Semantic Web Interest Group IRC Chat Logs for 2004-01-07

03:32:41 * sandro pokes at treehugger and rdftwig and tries to understand their differences, and see if there's a non-java solution.

04:08:35 <DanC> ok, apple's Garage Band looks way, way cool.

04:09:05 <DanC> I bought a rhodes stage piano at an auction a while ago for $50 because I knew that was a rediculously low price...

04:09:39 <DanC> ... when thinking about where to get the money for iLife 4 and the keyboard, I checked the going rate on ebay... $400 to $600! yeah baby!

04:17:23 <_deusx> _deusx is now known as deusx

05:36:41 * DanC starts a game of window-whack-a-mole, trying to close windows, hoping more don't spring up...

10:36:12 <dajobe> dajobe is now known as djb

10:37:47 <djb> djb is now known as daj-no-obe

10:38:08 <daj-no-obe> daj-no-obe is now known as dajobe

11:50:57 <deusx> deusx is now known as _deusx

12:40:22 <bryce_afk> bryce_afk is now known as bryce

13:28:02 <_deusx> _deusx is now known as deusx

13:50:24 <Davey> Davey is now known as D[a]vey

13:53:32 <libby> BLURB:photo and conference annotation UI meeting, 2004-01-07 1600UTC, #rdfig

13:53:33 <dc_rdfig> A: photo and conference annotation UI meeting, 2004-01-07 1600UTC, #rdfig from libby

13:53:45 <libby> A:i.e. here, today, about 2 hours' time

13:53:46 <dc_rdfig> Added comment A1.

14:37:59 <D[a]vey> libby: what exactly does the latter topic (conference annotation) cover?

14:38:50 <libby> sorry that's not a good phrase

14:39:35 <libby> the idea I think is to link together a descripotion of a conference (say WWW2003) with various annotations of it

14:39:49 <libby> I guess one UI for the annotatiosn could be the chump

14:40:04 <libby> probably need another sort for adding images of the conference

14:40:06 <D[a]vey> the annotation using Annotea?

14:40:33 <libby> it's not clear whether you'd need a UI to describe the conference itself and the paper etc, or whether you could just scrape that from somewhere as a one-off

14:40:43 <libby> not particular;y using annotea, no...

14:40:50 <D[a]vey> this is obviously of direct interest to me, and I'm working on something to do with this... sort of

14:40:57 <libby> sure...

14:41:09 <libby> did you see yesterday's discussions?

14:41:14 <D[a]vey> not really :/

14:41:42 <libby> well, there wasn;t a great deal of conferenecs per se but there was a proposed ontology for conferences

14:41:48 <libby> - see yesterday's chump

14:42:39 * D[a]vey looks

14:46:37 <D[a]vey> Hmm, found talk about it, but no link to info about it :/

14:48:31 <libby> one of the cchump items is it

14:49:11 <D[a]vey> guess I'm looking in the wrong place :/

14:49:17 <libby> - http://www.mindswap.org/~golbeck/web/www04photo.owl

14:49:37 <D[a]vey> thanks

14:49:52 <D[a]vey> of course, I have no idea how to understand OWL ;)

14:50:13 <D[a]vey> looks sorta like RDF Schema though...

14:50:23 <libby> yeah

14:50:46 <libby> I *think* owl classes are rdfs classes

14:51:17 <D[a]vey> well, I'm going to be working on XML_FOAF_Parser today, I need to get it working today :)

14:51:39 <libby> it bears some resemblance in part to bits of foaf and RDFical, but there's also some conference-specific stuff at the end

14:51:58 <D[a]vey> seems to use some of FOAF in it

14:52:07 <D[a]vey> like:

14:52:07 <D[a]vey> <owl:Class rdf:ID="Person">

14:52:08 <D[a]vey> <owl:equivalentClass rdf:resource="http://xmlns.com/foaf/0.1/Person"/>

14:52:08 <D[a]vey> </owl:Class>

14:53:17 <libby> right yep

14:53:59 <libby> we were arguing yestrday about whether to use this approach - which defines a new vocan using parts of others but declaring equivalencies; or to use the other vocabs directly

14:54:19 <libby> if you have an owl reasnoer, it's neater to do it the way they've done it there

14:55:13 <libby> if not, it won;t connect up

14:55:23 <D[a]vey> why reinvent the wheel? or re-type anything.

14:56:31 <libby> well, I think the idea is to have a convenient, small vocab to work with on a specific project, rather than picking bits and bobs from everywhere which I suppose might be harder to understand

14:57:21 <libby> from an OWL point of view, there isn;t any difference between foaf:Perso and golback:Person

14:58:30 * D[a]vey is going to have to rewrite XML_FOAF to use RAP methinks

14:58:40 <D[a]vey> not a thought I relish :/

14:58:51 * libby perfers to reuse and I don;t have any owl reasoning capability, so i prefer to mix and match existing vocabs. although sometimes they don;t do exactly what you want (we were arguing about depicts in foaf not applying to parts of images or other multimedia ibjects etc)

14:59:09 <libby> anyway, we're still discussing this

15:00:31 <bryce> if they are truly equivalent, wouldn't *both* (reusing vs. reinventing) approaches work?

15:00:38 <DanCon> does RDF Core meet this fri?

15:02:22 <libby> bryce, yep, if they are equivalent and your software understands equivalency

15:06:02 <nmg> libby: you don't actually need OWL reasoning capability

15:06:06 <bryce> additionally, if our *users* understand equivalency... meaning that good documentation will help people understand how to use (and extend) the vocab properly. (not a new thought, i realize... just reiterating)

15:06:11 <nmg> mutual rdfs:subClassOf will suffice

15:06:29 <DanCon> dajobe, do you expect an RDF Core WG meeting this fri?

15:06:37 <libby> ah, guess so nmg

15:06:49 <DanCon> seems to me the WG can't meet, officially, since 7 days notice hasn't been given

15:06:51 <nmg> ditto for properties

15:06:54 * libby isn;t anti-owl; don;t have any rdfs reasoning either :)

15:07:28 <libby> bryce, I agree; I'm struggling as to what is 'best practice' here...

