TOOLBOX.ACADEMY: CODING & ARTIFICIAL INTELLIGENCE MADE EASY FOR KIDS, BIG DATA FOR EDUCATORS

Abstract

Society has reached the point where there is considerable awareness about the need of teaching coding skills at the schools. Both, political leaders and educational systems start taking positions in this strategic new arena, but they lack from much of the expertise and necessary tools about how to teach this complex (but also intuitive to children and youngsters) field of knowledge. In this context, ToolboX Academy is a programming platform which subscribes a slightly new approach to attack the problem, in four differential aspects: 1) While most attention focuses on block-based approaches (Scratch and Code.org being the most popular initiatives), ToolboX embraces coding as it works for developers: text-based programming; 2) ToolboX embeds mainstream programming languages, which are widely used, are supported by big communities, are weakly typed, and closer to human natural language, like it is Javascript, for web applications, and GNU Octave, in engineering and scientific research; 3) it provides user-oriented problem-based learning, designed upon educational research results; and 4) it integrates curricular subjects in learning to code. In this paper, we review the design constraints taken into account to develop ToolboX Academy, and how it has been tested in the classroom. Some features are particularly stressed, like the use of open formats for content representation (definition of tasks or problems); the design of a simplified environment to represent most CS concepts, from basic loops to complex graph traversal strategies; and the use of learning analytics based on big data algorithms that can help educators and managers to learn from the interaction of their students with this environment.