Computer Vision Based Assessment of University Websites' Accessibility and Inclusiveness

One of the directions for the development of assistive technologies for people with disabilities (visually impaired, deaf and other categories) is the adaptation of university websites to their needs, often used in an inclusive form of education. According to the law, educational websites in Russia must have dedicated versions for the visually impaired. Our research objective is to evaluate accessibility of the websites and the quality of the versions for visually impaired, mainly automatically, and also to identify the best way to implement such solutions or create accessible websites in general. We have developed a software system for collecting metrics from the two versions of websites: the general one and the one for the visually impaired users. Quantitative metrics based on the HTML code of homepages and based on visual analysis of the homepages screenshots are compared. Metrics are also collected via manual testing. The data was collected for 572 universities websites and was made openly accessible. Correlation analysis of the metrics was carried out. The quality of the versions for the visually impaired users was assessed, recommendations were made for the technical implementation and a standard set of settings was proposed. We collected statistics of used ready-made solutions, as well as of website development tools. We also performed an analysis of the readability of the Russian text in comparison with English on the websites of universities. The paper presents current issues in the field of accessibility of educational sites in Russia, and proposes certain means for their evaluation and guidelines for development of accessible websites.