Tailoring Feedback by Correcting Student Answers

Constraint-based models [7] represent the domain by describing states into which a solution may fall, and testing that solutions in a state are consistent with the problem being solved. Constraints have a relevance condition (which defines the state) and a satisfaction condition (which tests the integrity of the solution.) In this paper we present a purely pattern-based representation for constraints, and describe a method for using it to generate correct solutions based on students' incorrect answers. This method will be used to tailor feedback, by presenting the student with correct examples that most closely match their attempts.