Three web accessibility evaluation perspectives for RIA

With the increasing popularity of Rich Internet Applications (RIAs), several challenges arise in the area of web accessibility evaluation. A particular set of challenges emerges from RIAs dynamic nature: original static Web specifications can change dramatically before being presented to the end user; a user triggered event may provide complete new content within the same RIA. Whatever the evaluation alternative, the challenges must be met. We focus on automatic evaluation using the current WGAG standards. That enables us to do extensive evaluations in order to grasp the accessibility state of the web eventually pointing new direction for improvement. In this paper, we present a comparative study to understand the difference of the accessibility properties of the Web regarding three different evaluation perspectives: 1) before browser processing; 2) after browser processing (dynamic loading); 3) and, also after browser processing, considering the triggering of user interaction events. The results clearly show that for a RIA the number of accessibility outcomes varies considerably between those tree perspectives. First of all, this variation shows an increase of the number of assessed elements as well as passes, warnings and errors from perspective 1 to 2, due to dynamically loaded code, and from 2 to 3, due to the new pages reached by the interaction events. This shows that evaluating RIAs without considering its dynamic components provides an erroneous perception of its accessibility. Secondly, the relative growth of the number of fails is bigger than the growth of passes. This signifies that considering pages reached by interaction reveals lower quality for RIAs. Finally, a tendency is shown for the RIAs with higher number of states also exposing differences in accessibility quality.