Christophe Després

A plan recognition process, based on a task model, for detecting learner's erroneous actions

When a tutoring system aims to provide learners with accurate and appropriate help and assistance, it needs to know what goals the learner is currently trying to achieve, what plans he is implementing and what errors he is making. That is, it must do both plan recognition and error detection. In this paper, we propose a generic framework which supports two main issues (i) the detection of learners unexpected behavior by using the Hollnagel classification of erroneous actions and (ii) a recognition process based on a task model METISSE that we propose. This model, which is used to describe the tasks the learner has to do according to pedagogical goals, allows learners unexpected behavior to be detected. The solutions proposed are generic because not dependent on the domain task, and they do not relate to a particular device.

Supporting learners' activities in a distance learning environment

In this paper, we describe the social context of a distance learning situation between teachers and learners and its cooperative learning activities. The objective is to make learners cooperate through project pedagogy. The project pedagogy is based on experiential learning and consists of learners working together in a team. During their cooperative activities, learners are supported by the system in a way that depends on the difficulties detected and the support style. When the system is no longer able to provide useful advice, learners are supported remotely by the teacher. We describe a model of multi-agent architecture providing this support. This architecture is made up of various communicating agents distributed across a network like the internet. We have developed a distance learning environment, based on this multi-agent architecture, in the field of educational robotics. We give an example to show how teachers can support learners remotely by modifying their interface or by acting on the distant physical device (a micro-robot).

Supporting Teacher's Self-Regulation Process

In this paper, we focus on teachers instrumentation. Previous works on this field mostly propose models and tools to teachers in order to follow the learners activity. By using indicators, we intent to go further in this research area in order to assist teachers in their activities. Indeed, giving a feedback to teachers about their works thanks to indicators, will enable them to measure their effectiveness (i.e if the effects of their actions meet their original intentions). Depending on this latter, teachers can assist learners and adapt their pedagogical strategies by trying to improve the positive effects of their interventions.

Proposing the underlying causes that lead to the trainee's erroneous actions to the trainer

When a trainer aims to provide trainees with appropriate help and assistance, she/he needs to know what errors the trainee is making and furthermore what causes lead to these errors. In this paper, we propose a mechanism which infers the underlying causes that lead to the production of the trainees erroneous action in order to support the trainer in her/his monitoring activity. This mechanism is based on and uses CREAM, a second-generation method of Human Reliability Analysis (HRA). We implement this mechanism by using Dempster-Shafers theory.

Engineering of Open Learning Scenarios - The Case of Hop3x Learning Scenarios

Within an educational context, adaptation could improve the learning’s quality. Many researches are done in the learning-situations adaptation field. The Educational Modeling Languages (EML) and tools provided currently to the teacher for a learning design, as preexistent means in our sense, remain useless by practitioners. In our work we aim to support the practitioner teacher to design and adapt his/her learning scenarios which are by nature open, influenced by the context in which they are executed. The approach adopted is the combination between Model-Driven Engineering (MDE) and Domain-Specific Modeling (DSM). In this paper we present a model of open learning scenario. So we take the learning scenarios of Hop3x as a case study.