the properties have labels yet?

00:00:57 <Ian_> yes

00:01:02 * ndw wonders why telnum.py turns 5551212 into 011,551212

00:01:15 <Ian_> hey ndw

00:01:20 <ndw> Hi Ian_!

00:01:29 <DanC> trying this: [[[

00:01:31 <DanC> { ?G g:hasNode ?C, ?C2.

00:01:31 <DanC> ?C ?P ?C2.

00:01:31 <DanC> ?P rdfs:label ?TXT }

00:01:31 <DanC> => { ?C [ a g:EdgeProperty; g:label ?TXT ?C2 }.

00:01:32 <DanC> ]]]

00:01:59 <Ian_> So not looking specifically that owl:ObjectProperty?

00:02:04 <Ian_> Just that has a label?

00:02:27 <DanC> not working yet

00:02:49 <DanC> got an example?

00:03:03 <Ian_> of?

00:03:11 <DanC> ?C ?P and ?C2

00:03:33 <Ian_> <owl:ObjectProperty rdf:about="#representation">

00:03:33 <Ian_> <rdfs:label>Representation</rdfs:label>

00:03:33 <Ian_> <rdfs:domain rdf:resource="#Resource"/>

00:03:33 <Ian_> <rdfs:range rdf:resource="#Representation"/>

00:03:33 <Ian_> </owl:ObjectProperty>

00:03:36 <DanC> ah! I'm barking up the wrong tree...

00:03:44 <Ian_> bark!

00:04:14 <DanC> trying [[[

00:04:14 <DanC> { ?G g:hasNode ?C, ?C2.

00:04:14 <DanC> ?P rdfs:domain ?C; rdfs:range ?C2.

00:04:14 <DanC> ?P rdfs:label ?TXT }

00:04:14 <DanC> => { ?C [ a g:EdgeProperty; g:label ?TXT] ?C2 }.

00:04:15 <DanC> ]]]

00:04:32 <DanC> still no joy :-{

00:04:48 <Ian_> I don't know g:EdgeProperty

00:05:09 <DanC> g:EdgeProperty's show up in the .dot file

00:05:14 <Ian_> ok

00:05:15 <DanC> not all properties do.

00:05:21 <Ian_> That's what I found out...

00:05:30 <Ian_> I had started to man dot, etc.

00:05:57 <Ian_> What is ?C [...] ?C2 saying?

00:06:49 <Ian_> Do you need "." after ?TXT ?

00:07:13 <Ian_> (In first part)

00:07:50 <DanC> phpht. sorry, gotta go.

00:07:52 <Ian_> ok.

00:07:58 <Ian_> I'll try to work with.

00:08:00 <Ian_> Thanks for the lead.

00:08:04 <Ian_> can you check in?

00:08:53 * Ian_ gonna go to pool

00:15:46 * ndw giggles uncontrollably. It works!

00:15:58 <ndw> That's great fun, DanC!

00:16:51 <ndw> Anyone know how to add a protocol to Firebird?

00:17:05 <Ian_> Do you do it at the gnome level, e.g.?

00:17:54 <ndw> I did it in Gnome and now Epiphany uses it. Hmm. I suppose I should restart Firebird...

00:18:14 * ndw mutters curses about the fact that running the control center applet steals my desktop

00:23:09 * ndw dials an ATT Wireless by typing tel:+1.866... right into the Epiphany browser bar. Whee!!!

00:23:14 * ndw thanks DanC for the new toy!

00:23:17 <Ian_> :)

00:23:23 <Ian_> Now you can put tel uris in pages, etc.

00:23:34 <Ian_> I haven't set this up...

00:23:48 <ndw> I've been doing that for ages, it just hasn't been valuable to click on them before!

00:23:52 <Ian_> right

00:23:54 <Ian_> URI?

00:24:06 <ndw> rdfig.xmlhack.com

00:24:09 <Ian_> :)

00:24:14 <ndw> Dan just put a bunch of stuff up there

00:24:19 <Ian_> ok

00:24:46 * ndw shoved 8 large specs through W3C pubrules today. Is going to go find a...beverage and tear down VPN. bbiab.

00:24:54 <Ian_> well done!

00:25:04 * Ian_ comm team still feeling it ;)

00:25:37 <mdupont> mdupont is now known as md-ZZZ

00:41:49 <eaon> eaon is now known as eaon|zZz

01:07:20 <ChanServ> [#rdfig] This channel is logged and blogged: http://logicerror.com/rdfIRCWelcome

01:08:27 <dmwaters> {global notice} Hi all, it appears we're still having problems with one of our main EU hubs. I'm still looking for the cause of the problem, and appologize for the problems. I've rerouted around it for the time being.

03:12:13 <grrrrr> grrrrr is now known as gmm

05:24:35 <D[a]vey> D[a]vey is now known as Davey

05:37:18 <gmm_> gmm_ is now known as grrrrr

06:57:14 <md-ZZZ> md-ZZZ is now known as mdupont

07:33:03 <GNUPredator> GNUPredator is now known as mdupont

07:49:04 <mdupont> mdupont is now known as md-afk

08:10:45 <esigler> esigler is now known as esigler_away

08:58:43 <md-afk> md-afk is now known as md-work

09:06:47 <JimJibber> JimJibber is now known as JibberJim

10:35:20 <ericP> hey jeen, did i ever reply about your corrections to the query survey?

10:36:12 <josek> Hi EricP

10:36:21 <ericP> heya, how do?

10:36:28 <josek> long time no seen :)

10:36:51 <ericP> my invisible elixer works well

10:37:13 * danbri uhohs

10:37:31 <ericP> i saw you typing that before you hit return

10:37:42 <ericP> got a hair-brained scheme for y'all

10:38:01 <ericP> kind of RDF-related, or at least enabled:

10:38:22 <ericP> s'pose i want to use UUIDs to invent URIs while on a plane

10:38:39 <ericP> and i want to share the stuff i'm talking about with other machines and people

10:39:53 <ericP> i could itereate through all the statements in the database, pit them out in ntriple form, SHA-1 the lot, and i'd have one of many byte sequences that would uniquely identify that doucment

10:40:22 <ericP> can use that as the seed for a series of UUIDs in that document

10:41:20 <ericP> the implication being that collision will only occur between two identical documents (where the staements happened to be ntriple-serialized in the same order)

10:42:09 <ericP> now i ahve an acceptable UUID scheme i can use on a plane or metal tube orbiting pluto

10:42:52 <ericP> what i miss, compared to URL-identifying the nodes, is the ability to have someone discover such a node and do a GET to learn more about it

10:43:57 <ericP> so now i want to add that functionality to my corpus of data, i publish it to some site that gives the interesting nodes URLs and assert owl:sameAs to associate them with the orginal UUIDs

10:44:02 <ericP> tada

10:44:07 <ericP> is anybody listening?

