A Data Mining Approach to Intelligent Tutoring
Ivon Arroyo, Hasmik Mehranian, Beverly P. Woolf Journal of Educational Data Mining (JEDM). In Press
We describe pedagogical and student modeling based on past student interactions with a tutoring system. We model student effort with an integrated view of student behaviors (e.g. timing and help requests in addition to modeling success at solving problems) and parametrize an algorithm for adaptive tutoring that heavily depends on students data and estimations of problem difficulty. We argue that methods based on this integrated and empirical view of student effort at individual items accurately represent the real way that students use tutoring systems. This integrated view helps to discern factors that affect student behavior beyond cognition (e.g., help misuse due to meta-cognitive of affective flaws). We specify parameters to the pedagogical model in detail, and specify a variety of tests to verify that estimates of those parameters are accurate, and become more accurate as new data arrives to the system.