Determining Problem Selection for a Logic Proof Tutor

When developing an intelligent tutoring system, it is necessary to have a significant number of highly varied problems that adapt to a student’s individual learning style. In developing an intelligent tutor for logic proof construction, selecting problems for individual students that effectively aid their progress can be difficult, since logic proofs require knowledge of a number of concepts and problem solving abilities. The level of variation in the problems needed to satisfy all possibilities would require an infeasible number of problems to develop. Using a proof construction tool called Deep Thought, we have developed a system which chooses existing problem sets for students using knowledge tracing of students’ accumulated application of logic proof solving concepts and are running a pilot study to determine the system’s effectiveness. Our ultimate goal is to use what is learned from this study to be able to automatically generate logic proof problems for students that fit their individual learning style, and aid in the mastery of proof construction concepts.