Avaliação automática de acessibilidade em RIA

Abstract

Agradecimentos À professora Dra. Renata Pontin de Mattos Fortes, por todo seu apoio durante todo o projeto; aos colegas e amigos, por suas valiosas contribuições e enriquecimento da tese; aos pais, pela compreensão e apoio presentes em todos os momentos; a Rejane Kiyomi Furuya, pelo amor; e a Deus que iluminou meu caminho durante esta caminhada.ii Often it appears that there is no better way to solve a problem than to try all possible solutions. This approach, called exhaustive search, is almost always slow, but sometimes it is better than nothing. Ian Parberryiii tese contribuem para o processo de Engenharia Web de aplicações ricas de Internet acessíveis. Palavras-chave: Acessibilidade na Web, Engenharia Web, ARIA, RIA e Web 2.0. v Abstract W ith the increased popularity of the Web 2.0 and RIA -Rich Internet Applications, web applications rely more and more in Ja-vaScript to implement richer and more complex interaction mechanisms in the Web platform. Since these mechanisms, frequently, generate dynamic updates to the DOM -Document Object Model structure of a webpage and require visual perception of users to notify them about these changes in the interface, users that interact with the Web through usage of Assistive Technologies such as a screen reader are not capable of correctly identifying and interacting with the interface components built with these technologies -the widgets. In this context, the WAI -Web Accessibility Initiative created the ARIA -Accessible Rich Internet Applications specification which describes the use of properties that add semantics to elements that compose a widget, allowing Assistive Technology to priorly identify the behavior of a widget and inform the user about changes that might be made to the DOM structure of the webpage. This thesis' goal was to elaborate automatic evaluation strategies for accessibility requirements of the ARIA specification. Three evaluation strategies were elaborated following the research-action methodology with the conduction of three cycles of the activities of planning, acting, describing and evaluating. The strategies were implemented based in Acceptance Tests and verifications on specific details of the technology that compose web applications, considering specifically blind users interaction patterns through the use of screen readers. Each strategy was separately validated and the results show tendencies that the strategies were capable of correctly evaluating the behavior of web applications, considering their conformance with ARIA recommendations for blind users. The development and validation of the evaluation strategies also presented contributions by: including user interaction models in the automatic evaluation approaches and presenting a survey on web applications and JavaScript toolkits conformance rates with the ARIA specification. Thus contributing to the Web Engineering process of ARIA.