Judging Granularity for Automated Mathematics Teaching
Marvin R. G. Schiller, Christoph Benzmüller, Ann van de Veire · 2006
In proof tutoring, human maths tutors are observed to reject correct proof steps if they are not at the expected level of granularity, i.e. if they are too detailed or too coarse-grained. We investigate how the judgments on granularity as observed from human tutors can be automated with the help of automated reasoning techniques. We evaluate our approach with data collected in an empirical study.