Nguyen-Thinh Le, Niels Pinkwart
Problems for which multiple solution strategies are possible can be challenging for intelligent tutors. These kinds of problems are often the norm in exploratory learning environments which allow students to develop solutions in a creative manner without many restrictions imposed by the problem solving interface. How can intelligent tutors determine a student’s intention in order to give appropriate feedback for problems with multiple, quite different solutions? This paper focuses on improving the diagnosis capabilities of constraint-based intelligent tutors with respect to supporting problems with multiple possible solution strategies. An evaluation study showed that by applying a soft-computing technique (a probabilistic approach for constraint satisfaction problems), the diagnostic accuracy of constraint-based intelligent tutors can be improved.
The final publication is available at Springer via https://doi.org/10.1007/978-3-642-30950-2_42.