A multi-agent framework to support user-aware conversational agents In an e-learning environment

Abstract

E-learning systems based on a conversational agent (CA) provide the basis of an intuitive, engaging interface for the student. The goal of this thesis is to propose an approach to improve the way that students interact with conversation-based e-learning applications. It attempts to do this through three contributions. The first is the design of an adaptable agent-based framework for improving interactions with conversation-based learning applications. The second contribution is to put forward a new approach to detecting user engagement based on real-time detection of conversational behaviour using the ongoing transcript of the interaction. The last contribution is to validate the approach by using volunteer students to test a proof-of-concept implementation of the framework. The observational and self-report data collected from the student testing provides new insights into how student interact with, and what their priorities are in evaluating, a pedagogical CA. This has implications for future development and research.