A Formal Methods Approach for Predicting How Users will Utilize System Features

Abstract

Model checking is increasingly being used with task analytic behavior models to prove whether models of human-interactive systems are safe and reliable. Such methods could be used to predict how different types of users will choose to use system features. However, existing methods focus on modeling the full space of possible human behaviors without considering how users will choose to navigate this space. In this work, we present a new approach that enables model checking to predict how different types of users will use features of an interactive system by employing a novel combination of task analytic modeling and utility theory. This paper presents this method and illustrates its power with a smart thermostat application. The results of the application analysis and its implications for future research are discussed.