Intelligent Tutoring System (ITS) for Pre-University Mathematics

Abstract

Our goal is to develop an adaptive learning system which is tailored to the specific curriculum of high school mathematics taught in Switzerland.

An adaptive learning system, also known as Intelligent Tutoring System (ITS), must be capable of managing learning paths adapted to the users by monitoring their activities, inferring their needs and preferences, but most importantly building on top of the users’ knowledge to continuously facilitate the learning. Creating an efficient ITS is of course a very challenging task, given that it is almost impossible to develop an adaptive learning system which is applicable across subjects or even across age groups within the same subject.

In order to create an ITS which is tailored to the needs of high school students who are learning pre-university mathematics, we want to focus on developing an AI-powered knowledge tracing model. For adaptive learning systems, models of student knowledge have in fact become an essential component in tracking student behaviour within ITS.

In this project, we want to focus our knowledge tracing study around predicting and analysing two fundamental states of the learning process: blocking states and learning moments.

One fundamental aspect of an ITS is the ability to act upon instances where the learner needs help. These are the so-called blocking states, or momentary situations in which the learner cannot progress in completing an exercise. Additionally, it is very important to understand when a new piece of knowledge is acquired. Being able to infer the probability that a student learnt a skill during a specific step would allow us to study the differences between gradual learning and the eureka moments, when a skill is suddenly understood.

We are currently in the process of building our own ITS focusing on some specific mathematical problems with the help of Taskbase from the Swiss EdTech Collider, who already have extensive knowledge and experience in developing adaptive learning platforms. Once a first prototype has been built, we will be working together with PHBern to connect with the high school network within the Canton of Bern in order to conduct in-school trials and collect data for our study.