Personalization
and adaptivity are invaluable features in modern intelligent tutoring
systems. However, it is not easy for software to decide how to adapt
its performance to meet individual student's needs, partially because
usually the uncertain nature of the correspondences between what we
observe and what we need to know for making the best decisions for
intelligent tutoring. For instance, students that are competent in the
subject matter being tested can make unintentional mistakes. It is also
difficult to infer students' learning needs by observing how they
explore educational web sites. Building operational models that can
capture and reason about the uncertainty is thus important for
intelligent tutoring that aims at meeting individual student's needs.
Bayesian
networks offer a great formalism for modeling under uncertainty, and
have been adopted in many applications including intelligent tutoring
in the past decade or so. This satellite workshop of the Eighth
International Conference on Intelligent Tutoring Systems aims at
providing a forum for interested researchers and practitioners to
discuss and share methodologies and applications of Bayesian networks
in all aspects of intelligent tutoring.
Topics
of interest include, but are not limited to:
- Building networks: using expert's opinion, learning from
data, theoretical models, etc.
- User modeling: diagnosis of knowledge/skills,
multi-layered models, affective computing, emotional monitoring, etc.
- Task selection algorithms: curriculum sequencing, testing
knowledge, problem selection, etc.
- Displaying results to users: inspectable models, open
models, etc.
- Reusability, maintenance and upgrade of systems
- Practical barriers to adoption
|
