Abstract
Lifelong learning recommendation systems face unique challenges. Not only must they provide relevant and engaging material but also must consider background knowledge, changing interests and the materials novelty which must be maintained ideally for a lifetime.
TrueLearn, a family of Bayesian algorithms tackles this head on by modelling learners’ knowledge and interests from their interactions with educational resources to provide predictions about future engagement. Requiring minimal data, it offers a transparent solution that respects the privacy of its users. The development of the TrueLearn library aims to refactor the existing codebase whilst adding visualisations to help learners study more effectively.
The overarching goal is to make its API publicly available, allowing other developers to use its powerful algorithms. With the library being officially released at the end of the project under the MIT license, future work can begin immediately on implementing its Machine Learning solution into online platforms such as X5Learn.
Portfolio Video
Project Details
Project Timeline
Our Team
Karim Djemili
Team Lead, Core Developer
Tim Qiu
Lead Developer, Researcher
Denis Elezi
Visualisations Developer, Pre-processing Lead