IRC log of erswad on 2002-06-26

Timestamps are in UTC.

08:13:54 [RRSAgent]
RRSAgent has joined #erswad
08:14:30 [maxf]
maxf has joined #erswad
08:15:51 [chaalsXXX]
In the room:
08:16:23 [chaalsXXX]
Charles McCathieNevile, Jim Ley, Nick Gibbins, Paul Shabajee, Jill Evans, Liddy Nevile,
08:16:48 [chaalsXXX]
Max Froumentin, Wendy Chisholm, Jonathan Chetwynd, Bert Bos, Dave Pawson, Nick Kew
08:17:26 [chaalsXXX]
Libby Miller, Nadia Heninger, Ian Hickson, Dan Brickley
08:19:10 [Bert-lap]
Bert-lap has joined #erswad
08:20:47 [wendy]
wendy has joined #erswad
08:21:16 [chaalsXXX]
CMN Topics for today:
08:21:41 [chaalsXXX]
What kind of interfaces are there for making annotations - demos please
08:21:56 [Zakim]
Zakim has joined #erswad
08:22:12 [chaalsXXX]
agenda+ What kind of interfaces are there for making annotations - demos please
08:22:27 [chaalsXXX]
agenda+ What uses are we making of annotations?
08:22:58 [chaalsXXX]
agenda+ how is information stored? Does this meet our needs, and can we interoperate between different storage systems?
08:23:59 [chaalsXXX]
agenda+ Can these storage mechanisms cope with other kinds of annotations, and can we interoperate with those.
08:24:16 [nadia]
nadia has joined #erswad
08:24:30 [libby]
libby has joined #erswad
08:25:46 [libby]
annotations creation systems: libby, dan, nadia, jim, niq, dp
08:26:18 [libby]
(for images)
08:27:11 [Hixie]
Hixie has joined #erswad
08:27:14 [danbri]
danbri has joined #erswad
08:27:22 [nmg]
nmg has joined #erswad
08:28:04 [chaalsXXX]
Libby MIller: Codepiction information
08:34:25 [chaalsXXX]
http://swordfish.rdfweb.org/rweb/imgmeta
08:34:48 [chaalsXXX]
adding data for co-depiction - two or more people in a picture.
08:35:33 [chaalsXXX]
We created the tool to search information and then realised that it was difficult to add information to the database.
08:35:43 [chaalsXXX]
So this is a webform that collects information.
08:35:59 [chaalsXXX]
This system doesn't actually store the information - we don't want to own it.
08:36:38 [chaalsXXX]
So we persudae other people to store their information, and submit it to the database.
08:41:48 [chaalsXXX]
[Libby goes through the process of adding a codepiction]
08:42:36 [chaalsXXX]
LM: We copy the RDF and store it somewhere, then give the URI of that somewhere to the database.
08:42:50 [chaalsXXX]
...it's a very simple system that gives an easy way to get the data.
08:43:43 [chaalsXXX]
Questions...
08:43:57 [chaalsXXX]
JC It isn't possible to change the input fields easily is it?
08:44:09 [chaalsXXX]
... I have different requireemnts for things I want to find.
08:44:28 [chaalsXXX]
LM You can store any RDF you want in the database and the query can get anything out.
08:44:39 [chaalsXXX]
JC So I just need to work out how to add something.
08:44:56 [chaalsXXX]
LM The issue is that generating the forms is really hard work - I don't have a schema for it.
08:45:24 [chaalsXXX]
DB When Libby demonstrated you go through forms and get RDF, and then you put the RDF into the Web.
08:45:45 [chaalsXXX]
...ther was one thing that was dealing with the data for ages - now there are two.
08:46:33 [chaalsXXX]
DP What is the objective, other than playing with RDF, to have the data elsewhere than with the image you are describing?
08:47:03 [chaalsXXX]
JL So you can annotate pictures that you don't own.
08:47:29 [chaalsXXX]
CMN The system doesn't say anything about whether it should be in the same place or a different place.
08:49:01 [chaalsXXX]
Dan Brickley: Amaya - demo by screenshot
08:49:20 [chaalsXXX]
Amaya is an authoring tool produced by W3C http://www.w3.org/Amaya
08:49:43 [chaalsXXX]
DB This stuff was a toy to play with things, but it turns out to be useful for playing...
08:50:05 [chaalsXXX]
we wanted to say a bit more about photos than we had - what other things are in the photo.
08:50:29 [Hixie]
DB some people on the web are going to be obsessed with ash trays
08:50:49 [chaalsXXX]
The early stuff, like libby's demo, said "there is a photo and it depicts chaals and dan"
08:51:09 [chaalsXXX]
(it also depicts an ashtray and a table. Some people are interested in that but not everyone)
08:51:31 [chaalsXXX]
This also comes from a project collecting information about pictures from the medical world.
08:51:58 [chaalsXXX]
Creating authoring tools for web content is expensive, but we want to have tools that people can use - even the form thing is slow.
08:52:21 [chaalsXXX]
So what we want is an image overlaid with regions. We had a way of doing that in HTML by using imagemaps.
08:52:42 [chaalsXXX]
They are used in an odd way, but they basically overlay polygons on an image.
08:53:04 [chaalsXXX]
We have revisited this in Amaya, which is an SVG editor.
08:53:35 [chaalsXXX]
We create an image with a photo as a background, and trace out paths over it to identify the things that are in it.
08:53:51 [chaalsXXX]
Amaya let us do this easily, and save the new image to the web.
08:54:31 [danbri]
http://www.w3.org/2001/08/rdfweb/rdfweb-svg-amaya.jpg
08:54:33 [chaalsXXX]
Then we had to try and add the RDF data...
08:54:45 [danbri]
http://www.w3.org/2001/08/rdfweb/foafwho/imagemap/gimp-w34.imagemap.html
08:54:46 [chaalsXXX]
previous URI: screenshot of Amaya
08:55:39 [chaalsXXX]
then an HTML image map that does something similar.
08:56:04 [chaalsXXX]
Created by GIMP (an image tool). Usually the image map is for clicking on to follow as links.
08:56:29 [chaalsXXX]
I ran the HTML from GIMP through HTML tidy - http://www.w3.org/People/Ragget/Tidy
08:56:37 [chaalsXXX]
to clean up the HTML.
08:57:03 [chaalsXXX]
Then Max used XSLT http://www.w3.org/Style/XSL to create a transformation
08:57:06 [danbri]
http://www.w3.org/2001/08/rdfweb/foafwho/imagemap/imagemap2svg.xslt
08:57:29 [chaalsXXX]
that generates SVG - like what we used with Amaya
08:57:57 [chaalsXXX]
DP It was GIMP that let you pick out the outline?
08:58:34 [chaalsXXX]
DB Yes. but there are hundreds of tools for editing HTML imagemaps. So the smart thing was to work out how to use existing tools - take their output, transform it through a couple of tools and end up with something.
08:58:46 [chaalsXXX]
JC Is this available on the Web as a demonstration?
08:59:00 [chaalsXXX]
DB No, it is scattered around...
08:59:09 [chaalsXXX]
... maybe this record helps a bit.
09:01:16 [danbri]
http://www.w3.org/2001/08/rdfweb/foafwho/imagemap/imagemap2svg.xslt
09:01:21 [chaalsXXX]
So we can create data that says "there is a rock rabbit" (by using an RDF dictionary), that is depicted in an image, and there is a person in that image - Damian Steer
09:02:05 [chaalsXXX]
Each of the individuals can be noted by email address - so we can play with other tools.
09:03:08 [chaalsXXX]
Wordnet: a dictionary based on an electronic dictionary that provides class hierarchies of things.
09:03:19 [chaalsXXX]
CMN Is there an interface for dealing with this stuff?
09:03:21 [chaalsXXX]
DB No.
09:03:45 [nadiaa]
nadiaa has joined #erswad
09:04:45 [chaalsXXX]
We can trace the things in the image, but if we had a way to add wordnet definitions easily that would be helpful - we can easily create clip-art...
09:05:06 [chaalsXXX]
we look for a mammal, and we find out that there is a picture of a rock-hyrax, which is a type of mammal...
09:06:06 [danbri]
http://xmlns.com/wordnet/1.6/Rock_hrax
09:06:11 [chaalsXXX]
What is fun about this is that it isn't massive heavily-funded projects - these things work on very simple tools that do just enough to make useful things
09:06:24 [chaalsXXX]
http://xmlns.com/wordnet/1.6/Rock_hyrax
09:06:50 [chaalsXXX]
Jim Ley: More demonstrations
09:07:01 [chaalsXXX]
A tool with no name?
