Machine Learning Contributions to a Guided Discovery Tutoring Environment for Chinese Characters

This paper presents an example of how Machine Learning tools can effectively contribute to Guided Discovery Tutoring Environments. MEMOCAR is that kind of environment, developed to evaluate the validity of a cognitive hypothesis concerning the teaching of Chinese characters to non native speakers. This hypothesis states that an early familiarization with Chinese character features is a pre-requisite for an efficient learning of characters. Three main types of activity are identified in MEMOCAR: familiarization, cooperative discovery of similarities between characters, and drilling exercises. Familiarization is supported by an exploration of hyperdata whilst cooperative discovery and exercises’ diagnosis make use of tools that learn concepts inductively from a set of examples. They exemplify how Machine Learning tools can offer new alternatives to the complex problem of making tutoring systems more cooperative.