<raphael> Scribe: Raphael
Problem: it is difficult for users and applications to identify, manage and apply algorithms for image analysis
1/ low level: data formats and algorithms, their transition and selection
2/ high level: defining domain semantics, learning from similar problems
Solution: an algorithm ontology
The algorithm ontology should consist of information such as: name, informal natural language description, formal description, input format, output format, goal of the algorithm, example media prior and after the application, etc.
Challenges: to articulate and qualify the visual result of applying algorithms, to integrate and harmonize the ontologies, to reason with and apply the knowledge in the algorithm ontology
Technology: potential overlap with web service choregraphy domain
Problem: see wiki page (pneumo thorax)
Suzanne describe the outline of a possible solution
The use case focus on high-level interoperability
The driving question: how to describe media analysis algorithms ?
Michael: the problem reminds me
the pre-condition / post-condition parameters in the software
enginnering problems ?
... why not specifying the possible and impossible chaining of various algorithms ?
Suzanne: yes, exactly, this is where we would like to go
Susanne: I see links with my problems
Suzanne: it is about the
description of a workflow at a very abstract level
... we can work in the medical domain, and/or in the manufacturer workflow
Oscar: what is the state of the art ?
Suzanne mention some work I didn't get ...
<ocelma> some people from Russia...
Raphael: seems very close to the
SW web services
... proposel also to take a look at the Multimedia Ontology made of several design patterns ... among them, an algorithm design pattern
<scribe> ACTION: Raphael to send the paper describing this work to Suzanne [recorded in http://www.w3.org/2006/12/09-mmsem-minutes.html#action01]
Sophia mention an interesting work about an application toolbow where the user can select some algorithm to do feature extraction on music
<scribe> ACTION: Sophia to find and send a documentation about this work [recorded in http://www.w3.org/2006/12/09-mmsem-minutes.html#action02]
Ioannis: I present an image
analysis use case, with some similarities
... warn about the huge number of parameters that are generally use in image analysis
... sounds too ambitious for me
... propose to focus on a single (simple) application
Suzanne: I tend to agree
... difference between the complexity of the algorithm and the generic usefulness of such an ontology
Ioannis: what is useful is a framework that could allow to chain algorithms, and do optimization
Suzanne: we would like to do that ... close to what do the Web Services
Vassilis: is it worth to merge the use case
Suzanne: don't want too specific
<scribe> ACTION: Suzanne, Massimo to further exemplify and sketch the solution: what will be needed to model in the ontology, and how it would be used [recorded in http://www.w3.org/2006/12/09-mmsem-minutes.html#action03]
Issue: merge of these two use cases ?
Proposal: the two use cases continue on their own
Sophia will join Ioannis to work on the semantic retrieval use case
<mhausenblas> ScribeNick: mhausenblas
Thomas: Demo using flickr,
del.icio.us, YouTube, Technorati
... e.g. there might be a user Raphael that has some pieces on flickr
... other stuff (video) might be located on YouTube
... but not connected
... OR there are two users
... using different platforms
... may share a location, event, etc. - how to share them accross system boundaries
... Proposed solution: ontology describing personomies and tagging
... platforms offer a Web-API, helps accessing it
... no standards given
... tagging ontology (Koblenz, Newman)
... we opt to go for SKOS
Thomas shows a RDF graph from the SKOS guide
Thomas: We use SKOS to model
... also personomie terms are somehow available in SKOS
... multilingua aspects
... learn relations between tags
Jeff: Common sense reasoning?
Thomas: we can relate different modalities
Michael: Relation to Tom Grubers approach?
Thomas: Work in parallel
Raphael: Explains connection
between the lightweight way and the SKOS schemes
... one SKOS schema per user accross all systems
Susanne: How to integrate with the platforms _I_ use ?
Raphael demos the "Personomy Manager"
scribe: currently flickr and
del.icio.us are supported
... available operations: Config, Tag cloud (per platform or merged)
... operations cntd: download personomy
... how to tag a resource is just as easy
... enter URL, descr., tags, etc.
... everything done within the PM, using the Web-API of the diverse platform
Susanne: Usability issues. How about "reusing" flickr, etc. ?