15:07:33 <dajobe> DanCon: not me

15:08:09 <nmg> libby: you could always ask a third party's reasoner to do the hard bits for you...

15:08:26 <libby> yep true

15:08:56 <nmg> gosh...distribution...that would almost be webby! ;)

15:09:03 <bryce> :)

15:09:55 <libby> I'm very happy to use owl to define ontologies - i think it makes for greater precision. I'm not sure exactly what the benefit of having a new vocab with equivalencies is though, other than readbility

15:10:06 <libby> 'course, readability is useful

15:10:36 * libby possibly missing the point

15:12:34 <bryce> i like the medium ground of "firm, but extensible". with a very specific vocab that borrows from other vocabs and possibly adds a few new terms, we identify the vocab terms we expect to be interested in... it's like highlighting vocabs to encourage the use of certain terms.

15:13:15 <bryce> but not limiting it, either.

15:14:23 <DanCon>http://esw.w3.org/topic/BuildOrBuyTerms

15:14:23 <dc_rdfig> B: http://esw.w3.org/topic/BuildOrBuyTerms from DanCon

15:14:36 <DanCon> B:|BuildOrBuyTerms in the ESW Wiki

15:14:37 <dc_rdfig> Titled item B.

15:14:46 <libby> so is the idea that if(say) you wanted people to write data that looked like http://swordfish.rdfweb.org/discovery/2004/01/www2004/long-eg.rdf, then you would write a new schema expressing euqivalencies to those vocabs?

15:14:48 <DanCon> B:*Should I use somebody else's vocabulary of terms or make my own?*

15:14:48 <dc_rdfig> Added comment B1.

15:14:58 <DanCon> logger, chump B

15:14:58 <DanCon> B:See [http://www.ilrt.bris.ac.uk/discovery/chatlogs/rdfig/2004-01-07#T15-14-58|discussion]

15:14:58 <dc_rdfig> Added comment B2.

15:15:07 <libby> cheers DanCon

15:16:22 <bryce> libby, if there is a specific application (such as www2004)... then maybe it is a good idea.

15:16:27 * DanCon encourages folks to record any suggested conventional wisdom in http://esw.w3.org/topic/BuildOrBuyTerms

15:18:49 <DanCon> hmmm... on "whether to define a new vocab using parts of others but declaring equivalencies; or to use the other vocabs directly"

15:19:07 <libby> I think you could write a schema that did that without using equivalencies

15:19:13 <DanCon> I think equivalencies are for when you discover that they're the same after-the-fact; it's antisocial to make aliases.

15:19:22 <DanCon> ... to make aliases on purpose, that is.

15:19:34 <libby> this is my instinct but I'm not sure how to argue for it

15:19:48 <libby> except that many tools won;t handle it currently

15:19:54 <DanCon> it increases the cost of dealing with the data; puts the burden of equivalence processing on the consumers.

15:20:08 <DanCon> and even tools that do handle it take more CPU time to do it

15:20:27 <libby> that was bwms argument I think

15:21:03 <libby> I tend to work the other way around anyway - start with some data and then work up to a schema

15:21:07 <DanCon> so unless the vocab you're importing from has license conditions that are too scary, or you expect it to become obsolete, you shouldn't make aliases.

15:21:21 <bryce> danCon... antisocial, indeed. so maybe there is a better way to encourage the use of certain terms... maybe just by docementation: "Don't forget about foaf:x and y:This"

15:22:43 <libby> I think one interesting thing is how to make links that don;t alreday exist between vocabs, e.g. rdfical and geo

15:23:43 <libby> the author didn;t intend there to be a link; but it seems to make sense to link them; say I describe some link in a schema document...I wonder what the relationship is between my schema and the intial ones

15:24:15 * DanCon reflects his view in http://esw.w3.org/topic/BuildOrBuyTerms of 2004-01-07 09:23:50

15:24:26 <libby> maybe some are trustworthier than others

15:25:13 * bryce is becoming convinced.

15:26:16 <bryce> there isn't a need to abstract to another level. this should be done on the UI level, instead.

15:26:30 <bryce> (the abstraction, that is)

15:27:15 * bryce wonders if that makes sense.

15:27:17 <DanCon> yes, good UIs can make a blizzard of namespaces look nice

15:43:18 <nmg> libby: in the OWL worldview, the axioms which form the links between two existing ontologies are themselves another ontology which extends both of the ontologies it links

15:44:02 <libby> right, makes sense, I think

15:54:38 <GregElin> Morning.

15:55:44 <bryce> greg, i'm very interested in the UI discussion at 1600utc, but I may be called out into a meeting any minute.

15:57:27 <GregElin> Well, do you want to share any thoughts Bryce?

15:57:53 * bryce is thinking.

15:58:52 <GregElin> I've been working on some interface improvements to Fotonotes. The mock up is here (just click on the view notes link under the image): http://www.gregelin.com/blogpage3.html

16:00:12 * bryce is quite impressed with the fotonotes.

16:00:24 * GregElin blushes

16:00:28 <GregElin> Thanks.

16:00:44 <GregElin> Did you click and drag over an area of the image.

16:01:05 <mc> greg, greetings

16:01:31 <GregElin> Greetings, monica.

16:01:46 <mc> Greg, i'm not sure where to start...

16:02:00 <GregElin> hmmm..

16:02:08 <mc> i know we can do a UI for the exploring

16:02:14 <mc> of the conference/pic stuff

16:02:32 <mc> That will fall out of the southampton team work

16:02:35 <GregElin> Let's iterate what we need, specifically, then take inventory of what we have. Then look at some things and discuss?

16:02:40 <mc> ok

16:02:46 <mc> let me start

16:02:49 <GregElin> Okay.

16:02:59 <GregElin> (libby, can you help me archive this?)

16:03:14 <mc> we need two things: 1) a ui for annotations of photos

16:03:32 <mc> and 2) an explorer for the photos and all the other associated dimenstions

16:03:51 <mc> we can do the explorer of the dimensions - at least a first one

16:04:01 <mc> if that sits well with you

16:04:10 <mc> which leaves the ui for the annotations

16:04:12 <GregElin> Do we need a third UI, a UI to annotate/comment on the other dimensions? Or a UI to link images to other dimensions?

