Towards an Adaptive Pedagogical Agent in a Reading Intervention Using Reinforcement Learning

Learning to read can be difficult, but is a crucial skill that is needed for the whole lifespan. Deficiencies in reading capabilities can often be traced back to the initial attempts to learn reading during primary school. While there exist a number of analogue and digital tools to support reading acquisition for primary school children, they often don't account for their heterogeneity. This paper presents work in progress on integrating an adaptive pedagogical agent in a mobile reading app designed for second graders with reading difficulties. We apply Reinforcement Learning to adapt the feedback behavior of the agent to each child's learning needs and elaborate on our methodological approach. With it, we contribute a novel concept for adapting a pedagogical agent's feedback behavior within the critical context of reading development.