Postdoc position available at GREYC - University of Caen Basse-Normandie
- France

Event prediction for dialogue modelling in embodied conversational agent

Location: Caen (France)
Duration: 15 months (potentially 18 months)
Keywords: sequence mining, knowledge discovery, event prediction, 
dialogue modelling

Description of the task:

Designing a model of dialogue is a difficult and often multidisciplinary
task. Whether it is dedicated to interactive storytelling or not, it
involves many algorithms: multi-modal input recognition (utterances,
gestures, gazes, vocal inflections), natural language understanding and
generation, dialogue management, planning and cognitive capacities,
emotion modelling, prosodic speech generation, non-verbal behavior.

In this context, the post-doc fellow's research will focus on designing
algorithms for modelling the interactions (turn of speech) during the
conversation, starting from real data gathered during dialogues between
adult (or Wizard of Oz) and child, that are annotated by
psychologists. This task can be viewed as predicting a particular event
in a sequence of itemsets. The model of the dialogue will be implemented
in an embodied conversational agent.

Application:
The candidate must prepare a detailed CV including a complete
bibliography, a motivation letter and recommendation letters as a single
pdf file. This file should be sent by email to the contact below. French
people may apply using French language.

Contact:
François Rioult
CNRS UMR6072 GREYC
Université de Caen Basse-Normandie
francois.rioult@unicaen.fr