16:04:23 <mc> maybe not

16:04:40 <GregElin> I think I see why not, but lets' review, okay?

16:04:44 <mc> ok

16:04:48 <mc> here's a couple thoughts

16:05:08 <mc> right now, we have a lot going on to think about how we'll manage all the possible annotations

16:05:11 <GregElin> You are gathering most of the information for conference material via what? existing sites?

16:05:23 <mc> yes

16:05:31 <mc> so while i can see

16:05:32 <GregElin> And that information becomes self-explanatory b/c it is already "semanticized"?

16:05:39 <mc> that's it yes

16:05:49 <mc> so while holes can be filled by annotation

16:05:57 <mc> maybe that's not a priority right now

16:06:03 <GregElin> I agree.

16:06:06 <mc> since there's annotation being demo'd on the photos

16:06:13 <mc> May i ask tho

16:06:17 <GregElin> yes?

16:06:25 <mc> if Jim's group has a conference annotator hunk of thing,

16:06:37 <mc> then perhaps that's not a biggie to make available?

16:06:47 <mc> if it comes on stream that's great

16:07:03 <mc> i'm not exactly clear what jim's offering???

16:07:07 <GregElin> what's not a biggie to make available? Additional annotations using Jim's conference annotator?

16:07:08 <mc> could you clarify?

16:07:15 <mc> i mean beyond the ontology

16:08:17 <GregElin> Jim and I had a good conversation Monday. He first off sees what they have as an additional/alternative tool set in the spirit of the repository supporting different interfaces.

16:08:48 <GregElin> And he agrees with the idea that default tools are pointed to as to not confuse things, especially in the beginning.

16:09:14 <mc> right. what do his tools/alternatives do?

16:09:32 <GregElin> From what I've seen and gathered speaking with Jim, the tools they have, while not refined or pretty yet have bits that overlap with photo annotation and conference thingies.

16:10:02 <mc> hmm.

16:10:04 <GregElin> Or more specifically he has an annotation for conferences more than conference materials, that is, the *event* itself and subdivisions of the event.

16:10:32 <mc> could you give me an example? i'm not grokking this

16:10:59 <GregElin> Well, I'm learning all this, too.

16:11:04 <mc> ok

16:11:25 <mc> i think these tools may be more back endy

16:11:34 <GregElin> As I understand it Aktive space will help organize conference materials like presentations, powerpoints, papers, posters, etc.

16:12:01 <mc> let me send you a url for cs aktive space

16:12:06 <GregElin> That is, your work tackles alot of the conceptual space of a conference.

16:12:15 <GregElin> the ideas that are being explored.

16:12:15 <mc> so you can get a sense of an instance

16:12:32 <mc> Right. it's about context and association

16:12:35 <bryce>http://eprints.ecs.soton.ac.uk/archive/00007440/

16:12:35 <dc_rdfig> C: http://eprints.ecs.soton.ac.uk/archive/00007440/ from bryce

16:12:46 <mc> how does this thing here relate to that thing there?

16:13:24 <mc> so you might see a photo, see who's in it, but also see what session, the papers, or whatever info available about that thing

16:13:31 <mc> alternately you might start with a paper

16:13:35 <mc> and get to a photo

16:14:29 * GregElin looks at link

16:14:29 <mc>http://triplestore.aktors.org/SemanticWebChallenge

16:14:29 <dc_rdfig> D: http://triplestore.aktors.org/SemanticWebChallenge from mc

16:14:48 <libby> could you guys title them when you chump the links, e.g.

16:15:02 <libby> C:|CS AKTive Space: or How We Stopped Worrying and Learned to Love the Semantic Web

16:15:03 <dc_rdfig> Titled item C.

16:15:18 <GregElin> And AKtive space handles the organization of information so it can be "flattened" into different hierarchies?

16:15:32 <mc> Greg

16:15:34 <mc> yes

16:15:38 <mc> it's like this

16:15:46 <mc> you have a multidimensional space

16:15:51 <mc> you take a projection

16:15:56 <mc> flatten the projection

16:16:01 <mc> you get a hierarchy

16:16:05 <mc> mspaces let you

16:16:22 <mc> manipulate the hierarchies and define the slices

16:16:41 <mc> by slice we mean a slice is a projection through the n-dimensional space

16:16:53 <mc> the idea is to support the user determining

16:17:02 <mc> what in that space is relevant important to them

16:17:11 <mc> and letting them orient the universe around that

16:17:26 <mc> Applying mspaceinterfaces to the semantic web describes this:

16:17:28 <mc>http://eprints.ecs.soton.ac.uk/archive/00008639/

16:17:28 <dc_rdfig> E: http://eprints.ecs.soton.ac.uk/archive/00008639/ from mc

16:17:42 <mc> the title is Applying mspace interfaces to the semantic web

16:17:52 <mc> so ya, we leverage the semantics of the space

16:18:01 <mc> to represent the relations among the dimensions

16:18:07 <mc> for example

16:18:16 <mc> one projection through classical music is

16:18:29 <mc> period, composer, genre, arrangement, piece

16:18:50 <mc> a user who knows nothing about classical music

16:19:07 <mc> but who plays the guitar may want to reorganize that space from instrument

16:19:24 <mc> so the hierarchy changes when it's instrument, period, composer,

16:19:30 <GregElin> Okay. I think I got that.

16:19:37 <mc> the person may then wish to add a new dimesion, like recording

16:19:42 <mc> etc

16:20:04 <mc> in this case, with the conference photo site

16:20:22 <mc> users would be able to focus on the attributes

16:20:27 <mc> important to them

16:20:31 <mc> get at those resources

16:20:40 <mc> in ways that are meaningful to them

16:20:54 <GregElin> So I think two things: (a) Jim's ontology is more focused on the logistic of a conference/event: where, who, schedule, etc. (b) Jim's ontology may be the words that are used in the photo annotation's that help link the photos to mspace's ontology of the conference concepts.

16:20:55 <mc> in the default orientation, we'd privilege photos first

16:21:09 <mc> gerg re Jim, yes

16:21:15 <mc> and that's what we'd hook into

16:21:21 <mc> to represent the space

16:21:37 <mc> along with the ontology for the photos

16:22:06 <mc> If jim's tools let you revise the ontology

16:22:16 <mc> it would be cool to add that to the mspace

16:22:27 <mc> so that you can define new possible associations

16:22:32 <mc> as data becomes available

16:22:44 <mc> So, timelines?

