MobileVisFixer: Tailoring Web Visualizations for Mobile Phones Leveraging an Explainable Reinforcement Learning Framework

We contribute MobileVisFixer, a new method to make visualizations more mobile-friendly. Although mobile devices have become the primary means of accessing information on web, many existing are not optimized for small screens and can lead frustrating user experience. Currently, practitioners researchers engage in tedious time-consuming process ensure that their designs scale different sizes, toolkits libraries provide little support diagnosing repairing issues. To address this challenge, MobileVisFixer automates mobile-friendly visualization re-design with novel reinforcement learning framework. inform design we first collected analyzed SVG-based identified five common addresses four these issues single-view Cartesian linear or discrete scales by Markov Decision Process model is both generalizable across various fully explainable. deconstructs charts into declarative formats, uses greedy heuristic based Policy Gradient methods find solutions difficult, multi-criteria optimization problem reasonable time. In addition, be easily extended incorporation algorithms data visualizations. Quantitative evaluation two real-world datasets demonstrates effectiveness generalizability our method.