09:07:19 [chaalsXXX]
http://www.jibbering.com/svg/AnnotateImage.html
09:07:34 [chaalsXXX]
Lucky for me Dan and Libby have explained the rationale
09:07:52 [chaalsXXX]
So I saw the demos that we have seen, and worked out how to do those in SVG.
09:08:01 [danbri]
http://www.jibbering.com/svg/AnnotateImage.html
09:08:10 [danbri]
http://jibbering.com/imgs/shepherds.jpg
09:08:11 [chaalsXXX]
DB Amaya is evocative and inspiring, but not a mass-market tool.
09:08:33 [chaalsXXX]
JL I try to do everything so it works in Internet Explorer - which millions of people do use.
09:09:08 [chaalsXXX]
There is a form that asks for some information about an image. The text interface
09:09:17 [chaalsXXX]
is clunky - on the TODO list.
09:09:39 [chaalsXXX]
In theory you can also add RDF in nTriple format (in theory - that bit doesn't work yet)
09:11:07 [chaalsXXX]
CMN So the interface requires explorer plus an SVG plugin
09:11:21 [chaalsXXX]
JL Don't need explorer - can use any browser with the plugin.
09:11:39 [chaalsXXX]
JC Can we have some more documentation in the tool?
09:12:04 [chaalsXXX]
DB In terms of things we have noticed, wordnet is cool but typing URIs is a pain - we need a tool that does it better.
09:13:14 [chaalsXXX]
...Documentation has been weak - a lot of this is done as fast development. It would be nicer to
09:13:22 [chaalsXXX]
...get more of that done.
09:13:38 [chaalsXXX]
(dan finishes tracing out the sheep)
09:14:44 [chaalsXXX]
JL The tool then gives you some options - the SVG with the traced line. You can also add text to the image that labels the sheep and the shepherd (or whatever you labelled the things)
09:15:23 [chaalsXXX]
You can get the RDF that describes the picture. And then I alllow you to store the RDF on my server to re-use it later on.
09:16:44 [chaalsXXX]
So I can search through the annotations and say "is there a thing which has the wordnet label of being a sheep?" and then shows the thing it finds in its database.
09:18:29 [chaalsXXX]
I have also created something that asks for the same information that libby does, but also lets you trace out the person. Again, this is in SVG. When you generate it, I store it locally and submit it to Libby's codepiction system.
09:20:31 [danbri]
http://rdfweb.org/people/danbri/2001/11/rabbit/rock.rdf
09:23:09 [chaalsXXX]
Coffee break
09:25:50 [danbri]
http://jibbering.com/rdfsvg/1025083477768.rdf -- I just annotated the rock hyrax w/ jim's tool
09:26:46 [danbri]
http://jibbering.com/rdfsvg/search.1?noun=Rock_hyrax :)
09:36:44 [chaalsXXX]
Coffee break...
09:39:04 [chaalsXXX]
DB Now you can find a picture of a Rock Hyrax on the Web...
09:39:54 [chaalsXXX]
JL well, it stores an outline, not a region as such. Minor tool bug.
09:40:27 [chaalsXXX]
Nadia Heninger: This is RDFPic - a standalone application in Java that stores its data inside the image.
09:40:58 [chaalsXXX]
This started as a tool for entering other types of information - Dublin Core information, a schema containing technical information about a photo, etc.
09:41:40 [chaalsXXX]
The tool lets you add information about a person in the picture, and make an outline, and stores SVG information
09:42:11 [chaalsXXX]
(not directly as SVG, but as RDF that can be used to generate SVG).
09:42:45 [chaalsXXX]
It can read EXIF information - a binary format that contains information about photos, automatically generated by some digital cameras
09:43:48 [chaalsXXX]
This demo is of a development update - the original version is available, and this update will be made "soon"
09:44:48 [chaalsXXX]
http://jigsaw.w3.org/rdfpic/
09:45:29 [chaalsXXX]
JL We can store the information and submit it - for example to the codepiction database.
09:45:47 [marja]
marja has joined #erswad
09:47:00 [chaalsXXX]
Libby, it's handy having a list of the people available.
09:47:54 [chaalsXXX]
Nadia: Yes, although it would be nice to have a way of identifying them (at the moment there are just boxes and you need to click on wach one to see what it refers to)
09:48:41 [chaalsXXX]
The interesting thing about this is that it uses XMP - Adobe's format based on RDF. They have some tools that can scan that and put information into different formats like PDF...
09:48:53 [chaalsXXX]
LM Can you save the RDF locally? Remotely?
09:49:41 [chaalsXXX]
NH: You can cut and paste to save remotely ;-) It saves the information into the image. If it moves around then you need to deal with the thing that is being described in a bit of editing trickery.
09:50:13 [chaalsXXX]
NH This uses a different model for the information.
09:50:23 [chaalsXXX]
JC You put this on the Web - what can people do with it?
09:50:29 [chaalsXXX]
NH it comes out as RDF.
09:51:09 [chaalsXXX]
JL I have tools that can read that RDF and do some things with it. But it won't consume the data that this tool produces, but that isn't going to be a lot of work to make it happen
09:51:50 [chaalsXXX]
DB It turns out a lot of the photos that have been described are easy to find as images from google - not because it reads RDF but because it reads HTML pages generated.
09:52:01 [chaalsXXX]
Bert Bos: I can show two things that use this information.
09:53:15 [chaalsXXX]
One is a thumbnail tool - it makes a page of thumbnail images that includes the photo, a thumbnail, pointers to the RDF, and you can get to the image in a page that includes other data extracted from the RDF (title, creator, etc).
09:54:59 [chaalsXXX]
This is done with a little bash script...
09:55:06 [danbri]
Oh, wanted to mention Gerald's photo/rdf tools for the record: http://impressive.net/people/gerald/2000/09/photo.html
09:55:44 [chaalsXXX]
http://www.w3.org/People/Bos/thumbnails-rdf - Bert's script
09:56:36 [chaalsXXX]
DP There is a question of whether the information is stored in the image, or the annotation is somewhere else
09:56:58 [chaalsXXX]
BB: Yves Lafon has a different thubnailing tool that uses the same information in a differnt style
09:57:22 [chaalsXXX]
DB wonders if we are getting enough of these tools that we should be concerned about vocabularies being interchangeable
09:57:59 [chaalsXXX]
LM There is an issue here. I was talking to someone who wants to store information inside the images so he can move them around later.
09:58:17 [chaalsXXX]
DP What kind of format should we be using for these annotations to package them up?
09:58:41 [chaalsXXX]
JL There is RDF, which lets us pass it around. XMP is a good way to do it.
09:59:12 [chaalsXXX]
NH There are tools that can extract XMP from different formats. DanBri has made a tool that can extract that information.
09:59:37 [danbri]
http://www.w3.org/2001/sw/Europe/200206/imagemeta/extract/extract
09:59:41 [chaalsXXX]
DB I made a little web service - give it a URI and it extracts data.
10:00:22 [chaalsXXX]
DB The interesting thing about XMPis that it ignores information it doesn't care about - it is very standard in what it passes around and ignores stuff being used by particular communities.
10:01:08 [chaalsXXX]
Adobe have made lots of products based on this. But we can extract that information into RRDF and from there into a Web page.
10:01:29 [libby]
nadia: you don;t have to erase the xif data with rdf data - both can coexist
10:01:38 [libby]
...xmp is not a full rdf inplementation
10:01:42 [libby]
sure
10:02:05 [libby]
....this means that nadia is not corect in addiung teh more complex codepiction rdf in xmp
10:02:24 [libby]
..although it is a really uswefl way of storing arbitrary RDf in documents
10:02:38 [libby]
danbri: mixes up storage and the format [?]
10:03:13 [libby]
...would be good to show them this sort of thing - they're not really expefring this. phpotoshop 7 has the XMP in it
10:03:41 [libby]
ps: for museums community would be great to store the rdf in the pic so it never gets lost
10:03:52 [libby]
dp: standardize?
10:04:06 [libby]
chaals: usecases are great, esp the museums one
10:04:49 [libby]
niq: classroom situation - submitting the images to teacher
10:05:50 [marja]
I think your should store to the image when you can - but if cannot then you can still store elsewhere
10:05:54 [libby]
jc: if you clicked on the images it offers you 'more of the same' e.g. more popgroups with similar attributes in an image-based jukebox
10:06:46 [libby]
danbri: depends on how detiled the metadta is, e.g. colour
10:06:52 [libby]
jc: yes, image based
10:07:29 [libby]
ps: images have stories; what makes an image really great is the story behind it. all this info is text; how do you make that information avilable as non-text?
