Adapting data table to improve web accessibility

Web table understanding is challenging for people with visual disability. They depend on screen readers to convey the table information. Screen readers present content linearly to users, but if the table is large, the user may have long forgotten the heading before the last row is read. Even in table navigation mode, it can still be confusing if the table is not marked up properly. Though there are guidelines for web developers to create accessible web tables, some authors may still not properly mark up the web tables. There are also lots of legacy web tables that are not designed with accessibility in mind. These unstructured web tables arouse a need for web accessibility improvements. Existing solutions mainly focus on interpreting tables by screen readers and providing guidelines to create accessible web table, so there is a research gap on how to adapt unstructured table to improve web accessibility. In this regard, we propose a method to extract the structure from these tables and re-organize them into multiple levels of abstractions so that the visually impaired users can access the tables level by level by selecting the corresponding option number. This has enhanced the table content understanding for people without visual perception and has greatly improved web accessibility of unstructured web table for both PC users and mobile users with visual disabilities.