Predictive, accessible web automation: a longitudinal study

The problem of efficient, usable non-visual web access remains inadequately solved, despite its critical importance to a significant number of visually-impaired existing and potential Web users. Web automation, a process of automating browsing actions on behalf of the user, has the potential to significantly improve the usability and accessibility of non-visual web browsing. Automation Assistant is an accessible web automation system designed specifically for this purpose. Until now, however, Automation Assistant was only evaluated in a short-term, controlled experiment. In this paper we report preliminary results on a longitudinal study designed to explore the long term practicality of the approach to accessible web automation taken by Automation Assistant. The study involves a single participant and can be considered a pilot of a future experiment involving a large number of participants and benefiting from the lessons learned during the pilot.