Dr. Volker Sorge - Seminar Talk - 14th June 2016
Title: Making STEM Content Accessible in the Age of the Web
Sponsored by the ADAPT Centre and the Faculty of Engineering &
Computing Enhancing Performance Fund.
Making STEM content accessible for visually impaired or blind students has always been challenging. Much of the information is conveyed via formulas and graphics that need specialist knowledge to be explained correctly. While traditionally textbooks could be made accessible by providing students with recordings of expert readers, with more teaching content moving to the Web, traditional means of making teaching material in the sciences accessible are no longer feasible. Teachers can easily prepare their own content, custom handouts are produced on the fly, and students are expected to work through online material independently. This raises the danger that the accessibility gap for STEM subjects will actually widen in the age of the web.
I present my approach to making STEM material web accessible, which aims to eliminate the two most important obstacles to the production of accessible content: Relying on authors to produce accessible content and requiring readers to use special software or restricting them to a particular choice of platform to get access to content. Instead my work relies on automatic procedures for the recognition of content regardless of input format and its transformation into accessible and interactive web content using standard web technology, that is accessible to visually impaired readers regardless of their choice of platform or assistive technology. Since deep understanding of STEM content is often required to present it adequately, the most challenging part of the transformation process is the correct semantic interpretation of content and how it can be automated.
I shall present examples on how we make chemical diagrams and mathematical formulas fully automatically web accessible and interactive, that demonstrate the power of the approach. Although most of my work aims at producing web content and supporting screen readers, the techniques can also be exploited for the generation audio-tactile graphics, Braille output or 3D models.