10:44:56 <dajobe> hello

10:45:13 <ericP> sweet, i'll take that as a ringing endorsement

10:45:18 <dajobe> nope

10:45:36 <ericP> got rocks to throw?

10:45:49 <dajobe> bnodes

10:46:28 <ericP> sticks and stones will break my bones, but bnodes have no mass

10:46:30 <dajobe> ntriples was designed to be 1<>1 with rdf statements in the graph, except for bnodes

10:46:43 <ericP> bnodes obviate the need for tbe UUIDs?

10:46:46 <dajobe> anyway, I'm suspicious of uuids being good enough

10:46:48 <danbri> what's wrong w/ pure random?

10:46:56 * danbri prefers bnodes too

10:47:06 <ericP> i think there's not prob with bnodes

10:47:53 <ericP> i'm not trying to assert that two differently hahsed documents are the same, just that no two documents with the same hash are the same

10:48:00 * dajobe must focus on workshop stuff, relurks

10:48:10 <ericP> _:1 p1 _:2.

10:48:18 <ericP> _:2 p2 _:3.

10:48:26 <ericP> _:3 p3 _:1.

10:48:38 * danbri not really following the point

10:48:51 <ericP> has three possible hashings given possible statemen orders

10:48:55 <danbri> did you read jjc's stuff on signing rdf graphs? that had some cute tricks in it. don't have uri handy though.

10:49:08 <ericP> nope

10:49:19 <dajobe> .google signing rdf graphs carroll

10:49:20 <datum> signing rdf graphs carroll: http://www.hpl.hp.com/techreports/2003/HPL-2003-142.pdf

10:49:22 <danbri> .google jeremy signing rdf graphs

10:49:23 <datum> jeremy signing rdf graphs: http://www.hpl.hp.com/techreports/2003/HPL-2003-142.pdf

10:49:27 <danbri> :)

10:49:52 <ericP> yeah, i fgoured jeremy would have something

10:50:17 * ericP runs out for a minute hoping the "late night" canteen won't be a complete overstatement

10:50:35 <dajobe> you better hope "canteen" isn't also

10:51:08 * ericP gigles

10:51:34 <dajobe> jjc had something at the ontology conference too

10:52:14 <ericP> parting thesis: any two documents that have the same ntriple has (even with pessimal assignment of bnodes) are the same graph

11:02:03 * ericP returns more than a litle victorious

11:49:21 <josek> Hello, I was wondering if I could ask a question about URIs that are assigned to statements. I'm trying to find out if it's possible to move one of my RDF bookmark files around, and stil have URIs that make sense

11:50:17 <danbri> Just try asking! You never know... (if people don't answer they're probably busy or don't know)

11:50:27 <danbri> (your Q will be in the archives too, so someone might find it later)

11:50:43 <swh> josek: what do you mean by statement?

11:50:48 * danbri is btw too busy today to pay attention to irc much, but interested to know what you've been building

11:52:09 * ericP is looking forward to slamming josek for asking questions

11:52:33 <danbri> me too, but don't tell him that yet ;)

11:52:37 <danbri> ask away!

11:53:13 <josek> ok, thanks

11:54:04 <josek> we're making an RDF bookmark schema to ease the sharing of bookmarks among browsers and users: http://www.w3.org/2003/07/Annotea/BookmarkSchema-20030707

11:54:23 <josek> And I've a running implementation in Amaya.

11:54:59 <josek> when I make a bookmark, I attribute it a URI:

11:55:00 <josek> ----

11:55:02 <josek> <rdf:Description rdf:about="file:///home/.amaya/bookmarks.rdf#Actors">

11:55:03 <josek> <rdf:type rdf:resource="http://www.w3.org/2002/01/bookmark#Topic"/>

11:55:03 <josek> <b:subTopicOf rdf:resource="file:///home/.amaya/bookmarks.rdf#FamousPeople"/>

11:55:03 <josek> <dc:title>Actors</dc:title>

11:55:03 <josek> <dc:description>Actors</dc:description>

11:55:11 <josek> <dc:creator>Marja</dc:creator>

11:55:13 <josek> <a:created>2003-05-16T00:22:32-05:00</a:created>

11:55:15 <josek> <dc:date>2003-05-16T00:22:32-05:00</dc:date>

11:55:17 <josek> </rdf:Description>

11:55:22 <ericP> ooo, the question mark will be any moment now...

11:55:24 <josek> ---------

11:55:27 <josek> the URI in this case is file:///home/.amaya/bookmarks.rdf#Actors

11:56:00 <josek> I also have the possibility of opening bookmark files that are stored on the web and saving them elsewhere too

11:56:33 <josek> And bookmarks belong to topics. Those topics also have attributed URIs: file:///home/.amaya/bookmarks.rdf#FamousPeople

11:57:22 <swh> [what is the right word for a blog of triples all with the same subject BTW]

11:57:26 <josek> So the question is, if I save my local file to an HTTP server, what sense do those attributed URIs make

11:57:40 <ericP> what are you doing Asking QUESTIONS!?! what do you think this channel is FOR!? not INFORMATION! not for YOU! /kickban!

11:57:41 <swh> I call them objects sometimes, but nmg glares at me

11:57:50 <danbri> swh, there isn't really a word (blob of, you mean). A 'description', informally?

11:57:56 <danbri> 'object' is overloaded horribly

11:58:06 <danbri> subject/predicate/object is named the other way, for eg.

