Scratch-NB: A Scratch Extension for Introducing K-12 Learners to Supervised Machine Learning

Abstract

Artificial Intelligence has rapidly captured-again-the attention of practitioners and the CS Education community. Several initiatives have been spawning to identify effective ways to introduce learners at college and high-school levels to the basics of AI. In this context, we developed Scratch-NB, an extension to the Scratch programming language to equip K-8 learners with foundational tools for developing a Naive-Bayes classifier, explicitly transparenting its internal components. We opted to use Scratch due to its popularity within the CS4All movement and its predominance in schools, and extend the language with a simple implementation of the Naive Bayes classifier as a gateway for introducing basic notions of supervised learning and AI. In this paper, we report the design rationale of our tool and our experience of using it in an informal workshop with children aged 10-12 with no prior knowledge of computational thinking, coding, or AI. The obtained results show that Scratch-NB showed high levels of acceptance among the participants, raised interest in AI, and effectively provided basic foundations, particularly when comparing Scratch-NB with other state-of-the-art tools targeted to this learner group. We argue that our tool has the potential to illuminate further practical and research endeavors when exploring how to effectively introduce younger populations to the foundational notions of AI in practical contexts.