Improving the Accessibility of Online Data Visualizations for Screen-Reader Users and Visualization Creators

Abstract

This dissertation introduces a novel interaction technique for screen-reader users to interact with and extract information from online data visualizations using voice-activated commands. To this end, this dissertation pioneers VoxLens, a multi-modal open-source JavaScript plug-in that--with a single line of code--improves the (1) information extraction experiences of screen-reader users with online data visualizations and (2) understanding and knowledge of visualization creators to make data visualizations accessible. I present versions of VoxLens and independent artifacts, including VoxEx--a system that enables screen-reader users to customize the information they consume from online data visualizations. These artifacts collectively enable these users to extract data from simple and complex online data visualizations, both holistically and granularly, in the manner they prefer. The artifacts also provide these users with information on data uncertainty. VoxLens increased their accuracy of information extraction by 164% and reduced their interaction times by 50% over conventional methods to consume information from online data visualizations. Additionally, I present five interventions that minimize creators' challenges with accessibility. This work provides empirical and artifact contributions to the domains of accessibility and visualization. The thesis of this dissertation is as follows: A multi-modal, customizable, and interactive JavaScript plug-in called "VoxLens" improves the experiences of screen-reader users in extracting information from simple and complex online data visualizations while also enhancing the knowledge of visualization creators to make online data visualizations accessible.