16:22:59 <mc> oh - question!

16:23:04 <mc> would it be possible

16:23:16 <mc> to focus on the photo annotator first as the main UI

16:23:18 <mc> and then bring

16:23:26 <mc> the mspace online as we get the conference data?

16:23:34 <GregElin> I think that's the idea...Jim's stuff could be an nice bridging tool and also just another alternative tool set.

16:23:54 <mc> are you thinking of using your tool for the photoannotation?

16:23:57 <mc> that would be cool

16:24:02 <libby> GregElin: are you going to build on fotonotes for catalogiung photos?

16:24:08 <GregElin> Yes, mc. I think that is a good idea. We first need a way to gather and annotate photos. We need to get that going pronto!

16:24:26 <GregElin> Especially collecting photos!

16:24:29 <mc> to hook the data it saves into a triplestore

16:24:33 <mc> so have you

16:24:41 <mc> talked with folks about

16:24:58 <mc> storing the data from fotonotes as triples?

16:25:05 <mc> as rdf?

16:25:08 <GregElin> Yes, monica, I'd like to use Fotonotes (but also don't want to say ONLY fotonotes...want it to be open).

16:25:12 * libby would like to propose this as an alternate intereface for catalogiung photos: http://swordfish.rdfweb.org/discovery/2003/06/codjsform/shell.html

16:25:18 <libby> - ugly but workie

16:25:21 <GregElin> yes again.

16:25:38 <mc> as long as the data gets dumped to rdf ...

16:25:42 <GregElin> Alternative as in additional Libby or as instead of?

16:26:00 <mc> i think we need to foreground one tool, no?

16:26:01 <libby> additional I meant

16:26:05 <mc> right

16:26:06 <GregElin> Cool.

16:26:17 <GregElin> I like it Libby.

16:26:24 <libby> can you get fotonotes to emit rdf in time greg?

16:26:24 <mc> can fotonote store to rdf?

16:26:29 <GregElin> Let me point you to this link:

16:26:59 <mc> it's not just rdf is it. it would be nice if that rdf used one of the ontologies we're discussing, eh?

16:27:21 <GregElin> Yes, just one.

16:27:23 <GregElin>http://www.fotowiki.net/ElinFamily/tablescreen.ftw

16:27:23 <dc_rdfig> F: http://www.fotowiki.net/ElinFamily/tablescreen.ftw from GregElin

16:27:36 <mc> So path would be, get fotonotes+ up as the front end for the photo annotation, 2) make

16:27:43 <GregElin> This version is simple click-drag over area of photo.

16:27:46 <mc> sure it's talking the right ontology

16:27:59 <mc> and then get that data into the explorer

16:28:08 <mc> so three phases

16:28:23 <mc> what help would you like with fotonotes for this project?

16:28:25 <libby> is it relatively briwser-compatible greg?

16:28:36 <GregElin> Right now, selecting a region brings up a form with a single text area. But I was thinking we could add some additional fields specific to the context of W3 conferences.

16:28:42 <libby> (cos WWW2004 will have a lot of accessibility ect stuff in it from W3C)

16:29:23 <mc> like basic alt= image tags

16:29:32 <GregElin> Relatively, libby. Breaks in Apple Safari and Opera, but I think that is fixable.

16:29:38 <libby> cool

16:29:52 <mc> and the rdf issue?

16:30:11 <libby> yep alt tags would be cool

16:30:20 <libby> so you could just have a bunch more boxes?

16:30:35 <teefal> '/clear

16:30:37 <mc> also

16:30:41 <GregElin> The way the current code behind is this designed is that you could put different "display engines" in. So instead of javascript rollovers and dhtml, you could use render the image as an image map and use alt tags.

16:30:54 <mc> the current foton only has one field

16:30:56 <libby> like person/email/homapge whatever/ conference url etc to fill in?

16:31:10 <mc> it doesn't categorize/metadata the data

16:31:14 <libby> also annotator etc?

16:31:26 * libby thinks it could work. looks nice :)

16:31:27 <GregElin> Right. But we could (a) add additional fields or (b) make the single field wiki like. or both.

16:31:39 <GregElin> Right!

16:31:45 <mc> this is where the ontology comes in

16:31:53 <mc> maybe we can provide category fields

16:31:58 <mc> users pick those

16:32:01 * libby thinks (a) is less awkward, less prone to error, though harder to make look nice

16:32:02 <mc> and then add annotations

16:32:24 <GregElin> One thing I do like about Fotonotes approach is that it doesn't force the person to fill out a big form. It accepts informal chatter. I'd like to keep that element of informal chatter a part of this.

16:32:35 <mc> right

16:32:40 <mc> but we also need metadata

16:32:47 <libby> what about making that a freetext dunlin core description?

16:32:51 <libby> dublin

16:32:57 <GregElin> I'd also like to see tools that look at what's in the single text area field and try to CREATE triples, foaf, etc. and propose that back to the user.

16:33:07 <mc> oh gawd

16:33:10 <libby> that would be neat

16:33:14 <GregElin> Yes. All that.

16:33:17 <libby> I don;t know of any like that though...

16:33:47 <mc> we need alternative for version 1, gang

16:34:09 <mc> that means some kinds of pick lists, or keyword:pair

16:34:16 <GregElin> Hey, I'm up for anything! I've been thinking about this piece of making the selection easy, storing it the photo. I need all your help to glue to the semantic stuff, Dublin Core, Exif, yada-yada.

16:34:42 <mc> maybe we can get a parser happening

16:34:49 <mc> so that if we have simple syntax

16:35:14 <swh> swh is now known as swh_afk

16:35:16 <GregElin> you lost me on "mc: we need alternative for version 1, gang"

16:35:27 <libby> there are many wiki-like syntaxes available. I'm not convinced that people will be interested in learning another

16:35:42 <GregElin> Do you think a variation of Fotonotes with a few recommended fields in addition to the text area gets us going?

