Q-Mapping: Learning of Interpretation for User Operation Using Action Value Function

Abstract

Conventional user interface operation methods have been designed with the aim of improving user evaluation on average with one single design. However, if the interpretation for the operation is not fixed and the system adapts to the user's operation and can make personalized interpretations, there is a possibility that more people can easily use the system. In this study, we propose a Q-Mapping algorithm that adaptively obtains the interpretation for the user operation by using the action value function (Q-value) in reinforcement learning. We conducted a case study in which a user operates a system with Q-Mapping, and examined the user's response to consider the requirements of a system that can adapt to user operations.