Semantic Web Interest Group IRC Scratchpad

Welcome to the Semantic Web Interest Group scratchpad generated automatically from discussions on IRC at irc.freenode.net port 6667 channel #swig by the chump bot, instructions in the chump user manual. Please use UTF-8 charset on IRC and pastebin for code or data more than 10 lines long.

Nearby: IRC logs (Latest) | semantic-web list | W3C Wiki (Recent changes) | delicious swigbot

last updated at 2004-06-07 21:37
sbp: Source. Powered by garner.py. Sent a message about this to pubrixt, but it didn't seem to get through; probably need to be subscribed.
 
JibberJim: from LauraH
 
Namespace list from whiteboard
danbriMAD: XMP, Wordnet-as-thesaurus-ish-thing, wordnet-nouns-as-classes, PRISM, DC+DCterms, FOAF,RSS1, CreativeCommons, GeoLatLongWGS84, RegionsSVGstuff, ical, SUMO, ICONCLASS, TGN, MattB'sMainPicProperty ...
 
Idea: could someone with an RDF harvester run queries to find all namespaces used against, say, JPGs, GIFs, PNGs, SVGs...?
libby: sounds plausible with a bit of fiddling
mattb: i did a quick and dirty grep for 'jpg' against an old scutter run from last year and got this list of predicates (jpg may have been subject or object)
 
FAQs from the Image annotation meeting in Madrid
libby: should we assume that tools are getting better and so be happy with using RDFS and OWL to map vocabularies rather than reusing terms form other vocabs directly?
libby: should we map vocabulary terms to each other even if they don't map exactly, or pretend that they don't have a relationship
libby: how do we express geographical location of an image? - where it was taken, where it is a picture of
libby: is it better to write a new vocab or reuse ones?
libby: how do you handle stablity of schema, terms etc adequately?
libby: how do we know what vocabularies are out there? and which are stable, testing etc?
 
 
Why are we making stuff up?
danbriMAD: Notes: We're early adopters. We're adding detail to other vocabs. We're prototypign first and merging things that we know work later. We are experimenting with lightweight processes for semi-standards. We're porting pre-RDF systems to the Web and exploring different designs (eg. EXIF).
 
Vocabularies we are using
chaalsMAD: A rough list... feel free to add to it
chaalsMAD: Dublin Core
libby: w3photo vocabs
libby: some more vocabs in use
chaalsMAD: Morten is using Matt Biddulphs property for highlighting a picture out of a collection
chaalsMAD: I have used the annotea schema for talking about things I am annotating
chaalsMAD: XMP, Creative Commons, Prism
libby: Masahide kanzaki's EXIF schema
libby: SUMO ontology
libby: AAT - Art and Architecture Thesaurus
libby: Getty union list of artists names
libby: TAP - I want to use this as identifiers for dead people especially e.g. foaf:tapID
libby: <danbriMAD>Hmm.. Wordnet (various flavours), EXIF, DC, RSS, Wordnet, Geo/wgs84, path outlines
mof-alice: Photography - a vocabulary for photography and equipment
mortenfMAD: Also: OpenCyc
 
 
Maps of where people are use case
danbriMAD: From charles. Email from travel agent -> map. [per danc tools etc., @@url?]
danbriMAD: Looking at the map, is a prompt for thoughts such as 'someone close by I could go see', esp w/ overlays of other data.
logger: See discussion
 
danbriMAD: Relates to discussion of IFLA Work/Expression/Manifestation/Item in RDF (in this meeting and on www-rdf-interest lately).
 
