Vacancy PostDoc, Statistical NLP

The Institute for Logic, Language and Computation
http://www.illc.uva.nl

Postdoctoral Researcher

1.0 fte (38h per week)

For both internal and external candidates

Summary

• Position: POSTDOC researcher
• Duration: 2 years (full time, 38hrs per week)
  
• Salary (gross per month): minimum Euro 2379 and maximum Euro 3755 in the first year.
• The Collective Employment Agreement of the Dutch Universities is applicable.
• Last date for application: February 15, 2009.
  

Project

Priors for the Estimation of Probabilistic Grammars from Incomplete Natural Language Data
VIDI project Sima'an [2007-2011]

Abstract: The project aims at exploring the utility of available prior knowledge in specifying the model (hypothesis space) and the constraints on the objective estimation function for unsupervised estimation of probabilistic grammars for Statistical Machine Translation and Statistical Parsing.
This project addresses problems of unsupervised estimation in the face of different kinds of incomplete data (parallel and monolingual corpora):

1. A raw text corpus,
2. A partially annotated corpus possibly with negative examples, and
3. A raw in-domain corpus and an out-of-domain annotated corpus.
   

Corpus annotation is e.g., word-alignment, syntactic or semantic structure.

Example topics studied within the project (see project related publications):

• Statistical machine translation with hierarchical/syntactic structure, e.g., statistical estimation [Mylonakis and Sima'an 2007 and 2008], lexical syntactic structrure [Hassan, Way and Sima'an 2007 and 2008]
  
• Statistical machine translation with morphological structure -- Work in progress 2009,
  
• Inducing hidden compositional structure for machine translation (possibly using syntactic structure) -- Work in progress 2009.
  

Project team: Markos Mylonakis, Tejaswini Deoskar, and Khalil Sima'an (project leader).
Affiliated researchers: Reut Tsarfaty.
External collaborators: Hany Hassan , Barbara Plank and Andy Way.


Requirements

We seek a cooperative yet independent researcher who is passionate about empirical research, statistical methods and NLP/CL.
The following are the main requirements:

• obtained a Ph.D. recently (or will obtain it in the very near future, in which case testemoney from the PhD thesis Supervisor is needed)
• a strong background, experience in and affinity with the largest subset of the following areas:

1. Computational Linguistics/Natural Language Processing working with statistical methods
2. Machine Learning Applied to NLP with good knowledge of unsupervised statistical methods.
   
3. Statistical Machine Translation.
4. Unsupervised statistical methods for statistical Parsing.

Appointment

The appointment will be on a temporary basis for a maximum period of 2 years. Based on a full-time appointment (38 hours per week) the gross monthly salary will range from € 2379,- to, at the most, € 3755,- , in the first year depending on expertise and experience.

The Collective Employment Agreement of the Dutch Universities is applicable.

Application
To Dr Khalil Sima'an by email (k.simaan@uva.nl with subject heading: VIDI-POSTDOC).
The following documents (in pdf):

1. Letter expressing interest
   
2. Curriculum Vitae
3. List of publications (possibly including a downloadable pdf of your thesis)
   
4. Three letters of reference sent directly by the referees to Dr Khalil Sima'an
   
5. Optional: Research plan (1 page)

Last date for application: February 15, 2009.

For inquiries contact

Dr Khalil Sima'an.
Language and Computation
Computer Science
University of Amsterdam