Raphael: Back to interop issue
... guess relations between tags
... SKOS enables to do "small things" like broader, wider, etc.
... SKOS is not for the interop issue
Thomas: the ontology makes the
... so, interop is somehow supported with this
Raphael: The ontology focus on: what is a tag, a user, etc.
Suzanne: How to do merging when having different organisational/structural patterns
Raphael: Cf. work at Yahoo that
looks at tag characteristics
... have a look at http://rawsugar.com/index.faces as well
<ocelma> Re www2006 tagging workshop::http://www.rawsugar.com/www2006
Raphael: The proposed solution will contain the common format (SKOS-based), "core"-ontology, features
<raphael> The participants split into 2 groups and meet outside on the sunny terrasse (in front of the Parthenon
<raphael> ... to discuss the commonalities between the various use cases
<raphael> ... and see how they are related to, how they can be clustered, etc.
Discussion on a "common framework" that support multi modalities
propose to use RDF/XML as base
core ontology may also contain guidelines how-to use
Thomas: feasible to align the concepts in the UC used
Suzanne: from ABC model experience, might yield some 10 classes
Jeff: break it down now, make it
... good to base on existing Standards (RDF, bit-of-OWL, etc.)
... SKOS sitll not a standard
Thomas: Concerns about reinventing the wheel
Jeff: Our vocabulary should be more domain-specific
Suzanne: We got to be careful, cf. MPEG-7 concepts, etc.
Group now puts each of the 7 UC on a card (Music, News, Authoring, Algorithm, Retrieval, Photo, Tagging)
scribe: discusses common characteristics, goals, solutions
We now try to put UC into certain categories
scribe: e.g.: Modality (audio, vision, AV, etc.), Retrieval/Process, Solution-Exist,
Michael: Might be good idea to use the categories proposed in http://www.w3.org/2005/Incubator/mmsem/wiki/Vocabularies#head-5db084087806d9105c4f4ebd094b517b20913c5e
<scribe> new proposal: 3 layers (concepts/process, medium/instances, applications domains)
<raphael> Oscar, Raphael, Susanne, Giorgos, Thierry, Erik, Ioannis
<raphael> Picture showing how the use cases can be grouped together
<raphael> Common framework: describre how all these use cases can work together ... towards integration of RDF data
<raphael> Commonalities: decomposition of an image vs temporal sequence of audio
<raphael> ... all use cases make use of some common schema
<raphael> ... needs ontologies to model some knowledge
<raphael> ... integration of data among the use cases
<raphael> Some issues: cultural heritage use case ? liaison with Content labelling ? SKOS vs OWL ?
<raphael> Consensus on interoperability: common ontology ? using SKOS as a glue?
addendum to "Breakout Group JEFF":
scribe: participants were: Jeff, Thomas, Stamatia, Suzanne Little, Sophia, Michael
<raphael> Drawing of two schemas !
<raphael> There are pretty much the same thing
<raphael> Common framework:
<raphael> Syntax: RDF
<raphael> Common schema / ontology
<raphael> Schema / ontology harmonization and extension
<raphael> Guidelines: what to use in which context ?
<raphael> Interoperability problems between layer
<raphael> Table of Contents:
<raphael> Section 1: Introduction / Motivations
<raphael> Section 2: Summary of the interoperability problems of the use cases
<raphael> ... the schemas drawn, categorization of the use cases
<raphael> Do we need a state of the art ?
<raphael> Jeff: could be an Appendix, mentionning the related work
<raphael> Suzanne: could be in the vocabularies / tools / resources pages
<raphael> ... and just refer from the document
<raphael> Vassilis: issue about multilinguility ?
<raphael> Could be an open issue
<raphael> Section 3: Use cases
<raphael> Section 4: Common Framework (putting all together)
<raphael> Section 5: Open / Other problems
<raphael> For every layer: what are the similarities between the use cases ?
<raphael> Jeff: section 3 and 4 or the other way around ?
<raphael> Having a scenario for putting all together ...
