Introduction
Free-text Computer Assisted Assessment has not received much
attention till the nineties, apart from the works by Page in the
mid-sixties. Recently, given the limitations of test questions such as
Multiple-Choice Questions of fill-in-the-blank, there is an increasing
interest in this field. The final aim of these systems is not to
substitute the teacher in assessing exams, but to offer more
possibilities for self-assessment to students using e-learning
systems.
Some techniques that have provided good results are:
- Statistical analyses: word frequencies, Vector Space Model, n-gram
co-occurrence, etc.
- Linguistic analyses: syntactic, rhetorical analyses, etc.
- Machine Learning procedures: neural network, Bayesian networks,
etc.
Our work in this area is the design and deployment of a free-text CAA
system that combines n-gram co-occurrence metrics with partial
syntactic analyses and adaptation to user profiles. This system has
been integrated with the e-learning system Tangow/Wotan.
Publications
Click here to see our publications on
free-text CAA.
On-line demos
Coming soon...
Some external links
|