Crossing Barriers in Text Summarization Research
Workshop to be help in conjunction with
*** RANLP 2005 ***
Borovets - Bulgaria
http://www.lml.bas.bg/ranlp2005
*** 24th of September 2005 ***
Third and Last Call for Papers
*** Submission: 3 June 2005 ***
An abstract or summary is a text of a recognisable genre with a very
specific purpose: to give the reader an exact and concise knowledge of
the contents of a source document. In most cases, summaries are
written by humans, but nowadays, the overwhelming quantity of
information and the need to access the essential content of documents
accurately to satisfy users' demands has made of Automatic Text
Summarization a major research field.
Most summarization solutions developed today perform sentence
extraction, a useful, yet sometimes inadequate technique. In order to
move from the sentence extraction paradigm to a more challenging,
semantically and linguistically motivated 'abstracting' paradigm,
significant linguistic (i.e., lexicons, grammars, etc.) as well as
non-linguistic knowledge (i.e., ontologies, scripts, etc.) will be
required. Some 'abstracting' problems like 'headline generation', have
been recently addressed using language models that rely on little
semantic information, what are the limits of these approaches when
trying to generate multi-sentence discourses? What tools are there to
support 'text abstraction'? What type of natural language generation
techniques are appropriate in this context? Are general purpose
natural language generation systems appropriate in this task?
Professional abstractors play a mayor role in dissemination of
information through abstract writing, and their work has many times
inspired research on automatic text summarization, they are certainly
one of the keys in the understanding of the summarization
process. Therefore, what tools are there to support Computer-Assisted
Summarization and more specifically how these tools can be used to
capture 'professional summarization' knowledge?
In a multi-lingual context, summaries are useful instruments in
overcoming the language barrier: cross-lingual summaries help users
assess the relevance of the source, before deciding to obtain a good
human translation of the source. This topic is particularly important
in a context where the relevant information only exists in a language
different from that of the user. What techniques are there to attack
this new and challenging issue? What corpora would be appropriate for
the study of this task?
The ``news'' has been a traditional concern of summarization research,
but we have seen, in the past few years, an increasing interest for
summarization applications on technical and scientific texts, patient
records, sport events, legal texts, educative material, e-mails, web
pages, etc. The question then, is how to adapt summarization
algorithms to new domains and genres. Machine learning algorithms
over superficial features have been used in the past to decide upon a
number of indicators of content relevance, but when the feature space
is huge or when more ``linguistically'' motivated features are
required, and as a consequence the data sparseness problem appears,
what learning tools are more appropriate for training our
summarization algorithms? What types of models should be learned
(e.g., macrostructures, scripts, thematic structures, etc.)?
Text summarization, information retrieval, and question answering
support humans in gathering vital information in everyday activities.
How these tools can be effectively integrated in practical
applications? and how such applications can be evaluated in a
practical context?
We call for contributions on any aspect of the summarization problem,
but we would like the workshop to give the research community the
opportunity for discussion of the following research problems:
* Crossing the language barrier: cross-lingual summarization; corpora
to support this summarization enterprise;
* Crossing the extractive barrier: non-extractive summarization (i.e.,
text abstraction); resources for capturing abstraction knowledge or
expertise;
* Crossing genres, domains, and media: adaptation of summarization to
new genres, domains, media, and tasks.
* Crossing technological barriers: integration of summarization with
other NLP technologies such as Question Answering and Information
Retrieval.
The workshop will be organized around paper presentations, panel
discussions, and one invited talk.
Important Dates:
Deadline for submission: *** 3 June 2005 ***
Notification of acceptance: 29 July 2005
Camera-ready copy due: 19 August 2005
Workshop: 24 September 2005
Important Announcements:
*** Invited Speaker ****
Dragomir R. Radev
School of Information and Department of Electrical
Engineering and Computer Science, University of Michigan, USA
Title: *** Graphs everywhere: novel methods for summarization and natural
language processing ***
If the workshop is successful, we will issue an special call for a
thematically focused volume on text summarization. Workshop authors
will be invited to submit extended versions of their papers for this
purpose.
Submission guidelines:
Submissions should be A4, two-column format and should not exceed
seven pages, including cover page, figures, tables and references.
Times New Roman 12 font is preferred. The first page should state the
title of the paper, the author's name(s), affiliation, surface and
email address(es), followed by keywords and an abstract and continue
with the first section of your paper. Papers should be submitted
electronically in **PDF** format to saggion@dcs.shef.ac.uk. For up to
three free conversions to PDF see
http://192.150.14.203/index.pl?BP=NS. Guidelines for producing
camera-ready versions can be found at the conference web site:
http://www.lml.bas.bg/ranlp2005.
Each paper will be reviewed by up to three members of the program
committee. Authors of accepted papers will receive guidelines
regarding how to produce camera-ready versions of their papers for
inclusion in the proceedings.
Parallel submissions to the main conference and the workshop are
allowed but the review process will be coordinated. Please declare
this in the notification form.
Organization
*Horacio Saggion
NLP Group
Department of Computer Science
University of Sheffield
Sheffield - UK
*Jean-Luc Minel
LaLLIC
Universite de Paris IV-Sorbonne
Paris - France
Program Committee:
Gustavo Crispino, LaLLIC, Universite de Paris IV, France
Hercules Dalianis, Stockholm University, Sweden
Brigitte Endres-Niggemeyer, University of Applied Sciences and Arts,
Germany
Donna Harman, National Institute of Standards and Techology, USA
Hongyan Jing, IBM T.J. Watson Research Center, USA
Min-Yen Kan, School of Computing, National University of Singapore,
Singapore
Choy-Kim Chuah, Universiti Sains, Malaysia
Guy Lapalme, Departement d'informatique et de recherche
operationnelle, Universite de Montreal, Canada
Lehmam, Abderrafih, Pertinence Mining, Paris, France
Chin-Yew Lin, Information Sciences Institute, University of Southern
California, USA
Inderjeet Mani, Department of Linguistics, Georgetown University, USA
Jean-Luc Minel (Co-organizer), LaLLIC, Universite de Paris IV, France
Marie-France Moens, Interdisciplinary Centre for Law & Information
Technology, Katholieke Universiteit Leuven, Belgium
Constantin Orasan, School of Humanities, Languages and Social Studies,
University of Wolverhampton, UK
Dragomir Radev, School of Information and Department of Electrical
Engineering and Computer Science, University of Michigan, USA
Horacio Rodriguez, Department de Llenguatges i Sistemes Informatics,
Universitat Politecnica de Catalunya, Spain
Horacio Saggion (Organizer), Department of Computer Science,
University of Sheffield, UK
Stan Szpakowicz, School of Information Technology and Engineering,
University of Ottawa, Canada
Simone Teufel, Computer Laboratory, University of Cambridge, UK
Dina Wonsever, INCO, Universidad de la Republica, Uruguay
*** Please send your submission to:
Horacio Saggion Email: h.saggion@dcs.shef.ac.uk
Please use the subject line: "Summarization Workshop/RANLP2005"
and include in your message the following information:
# NAME: Name of author for correspondence
# TITLE: Title of the paper
# KEYS : Keywords
# EMAIL: Email of author for correspondence
# PAGES: Number of pages (including bibliographical references)
# FILE : Name of PDF file
# ABSTR:
# Abstract of the paper
# ...
# OTHER: Under consideration for other conferences? (please specify)
# NOTE : Anything you would like to add
*** For any further information please contact
Horacio Saggion at h.saggion@dcs.shef.ac.uk
This archive was generated by hypermail 2b29 : Sat May 21 2005 - 16:11:55 MET DST