10:08:02 [libby]
danbri: e.g. mixing with calendar type rdf
10:08:44 [libby]
ps: data protection act?
10:09:21 [libby]
..explains - have to contact the person, check ok or not
10:09:27 [chaalsBRS]
... if I collect data on anybody, I am suposed t register the fact, get your agreement that I can use information about you and tell you why I want it and have permission for each user
10:09:30 [libby]
...uk thing - personal data
10:10:05 [libby]
danbri: subscribe to my yahoogroups list i'm gettign data on you
10:10:12 [libby]
...its going to be a problem
10:10:51 [chaalsBRS]
LM The issue is when somebody makes information available about someone else.
10:11:12 [chaalsBRS]
LN When you aggregate information you are doing something different to information that is already out there, and would bneed permission
10:11:27 [libby]
dnbri: data oriented search
10:11:44 [libby]
...maybe how to draw the distinction. amybe rdf is same as google
10:11:53 [libby]
chaals - google could fall foul too...
10:12:10 [libby]
...answer - we don;t know - build these systems to test it
10:13:49 [chaalsBRS]
Libby Miller: Codepiction
10:14:17 [chaalsBRS]
LM There is a demo I am trying to do where you embed information as XHTML, which can be harvested by a tool to generate RDF.
10:14:51 [chaalsBRS]
http://swordfish.rdfweb.org/discovery/2001/08/codepict/
10:15:14 [chaalsBRS]
You give an email address and it looks for all the pictures that say they depict the person who has that email address
10:16:29 [chaalsBRS]
LM This is a simple RDF database of images, and who is in them.
10:16:40 [chaalsBRS]
The second demo tries to work out the paths between people.
10:18:22 [liddy]
liddy has joined #erswad
10:19:20 [libby]
http://swordfish.rdfweb.org/discovery/2002/02/paths/
10:19:35 [libby]
http://swordfish.rdfweb.org/discovery/2001/08/codepict/
10:20:06 [libby]
LM: the paths things is nice and fast becasue it uwes an RDF query to build a relational database optimised
10:20:31 [libby]
chaals: usecase - only let me see this (rude/strage) photo if you have the right access
10:21:14 [chaalsBRS]
LM This path thing is nice and fast, becuase it doesn't directly use the RDF - it runs its queery and creates a fast XML database for this one question.
10:21:35 [libby]
danbri: normally you expect to ahve control over images, but thses images and their annotatiosn can go wandering round the web - its unexpected and unsettling
10:22:02 [libby]
....fair use requirements for the image - langauge doesn;t exist yet - privacy and rights information
10:22:24 [chaalsBRS]
chaalsBRS has changed the topic to: logs: http://www.w3.org/2002/06/26-erswad-irc info http://www.w3.org/2002/05/er-
10:22:52 [chaalsBRS]
chaalsBRS has changed the topic to: http://www.w3.org/2002/06/26-erswad-irc logs http://www.w3.org/2002/05/er-swade-
10:23:04 [libby]
bb: agenda - a standardriosed machine-readable api for accessing the data
10:23:08 [libby]
?
10:23:08 [chaalsBRS]
chaalsBRS has changed the topic to: http://www.w3.org/2002/06/26-erswad-irc - http://www.w3.org/2002/05/er-swade-f2f
10:24:08 [libby]
bb: a machine can;'t deduce how to fill in a web form
10:24:34 [liddy]
liddy has joined #erswad
10:24:38 [libby]
bb: form exprtessed in RDF?
10:24:55 [wendy]
lm i'd like to do that.
10:25:04 [wendy]
.. say that in a human and/or machine readable way.
10:25:20 [wendy]
.. have the database say "i can tell you all this stuff"
10:25:33 [wendy]
.. not sure how to do machine-readably.
10:25:44 [libby]
...maybe daml?
10:26:05 [libby]
dp: xforms?
10:26:18 [libby]
...could you generate an xform?
10:26:40 [libby]
bb: yes maybe - that's what I was thinking
10:27:04 [libby]
LM: don;t know how to describe the schemna to do this - woudl e very cool
10:27:28 [wendy]
bb web services? perhaps in description lang
10:27:31 [libby]
danbri: we have a webservice...not sure what else you need
10:27:44 [libby]
libby: but you don;t know what's in the database
10:27:50 [libby]
dp: rddl?
10:28:10 [libby]
danbri: not sure if robust enough for very larag ethings like Wordnet
10:28:32 [marja]
generating an Xform from RDF beneficial also to annotations - need application profile and some presentation description
10:29:44 [marja]
would be able to show new properties without programming the interface
10:29:50 [libby]
I thinjk its worth investiagting
10:30:04 [libby]
danbri demos RDFAuthor querying over soap
10:30:23 [libby]
...a particla description of something can be used as a query (visually)
10:30:35 [libby]
s/particla/partial
10:31:02 [libby]
..craete a fragmented description of your area of interest and then say what bits you don;t know
10:31:45 [libby]
..this is basically the usual query structure for RDF
10:33:49 [chaalsBRS]
another demo of using information: http://rdfweb.org/2002/svgsemantics/picsng-demo.html
10:34:56 [libby]
danbri talks about PICS
10:35:05 [libby]
...e.g. as embedded in teh w3c homepage
10:35:31 [wendy]
danbri: "i can't read b/c it's curly brackets instead of pointy brackets" :)
10:35:39 [libby]
:)
10:35:59 [libby]
chaals - ncludes a rating for canadianness
10:36:40 [libby]
...rating true for w3.org/ and anything else that starts with the url - no sex, no violence...
10:37:17 [libby]
...also says who makes the claims
10:37:24 [libby]
...ratings are numbers
10:37:36 [libby]
...gives you a url for the ratings definietion place
10:38:36 [libby]
danbri: pics agressed a difficult social problems: don;t want a single committee saying what is rude - allows different communites
10:38:53 [libby]
...differenbt communites could also make claims according to teh categires
10:39:43 [libby]
...there was a spec for a pics label filter protocol - a simple machine-intereface for porography description " get me the lebel for playboy.com'
10:39:46 [marja]
with Annotea can choose different communities by choosing annotation servers
10:39:52 [libby]
...no business model
10:39:57 [libby]
(foir pics)
10:40:07 [libby]
...this might also be an issue for rdf remote access, servers
10:40:13 [libby]
...pics is dead...
10:40:49 [libby]
...pics descriptions could not be mixed well. with RDF you can mix them better - people, images, cvs etc etc
10:41:02 [libby]
... a big architectural improvement
10:41:50 [libby]
...can say - not just - is payboy rude, but find me all the things with a rude rating between 7 and 9
10:42:15 [libby]
...comes doen to repurposig the data - add the bits that you want to know
10:42:46 [libby]
...pic: http://swordfish.rdfweb.org/photos/2002/05/31/002120.JPG
10:42:59 [libby]
== a guy pulling a truck along with his genitals, in bath
10:43:19 [libby]
...not intrinsically a teaching resource, a piece of art, a piece of bad photography...
10:43:29 [libby]
...but could be any of these things...
10:44:11 [libby]
...can be annotated using Jim's tool: http;//rdfweb.org/2002/svgsemantics/picsdemo/rdf-taboo.rdf
10:44:18 [libby]
annotates the parts of the image
10:44:28 [libby]
...pics architectiure in the RDF world
10:44:55 [libby]
...can be annotated in lots of places. e.g. society for banning rude things could say somewhere that penises are rude
10:45:27 [libby]
...no one langauge for saying this - can have your own langauge
10:47:07 [libby]
...ruby script, which iterates through the contents of the image, for example for translating these categories terms into different languages - images are interestly non-langauge speciaic
10:47:49 [libby]
danbri shows: http://rdfweb.org/2002/svgsemantics/picsdemo/sample_filtered_output.svg
10:48:16 [libby]
...the tool finds the rude things in the database and displays them i9n a different way
10:48:32 [libby]
...but can do this in reverse " show me pictures of all the rude things"
10:49:16 [liddy]
liddy has joined #erswad
10:49:26 [libby]
...this dual use also applies to rights management information - its really useful to have one single name for a movie whether you are a p2p file ripper or you own the rights to the film, to identify it
10:49:53 [libby]
danbri: would like to take lots of clipart, markup in RDf using different tools and see if they interoperate.
