Universiteit van Amsterdam

Events

Institute for Logic, Language and Computation

Please note that this newsitem has been archived, and may contain outdated information or links.

18-22 August 2003, Adaptation of Automatic Learning Methods for Analytical and
Inflectional Languages

Date: 18-22 August 2003
Location: Vienna, Austria

Automatic (machine) learning approaches to any NLP task became a rich area with a variety of methodologies. During the last years, its development made significant progress in the direction of presenting new methods and, at the same time, their modifications. These modifications are of different nature and dependent on the language under consideration. The aim of the workshop is to present and evaluate various modifications of the automatic learning methods originally developed for English and declared as language independent. We are especially interested in automatic learning methods for the problems of morphological tagging and parsing across languages with high level of inflection. Further, we encourage quantitative and qualitative comparison/evaluation studies across languages on the inputs and the outputs of the mentioned procedures. The workshop encourages reports of work on:

  1. Summarization of morphological and syntactic features relevant for various automatic learning procedures.
  2. Tendencies of improvement of the automatic learning methods. Presentation of implemented modifications and their cross language evaluation.
  3. New/Latest algorithms for automatic learning.
  4. Hybrid approaches (Although, there are trials to apply hybrid approaches, it seems that the true key of how to combine the various parts has still not been found and lies mainly in the success of analyzing the errors of each single component. Studies which present the connection elements for a successful combination of diverse approaches are invited.)

This workshop is part of ESSLLI'03. For more information, see http://ckl.mff.cuni.cz/~alaf03/

Please note that this newsitem has been archived, and may contain outdated information or links.