Model-based cognitive diagnosis of students' test performance in an E-learning environment

Abstract

Cognitive diagnosis is the process of inferring a cognitive state from observations of performance. This paper considers the problem of cognitive diagnosis as an instance of model-based diagnosis, as studied in artificial intelligence for many years. The model-based cognitive diagnosis we present runs on a model of students' courses in terms of knowledge items that they may learn, tests them and helps them to understand their faults in cognition, and thus improve their learning performance in an E-learning environment. To do so, courses are formally defined as set of knowledge items with requirement constraints, knowledge items are associated with a set of exam questions. Moreover, diagnostic algorithms are used to help a student understand what knowledge item within a course the student does not master, the root reason of his/her test errors, and the recommendations like what should be done next. Experimental results show that the group of students with such understanding can improve their testing performance greatly in an E-learning environment. © 2011 Springer-Verlag.