16:35:48 <mc> ya, sure but wiki is intense

16:35:53 <libby> I do like the idea of natural language parsing, but I would simplify for now

16:36:09 <libby> you can contextualize the boxes in the simplest case

16:36:14 <mc> indeed

16:36:15 <GregElin> Oh yes. Simple to start. free form text area, 2-4 fields.

16:36:20 <libby> e.g. by assuming the box is a person

16:36:41 <libby> - and asking for name/identifer/something

16:36:46 <mc> ya, with "optional" strongly encouraged fields available

16:37:03 <mc> what are the core usual values?

16:37:05 <libby> then various fields for whlole picture - freetext description, date

16:37:09 <mc> provide those as defaults

16:37:10 <GregElin> Given the narrow context (conferences), I think we could try a drop down box for roughly categorizing things: person, group, panel, event, poster, slide, other.

16:37:30 <mc> sure, with appropriate defaults selected

16:37:45 <mc> so drop down matched with empty field

16:37:50 <GregElin> I think we might look at how your tool, libby, could help with photo overall.

16:37:54 <mc> then move to natural language parsing

16:38:25 <mc> i'm sorry but i have to go to another meeting. can we reconnoiter again at this time tomorrow?

16:38:34 <mc> i'll have an hour then

16:38:41 <GregElin> Another option is dragging icons onto photos and allow placement of icon to create a fotonotes selected area of a default size.

16:38:58 <GregElin> The possibilities are ENDLESS! Mwah-ha-ha!

16:39:02 <mc> could we get the url for libby's tool?

16:39:18 <mc> greg, could you do a draft ui idea in power point and send it to us?

16:39:23 <libby> where my tool tries to help is for finiding IDs for people where they exist (and other stuff like places)

16:39:27 <GregElin> Okay, libby.

16:39:49 <mc> we're getting somewhere.

16:39:55 <libby> which is potentially a reasonable-sized problem - not decided how to ID people yet

16:40:10 <GregElin> BTW...meet tomorrow is good. However, I may be in transit at that time tomorrow. Feel free to continue forth if I don't show. I'll get caught up.

16:40:17 <GregElin> Great.

16:40:34 <mc> greg, we don't have to meet - it would be better to get a draft to discuss

16:40:39 <libby>http://swordfish.rdfweb.org/discovery/2003/06/codjsform/shell.html

16:40:39 <dc_rdfig> G: http://swordfish.rdfweb.org/discovery/2003/06/codjsform/shell.html from libby

16:40:42 <mc> is that workable for you?

16:40:48 <libby> G:|javascript photo annotation tool

16:40:49 <dc_rdfig> Titled item G.

16:41:11 <GregElin> My big agenda is keeping things easy to show people that image annotation --and the semantic web-- can be easy.

16:41:36 <mc> can you do the power point mock up, email us, and then we'll reconnoiter?

16:41:42 <mc> must dash...

16:41:50 <GregElin> I like to think HTML 1.0 succeeded in mass adoption b/c it was a subset of SGML and was digestable for newbies.

16:41:59 <libby> G:uses lightweight webservices, and jim ley's js rdf parser. IE and moz only. [http://www.ilrt.bris.ac.uk/publications/researchreport/rr1065/report_html?ilrtyear=2003|more info]

16:41:59 <dc_rdfig> Added comment G1.

16:42:05 <GregElin> I'll do my best, mc!

16:42:17 <libby> any chance not power point?

16:42:23 <mc> what would you prefer?

16:42:28 <libby> html :)

16:42:37 <mc> you can't draw fast in html

16:42:41 <libby> true

16:42:55 <mc> this is just a sketch. draw it on paper and scan it!

16:43:02 <GregElin> Look, I can mock up a form in working HTML and attach it to Fotonotes. It just might be the form doesn't process.

16:43:08 <mc> it's just so we know what we're talking about.

16:43:10 <mc> ok

16:43:12 <nmg> [an SGML pedant's aside - HTML is an application fo SGML, not a subset]

16:43:18 <mc> bye for now...

16:43:41 <libby> cheers :)

16:44:29 <GregElin> Thanks, nick. That is better language. And I think we are looking here at an application of semantic web/photo annotation that is similarly an easy on ramp for people onto the possibilities.

16:45:10 <libby> G:[http://swordfish.rdfweb.org/discovery/2001/08/codepict/|searchig the image data]

16:45:10 <dc_rdfig> Added comment G2.

16:45:49 <libby> in a sense, the UI is the real innovation here. the rest has been done before pretty much by several different groups...

16:46:19 <libby> ...although that's not to say all the problems are resolved

16:48:06 <libby> do you have any idea how you will do access control GregElin?

16:48:36 <libby> I mean, it's nice to keep track of who has said what

16:50:02 <libby> also - at the moment (if I undertand it correctly) there's only one set of annotations per photo? do you are always editing/deleteing the same ones. would be nice to get different peoples' annotations. (maybe I misunderstood how it works though)

16:50:16 <GregElin> Gosh, libby. Such good questions.

16:50:58 <GregElin> The access control is the most serious issue to figure out. I haven't tackled it b/c (a) didn't have a great solution and (b) wanted to decide with a group.

16:51:15 <GregElin> Oh...the basic who said what is already built in if a login is used.

16:51:30 <libby> cool

16:51:44 <GregElin> We could take a blog-comment-form-trust stance and included a field to identify who you are.

16:52:11 <libby> yeah, by homepage/email is cool (I think)

16:52:19 <GregElin> +1 on different people's annotations. I'd like to get to a point where there are layers of annotation, like GIS.

16:52:24 <libby> be nice to avoid logins etc

16:52:46 <GregElin> Well, need a login for people who contribute photos. Have to be able to assign blame.

16:52:48 <GregElin> :-)

16:53:14 <libby> well, it might be that in this version there can only be one annotation oer pic. but if export as RDF, could combine wih other annotations of the same photo (as long as we allow for that in the modelling of the vocab)

16:53:22 <GregElin> Unless, people put photos up on a page they control and we spider and download them, kind of like how geoURL.org works.

16:53:37 <libby> I was wondering about that latter method

16:53:46 <GregElin> Authenticity is trusted via party's access control to their own URL.