danbriMAD: ImageDescription notes in the ESW wiki
 
Huridocs use case
danbriMAD: I went to meet huridocs recently. they document human rights abuses. go to trouble spots, teach people how to record the fact that somebody did something bad... so that in n years time, UN can set up a tribunal and the report be acted upon.
danbriMAD: ie. create evidence for future courts to consider. They have forms, multilingual... english/french/spanish versions currently.
danbriMAD: ...if u want to document a set of events, potentially two things. Photo snapshots w/ phone etc. Also saying what is happening. "This photo allegedly records some kind of event", and have onto/thes/vocab for event taxonomy, agents involved in events, etc.
danbriMAD: So one part of the use case... 'unauthorised shooting by member govt / paramilitary / ...'
danbriMAD: Looking through cartoon-style representations of event types can help people browse (arguably). Similar to wwac concerns.
danbriMAD: Using images to drive an image annotation system.
danbriMAD: Danbri: how deep/detail can you go? I imagine icons for 'killing' but not 'unauthorized killing by miltary police' so easy...
danbriMAD: Red Bus / Green Bus problem (per J Chatwin); people may take icons too literally. Cultural variation.
danbriMAD: Need to record who claimed what, can remine data figure out reliability etc.
danbriMAD: So links to access control, security, ...
danbriMAD: If the wrong people find out you're recording human rights violations, that's putting lives at risk.
danbriMAD: Using pictures to identify people is interesting, potentially quite tricky, ... photo might not be who you claim/believe it to be.
logger: See discussion
danbriMAD: See also Witness.org project, "Using video and technology to fight for human rights".
 
chaalsMAD: The idea is to collect concepts, and then associate labels with them (typically images) for people to use in communication.
chaalsMAD: danbri: Not just nouns, but entire language...
chaalsMAD: Andy - there are a couple of large systems in use for people who cannot talk and are using these symbols to communicate. They are proprietary, so there are problems if they want to send an email to someone who doesn't own the same symbol set.
chaalsMAD: A mapping system so that messages in one symbol set can be mapped to another.
chaalsMAD: Bengt - we had a real scenario where two people who were sharing a room couldn't communicate because they had bought different symbol systems.
chaalsMAD: Andy demos a much cooler intereactive version of chaals little SVG demo
 
Javier Candeira
chaalsMAD: I was on an IRC channel and they said there was a meeting I should attend. So here I am.
chaalsMAD: Help sustain barrapunto (like slashdot), and am a freelance journalist
 
 
chaalsMAD: See also another link
chaalsMAD: This stuff gets used in Jim's tool (previous chummp)
 
chaalsMAD: You get the person who annotated, what is being annotated, and some descriptions of image parts (including some stuff about where in the image we are talking about)
libby: annotated with Jim's SVG annotator
 
 
libby: including the picture described by dave Pawson
 
TOdo: dig out SMIL/audio anno from 1st w/shop
danbriMAD: Also EXIF schema on w3c (from masaka & karl)
 
danbri use case notes. Express access control policy of images (eg. baby pics); mix image with 'non-img' data, so can find images by matching data from elsewhere.
 
chaalsMAD: Allows for various description things on images - including saying who, what where in RDF as well as plain text descriptions and licensing/rights information
chaalsMAD: Because different people can put information into the same database you get a potentially multidimensional database of stuff.
chaalsMAD: Where people are moblogging (putting lots of pictures up from their cellphone) it is nice to be able to search things that are happening in a particular place, or at a particular time in different places, or involving different kinds of objects.
chaalsMAD: See also kanzaki demo of similar approach
chaalsMAD: another thing is the W3Photo where they were more interested in the story behind a photo
libby: lots of people are going to be moblogging soon
libby: easy (but different) annotation sytems producing similar data strutures
chaalsMAD: We tried to use existing vocabularies wherever possible, rather than invent new ones. That turned out to be hard. For example with ccreative commons they use birthdate as an identifier, which is more personal and harder to get than a hash of their email address, which FOAF uses.
 
 
Morten Frederiksen
chaalsMAD: Interested in photography,. wnated to show off my pictures, and find the ones I am intersted in
libby: use with gps
 
Eva Mendez
chaalsMAD: Work at Carlos III University, Madrid
chaalsMAD: Work on Metadata, Dublin Core (maintain SPanish Dublin Core mirror), translate RDF specs into spanish
 
Jim Ley
chaalsMAD: Don't work... for anyone particular. Have written some image annotation tools. Most of what I want from this meeting is to get motivation to develop a tool that I can demonstrate at the SVG Open conference
 
Laura Hollink
libby: things can't do with automatic image analysis
libby: ways of searching
libby: sorry - didn;t catch everything
libby: Laura's homepage
 
