It is now accepted that a learning activity could not be
scenarised/scripted in details: the learning situation is contextualised
and the pedagogical scenario, as well as the interaction, the software
environment, etc., should be adapted to the effective context of learning.
Some TEL environments delegate the adaptation decision making to humans,
as a tutor, a teacher, or the learner himself. Some others works explore
the possibility for the system to dynamically adapt its functioning and
its configuration, as in Intelligent Tutoring Systems or in Adaptive
These decisions of adaptation are based on the interpretation of the
learning context. Precisely, both humans and the TEL system need
information on the learning situation for taking the decision of
adaptation. This information could be, for a part of it, extracted and
calculated from tracks of the learning activity the TEL system collects.
Such information is usually named an indicator, which can be defined
from two different points of view:
- an indicator "stems from what is important" from a pedagogical and/or
psychological point of view;
- an indicator is a numeric or symbolic representation of what is
important, a significant (structured or not) datum which helps supports
the analysis of the learning activity.
We have defined a language for modelling such indicators. This language
(UTL: Usage Tracking Language) is designed for the capitalisation, the
reusing and the sharing of indicators. It allows their calculation at
real-time or after a learning session.
The aim of the study proposed here is to define a process and a
formalism for characterising the context where a decision of adaptation
is made. Part of this context is a specific set of indicators, but we
want to go farther in its description: the initial pedagogical aims, the
kind of interaction allowed, the nature of the activity, the learner(s)'
performance, etc. Finally, the context description, including the set of
indicators, could be linked to the decision of adaptation and then,
capitalised in such a way it can be reused.
The study will be made on the framework of a TEL system we have
developed for learning object oriented programming. In this system,
learners have to resolve JAVA programming exercises and their activity
is automatically tracked by the system. During or after the learning
session, the teacher could (1) visualise the code performed by the
students, (2) annotate it and/or (3) replay the learners' activity
sequences. Moreover, during the activity, the teacher could have to
intervene on the pedagogical situation by adapting the sequence of
exercises proposed, by providing hints or feedback, and so on.
Helped by corpus collected during previous tests of this TEL system, and
by the results of a test planned for the beginning of the next year
(Feb. 2010), the post-doc will have to propose some features of UTL
dedicated to the modelling of the context of the tutor's acts in this system. Depending of the nature of the results, the post-doc would be
able to propose a generic formalism for capitalising both these context
representations and the tutor's acts.
The candidate should have grounding in computer science, especially in
OO programming, in XML and in knowledge management.
English speaking is required, French speaking is better.