Using Bayesian belief networks for the automated assessment of students' knowledge of geometry problem solving procedures

Authors: Roccetti M.; Salomoni P.

Source: Journal of Experimental & Theoretical Artificial Intelligence, Volume 10, Number 2, 1 April 1998 , pp. 145-178(34)

Publisher: Taylor and Francis Ltd

Key:

- Free Content

- New Content

- Subscribed Content

- Free Trial Content

Abstract:

A probabilistic student model is proposed that is suitable for representing the uncertainty regarding the estimate of the student's knowledge of Euclidean geometry problem solving procedures. The probabilistic student model is based on the use of Bayesian belief networks. Several efficient computer-assisted procedures are presented that allow both the automated assessment of the Bayesian belief network's topological structure and also the automated computation of the conditional probability matrices. Numerical results (derived from real-world experiments) are also reported that prove the effectiveness of the use of these automated procedures in the process of construction of the student model.

Keywords: PROBABILISTIC; STUDENT; MODELLING; BAYESIAN; BELIEF; NETWORKS; INTELLIGENT; TUTORINGSYSTEMS; COMPUTER-ASSISTED; EDUCATION; PLAIN; EUCLIDEAN; GEOMETRY