Policies Selection for Pedagogical Agent Based on the Roulette Wheel Algorithm

Abstract

Pedagogical agents are computational entities that interact with users and facilitate learning opportunities. They usually need to be programmed to follow a set of commands for an effective personalized exchange of knowledge and tasks. In this study, we evaluate the effectiveness of a policy-based model and its level of satisfaction about the interaction without and with the pedagogical agent using the bio-inspired roulette selection algorithm. The approach is quantitative, with an exploratory and descriptive study. The results revealed that our agent achieved great acceptance among the users who rated it as intelligent, friendly, and reliable. It is evidenced that the agent can influence the attitude, perception, and behavior of the user to reach better self-regulated learning.