11:58:23 <swh> danbri: agreed, but theres not much else - I guess instance is close, but that will offend other people :)

11:58:56 <josek> we can say its an instance of a bookmark :) It doesn't offend me

11:59:01 <swh> danbri: I did mean glob, but blob will do too :)

12:00:16 <josek> I am wondernig if it makes sense to have those attributed URIs that are inside a bookmark to be resolvable into something. Once they are out of the local computer, they lose their meaning and become a bit like opaque URIs

12:00:47 <ericP> so if there is at least one file:///home/.amaya/bookmarks.rdf#FamousPeople on some mahcine somewhere

12:01:05 <josek> yes, unless you erased it in the meantime

12:01:09 <ericP> and you publish the topic to a bookmarks server

12:01:30 <ericP> and the bookmarks server agrees to give it a name in the bookmarks server's web space

12:01:37 <josek> no, I publish them as a stand-alone file

12:01:44 <ericP> what's the relationship tbetwen the two?

12:01:46 <ericP> oh, ok

12:01:58 <ericP> so i'd say that dangerous practive

12:02:06 <ericP> practice

12:02:10 <josek> publishing them to the server is another problem. But I'd still want to keep the original name I gave them

12:02:33 <ericP> i wouldn't wnat to depend on noone else using having used the same URL

12:02:36 <josek> so that I can e.g. make new topics that correspond to that URI

12:03:01 <josek> imagine that I migrate my bookmarks from your server to another one. I would need to retain the name, right?

12:03:22 <ericP> if i say file:///home/.amaya/bookmarks.rdf#FamousPeople means X and you say file:///home/.amaya/bookmarks.rdf#FamousPeople means Y, we get into trouble when both are visible

12:03:52 <ericP> retaining the anem during migration is no problem so long as it is globally unique

12:03:56 <ericP> file:///home/.amaya/bookmarks.rdf#FamousPeople is not

12:04:00 <josek> ericp, precisely. This is because the URI is not unique

12:04:27 <ericP> i'd say that as soon as you want it outside your machine, you'd better make it unique

12:04:40 <josek> it's stored locally. But this is because we want the URI to resolve to something. If I add more info like an IP@, user name, date, ... I can make a unique url

12:04:50 <ericP> this is why i thougth the UUID scheme that used the SHA-1 as a seed was cool

12:05:08 <ericP> yeah, so long as you ahve that information

12:05:11 <josek> why not make them unique from the moment you create them? Then you can publish them using scp, ftp, HTTP, webdav, whatever yuo want

12:05:39 <josek> you need to give proof that your UUID produces unique ids :)

12:05:49 <ericP> it is administratively simple to make sure that non-localhost non-NAT HTTP URLs are, at least fo rthe moment, unique

12:06:23 <ericP> make them unique on creation, i agree, i just thought that the hash of the graph was a good way to do that

12:06:32 <ericP> and kinda geeky too

12:07:02 <josek> geeky yes :) but I'm not sure how unique the ids couldb e

12:07:03 <josek> be

12:07:52 <ericP> i think they are theoretically (as opposed to just statistically) provably unique within the RDF domain

12:08:27 <ericP> now, someone could take the exact same string of ntriples and use it to mean something else, say, an image format, and the two could collide

12:09:00 <ericP> you'd be even better off prefixing the ntriple string with a URL for this hair-brained schema

12:09:05 <josek> ok, suppose I use UUIDS (however they are computed). This solves my problems with local files. What happens when I publish them to your server? I can't rewrite all the files that refer to those bookmarks

12:09:40 <ericP> it would take someone really dedicated to obfuscation ot use the same URL at the beginning of their, say, image format, and have it mean somethign completely different

12:10:13 <ericP> the server can tell you some owl:sameAs relationships

12:10:33 <ericP> whether your database will do the right thing with them is another matter

12:10:40 <danbri> ericP, does Annotea have a provence-exposing network interface?

12:10:41 <ericP> mine won't

12:10:42 <ericP> oof

12:11:04 <josek> what if I query the server and ask for for a list of topics. What URI will be used in the about: the server one or the one I defined? Because that's the one I need to use later on

12:11:05 <ericP> provence-exposing, not currently

12:11:13 <ericP> or at least it's not official

12:11:17 <danbri> Can I say 'give me matches for foaf:name(X Y), foaf:mbox(X, mailto:danbri@w3.org) FROM graphs written by danbri@w3.org

12:11:22 <ericP> it is in the RDF you get back, though

12:11:24 <danbri> guess not. wishlist!

12:11:32 * danbri forgot algae syntax but you get my meaning...

12:11:40 <josek> for example, if I take the hypothetical smithsonian living species ontology and decide to put a copy of it in your server, for quick access, how would you rewrite its urls? They are supposed to be unique already

12:11:40 <danbri> OK so could filter locally.

12:11:50 <ericP> algae will allow you to nail down provenance

12:12:06 <ericP> so it is implemented to some degree

12:12:13 <ericP> but not official

12:12:25 <danbri> nor is algae or annotea!

12:12:50 <danbri> (which is fine...)

12:13:04 <josek> danbri the only thing that's official is that bookmarks are annotations (groans)

12:13:08 <ericP> no, but our impelemtation offers an algae interface

12:13:17 <ericP> wait, they are?

12:13:19 <danbri> Bookmarks _are_ annotations, glad we agree :)

12:13:27 <ericP> i thought that was a temporary hack

12:13:34 <ericP> really? we want that?

12:13:39 <ericP> i certainly don't

12:13:42 <danbri> josek, you're right to point out that bookmarks are also a lot about UI and management, editing.

12:13:48 <ericP> why not make everything an annotation

12:13:53 <josek> I don't agree :) It's only people who made that official. Not me :)

12:13:56 <ericP> each RDF statement out there can be an annotation

12:14:31 <ericP> they have at least one common ancestor, but i don't think we've defined what it is

12:14:32 <danbri> well, bookmarking falls under the restricted notion of 'annotation' given on the Annotea site, even if you've modeled them differently in the RDF and protocol side. And even if you have a different app/UI.

12:14:41 <ericP> did this happen when i didn't call into a meeting?

12:14:52 * ericP will suffer for slacking

12:14:53 <danbri> All RDF description is annotation, since to describe something is to annotate something. But I wasn't talking in that extended sense...