10:49:58 [libby]
..some probs aleady:
10:50:03 [libby]
- people's arms
10:50:25 [libby]
- wordnet can be perjorative; there are also much better tools for certain domains
10:50:44 [libby]
...not clear hiow to connec teh generic tools and the vast thesauri
10:51:13 [libby]
ps: can you picck out images annotated by libby alone?
10:51:32 [libby]
...access contraol v important, e.g. just information catalogued by librarians
10:51:46 [libby]
jj: they have already addressed this in annotea - later
10:52:07 [libby]
danbri: we can do this in rdf
10:52:33 [libby]
dp: is it reasonable to assuem you could have many descriptions of the data - e.g. student and expert
10:52:52 [libby]
..if I'm desparate, give me anything; if there are experts, give me those
10:53:27 [libby]
chaals, yes, though expernsive, and depends on the iamge
10:54:13 [libby]
...an ex-art historian, wants to see how my friends describe images, then can get some idea of how they might describe a picture that I don;t know. everyone leaves out a differentb thing
10:55:42 [libby]
jc: has a bunch of images in a page that you can mocve around; the relations between teh images have semantic information. e.g. cat and dustbin, or a cat and a mouse, an immeduate connotation
10:55:56 [libby]
...any thoughts on this - a deep problem...
10:57:24 [libby]
dp: if W3C were to pick up on this, then Hatfield has a particular sklill in describing images for people; knowing why you are describing it and what for is very important
10:58:28 [libby]
jc: a few years ago - lots of auro-erotic asphyxiation - just because you ahve the rope, the organge etc, doesn;t tell you what is happening in teh information
10:59:00 [libby]
danbri: also the human-readbale prose is very important (and expensive)
10:59:08 [libby]
dp: and all the text tends to bve brief
10:59:44 [libby]
dp: my only request is to ensure you can insert paragraphed text
11:00:53 [libby]
[discussion of afternnon's agenda]
11:00:56 [libby]
chaals - annotea
11:01:14 [libby]
...also interested in whether the tools can talk to each other?
11:02:30 [libby]
bb: a tech question - a way of matching different fragments of RDF with different schemas and models, e.g. a photographer is just a string, or a firstname, last name.
11:03:25 [libby]
...I knwo I can match those myself. maybe this is a general operation we can namer and describe. maybe any node in rdf might be able to be replaced by an intermedate b-node which can help you find more matches
11:03:49 [libby]
chaals: also, what use is thiise stuff - playing about or further applicateion, usecases
11:04:17 [libby]
danbri: not just codepiction, but categorisation, parts of the pictures
11:04:22 [marja]
also I guess we need to add different contexts in the annotated resource e.g. not just XPointer but also SVG area or a time for movies
11:04:45 [libby]
chaasl: what do we want to do with image information ion general
11:04:53 [libby]
marja - interesting
11:05:19 [libby]
niq: large archives of satellite images, global positioning
11:05:34 [libby]
dp/niq demo
11:06:13 [libby]
dp: when cited people look at the picture of teh tndall family 9in the room, a painting, they get caertin information
11:07:08 [libby]
....blind people, how do they get the realtion between teh parts, the overall description of the painting - aim is to give the blind person a cmparable amount of information about the painting as signed people
11:07:38 [libby]
...not parts, but navigation within the image, not mouse-driven, a 'navigation map' of the iamge
11:08:29 [libby]
...get differet pieces of information abotu different parts of teh iamge, e.g. 'a 14 year old girl, 90% t th right and 50% up'; can get these faciliteis for description in svg, but not the navigation
11:09:35 [maxf]
maxf has joined #erswad
11:09:44 [libby]
jj: this is the same - if we can describe thatg image in a 'codepcitoopn areas' kind of way, then we can relate the parts of teh image together. it can help you with your application - other peopel can aid your usecase by doing the annotatiions for their own benefit
11:09:57 [marja]
if done right can get at least main parts/subparts from SVG but ui not usually supports this
11:10:41 [libby]
chaals: for info, lynx/linx has this vocabulary - behind, under
11:11:56 [libby]
...can extract information inn the svg for other uses. chaals thinks the soureccode is avilable somewhere. also someone in geneva, phd, can assign sound information and also mouse feedback information to regions of the image
11:12:34 [libby]
...at the moment uses binary content, proprietory
11:13:49 [libby]
danbri: thinks jim is rights in saying that dp's needs and codepiction things is quite similar
11:15:02 [libby]
...would be very cool to be able to grab the information dp siad about the iamges and associate them with a region
11:15:34 [libby]
dp: yes; sme people are specialised at describing things for blind eople; would be good to get this information, if only for the demos
11:15:55 [wendy]
describing images from the National Braille ASsociation. we were able to publish an excerpt from their manual: http://www.w3.org/2000/08/nba-manual/Overview.html
11:15:55 [libby]
danbri: this isn;t the only way to do this, we know this; we would liek help in making it usefgul for accessibility
11:16:32 [libby]
ps: a museum expert would describe the image in yet another way. would be graet to be asble to find the sorts of annotatios that you need
11:16:44 [libby]
dp: - totally differnt usecases
11:16:53 [libby]
danbri: lets's try and do that!
11:17:20 [libby]
chaals - "what questions have been answered about this image?"
11:19:06 [libby]
libby: posible output of this meeting - a multiply described image, uising different tools, for different communites
11:19:43 [libby]
ps: filter project - difefrent descriptions of images in teh filter project (jill evans) - would be even beter if could do this with tools such as jims
11:20:31 [libby]
jill would like to build up a database of how peopel have used these images (e.g. for teching and learning)
11:20:38 [marja]
marja has joined #erswad
11:21:06 [libby]
filter exercise in describing images: http://www.filter.ac.uk/exercise/
11:21:47 [libby]
...rating is also useful
11:22:02 [libby]
we have kit - maybe we could arrange a bridge
11:22:13 [libby]
or if it's just you maybe we could call you
11:22:38 [marja]
thats ok
11:23:02 [marja]
I'll take shower while you eat first :-)
11:23:11 [libby]
ps: show me all the pictures of images of things that fly - doesn;t work in current technology (noone says that birds fly; will usually pick up penguins)
11:23:25 [chaalsBRS]
we are wondering what happened to lunch. It doesn't seem to be arriving.
11:23:29 [libby]
...RDF/ontologies could help with this and make catloguig cheaper
11:23:54 [wendy]
we do have a bridge (just fyi) zakim.
11:24:13 [wendy]
when i requested it for monday, i got it for yest and today as well.
11:24:24 [marja]
when do you start again?
11:24:25 [wendy]
(i now realize i might not have passed that info along...)
11:24:52 [chaalsBRS]
1.45? (time here is now 12.30)
11:24:54 [wendy]
i have several possible use cases...but some are way out there.
11:25:31 [wendy]
same as the other day...
11:25:45 [chaalsBRS]
lunch. Back at 1.45 pm local time
11:25:49 [libby]
---lunch arrives----
11:26:41 [wendy]
it's in the log from monday, i'll have to locate after lunch.
11:54:15 [Zakim]
Zakim has left #erswad
12:30:40 [Hixie]
Hixie has joined #erswad
12:33:07 [marja]
marja has joined #erswad
12:37:11 [Hixie]
Hixie has joined #erswad
12:40:34 [Hixie]
Hixie has joined #erswad
12:45:00 [ericP]
hi all
12:45:34 [marja]
hi ericP - others are probably still eating
12:45:43 [ericP]
roger
12:48:39 [libby]
hia marja, ericp
12:48:48 [ericP]
hi libby
12:50:31 [chaalsBRS]
monday's minutes: http://www.w3.org/2002/06/24-erswad-irc
12:50:52 [chaalsBRS]
Tuesday's log (they aren't really minutes actually): http://www.w3.org/2002/06/25-erswad-irc
12:50:59 [wendy]
wendy has joined #erswad
12:51:03 [chaalsBRS]
Today's: http://www.w3.org/2002/06/26-erswad-irc
12:51:20 [chaalsBRS]
Do people want to dial in?
12:55:27 [danbri]
afternoon:
12:55:43 [danbri]
charles: would like to tlk about annotea
12:55:52 [danbri]
...about whether these tools can work together, whether they do
12:56:09 [danbri]
...of tools here, build a list, ask whether this works with that, and with other tools from elsewhere
12:56:17 [danbri]
....at use cases; things we need but don't have
12:56:37 [danbri]
...other folks views?
