ATALAWorkshop
Representation and Treatment of Syntactic Ambiguity in
Natural Language Processing
Paris, C.H.U.
Pitié-Salpétrière (Amphi
D)
91, Bd de l'Hopital
January 29, 2000
There are several approaches making it possible to treat syntactic ambiguity:
one can for example decide to apply disambiguation as soon as possible, in
particular using statistical methods. One can on the contrary choose to delay
to the maximum the disambiguation and to maintain the ambiguity, possibly
until the end of the processing. These two approaches are not necessarily
contradictory. However, the problem remains very difficult to solve and many
of the methods proposed amount to little more than the traditional
enumeration.
Factorizing the information (for example in using parse forests) can from
this point of view play a very significant role and even lead to effective
solutions.
The goal of this workshop is to propose an overview of the current techniques
for the representation and the treatment of syntactic ambiguity in natural
language processing.
Invited Speakers
- Bernard Lang -- INRIA
- David Weir -- Cognitive and Computing Science / University of Sussex
Proceedings
The proceedings of this workshop have been published by INRIA (ISBN 2-7261-1154-8).
- Syntactic Disambiguation: A Neural Network Approach
Martha Analía Alegre,
Dpto. LSI, Universitat Politécnica de Catalunya
Josep Ma. Sopena, Augusti Lloveras,
Lab de Neurocomputaciò, Universitat de Barcelona
- Stratégies linguistique de désambiguïsation syntaxique basée sur corpus
Didier Bourigault, Cécile Fabre,
Université de Toulouse Le Mirail
- Dealing with ambiguities in an answer extraction system
Diego Mollà, Michael Hess,
University of Zurich
- Shared semantic dependency representations for LTAG
Kim Gerdes,
TALANA, Université de Paris 7
Patrice Lopez,
DFKI Saarbrücken
- Factorisation d'arbres à l'aide d'arbres partagés
Philippe Blache,
LPL-CNRS, Aix-en-Provence
- Créer, extraire et manipuler des forêts partagées avec DyALog
Eric Villemonte de la Clergerie,
ATOLL, INRIA
Chair
- Philippe Blache -- CNRS / Université de Provence
- Eric de la Clergerie -- INRIA