- From: Nick Gibbins <nmg@ecs.soton.ac.uk>
- Date: Sat, 22 May 2004 18:08:16 +0100
- To: semantic-photolist@unitboy.com
- Cc: patrizia.andronico@iit.cnr.it, p.barna@tue.nl, himself@brycebenton.us, karl@w3.org, timothy@immuexa.com, nmg@ecs.soton.ac.uk, handschuh@aifb.uni-karlsruhe.de, swh@ecs.soton.ac.uk, hendler@cs.umd.edu, rlee@apotex.com, libby.miller@bristol.ac.uk
Photography markup Developers Day session at WWW2004
NYC, 2004-05-22
Chair: Greg Elin
Participants:
Patrizia Andronico <patrizia.andronico@iit.cnr.it>
Peter Barna <p.barna@tue.nl>
Bryce Benton <himself@brycebenton.us>
Karl Dubost <karl@w3.org>
Timothy Falconer <timothy@immuexa.com>
Nick Gibbins <nmg@ecs.soton.ac.uk>
Siegfried Handschuh <handschuh@aifb.uni-karlsruhe.de>
Steve Harris <swh@ecs.soton.ac.uk>
Jim Hendler <hendler@cs.umd.edu>
Richard Lee <rlee@apotex.com>
Libby Miller <libby.miller@bristol.ac.uk>
Mor Naarman
-----
Introduction
Greg Elin
- digital photography has outgrown conventional photgraphy
- pixel manipulation mature compared to 'story manipulation'
- non-pictorial information space
- infrastructure mature (image formats, wireless networking, affordable digital
cameras, blogs, etc)
- camera phone as key platform
Some examples:
- w3photo.org
- semantic web technology-based
- harlem.org
- archive of portrais of jazz musicians
- uses photo annotations as tour of information space
-----
Image Annotation as a Semantic Web Application
Libby Miller <libby.miller@bristol.ac.uk>
Where it started...
- depiction as specific type of image annotation
- grown from FOAF work
- person or thing is depicted in a photo - creates links between thingsa which
are depicted by the same images
- (shows codepiction path application - URL?)
- not just depiction, also identification
- SVG outlines to identify image regions
- depiction of things other than people (WOrdNet terms)
- SWAD-E workshop in june 2002 (URL?)
create a new vocabulary for image description
- draws on existing work
- mixes existing vocabularies
- community effort
RDF vocabularies used:
- Creative Commons
- image and metadata licensing
- licensing critical to reuse
- some modification of existing vocabulary (date of birth not used)
- Dublin Core
- bibliographic metadata
- textual description of image (dc:description)
- Annotea
- annotation vocabulary
- linking RDF annotation to the annotated image, some biblio metadata
- Image regions
- rectilinear regions, region depiction
- Conference ontology
- conference as epicted event, links with iCal RDF vocabulary
- FOAF
- identifying people throughout
Important/tricky areas
- identification of people and events (which don't have URIs) by key properties
- some others disagree ;)
- expressing location is difficult, was left out (implicit location at conference)
- annotation properties - merging graphs loses information (origins of annotations)
- reuse vs. creation of vocabularies. OWL capabilities (class/property equivalence)
difficult, most developers are only using RDF(S) tools
- validation
- required fields for descriptions
- must have identifier for people
- legal issues
- some countries (France, eg) person owns photographs of them
Demos:
- masahido's annotator - http://www.kanzaki.com/docs/sw/img-annotator.html
- libby's annotator -
http://swordfish.rdfweb.org/discovery/2004/03/w3photo/annotate.html
- lookups from text labels to resources
-----
AKTive Photo
Steve Harris <swh@ecs.soton.ac.uk>
- explorer interface to phot0 annotations
- photo.aktivespace.org
- AKTive Spaces, from mc schraefel's mSpace work
- previous Semantic Web application CS AKTive Space - cs.aktivespace.org
- AKTive Photo code is available under GPL
-----
Mor Naarman
- twofold value of photo annotation - personal vs. social
- social benefit through implicit sharing of information
- geographical location used to share labels between annotations
- different photos of same location may have similar labels
- ranking for suitable labels
- stanford campus experiment to gathr annotations (cameras and GPSs to visitors)
- general terms more common, 'float up'
- IDF for ranking terms
-----
Further demonstrations and comments
Greg Elin
Ron and Taylor's Road Trip - rzweb.com/roadtrip/
- GPS-located images, along with TerraServer satellite imagery tiles
Yafro - yafro.com
- MoBlog site
- example of threaded discussion attached to image of armed iraqis, taken by a
soldier
Flickr - flickr.com
- flash application, simple photo sharing, connected chat
The ESP Game - espgame.org
- competitive photo labelling, scores for getting th same labels as other people
--
Critical concepts for developers (greg's slides - didn't manage to capture all
of this)
why rectangular image regions (as opposed to SVG outlines)?
- similarity with existing (rectangular) images
Legal issues
(scribe had difficulty getting all of this down)
Greg Elin: is permission required for publishing annotated images?
Jim Hendler: approval and notification is a barrier to scalability
Karl Dubost: cultural issues - assumption that permission is not required
is not universal
Jim Hendler: default should be opt-in (on notification), not opt-out
(or, opt-out on automatic permission)
Tim Falconer: privacy issues involved
Jim Hendler: predominant view (from survey?) is that most people are
happy for photos to appear, but want the ability to remove those photos
that they don't want
-----
Tidepool
Tim Falconer
- image annotation for people that don't use folders in their email
- no difficult semantic concepts - who/what/where/when
- drag and drop of tags (annotations) from these four categories onto images
- help users to handle complexity of tags (large numbers)
- graph visualisation of tags?
- (JH: Ben Schneiderman says his studies show that people don't like
graph interfaces for large datasets)
-----
Jim Hendler
- presentation of SW information is as important as annotation
- generic SW portals (mindswap.org)
- domain-specific portals swint.mindswap.org
-----
Summary of participants' interests
- most participants were interested in photo annotation because
a) it was a nifty SW application and b) it addressed a need
- some extra points raised
Steve Harris: two main barriers to entry: rdf model, and rdf syntax
- need is for task-specific APIsfor generating rdf/xml
Karl (dubost?): rdf should be invisible to users and developers
Nick Gibbins: image annotations as requirement for ontology-based open
hypermedia - navigable links between image regions based on SW relations
and depicted things
Libby Miller: real value from SVG1.2 (Dean jackson and Jim Ley)
Bryce Benton: spectrum of access rights, from public to private
batch processing would also be a benefit
Mor Naarman: audio annotations (audio labels), storytelling (HP), geo/time
reference annotations
-----
The next step?
- SWAD-E workshop (contact Libby
==================================
This is the TEMPORARY discussion list for the W3 Semantic-Photo History
Project. For questions, contact greg@fotonotes.net.
Subscribe Instructions
To: semantic-photolist-request@unitboy.com
Body: subscribe
Unsubscribe Instructions
To: semantic-photolist-request@unitboy.com
Body: unsubscribe
Help
To: semantic-photolist-request@unitboy.com
Body: help
Received on Saturday, 22 May 2004 13:09:08 UTC