Personal tools
You are here: Home Workshops-Events LREC 2006 Cross Media Indexing Workshop

Cross-media workshop

Details on the LREC 2006 cross-media indexing workshop.

LREC banner
5th Language Resources and Evaluation Conference
May 24-26, 2006,
Magazzini del Cotone Conference Centre, Genoa, Italy






Crossing Media for Improved Information Access

Introduction ] [ Target audience ] [ Topics ] [ Submission format ] [ Important dates ] [ Registration ] [ Committees ] [ Contact ]

New: Workshop Programme and Presentations




The i2010 initiative, adopted under the renewed Lisbon agenda, concerns information, communication and media industries and caters, among others, for the possibility of European citizens to ?watch or listen to audiovisual content anytime, anywhere and on all technical platforms (TVset, computer, mobile phone, personal digital assistant, etc.)?. In fact, much of the information that reaches the user nowadays is in digital form: digital radio, music CDs, MP3 files, digital satellite and digital terrestrial TV, personal digital pictures and videos and, last but not least, digital information accessed through the Web. This information is heterogeneous, multimedia and, increasingly, multi-lingual in nature.

The development of methods and tools for content-based organization and filtering of this large amount of multimedia information that reaches the user is a key issue for effective content consumption. Advances in medium-specific (audio, image, text) processing have facilitated the development of tools for indexing multimedia content. It is no coincidence, for example, that text-based indexing methods of such content prevail; text processing has reached a level of maturity that enables shallow semantic analysis for identifying keywords, terms and named entities as indexing terms, with considerable progress being made in the extraction of events and facts, i.e. a wealth of information for more accurate indexing. Experiments are ongoing on applying this type of indexing on speech recognition output as such or/and on associating web text to such output (for recovering from ASR mistakes) and then performing text-based indexing (cf. work within the PRESTOSPACE project); speech processing can provide automatic speech transcriptions of good quality (in certain acoustic conditions), as well as speaker turn and identification information. On the other hand, image-based indexing methods for multimedia content rely on basic image processing and in particular on the extraction of keyframes, shotcuts and low-level image features, while progress in developing face detection, face identification and object recognition technologies contribute to a more promising future for such approaches.

However, competitions such as TRECVID and Image-CLEF have shown that some benefits in performance can be gained through the fusion of the results of visual and linguistic analyses of multimedia content. Research on the automatic association of images with corresponding textual data go beyond fusion of medium-specific results to multimedia integration for, among others, indexing and retrieval applications (cf. work on UP-TV, ACEMEDIA and BUSMAN projects), while a more general notion of -crossing media- within or/and across documents seems to emerge too (cf. the REVEAL THIS project).


Target audience

The workshop aims at exploring these new tendencies in accessing multimedia content by bringing together researchers working on the development of indexing technologies for archived and contemporary multimedia content.



The focus of the workshop is the notion of "cross-media" mechanisms for multimedia indexing and retrieval and its realisation in research prototypes. Topics of interest to the workshop include (but are not limited to):
  • Exploration of the notion of "cross-media" mechanisms
  • Modality/medium interaction relations in multimedia data collections
  • Methods for cross-media indexing within and/or across documents
  • Medium-specific processing for facilitating cross-media indexing
  • Resources needed/built for cross-media indexing
  • Annotation needs for cross-media indexing resources
  • Evaluation methods

Submission format

The workshop will consist of paper and poster presentations. The submissions should be sent by e-mail to no later than the 27th of February 2006 and should conform to the following formatting instructions
  • Papers should consist of max. 7 pages (inclusive of references, tables, figures and equations), in English
  • PDF files should be submitted, and should follow the traditional LREC formatting (with a slight modification in font size). Latex style files and a word document template are available:
All accepted papers will be published in the workshop proceedings.


Important dates

  • Deadline for workshop paper submissions: Monday, 27 February 2006
  • Notification of workshop paper acceptance: Monday, 20 March 2006
  • Deadline for camera-ready workshop papers: Monday, 3 April 2006
  • Workshop Date: Tuesday, 23 May 2006 (morning session)


Information on registration fees and procedures are/will be available at the main LREC 2006 conference pages.



Organising Committee:

Programme Committee (more to be added soon):



For any queries please send an email to the workshop organisers: