A dynamic bayesian network for inference of learners' algebraic knowledge
Henrique M. Seffrin, Geiseane Rubi, Patrícia A. Jaques · 2014
An Intelligent Tutoring System (ITS) is an educational software that provides personal assistance for students, allowing them to learn at their own pace. This is possible because ITSs are able to map the learners' knowledge to create a student model. Most of the tutors use a Bayesian Network (BN) to perform this task, due to their ability to deal with uncertain data. However, classic static BNs are unable to model data, such as the student's knowledge, that changes over time. Dynamic Bayesian Networks (DBN) are an interesting option in this case, because they are a special type of BN that reasons over time. This paper presents an architecture of DBN that aims at inferring student's algebraic knowledge. This network was constructed based on a concept map, which was developed with the goal of structuring the algebraic knowledge, i. e. defining relationships among concepts. The proposed DBN was evaluated with the help of an expert in order to verify the ability of the network to predict the student's knowledge on the application of operations to solve 1st degree equations. This DBN is being integrated into an web-based ITS for algebra learning.