Rico and Beyond: A Mobile App Dataset for Interaction Mining and Machine Learning

Abstract

Rico is a publicly available dataset of user interfaces from 9.7k Android apps produced in a collaboration between Google and UIUC. The dataset contains screenshots, user interface hierarchies, and interaction traces from approximately 72k unique screens across these apps, and was collected using a hybrid approach that employed both crowd workers and automated techniques. This talk will briefly describe the dataset, how it was collected, and some projects that were built using it. The focus will be on projects done at Google, but also touch on what others have done with the dataset since it was released publicly in 2017.