I would classify this as a relatively basic approach to content recommendation, though it is of course useful to select problems of appropriate difficulty for people who are struggling in a course. And it's pretty clear and offers an instance of the idea applied in EdX. This paper (16 page PDF) describes a collaborative recommendation mechanism based on people in the same class (or type of class) who are the 'nearest neighbour' based on a test of similarity. the idea is that 'people who like me struggled with this type of content were successful with this class of problems'.
Today: 1 Total: 15 [Direct link] [Share]
