School of IT Technical Report TARO: TEACHING WITH AN AUTOMATICALLY RETRIEVED ONTOLOGY

Ontologies are used in an increasing number of roles within Intelligent Tutoring Systems. In this paper, we describe TARO (Teaching with an Automatically Retrieved Ontology), a system which explores a role for ontologies that are externalised and used, in conjunction with a learner model, as a basis for independent learning. A novel and important feature of our approach is that the ontology is built automatically from a dictionary designed for people to read: the same dictionary is used, in conjunction with the ontology built from it to provide both a mechanism for explaining the meaning of the parts of a learner model and as a new learning source. We report a small-scale qualitative evaluation with five users, who self-assessed their knowledge before and after using TARO and who used TARO to assist them in refining their knowledge, externalising this in a concept-map. The study gives promising results both in terms of learner perceptions of their knowledge and with all learners demonstrating improved understanding in the concept-mapping task. TARO offers promise as a tool that can be used in conjunction with any teaching system that builds a learner model which can be shared with the learner.