12:58:42 [libby]
danbri talks about swad-europe a little
12:58:54 [libby]
...we have some time to carry this on
12:59:04 [libby]
...peiople could join the list to discuss things further
12:59:14 [libby]
...write upcommunity efforts, for example
12:59:27 [libby]
wendy asks about the list
12:59:32 [libby]
...and the members
12:59:40 [marja]
old slides: http://www.w3.org/2001/Talks/www10-Annotea-paper/Overview.html, KCAP: http://www.w3.org/Talks/2001/1021-KCAP/Overview.html, www2002: http://www.w3.org/2002/Talks/0511-annotea/Overview.html
13:00:08 [libby]
...5 partners: ilrt finance, project m,angement; W3C overall direction
13:00:28 [libby]
...hplabs, rutherford-appleton labs, Stlio, a little company in bristol
13:00:43 [libby]
..georgrphocally focused, dan/chaals honourary frenchies
13:01:04 [libby]
...previously e.g. swad-mit, v high-powered programming
13:01:16 [libby]
marja, try the minutes - chaals just posted them
13:01:35 [libby]
...writing up stuff already done
13:02:04 [libby]
chaals - so practically a report - finding out stuff from this meeting about how to do stuff
13:02:16 [libby]
wendy - for this report some discussion and usecases
13:02:44 [libby]
libby would like to do something oractivcal in thos meeting
13:03:10 [libby]
wendy - relation to earl - similarities
13:04:06 [libby]
danbri: we have some visaulisation workp[age. we are going to work in public as ffar as possible - working with communites
13:05:16 [libby]
wendy: what could we practuically do today?
13:05:27 [libby]
libby: annotate 2 pics from difeent points of view...
13:05:51 [libby]
wendy is interested in stepping through like a flowchart in svg - would lie to work on something like that
13:06:09 [libby]
s/lie/like
13:06:54 [libby]
danbri: we have different flavours of collaboration...being on the sane mailing list might be enough...
13:07:12 [libby]
----
13:09:06 [libby]
it worked monday
13:10:29 [libby]
we could call you direct marja
13:10:35 [libby]
want to give us a number?
13:11:30 [libby]
--chaals talking about amaya
13:11:36 [libby]
ok marja, just a sec...
13:12:26 [libby]
marja, can you hear chaals?
13:12:40 [libby]
cool :)
13:12:42 [ericP]
working on slides for O'Reilly Open Source Conv --> http://www.w3.org/2002/Talks/0722-W3COpenSource/
13:13:09 [wendy]
92479
13:13:15 [ericP]
re five digit passcode - i worked for me
13:13:27 [libby]
do we want to try the bridge again?
13:13:35 [libby]
ok
13:13:46 [libby]
what's the zakim number again
13:14:23 [wendy]
zaim +1 617.761.6200
13:14:38 [wendy]
s/zaim/zakim
13:15:34 [libby]
annotea:
13:15:53 [libby]
there's a query protocol, using the RDF query language algae
13:15:53 [Zakim]
Zakim has joined #erswad
13:16:13 [libby]
danbri demos making an annotation in amaya
13:16:28 [libby]
(the phone has not been a popular option :(
13:17:18 [libby]
you can create the annotation of partic;lar parts of an image, and then save it locally or remotely
13:17:20 [wendy]
the phone is next to the projector.
13:17:28 [wendy]
i bet all marja hear's in the projector.
13:17:28 [libby]
ah, good poiny wendy
13:17:54 [chaalsBRS]
zakim, who is here?
13:17:56 [Zakim]
sorry, chaalsBRS, I don't know what conference this is
13:17:57 [Zakim]
On IRC I see Zakim, wendy, Hixie, marja, maxf, nadia, nmg, danbri, libby, Bert-lap, RRSAgent, chaalsBRS, sbp, ericP
13:18:10 [libby]
wendy'd gone to get a little table marja - shoudl be quieter shortly
13:18:19 [chaalsBRS]
zakim, who is here?
13:18:20 [Zakim]
On the phone I see Marja, ??P1
13:18:21 [Zakim]
On IRC I see Zakim, wendy, Hixie, marja, maxf, nadia, nmg, danbri, libby, Bert-lap, RRSAgent, chaalsBRS, sbp, ericP
13:18:30 [chaalsBRS]
??P1 is bristol
13:19:02 [wendy]
how's that marja? moved it further from t he projector, but also further from cmn (unfortunately).
13:19:06 [Zakim]
+EricP
13:19:21 [libby]
...then from a page you can see other annotations, eityhjer of teh whole page or of part fo it yusing xlink
13:19:39 [libby]
chaals: ericP - what kind of queries can we ask annotea?
13:20:10 [libby]
ericP; the query for annotea is a thin veneer over teh rdf query language/database
13:20:47 [libby]
...you write the query in algae. it does graph match but also transitive closures
13:21:22 [libby]
danbri: if we loaded up teh wordnet heirarchy and put in snowleapard and askedd for mammal we would get it?
13:21:31 [libby]
marja, url?
13:21:35 [libby]
ericP thinks so
13:22:08 [libby]
chaals: in teh annotea db, just quering annotations?
13:22:14 [marja]
Algy interface http://www.w3.org/2002/Talks/0511-annotea/slide8-0.html
13:22:29 [libby]
ericP - in one interface can query anything in the database 9algae query)
13:22:58 [libby]
....also specific short queries sepcifically for annotations which expand to algae
13:23:17 [libby]
chaals: can you query 2 annotera databases at once?
13:23:46 [libby]
ericP: yes you can, we use this a lot. you can give it several servers to look in
13:23:55 [marja]
Algae with rules: http://www.w3.org/2002/Talks/0511-annotea/slide19-0.html
13:24:07 [libby]
chaals - what about 2 diffeent sources e.g. annotea and codepiction databaes?
13:24:35 [libby]
ericP has been playing with this
13:24:44 [libby]
chaals: any neat demos?
13:25:15 [libby]
....
13:25:17 [marja]
should document that too as updating the protocol draft right now - almost finished
13:25:43 [libby]
ericP - what about making an RDF payload of the annotation? have you lookmed at that?
13:27:02 [libby]
ericP: we've tried html; jim's rtied svg I think (in the body)
13:27:36 [marja]
also mathml etc.
13:27:37 [libby]
ericP: you can put anything in there, but ...[missed something important!]
13:27:56 [marja]
basically any XML
13:27:56 [libby]
jc: could you have a small image in the payload?
13:28:25 [ericP]
:27 <libby> ... but i would preserve the semantics of the Annotation type and the annotates property
13:28:41 [liddy]
liddy has joined #erswad
13:28:45 [libby]
danbri: bookmarks?
13:28:48 [danbri]
i'm interested in taxonomy/categories for bookmarks via annotea...
13:29:21 [libby]
ericP: soap over algae doing this for the server, but we don;t have a user agent for it.
13:29:52 [marja]
you can use any taxonomy
13:29:54 [libby]
danbri: if the bookmarks server had a taxonomy of some kind, could maybe associate the small image with it, a particular category within some taxonomy
13:30:04 [libby]
no, not really
13:30:25 [libby]
danbri: do you have a way of pluggig oin large taxonomies, e,f=g wordnet
13:31:15 [libby]
erip - you could say was a restuarant, and give it a ruile to use the img since a restuarnt. bookmarks thing has no idea of a rendering agent or anything
13:31:17 [wendy]
wendy has joined #erswad
13:31:37 [libby]
...prob with applying rules like that is that the rules are stored in perpetuity on teh server side
13:31:52 [marja]
we can annotate SVG
13:32:18 [marja]
or the whole document with an image - need URI
13:32:20 [libby]
chaals; can I query for an image uri in teh annotea?
13:32:51 [libby]
...will it return the actual annotationitself? - the annotation that is identified by some uri?
13:32:55 [libby]
ep: yes
13:33:19 [libby]
chaals: e.g. create a codepiction anjmotation, stote in annotea then query it
13:33:19 [libby]
?
13:33:21 [libby]
ep: yes
13:34:02 [marja]
if more properties added to annotation server about annotated image you get those too
13:34:12 [libby]
ca someone else make notes?
