Corpora: CFP - Workshop on Modular Programming for NLP at Eurolan'01

From: Catalina Barbu (C.Barbu@wlv.ac.uk)
Date: Wed Jan 10 2001 - 01:08:27 MET

  • Next message: l.hoxhallari@bangor.ac.uk: "Corpora: Albanian text corpus?"

                                    CALL FOR PAPERS
    Workshop on "Modular Programming applied to Natural Language Processing"
                Held as part of EUROLAN'01 Summer School
                            July 30 - August 11
                                Iasi, Romania

    The call for papers and comprehensive information can be found on line at www.wlv.ac.uk/sles/compling/news/workshop.html

    The effectiveness of modular programming in designing software has long been acknowledged by the computer science community. However, the computational linguistics community preferred to develop components in isolation, without integrating existing modules into proposed systems. There are several reasons for this. Firstly, integration of different modules is not a trivial task, requiring a lot of time. Usually the major problem is the loss of information caused when the output of one module has to be converted to the input of another. Most research projects do not have the time or resources to concentrate on a real modular architecture, using trade offs (such as manually created inputs) instead. Secondly, most of the work in the research community is directed towards proposing and demonstrating new hypotheses, and not building robust and fully automatic applications. In many cases preprocessing steps, which produce the input data for the tested method, are considered trivial and accurate, and as a result rep
    laced with hand produced data. Therefore, when a researcher needs a certain module for a method, s/he prefers to produce the output of that program manually, either because s/he is not aware of an existing implementation which performs the required task, or because the work involved in setting it up is greater than that involved in manually producing the output (usually because the implementation was developed and tested on a different platform).

    However, this situation has started to change rapidly. More and more researchers have appreciated the complexity of NLP tasks and the need to use modular programming. A quick look at the systems presented at the latest MUC indicated that they are complex systems which reuse previous research. Systems like GATE have been designed in order to help with the integration of different modules in a system. In addition, the research community is increasingly requiring the development of fully automatic applications.
     
    This workshop will provide a forum for discussion between researchers involved in the development of automatic NLP systems and leading names in the field. We would like to invite all researchers to submit their original and unpublished work to the workshop. Topics of interest include but are not limited to:
    - modular architectures for NLP
    - black/glass box evaluation measures
    - research on the influence of substitution and alternate combinations of modules on overall system performance
    - reusability
    - integration of resources (including conversion formats between modules)
    - platforms for developing modular applications
    - repositories
    Demos of the presented systems are encouraged.

    2. Submission Requirements

    Papers should describe original work in progress or completed work. The main purpose of presenting at the Workshop is to exchange ideas with other researchers and to receive helpful feedback for further development of the work.

    The submissions should not exceed 10 pages with 12pt Times New Roman font including figures, references, and appendices. Authors will be sent details of the final format for camera-ready versions together with notification of acceptance. Each paper will be reviewed by at least two members of the program committee.

    Electronic submissions are encouraged but hard copies are acceptable. For electronic submission, papers should be in Postscript, PDF, Microsoft Word, or RTF format. For other formats, please contact the organising committee. All papers should be sent to workshop-eurolan@wlv.ac.uk

    3. Contact Address

    Constantin Orasan
    HLSS
    University of Wolverhampton
    Stafford Street
    Wolverhampton
    WV1 1SB
    UK
    Email: C.Orasan@wlv.ac.uk (please use "EUROLAN WORKSHOP" in the subject line)

    4. Important Dates

    Submission Deadline: 1st April 2001
    Notification of Acceptance: 4th May 2001
    Camera-ready Papers: 1st June 2001
    Workshop: To be announced

    5. Registration

    People wanting to attend the workshop must be registered in the Eurolan'01 School. Participation to the workshop is open to all Eurolan'01 attendants. Copies of workshop proceedings will be made available.

    6. Organizing committee

    Constantin Orasan - University of Wolverhampton)
    Richard Evans - University of Wolverhampton)
    Catalina Barbu - University of Wolverhampton)

    7. Program committee

    Dan Cristea - University of Iasi
    Hamish Cunningham - University of Sheffield
    Le An Ha - University of Wolverhampton
    Nancy Ide - Vassar College, Poughkeepsie
    Ramesh Krishnamurty - University of Wolverhampton
    Daniel Marcu - University of Southern California
    Oliver Mason - University of Wolverhampton
    Ruslan Mitkov - University of Wolverhampton
    Maximiliano Saiz Noeda - University of Alicante
    Chris Paice - Lancaster University
    Valentin Tablan - Universities of Iasi & Sheffield
    Doina Tatar - Babes-Bolyai University
    Dan Tufis - Romanian Academy



    This archive was generated by hypermail 2b29 : Wed Jan 10 2001 - 09:00:36 MET