12:15:25 <ericP> 'annotation' given on the Annotea, don't follow

12:15:41 <josek> danbri: I can only agree that if using metadata to describe or classify something is an annotation, the bookmarks are annotations :)

12:15:50 <ericP> i'm sure that it's a slipper slope

12:16:03 <josek> a series of unfortunate events

12:16:10 <ericP> that it's best not to conflate things that have some similarities

12:16:16 <ericP> josek, oh well

12:16:37 <josek> ericp, I thought you were quoting that book

12:17:06 <ericP> "a series of unfortunate events" is a book?

12:17:16 <ericP> you forget how uncultured i am

12:17:24 <danbri> ericp, the Annotea site gives a few accounts of annotation to distinguish its goals from the broader woolly notion of 'annotation as description'.

12:17:31 <danbri> "By annotations we mean comments, notes, explanations, or other types of external remarks that can be attached to any Web document or a selected part of the document without actually needing to touch the document. "

12:17:59 <dajobe>http://www.w3.org/2001/sw/Europe/events/20031113-storage/

12:18:00 <dc_rdfig> A: http://www.w3.org/2001/sw/Europe/events/20031113-storage/ from dajobe

12:18:01 <danbri> bookmarking falls under that. The topic classification aspect is underplayed though. But annotations in the marginal note sense are classified too, and it doesn't stress that either.

12:18:06 <dajobe> A:|SWADE Workshop on Semantic Web Storage and Retrieval

12:18:06 <dc_rdfig> Titled item A.

12:18:14 <dajobe> A:now with [http://www.w3.org/2001/sw/Europe/events/20031113-storage/positions/|position papers]

12:18:14 <dc_rdfig> Added comment A1.

12:18:55 <danbri> <ericP> "a series of unfortunate events" is a book?

12:18:58 <ericP> if Bookmarks are under Annotations, then there is nothing i could want to say about an Annotaiton that i wound;t want to say about a Bookmark

12:18:58 <danbri> no, a biography.

12:19:10 <josek> http://www.amazon.com/exec/obidos/tg/detail/-/0064410137/104-3825315-5504700?v=glance

12:19:30 <danbri> ericP, I think you're being too literal. Maybe if we talked re verbs instead of nouns, it wouldn't map directly into subclassing.

12:19:49 <danbri> Bookmarking in the Annotea sense falls within the domain of Annotating, again in the Annotea sense.

12:20:14 <ericP> but is Bookark rdfs:subClassOf Annotation?

12:20:44 <danbri> Not quite, since bookmarking is as much about UI and editing as it is about a class of thing. So we're not comparing like with like.

12:20:49 <josek> isn't this a bit like saying "an image is a collection of bits that you can see in your browser"? Then under that definition anything that yuo can see in your browser is an image. The original definition wasn't good enough

12:21:05 <ericP> if there is a general sense of annotating, that includes making Annotations, then i don't mind it includeing makeing Bookmarks

12:21:07 <danbri> Annotating is about the metaphor of scribbling in the margin; bookmarking is about filing away in a card catalog. Metaphors.

12:21:42 <danbri> There is _also_ a more general sense of annotating which is basically 'describing'. Restaaurant reviews etc. (versus scribbles in the margin of Restaurant homepages, which is the Annotea notion of restaurant review).

12:22:02 <ericP> ok, i just got alarmed at the notion of the subClassOf relation

12:22:17 <danbri> If I say 'restaurant3--tasteyFood->true', that's out of scope for Annotea, since it is upre RDF description.

12:22:20 <danbri> I think we all agree that.

12:22:53 <danbri> If I bookmark restaurant3's homepage under 'Tastey Food Serving Restaurants' using Amaya/annotea, that's cool.

12:22:59 <ericP> i want annotea to have that power

12:23:09 <ericP> then we can chagne the world by making assertions

12:23:19 <danbri> If I scribble an annotea-annotation in the margin of that page, saying ''mmm, tastey!'' that's cool too.

12:23:57 <danbri> The last 2 use cases are 'annotations' in a more broad than 'scribbles in the margin' sense, but less broad than 'annotation==description' sense. They have an affinity, and some shared codebase. But different UI and some differing RDF vocab.

12:24:18 <danbri> Drag/drop into 'cheap food' and the restaurant drops its prices :)

12:24:48 <ericP> oh yes, i ahve no prob with that. just the subClassOf. as long as that isn't asserted, we have room to differentiate as the real-world implications occur to us

12:24:56 <danbri> Being able to mine the annotea representation to get to the RDF view that is couched in terms of the restaurant itself, now that'd be cool!

12:25:11 <danbri> eg. make explicit what it implies for something's homepage to be in a certain Topic.

12:25:26 <danbri> (that would btw help with mining the Open Directory too... there's a ton of data in there)

12:25:45 * danbri nods, no need for the subclass i think

12:25:46 <ericP> seems like we need a rule lang for that

12:25:53 <danbri> yes, nice use case huh? :)

12:26:06 <ericP> or create lots of one-off rules like rdfs and owl

12:26:38 <danbri> there is some real data around now too, http://esw.w3.org/topic/RestaurantRecommendation

12:26:44 <ericP> i'm not done arguing if you say "no need for the subclass" -- done when you say "subclass may not be true"

12:26:47 <danbri>http://esw.w3.org/topic/RestaurantsVersusTheirReviews

12:26:48 <dc_rdfig> B: http://esw.w3.org/topic/RestaurantsVersusTheirReviews from danbri

12:26:57 <danbri> B:|RestaurantsVersusTheirReviews (and homepages)

12:26:57 <dc_rdfig> Titled item B.

12:27:06 <danbri> B:Doc-centric vs World-centric description

12:27:07 <dc_rdfig> Added comment B1.

12:27:14 <danbri> logger, pointer?

12:27:14 <danbri> See http://ilrt.org/discovery/chatlogs/rdfig/2003-11-06#T12-27-14

12:27:23 <dajobe> you can use the new feature, danbri

12:27:30 <danbri> oh cool, how's it go?

12:27:33 <dajobe> tell "logger, chump b"

12:27:40 <danbri> logger, chump b

12:27:40 <danbri> B:See [http://ilrt.org/discovery/chatlogs/rdfig/2003-11-06#T12-27-40|discussion]

12:27:40 <dc_rdfig> Added comment B2.

12:27:42 <danbri> :)

12:27:45 <danbri> dajobe++

12:28:38 <josek> could we steer the discussion back to using UUIDS or resolvable ones? :)

12:28:44 <danbri> sorry!

12:28:52 * danbri goes back to typing elsewhere!

12:29:12 <josek> you can type here! didn't want to scare you

12:29:25 <swh> swh is now known as swh_lunch

12:29:31 <danbri> no prob, I've stuff to do :)

12:29:53 <josek> I'll go for lunch and come back later. Maybe nmb will be logged in then :)

12:30:04 <josek> thanks for your ideas wrt annotations

12:31:03 <danbri> ok, happy eating

12:31:40 <ericP> i'm pretty content with the semantics of the UUIDs, just not sure who implementes owl:sameAs

12:32:06 <danbri> Jena has some inference code.

12:32:11 <ericP> of course, the application can always implement such semantics

12:32:26 <mattb> owl:sameAs is such a useful basic statement to make

12:32:27 <danbri> mattb was reporting 'it just worked' re some joseki/foaf experiments with owl:InverseFunctionalProperty ...

12:32:35 <danbri> speak of the devil!

12:32:36 <ericP> i can put it into the bookmark server

12:32:46 <mattb> it'd be nice if RDF APIs provided some sort of "make_same_as(uri1,uri2)" call on a model

12:32:51 <mattb> even if they didn't implement a reasoner

12:33:05 <mattb> of course it's hard to communciate a sameAs from one system to another when bNodes are being merged

12:33:08 <ericP> does that include removing uri1 when done?

12:33:35 <mattb> doesn't uri1 has to stay around in case of sameAs(uri1,uri3) later?

12:33:59 <ericP> ie, it's easy to *change* a URI, which gives you all the implications of the rule at that time, but noo implications for which the ground facts become aparent later

12:34:19 <ericP> could a more akward voice have been used by myself?

12:34:47 <ericP> eveyplace you see uri1, change it to uri2

12:34:54 <mattb> (work intrudes)

12:35:06 <ericP> that's easy, but subsequent refrences to uri1 will be lost

12:35:13 <danbri> even in rdf:Statement descriptions (pred/subj/obj), which is why reification smells.

12:35:18 <ericP> work, huh?

12:35:34 <ericP> one reason reification smells

12:35:45 <danbri> mattb draws the little red dots in the corner of the tv screen

12:38:13 <ericP> can you draw circle and arc diagrams too?

12:38:23 <ericP> draw some constellations for RDF

12:44:11 * darobin longs for SVG on set-top boxes

12:44:45 <mattb> with the owl reasoner turned on in joseki+jena, introducing a sameAs works rather like a symbolic link afaics

12:45:31 <mattb> any triple (uri1,a,b) causes (uri2,a,b) to become present in the store

12:45:32 <mattb> and vice versa

12:45:38 <mattb> so it's not rewriting

12:45:48 <mattb> hence later sameAs (uri1,uri3) works cos uri1 is preserved

12:46:40 <ericP> do you know if it's doing forward chaining?

12:46:57 <mattb> turning on the owl reasoner in jena and feeding it the foaf schema caused only new triples to appear, as far as i can tell

12:47:16 <mattb> asserting a,owl:sameAs,b made a,owl:sameIndividualAs,b and so forth

12:47:24 <mattb> i don't have a solid understanding of reasoner internals i'm afraid

12:47:32 * mattb notes he must read about forward chaining and so forth

12:48:02 <mattb> not today, though

12:48:06 <mattb> we are having problems with the bbc red dots

12:48:07 <shellac> jena is hybrid, IIRC, backward/forward

12:49:13 * shellac rummages

12:49:27 <shellac> try http://jena.sourceforge.net/inference/index.html

12:50:05 <ericP> a owl:sameAs b. a p1 foo. => b p1 foo.

12:50:14 <ericP> forward chained reasoner stores b p1 foo.

12:51:00 <ericP> backward chained reasoner jsut stores a owl:sameAs b. a p1 foo. but knows to derive b p1 foo. when asked something like ?what p1 foo.

12:51:36 <mattb> thanks, useful explanation

12:51:54 <ericP> non-tabled prologs tend to be backward chaining, they eat their own tails when given non-optimal questions

12:52:11 <mattb> does forward chaining remember the proof of why b p1 foo. ?

12:52:26 <ericP> forward chaining doesn't allow you to answer x < 5 withing calculating all the things x <

12:52:34 <mattb> and can either type of reasoner cope with un-asserting a rule?

12:52:36 <ericP> that's impl-dependent

12:53:17 <ericP> i reckon one that stores proofs could clean up after the removal of a component in the proof

12:53:52 <ericP> don't know that it's implemented, but i'm neiter an expert nor an avid reader

12:54:52 <shellac> looking at the jena api it's possible, but not sure if it's implemented

12:55:24 <ericP> i really should play with jena.

12:55:32 <shellac> Resoners may be immutable

12:55:35 <ericP> oh, so much to do, so much to do,

12:55:55 <shellac> well link above is pretty informative

12:56:10 <mattb> considering the issue of whether a scutter should filter out any owl: or rdfs: statements in incoming scuttered models before storing them

12:56:32 <mattb> partly considering the fact that a scutter may well want to issue the instruction "delete all triples read from http://foo"

12:56:39 <mattb> because of a 404, document change, whatever

12:57:08 <mattb> i should do a long scutter run

12:57:13 <mattb> see what's in the wild

12:57:23 <mattb> i know i saw a few owl:things in random found documents on one run

12:58:32 <shellac> foaf groups might use OWL restrictions

12:59:05 <shellac> (but I don't think that works in jena properly)

13:40:22 <swh_lunch> swh_lunch is now known as swh

14:09:06 <arnarl> hi

15:02:35 * libby deletes 122 of 124 comments (all spam) from esw weblog

15:04:23 <Dorward> youwch!

15:04:56 <libby> sigh. I'd not been paying attention. about 6 were today!

15:06:26 <JibberJim> esw weblog?

15:07:42 <libby> yeah: http://esw.w3.org/mt/esw/ - we've not been using it much recently

15:17:21 <dajobe> you can turn off comments

15:25:47 <ericm> dajobe, you there?