13:34:21 [danbri]
ep: SW is a huge graph, but you don't want to slurp the entire thing into your query engine each time
13:34:55 [danbri]
charles: procees I'm interested in: Using Amaya/Annotea with an image. It creates an SVG which uses the image as a background, has paths in it + metadata about region/paths covering who-is-which
13:35:30 [danbri]
...if I build that svg as an annotation, can I get back from the annotea server the URI where it lives, so I can feed it to other RDF indexing tools? (eg. libby's db)
13:35:34 [danbri]
eric: yes
13:36:05 [danbri]
...get the annotation from the main server, use the body data which has the uri of the annotation
13:36:14 [danbri]
jim: the body has the uri in it
13:36:15 [marja]
we have been talking with Ralph to expand the context so that we could have other ways to get the place annotated - not only Xpointer
13:36:22 [danbri]
charles: so I can query for the uri of the body
13:36:38 [danbri]
(marja, that's interesting... relates to EARL discussion mon/tue...)
13:36:51 [marja]
so maybe the SVG should be in the context and not in the body?
13:36:55 [danbri]
jim: If you have the body URI, can you get back to the annotation URI(?!)
13:37:06 [marja]
the outline SVG
13:37:15 [danbri]
eric: you can find out whatever created each annotation or body?
13:37:42 [danbri]
eric: currently if you do a compound query aginst multiple sources, you _should_ get back multiple attributions, but you currently don't
13:38:03 [danbri]
...if you have the (aggregate) graph, you can find out whcih bit came from which sources (servers...)
13:38:24 [danbri]
charles: one other tool we discussed... Favourite Icon (favico)...
13:39:06 [marja]
what is favorite icon ? we do have ways to define new icons for different types
13:39:42 [danbri]
see http://www.peepo.com/
13:39:44 [danbri]
for an example
13:40:34 [marja]
oh you can add an icon to a fragment and not so nuch care about the note about that thing
13:40:56 [marja]
nice idea
13:43:31 [wendy]
http://www.favicon.com/
13:43:58 [libby]
jc: talks about problems of using sites if you can;'t read
13:44:06 [libby]
e.g. finding games
13:45:48 [libby]
there's a sound on mouseover
13:46:00 [libby]
you cvan use the keys
13:47:09 [marja]
I misunderstood favicon - but adding an image annotation to parts of pages and presenting them as embedded on the page might be helpful for users with cognitive problems
13:47:14 [libby]
...oooh a really cool thing!
13:47:33 [libby]
it makes noises and has pictures :)
13:47:44 [chaalsBRS]
danbri plays with a system that lets him make noises and put splashes of colour on the page...
13:48:54 [libby]
http://www.plmd.org
13:49:00 [chaalsBRS]
dan: I am not a web access person, and when I think about accessibility I tend to think about it in terms of blindness. Thanks for reminding us that there are reasons why sounds and fun is important.
13:49:24 [wendy]
important - but also part of accessibility.
13:49:39 [wendy]
accessibility is more than blindness and text, for some images and sounds are what make it accessible.
13:49:50 [wendy]
s/for some/for some people
13:51:20 [chaalsBRS]
CMN can we make favicon work with the tools that we have?
13:51:43 [chaalsBRS]
JL There is a problem of the file format - you have to convert them which loses a lot of the benefit.
13:52:01 [chaalsBRS]
... mozilla doesn't have that problem, because it can use any graphic type.
13:52:29 [chaalsBRS]
BB The CSS3 spec lets you associate an icon with an element...
13:52:32 [danbri]
bert: in css, we're adding an icon
13:52:34 [danbri]
thanks
13:52:38 [chaalsBRS]
CMN can you do that for sounds in the same way?
13:53:11 [chaalsBRS]
BB if you use an aural media type. There is interest in associating sound with visual things - like BGSOUND in some proprietary HTMLs
13:53:20 [chaalsBRS]
...but we don't have it yet.
13:54:32 [chaalsBRS]
DB Going through finding icons for sites is going to be slow. Looking at the games example, I followed my alexa toolbar to find out how the site is categorised in an open directory.
13:54:44 [chaalsBRS]
... then you can go to the site that the directory came from.
13:55:33 [chaalsBRS]
... So it seems that you should be able to drill down and find icons that match the category - it seems at least like a worthwhile student project to provide icons that can be used for the categories. is that plausible?
13:55:42 [chaalsBRS]
JC Makes sense if someone is able to do the work.
13:56:53 [libby]
http://pmld.org/
13:56:57 [libby]
was the blobs thing
13:58:38 [danbri]
urls I just showed: http:/dmoz.org/rdf.html and dmoz categories (plus image/icons per category).
13:58:42 [chaalsBRS]
At the annotea page http://www.w3.org/2001/Annotea/ there is the image http://www.w3.org/2001/Annotea/Images/annotea_main.png
13:58:57 [chaalsBRS]
is there any way of knowing that that icon represents anything?
13:59:42 [marja]
we could annotate it
14:03:44 [Bert-lap]
On W3C CSS WG pages I use REL attribute, e.g., <a rel="in-area">, <a rel="in-activity">, to label the icons
14:04:08 [chaalsBRS]
plan for after the coffee break: List the tools we have looked at and which works with what...
14:04:19 [chaalsBRS]
Use cases - do we have tools that can meet them.
14:05:07 [chaalsBRS]
Break for 15 minutes
14:05:43 [Zakim]
-Marja
14:05:58 [Zakim]
-EricP
14:06:05 [Zakim]
-Bristol
14:06:08 [Zakim]
WAI_(SW)9:00AM has ended
14:15:44 [Bert-lap]
non
14:16:31 [Bert-lap]
s/non//
14:21:30 [chaalsBRS]
we're baack...
14:21:45 [chaalsBRS]
dialling the bridge....
14:22:21 [chaalsBRS]
we don't know the magic dialout code :(
14:22:42 [chaalsBRS]
waiting for libby to return if you folks want to be on the phone.
14:31:14 [chaalsBRS]
Tools: Libby's codepiction form doesn't provide storage
14:31:26 [libby]
we're back on the bridge
14:31:54 [chaalsBRS]
Amaya in principle would allow annotations saved as annotea annotations.
14:32:07 [chaalsBRS]
Jim's SVG tool saves the data locally
14:32:29 [chaalsBRS]
RDFPic saves data into the image.
14:32:40 [chaalsBRS]
DB does it do that as EXIF and XMP?
14:33:03 [chaalsBRS]
NH no. EXIF is just used by digital cameras. I was just in the comment area.
14:33:37 [chaalsBRS]
CMN can a tool that reads EXIF still find the EXIF after you have put it back?
14:34:03 [chaalsBRS]
NH I don't change it. It can still be found as EXIF.
14:34:20 [chaalsBRS]
CMN how do you get at the data?
14:35:13 [chaalsBRS]
NH it can be sucked out of an image.
14:35:23 [danbri]
nadia: Adobe XMP technique is to scan thru looking for 'x packet' tags
14:35:45 [chaalsBRS]
CMN A jigsaw server can get RDF data out of a jpeg and give it to you, no?
14:35:53 [danbri]
nadia: I broke that currentyl while migrating to XMP
14:36:02 [danbri]
charles: are you going to fix it?
14:36:11 [wendy]
want me to take notes for a while?
14:36:12 [danbri]
nadia: ...if directed to the code
14:36:27 [wendy]
lm i can do this too.
14:36:35 [marja]
dynamic icons not published yet but look something like <rdf:Description rdf:about="http://www.w3.org/2000/10/annotationType#Question">
14:36:35 [marja]
<i:usesIcon rdf:resource="$THOTDIR/amaya/question.png"/>
14:36:35 [marja]
</rdf:Description>
14:36:38 [wendy]
jl using adobe's example code,
14:36:44 [wendy]
hard to hear down here!!
14:37:03 [wendy]
cmn bunch of ways to annotate images.
14:37:16 [wendy]
cmn points to marja's scheme
14:37:35 [danbri]
[[
14:37:58 [wendy]
cmn of the tools that query data...codepiction queries what type of database?
14:38:19 [wendy]
db lm's squish (for rdf dbs) i wrapped a layer around annotea so that it understands squish queries.
14:38:24 [wendy]
.. however the data is very diff.
14:38:39 [wendy]
.. not a common interface.
14:39:00 [wendy]
db we can do more thorough testing to make sure our tools speak the same language.
14:39:13 [Hixie]
Zakim: ??PI is Bristol
14:39:13 [wendy]
cmn (to jl) you query your own db?