16:54:00 <libby> we would have to validate tat there was a license attached somewhere I guess

16:54:13 <GregElin> Build a page that looks like X (web page, xml) and include this information and we'll get your images.

16:54:14 <libby> I don;t know what I mean by 'attached'

16:54:19 <GregElin> Maybe that is a better way to go?

16:54:23 <libby> yeah

16:54:26 <libby> I like that

16:54:41 <libby> can avoid lots of nasty infrastructure I think

16:55:09 <libby> hm, on the other hand, might be a pain for people to put their photos up in a particular way

16:55:16 * bryce is out of meeting now... lots of catching up to do in this channel.

16:55:29 <libby> on the other other hand if it's just a matter of a link to a CC, license, that's not so bad

16:56:13 <GregElin> Please do catch up bryce. Opinions and thoughts welcome. We are going to try and hook up again at 1600 gmt tomorrow.

16:56:33 <GregElin> Maybe we do both. Encourage the web page, but accept uploads with a login.

16:57:02 <GregElin> Hmmm...maybe fotonotes MAKES that web page!

16:57:20 <GregElin> me shivers

16:57:24 * GregElin shivers

16:57:30 * libby brings http://rdfig.xmlhack.com/2004/01/07/2004-01-07.html#1073488463.663254 to GregElin's attention, re ontologies and whatnot

16:57:41 <libby> I swear I'm not trying to be difficult :)

16:58:20 <GregElin> Shivering with excitement!

16:58:48 <GregElin> I was scanning back for "attached". Where did you refere to that?

16:59:15 <libby> <libby> we would have to validate tat there was a license attached somewhere I guess

16:59:20 <libby> to the photos I mean

16:59:36 <libby> or, somewhere in a thumbnail page, or something

16:59:42 <GregElin> I think eventually people will be able to create a set of annotations that point at a photo, so annotations can be created by parties distinct from the target object.

17:00:01 <libby> right....that would be ideal

17:00:21 <libby> the way I have done that is just to point to the photo on the web, and use its url as its identifier

17:00:41 <libby> I think you want to take a copy of the photo (which is fair enough, making it a dataset)

17:00:46 <GregElin> Ah. Well...we use Fotonotes to store/point to annotation right in jpeg (in fotonotes info, or EXIF data, or both, or other). And we also log in a database that photo/URL was donated with a particular license.

17:01:04 <libby> ...but it'll still need an identifier (a url? but then it may have 2 urls...)

17:01:21 <libby> cool re inside the image

17:01:43 <libby> ndw and the rdfpic folks have done that using rdf data

17:02:04 <GregElin> I think that works, libby. My only question about pointing at a photos URL is, in the long run, that method enough? Do people also want a second URI technique in case the photo moves. Kind of like dual entry accounting systems. Each transaction is stored in two places for integrity.

17:02:05 <libby> I'm not sure how we can refer to the identifier of the photo it is inside

17:02:31 <libby> I think in this case makes sense to take a copy

17:02:32 <GregElin> This might be the project where we figure out a framework to get all these approachs to play together.

17:03:07 <GregElin> In fact, my plan/hope is to see Fotonotes be part-of/base-of an open standard.

17:03:08 <libby> ...however the way I've tended to ID photos is by url, so if there are two that might be a headache. but maybe not if use owl

17:04:01 <libby> cool

17:04:10 <libby> what do you use as a backend currently?

17:04:27 <GregElin> Yeah, I don't feel like I know enough about different techniques out there to recommend solution. Want to learn more. My gut and head tells me something will emerge.

17:04:34 <GregElin> Someone will figure it our.

17:04:37 <GregElin> s/our/out

17:04:57 <libby> do you generate thumbnails btw?

17:05:50 <GregElin> PHP library that reads/writes jpeg, creates images on fly. Plan is to open source that library. It certainly needs to be extended. And to have a plug-in structure so it can read EXIF, OWL, rdf, whatever.

17:06:08 <GregElin> Yes. I can generate thumbs.

17:06:29 <libby> cool

17:06:33 <GregElin> go back to this link. And click edit button: http://www.fotowiki.net/ElinFamily/tablescreen.ftw

17:06:41 * bryce is caught up now. shewwwwwwwww!

17:06:54 <GregElin> shwiiing!

17:07:17 <GregElin> Additional thoughts welcome, bryce.

17:07:29 <libby> right yep GregElin, that'd be handy for other tools that want to create rdf for that same photo

17:07:52 <bryce> regarding dropdown boxes and libby's mbox lookup...

17:08:39 <bryce> it would be nice if the annotator "learns"... in the sense that if a given conf panel is id'd, then the panelists names appear in the drop down boxes.

17:08:46 <GregElin> the edit libby? or the fotonotes library being open source?

17:08:59 <libby> the edit

17:09:04 <libby> but also the open source :)

17:09:08 <GregElin> can you elaborate?

17:09:13 <libby> bryce, that'd be cool

17:09:28 <GregElin> It be great bryce. Do you have experience in programming things like that?

17:09:42 <bryce> the second section of the drop down would list people who have stated that they attended.

17:09:56 <GregElin> Seems we need a way to group photos and have individual images learn they are part of that group/context.

17:10:06 <bryce> greg... a bit, but only with php/mysql.

17:10:12 <libby> that would be useful but I'm not sure if attendee list will be avilable?

17:10:20 <bryce> right.

17:10:33 <libby> GregElin: there's no real need to group photos, or only in a very weak sense

17:10:39 <GregElin> Also interesting if system learns about the user...I'd have my list of people I know. Libby would have list of people she knows.

17:10:54 <bryce> but any that are known ta attend make useful list choices.

17:11:05 <libby> if a photo depicts a pannel, and a panel has members, then that's enough to retrieve that information

17:11:06 <GregElin> what a coincidence. Fotonotes is php/mysql...

17:11:14 <bryce> cool.

17:11:21 <GregElin> Duh. of course. you are right libby.

17:11:43 * libby keeps returnngto the problem of identifiers thouigh (this is nmg's doing...)

17:11:58 <GregElin> So I think we look, libby, how to pull in your stuff to the overall identification of a photo.

17:11:59 <bryce> also, if we can trust others' annotations, the attendee list grows organically.

