SBIR Phase I: Making Web Applications Accessible

The broader impact/commercial potential of this project include the development of a novel assistive technology that will make dynamic web applications accessible for people with vision impairments. The proposed technology will empower blind people to utilize the immensely popular web applications for social networking, email, online banking, and travel with the same ease-of-use as is experienced by sighted people. The resulting higher productivity of screen-reader users will lead to improved access to education and employment. The proposed technology will be commercialized as part of a screen-reader application for people with vision impairments, a population of up to 25 Million people in the U.S. and a quarter of a million worldwide. The proposed innovation can also be utilized in mainstream technologies that enable web automation and web crawling. This Small Business Innovation Research (SBIR) Phase I project will investigate the feasibility of making widgets in dynamic web application accessible for blind people. The dynamic nature of modern day web applications has become a serious barrier for blind people who use screen readers to access the Web. Unfortunately, while screen readers can usually narrate all the individual HTML elements in a webpage, they do not recognize widgets, which are a collection of HTML elements forming discrete user interface objects with which the user can interact, e.g., HTML windows, dropdown menus, calendars, suggestion boxes, etc. Inaccessibility of widgets in a web application can make the entire web applications unusable. This project will develop the Widget Accessibility Model that will employ machine learning techniques to identify and classify dynamic widgets in web pages and supply them with generic accessible interfaces. The Model will succinctly represent an assortment of widgets, of varying degrees of complexity, as objects associated with (inheritable) properties and interaction behaviors. WAM will drive the automatic detection of widgets used in a web application and stitch together their composite behavior from their components at runtime. The resulting Model will be incorporated into a commercial screen reader.