14:39:18 [danbri]
while($c =~ m/id='W5M0MpCehiHzreSzNTczkc9d'\s*(bytes=')*([^']*)'?\?>(.*)<\?xpack
14:39:19 [danbri]
et end='([^']*)'\?>/sg)
14:39:19 [danbri]
{
14:39:19 [danbri]
$rdf.= $3;
14:39:19 [danbri]
}
14:39:21 [wendy]
jl yes
14:39:29 [wendy]
jl can be fed any kind of rdf.
14:39:32 [danbri]
...is Perl code for extracting XMP
14:39:47 [wendy]
jl in principle. none are robust for large size of data.
14:39:57 [wendy]
jl 500 triples now, it's beg to get slow
14:40:03 [wendy]
cmn scalability problem for other tools?
14:40:12 [wendy]
nh a prob for every sem web tool?
14:40:26 [wendy]
lb there are some large ones, dave becket, and guha
14:40:48 [wendy]
lm it works w/lengthy queries (about 1000 triples). slow with text matches.
14:41:03 [wendy]
db we can build apps over rdf query.
14:41:09 [wendy]
.. it doesn't care how things are actually stored.
14:41:10 [libby]
s/1000/30,000/
14:41:14 [wendy]
.. we can rip out db and put in back end.
14:41:17 [wendy]
thx libby
14:41:25 [wendy]
db depends on the types of queries.
14:41:56 [wendy]
cmn do we have tools that can handle 100K triples
14:41:57 [wendy]
?
14:42:20 [wendy]
db started a project with dave becket. a sense of tools and their maturity. connect stores for use cases.
14:42:32 [wendy]
db substring searching requires brute force.
14:42:45 [wendy]
lb maybe ericp's does more than this?
14:43:05 [wendy]
eric - does your database handle lots of triples and queries?
14:43:26 [danbri]
see http://www.w3.org/2001/sw/Europe/reports/rdf_scalable_storage_report/ for draft towards swad-europe rdf/db report
14:43:39 [danbri]
[[
14:43:40 [danbri]
Providing fair, useful and timely summaries of the characteristics of RDF "database and storage systems" is a subtle problem. We need first to establish a clearer notion of what an RDF storage system amounts to. Are we considering just generic 'triple stores'? Which value adding features should be tested, benchmarked?
14:43:41 [danbri]
]]
14:43:50 [wendy]
ep query-wise - don't know. similar to yours, therefore probably scale like yours.
14:44:05 [wendy]
ep you're using a weak cache to determine # for a string or literal?
14:44:06 [wendy]
lb yep
14:44:08 [wendy]
s/lb/lm
14:44:09 [danbri]
s/cache/hash/
14:44:13 [danbri]
(sha1... fwiw)
14:44:35 [danbri]
eric: I think the queries (in our resp systems) scale the same way. A set of self-joins on a table of all integers.
14:44:47 [wendy]
cmn a database that does acl control.
14:44:53 [ericP]
acl --> http://www.w3.org/2002/05/24-RDF-SQL/
14:45:06 [danbri]
(eric's db also does ACLS == Access Control Lists)
14:45:11 [wendy]
ep that's an rdf interface to a triples database.
14:45:24 [wendy]
ep acls is interface to app-specific relational database.
14:45:37 [wendy]
ep on ly certain questions it can answer for certain properties.
14:45:48 [wendy]
ep a sql query that joins the tables
14:46:15 [wendy]
cmn if have database of triples, so can ask 10 questions...
14:46:21 [danbri]
(ie you're rewriting generic-looking RDF queries into some SQL representation focussed on an app-specific RDBMS schema...?)
14:46:46 [libby]
(which would be rather like the paths db I described earlier)
14:46:47 [wendy]
ep the queries in that paper executes in a millisecond.
14:46:57 [wendy]
ep on the order of several trillion.
14:47:12 [wendy]
ep largest table is ~x,000 (missed x)
14:47:25 [wendy]
cmn can take 1/2M datapoints and assume robust.
14:47:27 [db]
for notes: http://www.w3.org/2000/01/sw/#access points to RDF / SemWeb Access Control work at W3C.
14:47:37 [wendy]
ep ya.
14:47:42 [chaalsBRS]
x=600
14:48:16 [wendy]
ep if have something that includes daml, can have relational db tailored to that data.
14:48:25 [wendy]
ep build rules that are optimized.
14:48:28 [db]
cardinality contraints from DAML+OIL/WebOnt might help us create more optimised SQL representations.
14:48:33 [wendy]
cmn how diff to build the tailored db?
14:49:19 [wendy]
lm nightly requeries database and optimizes in db that is much faster.
14:49:36 [libby]
lM is building a relational database table using an rdf query over an rdf database
14:49:46 [wendy]
ep digital library folks looking at optimized database that does bulk of work. triple-store for incidentals that don't fit.
14:50:09 [wendy]
ep then reoptimize once ...
14:50:31 [wendy]
db dspace might be a swad-e demo possibility.
14:51:01 [wendy]
cmn we don't expect to see this in mid-august?
14:51:03 [wendy]
ep no
14:51:21 [wendy]
ep rdf -> sql, that is written.
14:51:39 [wendy]
ep will include linke to write-up
14:52:10 [wendy]
(:44)
14:52:37 [wendy]
jl some sort of web services
14:52:55 [libby]
(jj is on a different platform)
14:52:56 [ericP]
RDF queries on an application-optimized relational database --> http://www.w3.org/2002/05/24-RDF-SQL/
14:53:04 [wendy]
cmn nadia, your tool read its own data? if load an img, rdfpic read that data, does it read data from other places? pass it data as rdf?
14:53:37 [wendy]
nh there are many options to dump the contents of file into
14:54:13 [ericP]
nadia, project! tell the world!
14:54:27 [libby]
ericP are you happy to publicise that url?
14:54:44 [ericP]
sure
14:54:48 [wendy]
bb jigsaw server doesn't know about rdf, but finds info in jpg.
14:55:22 [wendy]
cmn jill, can you describe what you want to do. did you see anything today that looks useful?
14:55:36 [wendy]
jill bits. like users to link quality of an image.
14:55:49 [wendy]
jill self-sustaining system rather than input from library.
14:55:58 [wendy]
jill rating on visual aspects.
14:56:18 [wendy]
jill pick out regions of images and annotate those. perhaps get students to annotate.
14:56:25 [wendy]
jill microcommunities of users.
14:56:28 [wendy]
(ooh, neat word)
14:56:45 [wendy]
jill registered members or shared with anybody. requires registration and security.
14:56:56 [wendy]
jill comment on how they use the image.
14:57:24 [wendy]
jill in an img of 2 people, some pick out beer or ashtray...what is it they pick out (and why).
14:57:51 [wendy]
db i'm interested in clients for tools.
14:58:00 [wendy]
db jl are you planning on devving yours?
14:58:12 [wendy]
jl the main things are to combine codepiction w/?
14:58:32 [wendy]
jl improve so that it consumes other rdf sources. make authoring easier.
14:58:48 [wendy]
db need to know about instances that could be in things
14:59:02 [wendy]
jl look up in wordnet (for tree)
14:59:48 [wendy]
db what can we do?
14:59:52 [wendy]
jl suggest ui.
15:00:03 [wendy]
jl what would look good? how do you want it to work?
15:00:08 [wendy]
db test is as well?
15:00:09 [wendy]
jl ya
15:00:22 [wendy]
db grab a snapshot?
15:00:26 [wendy]
jl you betcha
15:00:57 [wendy]
cmn ln, any thoughts about how you can use these tools w/all the stuff you're wokring on?
15:01:22 [wendy]
cmn haven't talked about access control.
15:01:46 [wendy]
ln several things: 1. in the collection of cave paintings, vary in age. interesting to see how image changed over time.
15:02:13 [wendy]
ln rights mngmt interesting b/c as i understand it in aboriginal culture, it's not a matter of stating rights at beginning
15:02:40 [wendy]
.. distinguishing between sacred paintings....a case of user saying "x died, therefore all of these categories id on't wwant to have in my world for 6 weeks)
15:02:46 [wendy]
.. world changes as people die
15:03:20 [wendy]
ln when we show them a catalog of a paitning, depending on who looks at it, will describe it differently.
15:03:37 [wendy]
ln mtn folks talk very differently from river folks.
15:03:43 [wendy]
ln those need to have = status.
15:03:52 [wendy]
ln need to know who said it (will give you context)
15:03:56 [wendy]
db capture spoken word?
15:04:07 [wendy]
ln yes, they are illiterate. there is not a written ver of their lang.
15:04:14 [wendy]
ln some are 80 yrs old and only speakers of the lang.