15:25:51 <dajobe> barely

15:28:27 <josek> A very silli question... what is the correct way in RDF to do a bookmark property that says if a topic (our equivalent of folders) is expanded or collapsed? <b:collapsed>yes|no</b:collapsed> <b:collapsed url_to_logic_terms_schema:yes|no />

15:28:39 <josek> silly! oops

15:33:00 <danbri> sounds like a model-view-controller mixup...

15:33:26 <danbri> isn't the expanded/collapsed thing a description of a piece of UI?

15:33:49 <danbri> or i might be missing the goal...

15:34:14 <eikeon> Hum: rdflib is raising a "Not in NFC" error on line 3016 of http://www.w3.org/2000/10/rdf-tests/rdfcore/Manifest.rdf

15:35:47 <sandro> Hrm. I use rdflib to read that file all the time.

15:36:08 <eikeon> Might be my version of rdflib... /me tries non development version

15:36:35 <eikeon> Unless it's a Mac thing... /me is new to Mac

15:37:40 <dajobe> it's valid for me with rapper/raptor

15:38:16 <dajobe> and jena2/arp

15:38:49 <danbri> does Annotea/Amaya include just raptor now, or Redland too?

15:39:10 <dajobe> annotea comes with redland

15:39:16 <dajobe> not sure about amaya

15:39:52 <danbri> josek, is Annotea client stuff just in the normal Amaya dist's now?

15:39:53 <dajobe> in http://www.w3.org/Amaya/User/New.html "Annotea ... Upgraded to Redland 0.9.14"

15:39:58 <danbri> cool :)

15:40:14 <dajobe> yeah, amaya uses raptor

15:40:25 * danbri knew about raptor

15:40:50 <dajobe> I think josek adds one or both of them via CVS rather than releases

15:42:17 <josek> redland is only added to the tar file, but it's not on cvs. I point to your site from our pages

15:42:49 <josek> Raptor is inside amaya, because I've not had time to integrate redland into the the amaya win32. When I do that, I'll remove raptor from our source tree

15:42:53 <dajobe> I guess I was looking at http://dev.w3.org/cvsweb/Amaya/libraptor/

15:43:03 <josek> my prev. comment should answer that

15:43:46 <josek> amaya uses both raptor (for annotatons) and redland (for bookmarks). And one day when time lands on my hands, I'll rewrite the annotations code to work with redland

15:44:13 <dajobe> :)

15:44:47 <GregElin> danbri, libby, dajobe?

15:44:47 <josek> DanBri, the expanded collapsed thing is a piece of UI, but it's state you want to keep in your bookmark file

15:45:49 <libby> hello

15:46:12 <Ian> Dan, when I was looking at classTree.n3 yesterday,

15:46:16 <Ian> I was puzzled by this:

15:46:27 <GregElin> Hey Libby. The proposal for a semantic photo archive at WWW2004 made it passed the local hosts. It looks like it is a go!

15:46:42 <dajobe> hey, near

15:46:43 <dajobe> neat

15:46:51 <GregElin> yes. Very!

15:47:04 <Ian> { v:g g:hasNode v:c.

15:47:04 <Ian> v:p rdfs:subClassOf v:c2. } l:implies { v:g g:hasNode v:c2. }.

15:47:07 <GregElin> I'm excited to see the web page getting some attention, too.

15:47:11 <_joshua> hey greg

15:47:14 <dajobe> the WWW2004, yes

15:47:17 <Ian> Should that read v:c rdfs:subClassOf v:c2

15:47:18 <Ian> ?

15:47:42 <GregElin> Hey Joshua. Are we *ever* going to hang out? heh.

15:47:47 <libby> oh, nice one greg :)

15:48:19 <dajobe> yeah, well done for getting that organised

15:48:47 <GregElin> thank you. thank you. Couldn't have done it without everything I've learned here.

15:48:57 <josek> now if someone organizes an on-line annotation of all the papers by people who look at them using irc and chump....

15:49:07 <dajobe> well, we're thinking about it josek!

15:49:16 <dajobe> add your ideas to http://esw.w3.org/topic/WWW2004

15:49:20 <GregElin> Now comes the real work of organizing it. I'm going to extend that document into a description of scope.

15:49:38 <GregElin> Interesting idea josek!

15:50:36 <GregElin> Ben Shneiderman had some interesting ideas of the size of the archive. We need to pick brainstorm a good theme. But I'd love to see an archive of between photos of past events, this event, and history of the web.

15:50:38 <JibberJim> Does the chump bot still do the auto-post to annotea?

15:50:56 <GregElin> A big juicy data set with a creative commons license for continued research...

15:52:49 <GregElin> I think geoURL should be a part of all this, Josh.

15:53:49 <dajobe>http://www.w3.org/2003/11/results/rdf-core-tests

15:53:50 <dc_rdfig> C: http://www.w3.org/2003/11/results/rdf-core-tests from dajobe

15:53:57 <dajobe> C:|RDF Core Test Results

15:53:57 <dc_rdfig> Titled item C.

15:54:26 <dajobe> C:your chance to show off your RDF systems

15:54:26 <dc_rdfig> Added comment C1.

15:54:48 <dajobe> C:see the [http://lists.w3.org/Archives/Public/www-rdf-interest/2003Nov/0015.html|announcement] by Sandro

15:54:49 <dc_rdfig> Added comment C2.

15:55:10 <dajobe> C:not sure my *parser* is ever going to pass RDF entailment tests though!

15:55:10 <dc_rdfig> Added comment C3.

15:55:26 <dajobe> GregElin: licensing, yeah must be upfront with that too

15:58:18 <sandro> Is there a way you'd rather have the results presented, dajobe? Does RDF Core have classes of systems to which tests apply or do not?

15:58:38 <dajobe> not systems, no

15:59:07 <dajobe> the nearest is for rdf/xml only http://www.w3.org/TR/rdf-syntax-grammar/#section-conformance

16:03:52 <sandro> Yeah. Same with OWL. So my best answer was to just say -- if you don't submit any data on any tests in a section (eg Raptor on Entailment Tests), you're left out of that section.

16:06:05 <dajobe> my prob is that Redland doesn't pass any further tests since there aren't tests that just do rdf triples

16:06:29 <dajobe> ahah http://sesame.aidministrator.nl/testcases/parser/results.rdf

16:07:55 <sandro> Yay!

