With the emergence of Open Education Resources (OERs), educational content creation has boomed to a whole new scale. For AI-driven OER platforms such as X5GON, scalable quality assurance is highly impactful. As the quality of OERs could vary significantly, the quality assurance process plays a key role in maintaining a high-quality learner experience when using OERs. Managing this problem at large scale demands automating the quality assurance process as a whole or in parts. Prior research on automating quality assurance in the context of education is surprisingly scarce. We present our ongoing work of building Quality Assurance Models, a novel approach to using crossmodal features from OERs to predict quality using machine learning. While developing quality models, we extended our search beyond the education domain to identify features that indicate content quality that can be categorised into five main quality verticals. In the future, these features will enable us to leverage scalable quality assurance on OERs of different modalities. Furthermore, quality features will also become useful in learning quality preferences of learners when recommending content. Altogether, the expected outcomes of this research will mark a significant step towards Automatic, Scalable Quality Assurance in Open Education.
The first paper outlining the foundations of the what features can be used to automatically evaluate quality of educational resources at scale. In Proceeding of, 2020.