We apologize if you receive multiple copies of this message.

                      CALL FOR PAPERS
                   I-KNOW '04 Special Track:
                  Cross Media Knowledge Mining
                   http://www.i-know.at/cmkm

        June 30th - July 2nd, 2004, Graz, Austria 


Background
==========

This one-day special track, which takes place during I-KNOW '04 in Graz
- Austria, provides a unique opportunity for scientists working in
fields as diverse as information retrieval, multimedia knowledge mining
or metadata extraction to communicate and share experiences on
developments and achievements in the exciting new field of cross-media
knowledge mining.

Traditionally, large knowledge repositories demanding sophisticated
retrieval and mining mechanisms contain text documents as basic
information entities. Consequentially, the broad field of information
retrieval and knowledge mining has been shaped in many aspects towards
text-based approaches and algorithms. Today's knowledge retrieval and
visualisation systems easily deal with millions of information entities
structured in complex patterns, as long as the entities feature
significant textual content, or equivalent extracted or annotated meta data.

In another development, techniques like content-based image retrieval
have matured and are today quite capable of dealing with multimedia
content. Advanced algorithms allow automated extraction of content from
multimedia documents like images, sounds and videos. A variety of
standards has been devised for storing and organising extracted, or
manually annotated, meta data, enabling knowledge retrieval on
multimedia repositories. A growing number of systems and tools employ,
for example, the MPEG7 standard to annotate, organise and retrieve
multimedia documents.

Today's knowledge repositories usually contain significant amounts of
multimedia information in addition to text documents. This trend is
likely to continue and expand, as more and more digital cameras,
webcams, sound recording devices and other multimedia hardware becomes
available and widespread. Cross-media knowledge mining, the science and
art of retrieving images, text, sound and other multimedia documents
based on a single user query, is one of the key challenges for the
retrieval community. Topics like finding metrics for comparing
information entities of differing types or designing meta data
descriptions which allow unified annotation of text and multimedia
documents will provide advanced research topics for years to come.


Goals
=====

This special track aims at bringing into contact international
researchers from fields as information retrieval or multimedia knowledge
mining as well as practitioners in this areas to enable the exchange of
current research on cross-media knowledge retrieval. The special track
allows for networking among attendees to establish focused,
international relationships.


Topics of the Special Track
===========================

The special track seeks contributions that discuss expectations,
requirements, approaches and concepts as well as experiences and
practical results concerning cross-media knowledge retrieval. Topics may
include, but are not limited to:

    * Cross-media similarity metrics
    * Metadata standards for cross-media information entities
    * Inference by using cross-media information entities
    * Traditional data mining methods applied to cross-media
    * Indexing of cross-media repositories
    * Query formulation and query definition for cross-media searches
    * Machine learning and statistical methods applied to cross-media
      repositories
    * Automatic extraction of semantic information from cross-media content
    * User Interfaces for cross-media knowledge mining


Submission Instructions
=======================

Authors are encouraged to refer to the identified challenges and
questions of this special track in their contributions.

Please prepare your submission according to the J.UCS author guidelines
available at http://www.jucs.org/jucs_submit/style_guide.html. All
submissions, regardless of category, should include title, author(s),
abstract, keywords, body, and references.

Submissions are to be made electronically in PDF (Adobe's Portable
Document Format) or MS Word. Submissions are treated as confidential
communications during the review process, so submission does not
constitute public disclosure of any ideas therein. Submissions should
contain no information or materials that will be proprietary or
confidential at the time of publication, and should cite no publications
that are proprietary or confidential at the time of publication.


Submission Procedure
====================

Please submit your contribution via http://www.i-know.at/submission. In
case of problems or questions concerning the submission of papers,
please contact the track organizer Michael Granitzer at
[log in to unmask] <mailto:[log in to unmask]>.


Important Dates
===============

    * January 26th, 2004 - Submission of papers (4 pages)
    * March 22nd, 2004 - Notification of acceptance
    * April 26th, 2004 - Camera ready paper (8 pages)
    * June 30th - July 2nd, 2004 - The special track takes place during
      I-Know '04


Publication
===========

Accepted papers will be published within the I-Know'04 conference
proceedings in cooperation with J.UCS and Springer Co. Pub. In addition
to that, special track presenters have the possibility to publish an
extended version of their paper (up to 20 pages) in a special issue of
J.UCS to appear after the conference.

Further information
====================
By email to [log in to unmask] 

--

====================================================
Michael Granitzer
Knowledge Discovery
Know-Center
Inffeldgasse 21a, 8010 Graz, Austria
Email  : [log in to unmask]
Tel: +43 316 873 9263   Fax: +43 316 873 9252

http://www.know-center.at
====================================================