16:09:15 <dajobe> just parser though, jeen says

16:10:00 <ericm> thanks jeen! :)

16:18:29 <dajobe> C:[http://lists.w3.org/Archives/Public/www-rdf-interest/2003Nov/0033.html|parser data from RIO] from sesame (Aduna)

16:18:29 <dc_rdfig> Added comment C4.

16:30:25 <sandro> C: (now incorporated)

16:30:25 <dc_rdfig> Added comment C5.

16:43:01 <timbl> Ian, did you make an SVG of the owl stuff you did?

16:43:09 <timbl> (I couldn't read the ps)

16:43:24 <timbl> -Tsvg

16:43:26 <Ian> Yes, that's available if you type make -f 16-webarch-Make

16:43:32 <Ian> (PNG, SVG, PS)

16:43:38 <timbl> checked in?

16:43:47 <Ian> Actually, change ps/svg with URI I gave you

16:43:54 <Ian> I did check one in

16:59:53 <DanC> .google semantic web travel

16:59:54 <datum> semantic web travel: http://www.w3.org/2000/10/swap/pim/travel.html

17:27:30 <eaon> eaon is now known as eaon|away

18:29:05 <Dorward> Dorward is now known as DorwAFK

18:30:29 <md-work> md-work is now known as mdupont

19:02:31 <esigler_away> esigler_away is now known as esigler

19:19:41 <Davey> Davey is now known as D[a]vey

20:10:01 <DanC>http://www.w3.org/2001/tag/fdesc54/

20:10:02 <dc_rdfig> D: http://www.w3.org/2001/tag/fdesc54/ from DanC

20:10:38 <DanC> D:|toward formal description of Web Architecture

20:10:38 <dc_rdfig> Titled item D.

20:10:51 <DanC> D:been working with Ian on this.

20:10:51 <dc_rdfig> Added comment D1.

20:11:34 <DanC> D:I've got a couple diagrams I'm reasonably happy with: [http://www.w3.org/2001/tag/fdesc54/webarch-diag.png|webarch-diag.png] (.ps and .svg also in there)

20:11:35 <dc_rdfig> Added comment D2.

20:12:03 <DanC> D:and [http://www.w3.org/2001/tag/fdesc54/webarch-diag-frag.png|webarch-diag-frag.png] which models fragment identifiers

20:12:03 <dc_rdfig> Added comment D3.

20:13:36 <DanC> C:I thought I saw some Euler results in an earlier version

20:13:37 <dc_rdfig> Added comment C6.

20:16:52 <dajobe> C:that was in the OWL test case runs

20:16:53 <dc_rdfig> Added comment C7.

20:16:54 <DanC> C6:""

20:16:54 <dc_rdfig> Deleted comment C6.

20:17:01 <dajobe> heh

20:17:05 <dajobe> C6:""

20:17:06 <dc_rdfig> Deleted comment C6.

20:17:07 <DanC> no, there are Euler results here too

20:17:45 <DanC> Euler passes all the PETs. http://www.w3.org/2003/11/results/rdf-core-tests#ByTestType_table_2_Approved

20:17:48 <dajobe> oh right

20:17:50 <dajobe> it doesn't do parsing

20:18:29 <darobin> darobin is now known as anna

20:23:02 <sandro> I'm working on having Surnia pass them -- I'm kinda stuck on a charmod test looking like invalid UTF8.

20:23:20 <sandro> (which isn't something I want to think about, since I'm NOT doing a parser.)

20:23:52 <dajobe> is it a negative test case?

20:24:05 <sandro> that would make some sense, wouldn't it. Lets see.

20:24:14 <dajobe> you have to use the Manifest.rdf

20:24:40 <dajobe> does owl have must-fail tests?

20:24:46 <sandro> It's http://www.w3.org/2000/10/rdf-tests/rdfcore/rdf-charmod-uris/Manifest.rdf#003; rdflib is telling me UnicodeEncodeError: 'ascii' codec can't encode character '\ue9' in position 24: ordinal not in range(128)

20:25:09 <dajobe> it's not ascii

20:25:39 <sandro> Hrm. Lessee.

20:25:49 <dajobe> it's also Negative

20:25:53 <dajobe> test:NegativeEntailmentTest rdf:about="http://www.w3.org/2000/10/rdf-tests/rdfcore/rdf-charmod-uris/Manifest.rdf#test003">

20:25:59 <dajobe> ^<

20:26:21 <dajobe> what happened to the #test in your url?

20:26:29 <sandro> Negative Entailment doesnt mean it shouldnt parse. But I might just have my ascii confused with a hole in the ground; hold on.

20:26:51 <dajobe> I'm pretty sure none of the XML files are ascii

20:26:59 <sandro> the # part didnt get cut by my double-click and mistyped the rest

20:27:22 <dajobe> ah

20:27:42 <dajobe> but anyway, sounds like an app encoding problem

20:27:58 <sandro> it does.

20:29:12 <sandro> Ah. I changed s = str(f) to s = f.__str__() and it works. python's str() is doing extra encoding-related stuff I dont know about.

20:30:14 * DanC tries to run cwm against the entialment and non-entailment tests...

20:30:18 <dajobe> hmm, the __foo__ stuff's all internal to classes, not for typical app code?

20:31:50 <anna> anna is now known as darobin

20:33:07 <DanC> notation3.BadSyntax: Line 43 of <file:/home/connolly/w3ccvs/WWW/2000/10/swap/test/gentest.n3>: Bad syntax (unterminated string literal)

20:33:29 <darobin> darobin is now known as anna

20:33:46 <DanC> hmm... maybe a good time to try hugo's n3 mode?

20:34:15 * DanC is glad he tucked that away in http://esw.w3.org/topic/CwmTips ...

20:34:55 <DanC> phphe. elisp error

20:35:58 <DanC> aha! turning on "global font lock" option fixes it...

20:37:54 <DanC> well, I get pretty syntax-highlighting, but I still can't see the missing quote

20:38:07 <f8dy> f8dy is now known as f8dy|work

20:38:16 * DanC chacls http://www.w3.org/2000/10/swap/test/gentest.n3 to get more eyeballs

20:38:50 <dajobe> CWM="python

20:39:01 <dajobe> no

20:39:07 <sandro> A class can provide a __str__() method to tell the system how to implement the str(foo) function, in theory.