17:12:29 <libby> I don't think panels have homepages; eople do, some events do, but probably not subevents like individual papers...

17:12:33 <bryce> a person depicted in a photo was obviously at that event.

17:12:38 <GregElin> I don't know if I see the identifier as any major stumbling block.

17:12:51 <libby> hm, probbaly bryce

17:13:04 <libby> coulda snuck in :)

17:13:14 <libby> no GregElin?

17:13:36 <bryce> maybe not officially attending, but there... (barring photoshop)

17:13:42 <libby> well internally it's not a major problem, but as RDFit is I think

17:14:10 <bryce> can we hook into the www2004 registration ?

17:14:14 <GregElin> We have a specific contexts. We have a significantly size sample of images. I think errors, crossed-identifiers kind of show up quickly, especially since the identifiers point to visual objects. We need an identifier technique, but it doesn't have to be full proof.

17:14:19 <libby> so suppose jim hendler's group's tool wants to say that this is a pic of the panel that's also annotated by a fotonotes annotation - they can;t identify it

17:14:49 <GregElin> Wow...I didn't know that J.Lo use to attend W3 conferences....

17:14:51 <libby> no, I don;t think it has toi be foolproof either, but I do reckon it has to be globally unique

17:14:56 <libby> heheh

17:15:15 <GregElin> you lost me libby.

17:15:31 <libby> so we need to think of a way (or more than one way) to ID everything we wat to talk about - people, places, events,...

17:15:57 <bryce> isn't there any conference data we could use as ID's? why reinvent here?

17:16:17 <libby> sorry, I just meant that for people you could use mbox (which will occasioanlly fail) as an ID, but we shouldn't just use some application-specific identifier

17:16:19 <GregElin> Bryce, we can't guarantee that we hook into registration, but I think if the lists are public, we can get those lists. And as the project moves forward...things will come together and if we do well, more people will want to glue more things to it.

17:16:25 <libby> yeah maybe bryce

17:17:05 <bryce> even if we scraped it off the printed material... room #, day, time for a panel

17:17:05 <GregElin> I know what it my thoughts are on the ID issue.

17:17:26 <libby> so for www2003, all papers are online

17:17:34 <GregElin> The ID is an issue that *always* exist, especially with photos. One general/macro solution is URL as URI.

17:17:35 <libby> but a paper is not an event.hm

17:18:06 <libby> aure GregElin, but photos are more likely to have urls than events I think

17:18:07 <GregElin> Another approach is just trying to sure consistent Identifiers locally, within a specific context.

17:18:18 <libby> I'm sure we can come up with something, just wondering...

17:18:27 <bryce> +1 greg.

17:18:48 <GregElin> True, libby. But isn't this identifier one of the big challenges for the semantic web?

17:18:53 <libby> hm, but what then about the other tool that wants to refer to th event?

17:19:01 <bryce> we can always massage the data later. this is the approach i've taken in my own photo collection.

17:19:10 <libby> yep, bigger than we thought :) or I thought anyway

17:19:39 <GregElin> I'm not sure how we are going to id events. I'm hoping Jim's group has insight.

17:20:07 <libby> well, compromise is fine for speed. maybe when mc and the others get the data about the conferenecs togther, they could provide ids

17:20:33 <libby> events are notoriously hard to identify...

17:20:34 <GregElin> The most interesting thing we can do with this project, libby, is to create a significantly sized collection of photos that have enough semantic data that these issues can be studied properly.

17:21:04 <bryce> when we know where (room) and when (datetime), we make some naming convention, ie 204H-2004-05-17T12:30

17:21:07 <GregElin> My goal is to not solve them...just gather a worthwhile collection of images around a context to give us some real meat to chew on.

17:21:36 <GregElin> +1 for "maybe when mc and the others get the data about the conferenecs togther, they could provide ids

17:21:38 <GregElin> "

17:21:57 <libby> bryce's idea would work I think

17:22:22 * bryce 's photo collection is CC. could be used for example images. http://tenyearsof.us

17:22:39 <GregElin> Gosh, that would be great Bryce. And I think we are going to be surprised by even that b/c these conferences took place all over the world...

17:22:41 <libby> cool :)

17:22:49 <bryce> not conference based, mind you.

17:24:43 * libby better go. nice chatting to you guys :)

17:25:13 <GregElin> Such interesting things to figure out. I'm all for starting with convention X and seeing how it works. Very into decide-prototype-test-repeat.

17:25:28 <GregElin> I better be running too

17:25:32 <bryce> right. "don't worry, be crappy"

17:25:38 <GregElin>http://tenyearsof.us

17:25:38 <dc_rdfig> H: http://tenyearsof.us from GregElin

17:25:43 <bryce> nice chatting with you

17:25:49 <GregElin> I'll remember that Bryce.

17:25:50 <GregElin> Cute.

17:27:28 <GregElin> Good talking with you, Bryce.

17:29:44 <bryce> H:Volunteered as a repository of example images for experimenting with annotation.

17:29:45 <dc_rdfig> Added comment H1.

17:31:40 <bryce> F:Demo of Fotonotes

17:31:41 <dc_rdfig> Added comment F1.

17:32:46 <bryce> F:<GregElin> The way the current code behind is this designed is that you could put different "display engines" in. So instead of javascript rollovers and dhtml, you could use render the image as an image map and use alt tags.

17:32:46 <dc_rdfig> Added comment F2.

17:33:49 <bryce> E:|Applying mSpace Interfaces to the Semantic Web

17:33:49 <dc_rdfig> Titled item E.

17:34:26 <D[a]vey> D[a]vey is now known as Davey

17:34:36 <bryce> D:|CS AKTiveSpace Tour

17:34:37 <dc_rdfig> Titled item D.

17:51:37 <swh_afk> swh_afk is now known as swh

18:02:31 <bryce> bryce is now known as bryce_outtolunch

18:09:17 <tav> tav is now known as tav|offline

18:24:53 <dmiles> is there a KIf or CycL converter dirrect to RDF ?

18:33:45 <bryce_outtolunch> bryce_outtolunch is now known as bryce

18:43:36 <sandro> Give me the KIF that would produce this RDF: _:a <b> "c"

18:51:48 <sandro> (just curious what you'd want there.)

