How to devise a cognitive agent for distance language learning Jean Claude Bertin Patrick Gravé University of Le Havre France Introductio

Document Sample
How to devise a cognitive agent for distance language learning Jean Claude Bertin Patrick Gravé University of Le Havre France Introductio Powered By Docstoc
					How to devise a cognitive
  agent for distance
  language learning
         Jean-Claude Bertin
            Patrick Gravé
    University of Le Havre, France
                           Introduction
   An ongoing research program :
      linguist / educationalists / computer researchers


   Based on
       ODL didactic ergonomics model (CIRTAI, University of Le Havre)
       + LITIS team (computer sciences research laboratory Universities of Le
        Havre/Rouen)
            ( bibliography)


   Focus on learner follow-up

   Objective of communication : foster debate… and
    collaborations ?
                                A few references…

   Bertin J.C., Des outils pour des langues, Ellipses, Paris, 2000.

   Bertin J.C., Gravé P., « Didactic ergonomics and web-based materials design
    », CALICO ’04, Pittsburgh, 2004.

   Annoot E. , Bertin J.C. & Gravé P., « Quelles médiations dans les formations à
    distance dans l’enseignement Supérieur ? », Université du Havre, janvier 2005.
    Online at : http://perso.wanadoo.fr/jean-claude.bertin/SiteBertin.htm.

   Person P., Boukachour H., Coletta M., Galinho T., Serin F., « Data
    Representation Layer in a MultiAgent Decision Support System », LITIS,
    MultiAgent and Grid Systems – an International Journal, vol. 2, Special issue,
    2006.

   Gravé P. , Ennaji M., Boukachour H., «Une architecture multiagent pour la
    pédagogie de la formation à distance », Proceedings of MOSIM ’06,
    Modélisation,, optimisation et simulation des systèmes : défis et opportunités,
    Rabat, April 2006.
Starting point : an ODL « elaborate model »

                                             CONTEXT
                                      Integration in global course

                                        linguistic objectives
        Discipline                                                                   teacher
        language                                                                     designer

                                         nature/function
                                           of material




                                         computer
                                                                     Interactivity
                           teaching
                                                                         ???
                            agent

                                              virtual
                                             learning
                                           environment
                                                                        peers : virtual class



      other types
     of documents
                                           learner
               real-life
                  tutor
                                                                                           follow-up
        Focusing on the learner (1)
   Helping the learner in a distant learning setting :
    main objectives.
     Decrease in-training drop-outs
     Learner immediate feedback

     Learning process facilitation

     Individualized learner follow-up and scaffolding

     Development of learner autonomy (learning to learn)
       Focusing on the learner (2)
• Helping the learner in a distant learning setting :
  various types of help
    Operational
    Disciplinary

    Cognitive/metacognitive…



    What is required = a global definition of online
    tutoring
Research focus : tutoring modalities


   Human Vs machine tutors ?
     Feasibility ?
     Respective roles/functions

              of humans / machines ?
            Methodology


   2 main phases :
     Field study of online tutoring
     Design of agents
                        Field study
   Large-scale survey on tutoring practices and
    representations in ODL environments
       Technical, pedagogical & tutoring skills
       Tutors’ socioprofessional identity
       Learner strategies
       Learner practices (... tutors / peers)

       Questionnaires for tutors and learners in online campuses

   Comfort model of a complex human-machine tutoring
    system
               P. Hubbard’s “Virtual Tutor”
                     (Calico ’99, ’00)

   “teaching agent” should combine several
    features :
       -   physical presence and personality
       -   Expertise in the field of reference
       -   An aptitude for individualised instruction
       -   An aptitude to initiate learning
From Hubbard’s agent to the didactic ergonomic perspective

                                                  CONTEXT
                                           Integration in global course

                                             linguistic objectives
             Discipline                                                                   teacher
             language                                                                     designer

                                              nature/function
                                                of material




                                              computer
                                                                          Interactivity
                                teaching
                                                                              ???
                                 agent

                                                   virtual
                                                  learning
                                                environment
                                                                             peers : virtual class



           other types
          of documents
                                                learner
                    real-life
                       tutor
                                                                                                follow-up
            Translating the model into computer perspectives

                                      Computer research
Teaching agent (Hubbard)                perspectives
   Physical presence and             Flexible and intelligent HMI
    personnality                      Knowledge database
   Disciplinary expertise            Scenario database
   Aptitude for individualized       System features :
    instruction                              Autonomous and proactive
   Aptitude to initiate learning            dynamic
                                             flexible
                                             Open
                                         Complexity
                                         MultiAgent System
       Decision support system for dynamic situations
Machine tutor : model
                                       CONTEXT
                              Integration in global course

                                linguistic objectives
 Discipline                                                               teacher
 language                                                                 designer

                                 nature/function
                                   of material




                                 computer

                                                      virtual
               teaching                              learning
                agent               ?              environment


  follow-up
                          monitoring
                                                             peers : virtual class
              Decision
               node

                     machine
                     feedback
                                   learner

                real-life    human tutor
                   tutor      feedback
                                                        scaffolding
                                                    (learner/designer)
       Multi-layer architecture
                                    Scenarios
                                                Comparing
                                                Situation / Scenarios
     Level 3 : Prediction agents


                              USE
                                                Situation features


 Level 2 : Clustering agent



                                                Situation representation

Level 1 : Factual Agents
                Conclusion
   Finding common grounds between Language
    Learning and Computer Sciences
   Three main key words :
        COMPLEXITY
        DECISION

        MODELING
                Future developments expected...
   This presentation : first phase of a large-scale project

   Enrich this research thanks to new collaborations on (especially) :
        Building a model of Human-Machine relations in a mixed tutoring system
        Learner follow-up and guidance
        Peer-to-peer learning


   For further contact and cooperation

                           jean-claude.bertin@univ-lehavre.fr
                               patrick.grave@univ-lehavre.fr

                                  Presentation at :
               http://perso.wanadoo.fr/jean-claude.bertin/SiteBertin.htm

				
DOCUMENT INFO
Shared By:
Categories:
Stats:
views:51
posted:10/17/2008
language:English
pages:16