Formal Methods for Computer-Aided STEM Education

Providing personalized and interactive education (as in one-on-one tutoring) remains an unsolved problem for standard classrooms. The arrival of Massive Open Online Courses (MOOCs), while having provided a unique opportunity to share quality instruction with massive number of students, only exacerbates this problem with an even higher student to teacher ratio. We believe that automated intelligent tutoring can play a revolutionary role in this context. In this article, we motivate four problem definitions, namely problem generation, solution generation, feedback generation, and content authoring in the context of intelligent tutoring. We describe how formal methods can play a useful role in addressing these problems. We present some recent results that have been applied to a variety of STEM subject domains (including logic, automata, programming, arithmetic, algebra, and geometry) as illustrative examples.