Research

François Bouchet, Ph.D., is an assistant professor at Université Pierre et Marie Curie (UPMC) and a researcher at the Laboratoire d'Informatique de Paris 6 (LIP6) in the MOCAH team.

He received his M.Eng from ESIEA in 2005, and both his M.Sc. (2006) and Ph.D. (2010) in Computer Science from Université Paris-Sud 11, with a thesis supervised by Dr. Jean-Paul Sansonnet at LIMSI, centered on the use of conversational agents for assistance, which resulted in the design of (1) a custom natural language processing chain for assistance requests and (2) a novel agent cognitive architecture taking into account personality and emotions.

During his postdoc (2010-2013) at McGill University under Dr. Roger Azevedo's supervision, he has been the main architect of the Intelligent Tutoring System (ITS) MetaTutor and focused on applying data mining techniques (e.g., HMM, clustering, sequence mining) across multiple data channels (e.g., logfiles, facial expressions, eyetracking, electrodermal activity) to identify learners' profiles, in order to conceive a new generation of more adaptive ITSs.

His latest research interests revolve around serious games and long term interactions with pedagogical agents.

Defense

My thesis entitled "Design of a natural language processing chain for an assistant conversational agent" has been defended on Tuesday June 29th 2010 at 2:00pm in the conference room of the LIMSI-CNRS, in front of the following jury members:

• Guy Lapalme, Professor at Montreal University, RALI (reviewer)
• Catherine Pelachaud, Research director at LTCI Télécom ParisTech (president)
• Sylvie Pesty, Professor at University Pierre Mendès-France, LIG (reviewer)
• Jean-Paul Sansonnet, Research director at LIMSI-CNRS (thesis supervisor)
• Anne Vilnat, Professor at University Paris-Sud 11, LIMSI-CNRS (examiner)

Summary

With the increasing number of novice users of computer applications, the need for efficient assistance has become critical. To supply the need, we suggest using an Assistant Conversational Agent (ACA), an interface allowing the use of natural language (used spontaneously when a problem arises) and providing a reassuring presence to the users.

A preliminary study details the constitution (combining collection and the use of thesauri) of a corpus of requests, which need is justified. This corpus of 11,626 requests is compared with others, and we show that it covers the studied domain of assistance and moreover, contains requests regarding controlling of the application and chatting with the agent.
This corpus provides a sound foundation for the conception of a syntactico-semantic analyzer of natural language requests, using a set of semantic keys, a set of analysis rules and a set of transformation rules. In output, requests are expressed in a formal language (DAFT) for which we provide the syntax and the semantics.

The analyzer is evaluated by comparing a manual annotation and the requests automatically produced, and we consider the use of some supervised machine learning approaches in order to identify conversational activities.
The methodology followed is validated through the integration of an ACA into an existing Web application for cooperative music prototyping.

Finally, we describe the required architecture for the rational agent in charge of defining the reactions based on the formal requests expressed in DAFT and on the model of the assisted application, emphasizing the need for a cognitive model.

Educational data mining to identify learners' profiles

Adaptive Intelligent Tutoring Systems design

Cognitive architecture for rational agents with a personality

Embodied Conversational Agents for Assistance

My PhD main research topic was the Embodied Conversational Agents (ECA) and particularly their use to provide help to ordinary people using computers. My PhD thesis, which started in October 2006 and ended in June 2010, was part of the LIMSI DAFT project, in which I was interested in the construction of a formal representation for natural language requests semantic and in the definition of canonical heuristics for the rational agent.
I collaborated with David Leray who was working on applications symbolic modelization. Our work was supervised by Jean-Paul Sansonnet, who had initiated the project in 2004.

The following articles describe some specific aspects of my research work:

• Registering and analysis of a corpus of natural language assisting requests

DIVA

• CODIVA demo: in the context of my collaboration with Evandro Miletto Manara, we've been leading experiments in which an assisting agent was embedded into the collaborative music prototyping framework CODES (latest version here)
• DIVA toolkit: in the context of my works with Mao Xuetao, I've been working on the DIVA framework developped within the AMI group

DAFT

• An except of 1/10th of the full Daft corpus made of 1074 collected requests from users in situation of assistance:
  • ordered alphabetically: ISO8859-15 / UTF8
  • ordered by length: ISO8859-15 / UTF8
• GRASP semantic keys: created by Jean-Paul Sansonnet, I have associated DAFT schemes to them
• DAFT language schemes, used by DIG to generate formal requests
• GRASP & DIG interactive demo