CS-BKT: introducing item relationship to the Bayesian knowledge tracing model

ABSTRACT

Bayesian knowledge tracing model (BKT) is a typical student knowledge assessment method. It is widely used in intelligent tutoring systems. In the standard BKT model, all knowledge and skills are independent of each other. However, in the process of student learning, they have a very close relation. A student may understand knowledge B better when he masters knowledge A. Therefore, this work introduces a new student model based on BKT. It takes the relationship between knowledge into account. By doing this, the new model proves higher prediction accuracy and performs better. Then this paper uses the new model to make a cognitive diagnosis according to students’ test scores. The diagnostic results can help teachers provide personalized guidance to students and improve teaching efficiency.

KEYWORDS:

• Knowledge tracing
• student modeling
• cross skill

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

This work was sponsored by Peak Discipline Construction Project of Education at East China Normal University.

Notes on contributors

Lingling Meng

Lingling Meng is an associate professor of Department of Education Information Technology in East China Normal University. Her research interests include personalized learning, learning analytics and knowledge engineering.

Mingxin Zhang

Mingxin Zhang is a master student in East China Normal University. Her major is Educational Technology.

Wanxue Zhang

Wanxue Zhang is a master student in East China Normal University. Her major is Educational Technology.

Yu Chu

Yu Chu is a master student in East China Normal University. Her major is Educational Technology.