Khalil Sima'an


Position:
   Research Scientist 
  KNAW   Postdoctoral Fellow  [June 2002 - May 2005]
Affiliation:
   Computational Linguistics
  Institute for Logic, Language and Computation (ILLC)
  Faculty of Humanities            University of Amsterdam
Address:
  Room 2.34, Nieuwe Achtergracht 166, 1018 WV Amsterdam
Email:
  khalil.simaan@hum.uva.nl
Telephone:
 +31- 20 - 525 6573
Fax:
 +31- 20 - 525 2800
 
In short
I am a postdoctoral research scientist working on Statistical Natural Language Processing and Machine Learning.
Starting June 2002 I am an ``Akademie Onderzoeker" (KNAW - Royal Netherlands Academy of Arts and Sciences ).


Research 
 Courses
Brief CV
  Publications
Systems
    Links

Current research:

Probabilistic models of Natural Language Processing and their application to various language technologies.
Machine Learning of Stochastic Feature Structure Grammars from Phrase-Structure tree-banks.


 Research interests:

I am interested  in  computational models of intelligent behavior, including  the formal, computational and empirical aspects.
Some of the issues I currently work on include
  • Stochastic ambiguity resolution and probabilistic models of Natural Language Processing (NLP)
  • Machine Learning techniques, specifically specialization methods for  probabilistic models of  natural language parsing
  • Applications of the above techniques to e.g. Speech Understanding and Question Answering systems
  • Algorithms for and computational complexity of stochastic parsing problems
  • Applications of machine learning algorithms to general cognitive modeling problem


    • Other research activities:

    In cooperation with dr Yoad Winter and Prof. Alon Itai from the  Technion we are developing  A  Hebrew Tree-Bank
    within a project on Corpus-Based Analysis of Hebrew.


    Back to top



     Courses 2001/2002:

    Back to top


     
     Actively  used software system developed during my work on NLP:

  • Data Oriented Parsing and Disambiguation System (DOPDIS)
    •    An integrated training module for  obtaining a DOP model, and  a CYK-based parsing module .

    DOPDIS is being (has been) used by:

  • OVIS Project
  • CLASK Project
  • Rens Bod  for various experiments, especially for speech-understanding
  • Menno van Zaenen for experiments with DOP models for his Ph.D. thesis
  • Remko Bonnema for various experiments
  • Technion project
    •       
    Back to top

     Some Links:

     
     Nabil Sima'an  ILLC , UvA ILK, Tilburg University Computational Linguistics-UvA LIT Group
     Hebrew Tree-Bank Project Ronald M. Kaplan Christer Samuelsson

    FoLLI CLIN'00 BNAIC'00