Goal-Oriented Browsing: The Next-Generation Web Accessibility Technology for People with Visual Impairments.

This project investigates the feasibility of goal-oriented browsing. Goal-orientation is realized by enabling usable automation of browsing tasks. This technology helps users to stay focused on high-level browsing tasks, while the browser does low-level operations such as clicking on links, filling forms, etc., that are necessary to accomplish the tasks. To enable automation, a computational model is designed to encode users’ browsing history as sequences of browsing actions. Algorithms are developed to analyze this history and infer macros, which are useful sequences of browsing actions that users could replay to automate browsing tasks. To make the model resilient towards changes in structure and content of web pages, robust addressing techniques with self-repair capability are developed. End-user interfaces are designed for usable interaction with the model and for model validation. Finally, automated experiments and extensive user studies with blind subjects are conducted to determine project feasibility.