Drop-Down Menu Widget Identification Using HTML Structure Changes Classification

Abstract

Widgets have been deployed in rich internet applications for more than 10 years. However, many of the widgets currently available on the web do not implement current accessibility design solutions standardized in ARIA (Accessible Rich Internet Applications) specification, hence are not accessible to disabled users. This article sets out an approach for automatically identifying widgets on the basis of machine-learning algorithms and the classification of mutation records; it is an HTML5 technology that logs all changes that occur in the structure of a web application. Automatic widget identification is an essential component for the elaboration of automatic ARIA evaluation and adaptation strategies. Thus, the aim of this article is to take steps toward easing the software-engineering process of ARIA widgets. The proposed approach focuses on the identification of drop-down menu widgets. An experiment with real-world web applications was conducted and the results showed evidence that this approach is capable of identifying these widgets and can outperform previous state-of-the-art techniques based on an F-measure analysis conducted during the experiment.