20:39:15 <DanC> """ --ntriples|diff - ,temp)||echo "<""""

20:39:28 <dajobe> in practice, it's more complex presumably

20:39:59 <sandro> Arg. I have my test results from Surnia now, but they're not showing up. RDF graphs that dont merge when you expect them to are.... painful.

20:40:27 <DanC> yeah, OpenWorldDebuggingHurts

20:41:36 <dajobe> I can't read that gentest.n3 well, but the 'echo PPT' line has only three """ so that's unbalanced

20:42:30 <DanC> I see a few """-delimited strings that end with ", which doesn't work

20:44:17 <DanC> ok, got past that, but it's generating a mostly empty shell script...

20:44:24 <DanC> ah... new test schema URI... <http://www.w3.org/2000/10/rdf-tests/rdfcore/testSchema#>

20:46:46 * DanC is winning now...

20:49:36 <DanC> umm... how can I tell that there are no datatypes needed for a test?

20:50:15 * DanC thinks the manifest has a closed world assumption

20:52:49 <DanC> hmm... :status "APPROVED" is redundant w.r.t. :approval <http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2002Nov/0611.html>; no?

20:54:22 <dajobe> I don't think every test has an approval url

20:54:28 <DanC> ew

20:54:31 <dajobe> the test:status (statui?) are all present

20:54:44 <dajobe> anyway, this file doesn't need to be interpreted as rdf

20:54:52 <dajobe> I use perl regexes

20:54:52 <DanC> ?

20:54:57 <dajobe> so it doesn't even need XML

20:55:07 <dajobe> but that's an aside

20:59:52 <DanC> odd that the entailment tests use .nt rather than .rdf

21:00:18 <sandro> I was just thinking the same thing a few hours ago.

21:00:19 <dajobe> graph to graph tests

21:00:21 <dajobe> not parsing

21:02:13 <anna> anna is now known as darobin

21:04:27 <swh> dajobe: I'm glad you admitted to parsing the manifest with perl, I do too :)

21:04:59 <swh> will write a scirpt to turn my text file into RDF results tommorow

21:05:44 <dajobe> it's in CVS

21:05:47 <dajobe> has been for ages

21:06:02 <DanC> what's in CVS?

21:06:24 <dajobe> the perl script I use to run the tests

21:06:58 <swh> the bulk of mine is the n3 -> RDQL code

21:07:46 <swh> bye

21:08:09 * sandro is tearing his hair out over the graphs not merging. time to stop and think.

21:08:46 <sandro> which is to say -- time to figure out which debugging/monitoring tools might help.

21:38:55 <darobin> !pic Ley

21:39:12 <JibberJim> whwhwhwhw is on foaf

21:39:20 <darobin> oops, that should've been foaf

21:39:23 <darobin> yeah sorry :)

21:44:10 * sandro has some really weird bug stopping node merging in certain cases. Fooey.

21:44:54 <darobin> but they don't *want* to be merged! they want to keep their individuality!

21:48:23 * timbl tunes in, rewinds ....

21:50:12 <mattb> yeek

21:50:36 * mattb loads a couple of foaf files into jena2, switches on the owl reasoner and injects the foaf schema

21:50:50 <mattb> 589 basic triples expand to 2380

21:51:28 <mattb> oops, 444 of those 2380 are the foaf schema of course

21:51:45 <mattb> so only 1347 inferred triples

21:52:03 <JibberJim> what's it inferring?

21:52:10 <JibberJim> lots of sameAs's?

21:52:54 <mattb> and a bunch of differentFroms

21:53:02 <mattb> curious

21:53:54 <mattb> seems to be generating a differentFrom for each pair of distinct foaf:Persons

21:54:00 <mattb> true statement i guess

22:01:01 <sandro> My nodes were not merging because the URIs of some of them were no string object, but they would convert themselves to strings whenever I tried to look at them.,

22:01:28 <sandro> (really, they were subclasses of string.)

22:01:42 <sandro> (not my idea)

22:04:59 <sandro> Hrm. How do I get at the unicode URI essense of an rdflib.URIREF ?

22:31:41 <darobin> darobin is now known as an

22:31:49 <an> an is now known as annna

22:32:26 <annna> annna is now known as darobin

22:39:49 <DorwAFK> DorwAFK is now known as Dorward

22:54:52 <swh> possible rdf test results for 3store in RDF, if anyone cn check them: http://triplestore.aktors.org/rdf-tests/results.rdf

23:09:44 <dajobe> C:[http://lists.w3.org/Archives/Public/www-rdf-interest/2003Nov/0033.html|results from Sesame's RIO]

23:09:45 <dc_rdfig> Added comment C6.

23:10:01 <dajobe> C:[http://lists.w3.org/Archives/Public/www-rdf-interest/2003Nov/0036.html|results from 3store]

23:10:02 <dc_rdfig> Added comment C7.

23:10:12 <dajobe> darn

23:10:19 <dajobe> C6:""

23:10:19 <dc_rdfig> Deleted comment C6.

23:10:55 <dajobe> C:[http://lists.w3.org/Archives/Public/www-rdf-interest/2003Nov/0026.html|results from Euler]

23:10:55 <dc_rdfig> Added comment C7.

23:21:46 <dajobe> ah, I just worked out why I was getting redland clickthroughs today, amaya 8.2pre out

23:27:58 <dajobe>http://www.w3.org/TR/2003/WD-turingtest-20031105/

23:27:58 <dc_rdfig> E: http://www.w3.org/TR/2003/WD-turingtest-20031105/ from dajobe

23:28:15 <dajobe> E:|Inaccessibility of Visually-Oriented Anti-Robot Tests (W3C WD 5 Nov 2003)

23:28:15 <dc_rdfig> Titled item E.

23:28:50 <dajobe> E:not the 'robots.txt' type robot test

23:28:50 <dc_rdfig> Added comment E1.

23:30:03 <dajobe> E:as news.com.com.com* put it [http://news.com.com/2100-1032_3-5103639.html|W3C criticizes antirobot tests]

23:30:03 <dc_rdfig> Added comment E2.