Developing an Intelligent Personal Tutor to Support EFL Learning

Abstract

Intelligent personal assistants are critical to facilitate EFL learning efficiency. This study developed an intelligent personal tutor called Blingbang on Microsoft's Bot Framework, aiming to address students' puzzles and evaluate speech and writing from a pedagogical perspective. By following the goal-oriented principle, it was embedded some functional modules and a specific knowledge base to understand user's utterance, detect intent within the EFL domain, decide dialogue policy, and ultimately respond feedback to users. The validity of this system was verified by an empirical study. Results show the favorable performance, especially in answering questions, made participants satisfied. The behaviors users adopted to resolve communication breakdowns include rephrasing questions, changing keywords, or eventually turning questions to the teacher.