Poul Andersen * Cooperation with Central and Eastern Europe in Language Engineering

• BABEL - A Multi-Language Database

  Speech input/output is going to be a major growth area in information technology in the coming years. Scientific development is dependent on the quality and quantity of recorded speech material available for training and testing systems, and speech databases should be produced according to carefully defined standards, with high priority given to re-usability.

  As speech scientists in Central and Eastern Europe become more closely associated with their colleagues in Western Europe, it is vital that speech database standards should not be allowed to diversify. The BABEL project trains and equips a number of non-EU speech research laboratories to use existing EU speech database standards, with researchers from the C&EE partner institutions spending some time working in Western laboratories.

  The project will produce a database of high scientific quality that will be used for many years in the development of European speech technology.

  East European languages covered: Bulgarian, Estonian, Hungarian, Polish and Romanian.

  The Romanian partner is M. Boldea, Computer Science Department, Faculty of Computer and Control Systems, Timisoara.

  The 6 projects without Romanian participation are:

  Three projects concerned with Dictionary standards + coding :
  
  

  • CEGLEX - Central European GeneLEX model

    CEGLEX aims at extending the generic electronic dictionary model developed in the EUREKA GENELEX project to three central European languages and to:
    
    

    • give rapid access to a Western European Linguistic Engineering pre-standard model for Central European actors of the NLP scene,
    • extend the GENELEX model to new languages, evaluating its appropriateness and making it a stronger candidate for being an internationally recognised standard,
    • start a larger cooperation between the partners leading to industrial level applications.

    The work consists in:
    
    

    • identifying theoretical issues in Czech, Hungarian and Polish that may lead to specific adaptations and extensions of the model;
    • building a representative core lexicon that will conform to the elaborated model;
    • verifying the possibility to use the core dictionary in the context of an application.

  • GRAMLEX

    The aim of GRAMLEX is to facilitate the initiation, coordination and standardisation of the construction of morphological dictionary packages for French, Hungarian, Italian and Polish, including detailed formal description of the morphology of the languages. The major challenges in such an enterprise are to give the description the largest possible coverage, in order to be able to process unrestricted texts; to share as much as possible of the formats, methods and algorithms; and to improve time and space efficiency of programs.

    Keywords: lexical tagging, morphological dictionaries, lexical resources.

  • BILEDITA - Bilingual electronic dictionaries and intelligent text alignment

    The goals of BILEDITA are:
    
    

    • to provide a uniform dictionary format for all of the existing dictionaries constructed by the partners;
    • to provide a uniform lexical encoding scheme for both form and content of the entries in the electronic dictionaries. This has been systematically dealt with as far as the French and German dictionaries are concerned, and will be accomplished for the other project languages : Bulgarian, Polish and Russian;
    • to systematically construct and exploit bilingual corpora for the purpose of building bilingual basic dictionaries, terminology dictionaries and phrasal dictionaries.

17