15:04:18 [wendy]
ln good to capture that.
15:04:28 [wendy]
ln would we ever translate that into text or leave it as spoken word?
15:04:38 [wendy]
ln hearing it is the only thing that will make sense to them.
15:05:03 [wendy]
db sound clips + tech meta data...
15:05:12 [wendy]
ln might be video clip. gesture+sound.
15:05:19 [wendy]
ln just sound won't tell whole story.
15:05:57 [wendy]
db vocabularies and schemas? we tend to use combo of DC, goofy people stuff...
15:06:12 [wendy]
db +RDFPIC
15:06:22 [wendy]
nh i use depics and regiondepics
15:06:55 [wendy]
db rdfpic used to use dc, ... should we get together on mailing list, announce who doing what,
15:07:05 [wendy]
jl isn't annotea using other namespaces?
15:07:15 [wendy]
jl b/c of its age, using older namespaces.
15:07:43 [db]
two version of dc namespace: http://purl.org/dc/elements/1.0/ or /1.1/
15:07:45 [wendy]
jl sir name is sir name or full name. don't know if get sir or full.
15:07:50 [db]
...capitals vs lowercase
15:07:51 [wendy]
jl using dc 1
15:07:54 [db]
annotea: using dc1
15:08:13 [db]
nick: is there a list of what annotea's using?
15:08:20 [db]
jim: only by dumping the rdf
15:08:23 [wendy]
db - gonna minute for a while?
15:08:34 [wendy]
i'm happy to take a break. :)
15:08:39 [ericP]
nick, using for what?
15:08:57 [wendy]
db a w3c note?
15:09:00 [ericP]
i could minute
15:09:04 [wendy]
lol
15:09:07 [ericP]
phone: bip bp
15:09:22 [ericP]
phone: titter - click pop
15:09:25 [wendy]
asdfkljaseroipuab;ljk asdfasndfaweoiuraslkjdf asfjlweofu dfalhsdr
15:09:40 [ericP]
that;s my name, don't wear it out
15:09:48 [wendy]
:)
15:10:14 [wendy]
cmn selection of stuff looked at - some of tools that r aroond
15:10:26 [wendy]
.. stacks of annotation toools
15:10:35 [wendy]
.. rate folks photos
15:10:59 [wendy]
phone giving cmn a rating...
15:11:02 [db]
http://amigeekornot.com/?id=71589
15:11:07 [libby]
amihotornot.com
15:12:16 [ericP]
plus the t-inator and the pornolizer
15:12:17 [wendy]
cmn do they create data we can use? is the s/w avail?
15:13:18 [wendy]
action: db summarize in an email...something very special
15:13:53 [wendy]
db in hawaii adobe engs interested. i sent themnadia's examples. there will be a write-up.
15:14:01 [wendy]
action nh review db's write-up.
15:14:15 [wendy]
action all: read the xmp spec
15:15:05 [nadia]
on xmp... you have to register to get access to the xmp materials on adobe's site
15:15:24 [nadia]
the spec itself is quite long
15:15:53 [db]
see http://www.jibbering.com/svg/ under 'whiteboard'
15:15:57 [Zakim]
-EricP
15:15:58 [nadia]
there's a shorter document "embedding xmp" that is more practical, but only useful if you're familiar with the file formats in question...
15:16:00 [db]
jim: idea... you have a board with the background
15:16:13 [ericP]
oops
15:16:23 [ericP]
dropped the phone (literally)
15:16:32 [db]
charles: lots of univesrities want to do...
15:16:36 [db]
(i) shared drawing spaces
15:16:38 [ericP]
i wasn't hearing much anywyas
15:16:41 [db]
(ii) accessibility
15:17:04 [db]
lots of whiteboard tools don't capture the mening/semantics/stuff 'behind the scribbles'
15:17:17 [db]
q to jim: how much work technically to turn this kind of tool to that end?
15:17:26 [db]
jim: not much programming work, lots of usability, ui work
15:17:41 [db]
...i'd be interested. I'm not a UI person...
15:18:12 [db]
...suggestions?
15:19:40 [marja]
need to know what kinds of semantics to use - maybe different for different drawing categories or user groups
15:21:00 [db]
(discussion of military apps; daml etc)
15:22:11 [db]
charles: how many folk here now see themselves as users / potential uses f these tools...?
15:22:13 [libby]
chaals: how many people see themseleves as users of the tool and how many developers?
15:22:17 [db]
...and developers?
15:22:23 [db]
about 1/2-2/3
15:22:30 [db]
ok, thanks for coming marja, eric
15:22:31 [Zakim]
-Marja
15:23:29 [danbri]
danbri: I'm not a UI person but have thru using Jim's tools, various bits of feedback
15:23:46 [danbri]
niq: as a developer, hard to get quality feedback, easy to feel like working in vacuum
15:24:14 [danbri]
charles: people often acquire the software, use it but don't take time (or see utility) to send feedback
15:24:17 [danbri]
...takes time
15:25:58 [marja]
I can try some tools from user perspective if needed
15:32:05 [libby]
http://www.isketch.net/isketch.html
15:32:46 [danbri]
discussion of mailing lists
15:33:02 [danbri]
danbri: if I set up a specific list for tasks/actions outcoming from this meeting, who would join
15:33:17 [danbri]
(some reasonable amount of nodding, but not huge enthusiasm for Yet More EMail)
15:33:25 [danbri]
discussion of www-annotation list
15:33:35 [danbri]
niq: has flagged up prblem areas, eg xpointer stuff
15:33:44 [danbri]
jim: www-annotation interesting list
15:33:54 [danbri]
charles: should we be trying to find a mailing list
15:35:23 [danbri]
danbri: where to take, for example, the namespace divergence problem amongst our tools...
15:35:27 [marja]
for annotations why not use www-annotation
15:35:38 [danbri]
charles: diff people will take things in different directions
15:36:20 [danbri]
charles: some people likely to go away wondering 'which namespace to use x'... others looking at other things, eg scalability
15:37:21 [marja]
trying to write instructions for Annotea extensions - probably need feedback at some point
15:37:59 [danbri]
danbri: things will inevitably scatter to the winds, eg. Scalability work will be addressed in swad-e, more annotational stuff in www-annotation... in swad-e we'll try to keep pointers to latest work fresh on our website
15:38:27 [danbri]
charles: I'm looking fwd to the report, write up of today...
15:38:39 [danbri]
jill: what's the best way to keep up to date with what various folk in this room are doing?
15:38:43 [danbri]
...how we'll coordinate their efforts
15:39:39 [danbri]
...how to take my interest in these tools fwd.
15:39:43 [libby]
some ideas for annotatiuons from different points of view: http://swordfish.rdfweb.org/people/libby/rdfweb/2002/06/annotations/
15:40:28 [danbri]
danbri: keep an eye on swad-europe web pages ( http://www.w3.org/2001/sw/Europe/) which should be a good point of call. Also mailing list (public-esw@w3.org, archives will be public) for swad-e.
15:40:36 [danbri]
charles: and www-annotation is a good list
15:41:46 [danbri]
(some nodding re use of www-annotation)
15:42:15 [danbri]
charles: also the RDF Interest Group. Has mailing list(s) and an IRC chat channel, see http://www.w3.org/RDF/Interest/#irc
15:43:41 [danbri]
see http://rdfig.xmlhack.com/index.html for weblog from irc chat
15:44:10 [chaalsBRS]
irc.w3.org/#er chat about tools in accessibilty but a lot of stuff about annotations
15:44:16 [danbri]
danbri: also irc.w3.org 6665 #er
15:44:17 [danbri]
yup
15:44:23 [chaalsBRS]
there are other fora...
15:45:26 [danbri]
The participants record their thanks to Libby.
15:45:28 [danbri]
we're not worthy!
15:45:33 [danbri]
ADJOURNED.
15:45:33 [nmg]
nmg has left #erswad
15:45:35 [danbri]
------
15:45:41 [danbri]
RRSAgent, pointer?
15:45:42 [RRSAgent]
See http://www.w3.org/2002/06/26-erswad-irc#T15-45-41
15:45:47 [chaalsBRS]
Thanks to all....
15:46:01 [wendy]
wendy has left #erswad
15:50:53 [marja]
marja has left #erswad
16:03:43 [Zakim]
-Bristol
16:03:45 [Zakim]
WAI_(SW)9:00AM has ended
17:52:43 [Hixie]
Hixie has joined #erswad
18:00:58 [nadia]
nadia has joined #erswad