18:52:21 <sandro> (I've written a few such things, like an lbase parser.)

19:06:21 <Davey> Davey is now known as D[a]vey

19:51:21 <D[a]vey> D[a]vey is now known as Davey

20:05:11 <Davey> Davey is now known as D[a]vey

20:48:30 <Arnia22> Arnia22 is now known as Arnia

20:53:50 <D[a]vey> D[a]vey is now known as Davey

20:59:47 * dajobe looks around for shellac

21:11:43 <mortenf> sometimes this is outright scary...

21:12:01 <dajobe> ok, hi shellac

21:12:15 <dajobe> remember about N-Triples Plus

21:12:16 <shellac> hi dave

21:12:22 <shellac> oh yes

21:12:27 <dajobe> it has a new name

21:12:34 <dajobe> Terse RDF Triple Language

21:12:37 <dajobe> Turtle

21:12:41 <shellac> yay

21:12:44 <mortenf> heh

21:12:52 <dajobe>http://www.ilrt.bris.ac.uk/discovery/2003/11/ntriplesplus/

21:12:52 <dc_rdfig> I: http://www.ilrt.bris.ac.uk/discovery/2003/11/ntriplesplus/ from dajobe

21:12:59 <dajobe> I:|Turtle - Terse RDF Triple Language

21:12:59 <dc_rdfig> Titled item I.

21:13:01 <shellac> good name

21:13:05 <dajobe> I:it's turtles all the way down

21:13:05 <dc_rdfig> Added comment I1.

21:13:53 <dajobe> I:after [http://planetrdf.com/|Planet RDF] slogan

21:13:53 <dc_rdfig> Added comment I2.

21:15:02 <shellac> the name was worth the wait

21:15:17 <dajobe> heh, it came to me this morning just walking in to ilrt

21:16:07 <mortenf> any news on allowing - in qnames?

21:16:28 <dajobe> this is just a rename, no other update.

21:16:35 <mortenf> yeah

21:16:44 <dajobe> I:formerly N-Triples Plus (working name)

21:16:44 <dc_rdfig> Added comment I3.

21:19:49 <tav|offline> tav|offline is now known as tav

21:20:30 <dajobe> I:or the quote after [http://www.the-funneled-web.com/hawking.htm|Hawking or Russell]

21:20:30 <dc_rdfig> Added comment I4.

21:32:13 <bryce> cheers to TRTL and the slogan "it's turtles all the way down"... even comes with its own mascot!

21:32:29 <dajobe> yeah

21:33:00 <bryce> reminiscent of tortoise and hare fable... very apt.

21:33:27 <dajobe> I found it the right name one I got there.

21:33:32 <dajobe> s/one/once/

21:34:06 <bryce> tortoise -> rdf, hare -> html

21:34:39 <dajobe> hmm, turtle not tortoise

21:35:05 <bryce> i am referring to the fable.

21:35:35 <bryce> i'm just saying that the connotation is there... slow and steady wins the race.

21:38:17 <bryce> the www began relatively quickly due to html, the semantic web hasn't (yet) been as explosive in adoption. that's what i'm getting at. anyway, once again: cheers to the Turtle.

21:38:32 <dajobe> I see your analogy.

21:39:03 <dajobe> I:it is of course, [http://www.redland.opensource.ac.uk/raptor/|implemented in Raptor] but that was before I renamed it

21:39:03 <dc_rdfig> Added comment I5.

21:54:51 <Davey> Davey is now known as D[a]vey

22:38:47 <sandro> dajobe, Turtle looks good. But why no text/turtle mime type??

22:39:03 * dajobe shrugs, why not

22:39:15 <sandro> semantics. :-)

22:39:40 <sandro> I want my RDF browser to see it's RDF, not have to guess whether it's english or not.

22:41:22 <sandro> As it is, it's annoying getting RDF/XML feeds and half of them come as text/plain or text/xml. A lost cause, ... maybe.

22:41:45 <sandro> But surely mime types are useful.

22:42:05 <mortenf> oh, yeah, that reminds me: could rapper send an Accept header?

22:42:28 <dajobe> maybe, not all the www libraries let you set headers

22:42:34 <mortenf> hmm, bummer

22:42:49 <dajobe> I also have ideas for the heuristics code, guessing content types are needed cos of the mime types lunacy

22:43:03 <dajobe> ... even with TAG saying don't do that ;)

22:43:07 <mortenf> heh

22:43:12 <sandro> Agreed -- but let's not force people to be lunatics!

22:43:17 <dajobe> sandro: I've edited the word for mime type

22:43:34 <dajobe> I need to revisit that, I recall something about conflicting encodings elsewhere

22:45:03 <sandro> Also, do you know about @keywords recently added t n3? It lets you drop the colon on terms in the default namespace, so you can really have "sam foaf:knows fred."

22:45:33 <sandro> example: http://www.w3.org/2000/10/swap/test/syntax/keywords1.n3

22:46:42 <dajobe> timbl mentioned it

22:46:49 <dajobe> I didn't like it

22:48:03 <dajobe> what's python for joining a list of terms with a space to give a single string?

22:48:58 <dajobe> ok, string.join

23:04:21 <sandro> you like all those gawdawful colons? Ugh.

23:04:53 <dajobe> just that bare words clash with 'a'

23:04:56 <dajobe> @keyword :

23:04:57 <dajobe> a a a .

23:18:42 <deusx> deusx is now known as _deusx

23:41:13 <DanCon>http://philebus.tamu.edu/pipermail/scl/2004-January/000670.html

23:41:13 <dc_rdfig> J: http://philebus.tamu.edu/pipermail/scl/2004-January/000670.html from DanCon

23:41:22 <DanCon> J:|iSight and SubEthaEdit

23:41:22 <dc_rdfig> Titled item J.

23:42:33 <DanCon> J:Chris Menzel to the SCL folk about some OS X apple tools... one for [RemotePresence|http://dm93.org/z2001/RemotePresence] except... [http://dm93.org/z2001/ToMacOrNotToMac|ToMacOrNotToMac]

23:42:33 <dc_rdfig> Added comment J1.