Computational Linguistics & Phonetics Computational Linguistics & Phonetics Fachrichtung 4.7 Universität des Saarlandes

Spoken Dialogue Systems

Softwareprojekt: Computerlinguistik, B.Sc./M.Sc.

Leitung: Klakow, Wolska
Ort: wird noch bekannt gegeben
Zeit: wird noch bekannt gegeben
Beginn: wird noch bekannt gegeben
Geeignet für: B.Sc./M.Sc.

In this course we plan to explore methods to learn dialog strategies. For this, reinforcement learning has turned out to be very popular in the past years. The rough plan is

  • study methods for reinforcement learning
  • implement them for a text book example
  • adapt the method to a dialog task


  • R. Sutton and A. Barto "Reinforcement Learning: An Introduction" HTML
  • S. Singh, D. Litman, M. Kearns and M. Walker "Optimizing Dialogue Management with Reinforcement Learning: Experiments with the NJFun System"
  • James Henderson, Oliver Lemon, and Kallirroi Georgila "Hybrid Reinforcement/Supervised Learning for Dialogue Policies from COMMUNICATOR data", IJCAI workshop 2005
  • Kallirroi Georgila and Oliver Lemon and James Henderson, "Automatic annotation of COMMUNICATOR dialogue data for learning dialogue strategies and user simulations", 9th SEMDIAL, DIALOR, 2005

    If you are interested in this course, drop us a mail (magda@CoLi.Uni-SB.DE, dietrich@CoLi.Uni-SB.DE).

    Credits are awarded upon successful completion of practical work and a final presentation. For the software project, a written report together with a description of the code and the code itself are required.

    Bachelor: 12 CP; Master: 8 CP