<raphael> ACTION: Vassilis update the structure of the deliverable and fill it in a XG report template, HTML document in our W3C space [recorded in http://www.w3.org/2006/12/09-mmsem-minutes.html#action04]
<raphael> ACTION: Jeff to find the tool to move a wiki page into an HTML page [recorded in http://www.w3.org/2006/12/09-mmsem-minutes.html#action05]
<raphael> ACTION: Authors of each use case to fill the summary of their interoperability problems [recorded in http://www.w3.org/2006/12/09-mmsem-minutes.html#action06]
<raphael> ACTION: Vassilis and Jeff to start the Common Framework [recorded in http://www.w3.org/2006/12/09-mmsem-minutes.html#action07]
<raphael> Giorgos to help on this Common Framework
<raphael> Open Issue: Multilinguality ? Multmedia Authoring ?
<raphael> ACTION: Rapahel to finish the Image Annotation on the Semantic Web Deliverable [recorded in http://www.w3.org/2006/12/09-mmsem-minutes.html#action08]
<raphael> Approved: to have an MPEG-7 document
<raphael> Issue: do we want a frozen version of the Vocabularies document ?
<raphael> Vassilis: all vocabularies used in the use case should be described HERE !
<raphael> ACTION: all use case authors to check if their vocabularies are described in this page ... [recorded in http://www.w3.org/2006/12/09-mmsem-minutes.html#action09]
<raphael> ACTION: Michael to make testing how to transfert the wiki page into an XG report template [recorded in http://www.w3.org/2006/12/09-mmsem-minutes.html#action10]
<raphael> Schedule: be in position of reviewing a frozen version in End of February
<raphael> Michael: proposal to merge resource and tools pages
<raphael> ACTION: Sophia to be the main editor of a merge between the vocabularies and resources page [recorded in http://www.w3.org/2006/12/09-mmsem-minutes.html#action11]
<raphael> Leave the telecon as it is
<raphael> ACTION: chairs to give hints how it could be possible to join the telecon for free [recorded in http://www.w3.org/2006/12/09-mmsem-minutes.html#action12]
<raphael> and provide more phone numbers ...
<raphael> having more scribing turning
<raphael> Next F2F ?
<raphael> Beginning of April ?
<raphael> Transition to a WG ?
<raphael> Jeff shows the ususal W3C process diagramm
<raphael> What are the advantages to move to Rec Track ?
<raphael> An XG report has no W3C endorsment
<raphael> Recommendation means Specifications or Guidelines ...
<raphael> Michael: we still have till March 2007 to prepare such a document
<raphael> Vassilis: many work to do
<raphael> ... the WG could be a good idea, but make the work before
<raphael> Possibility to have a F2F in April 2007
<raphael> ... could be also a workshop to start a WG
<raphael> Jeff and Raphael will report on the teleconference with the SW Coordination Activity
<raphael> Week of 16 April for a possible of F2F
<raphael> no ... apart to make a photo all together !
<raphael> We close the meeting
This is scribe.perl Revision: 1.127 of Date: 2005/08/16 15:12:03 Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/ Guessing input format: RRSAgent_Text_Format (score 1.00) Succeeded: s/to/too/ Succeeded: s/Sussane/Susanne/ Succeeded: s/look/looks/ Found Scribe: Raphael Inferring ScribeNick: raphael WARNING: No scribe lines found matching previous ScribeNick pattern: <mhausenblas> ... Found ScribeNick: mhausenblas ScribeNicks: mhausenblas, raphael Present: Vassilis Susanne_Boll Suzanne_Little Sophia Stamatia Erik Michael Thomas Ioannis Raphael Jeff Thierry Oscar Agenda: http://lists.w3.org/Archives/Public/public-xg-mmsem/2006Dec/0014.html WARNING: No date found! Assuming today. (Hint: Specify the W3C IRC log URL, and the date will be determined from that.) Or specify the date like this: <dbooth> Date: 12 Sep 2002 Guessing minutes URL: http://www.w3.org/2006/12/09-mmsem-minutes.html People with action items: all authors case chairs each jeff massimo michael of rapahel raphael sophia suzanne use vassilis WARNING: IRC log location not specified! (You can ignore this warning if you do not want the generated minutes to contain a link to the original IRC log.)[End of scribe.perl diagnostic output]