Andy Judson
chaalsMAD: University of Dundee, work on the world wide augmentative & alternative communication project (wwaac) - particulary the concept coding approach to symbol language translation using rdf ... see http://dewey.computing.dundee.ac.uk/ccf
 
Use cases...
chaalsMAD: Describing things that are in images, so I can re-use them in authoring scenarios
chaalsMAD: Libby - annotating photos so I can find them, search them
danbriMAD: motivated to make simple practical authoring/annotation tools
chaalsMAD: demo
chaalsMAD: Re-uses data from an existing database.
chaalsMAD: Important to be able to use info from different sources, and avoid making people continually write the same stuff.
chaalsMAD: 3. Multiple descriptions of artworks or objects, to get a general perspective rather than "an authorative description"
chaalsMAD: Means that you need to be able to work out who said something, as well as what they are describing. This is different from the standard library idea of having one person describe things
chaalsMAD: 4. describing a picture from the Bristol workshop, which has a picture in it - what exactly are we describing?
chaalsMAD: Dave Pawson described in audio the picture. (He's a professional at describing things for blind people). I turned this into SMIL, so that we could identify bits of what Dave was saying
chaalsMAD: ... so we could search for things in the audio as well
chaalsMAD: 5. annotate multiple pictures at once (since I tend to take a lot at a time)
chaalsMAD: Would like to hook up nice interfaces to this
chaalsMAD: Laura - annotate spatial relations between parts of images
chaalsMAD: Bengt and Andy - Concept coding framework
chaalsMAD: Patrick Roth's work on presenting images via sounds and force-feedback
chaalsMAD: FoafMaps - is there an interesting person in some place?
 
Bengt Farre
libby: working with World Wide Augmentative & Alternative Communication project (wwaac) - mostly the concept coding and making a reference implementation and implementing that into UA and AT programs.
chaalsMAD: working with World Wide Augmentative & Alternative Communication project (wwaac) - mostly the concept coding and making a reference implementation and implementing that into UA and AT programs
 
Alan Chuter - Teleseervicios (owned by Fundación ONCE)
chaalsMAD: We work on accessibility, and are owned by a group working with visual disabilities. That's my interest
 
Dan Brickley
chaalsMAD: Used to work for ILRT, now works for W3C. Worked on image annotation projects
chaalsMAD: People were always talking about images, but it turned out what they meant was information that the images showed.
libby: very hard to scope
chaalsMAD: Lots of intertesting uses for image metadata. Nice fit for RDF work that I was getting invlved in. Nice accessibility applications, image searching, etc.
chaalsMAD: Unlike text searching, with images there is a big step forwards in having image metadata because you can't do free text searching
 
Libby Miller
danbriMAD: Work on SWAD-Europe at ILRT, Uni bristol. Like image apps a lot. Recently worked on www2004 w3photo project.
danbriMAD: Focussed recently on describing image regions/areas, contrib'd a schema to w3c site. Targetted 'ordinary photo taking users'. Also worked on FOAF.
 
Charles McCathieNevile
chaalsMAD: Chaals - intro
danbriMAD: Work at W3C. Also with Sidar, a Spanish language group for accessibility ie for people with disabilities.
danbriMAD: Interested in this for several reasons. That RDF makes certain hard things easier. The accessibility issue that motivates me: ppl who can't work easily with images getting text represntations...
danbriMAD: So involved w/ various projects ...] eg. [with dan that Jim rewrote much nicer.
chaalsMAD: I am particularly interested in making authoring possible for people who don't write code
 
Goals of the workshop
chaalsMAD: 1. collect some information about what projects people are acutally doing, what is already done, and what would be good to do.
 
JibberJim: Jim Ley
danbriMAD: Dan Brickley
libby: libby miller
chaalsMAD: chaals - Charles McCathieNevile
chaalsMAD: Bengt Farre, Andy Judson, Laura Hollink, Morten Frederiksen, Javier Candeira, Eva Méndez, Alan Chuter
 
Created by the Daily Chump bot. Hosted by PlanetRDF.