Les déficients visuels rencontrent de nombreuses difficultés lors de leur navigation web avec un lecteur d’écran, d’autant plus que les interfaces web deviennent de plus enplus riches de par leurs interactions et leur densité et diversité informationnelles. Cette thèse confronte ainsi 2 approches de l’accessibilité : l’approche dite « exhaustive » quivise à assurer l’accessibilité normative - garantissant l’accès à l’information par l’application de normes - et l’approche dite « holistique » qui vise à assurer l’accessibiliténormative mais également l’accessibilité effective - garantissant l’utilisabilité de l’interface - par la prise en compte du contexte d’utilisation des utilisateurs.Les 2 premières études ont analysé l’activité des utilisateurs déficients visuels qui a donné lieu à un modèle de la navigation web de ces utilisateurs. Ce modèle permet deconcevoir une interface suivant l’approche holistique par la prise en compte de leurs besoins, notamment celui de filtrage des informations non pertinentes et redondantes. Lorsde 3 expérimentations, les déficients visuels ont été confrontés à l'approche exhaustive par la retranscription exhaustive des informations et à l'approche holistique par l’accèsdirect au contenu associé à l’action réalisée. Les résultats montrent le bénéfice substantiel apporté par la version holistique concernant l’allègement de la charge cognitive etl’utilisabilité de l’interface selon les 3 critères : efficacité, efficience et satisfaction. Ainsi, cette thèse fournit des arguments probants pour promouvoir l’approche holistique pourguider la conception d’interfaces et l’élaboration de solutions d’assistance pour cette population. / Visually impaired people face many difficulties when navigating on the web with a screen reader, especially as web interfaces become increasingly rich because of theirdynamic interactions and their informational density and diversity. This thesis compares two approaches to accessibility: the current approach called "exhaustive" which aims toensure normative accessibility - guaranteeing access to information through the application of standards - and the approach called "holistic" which aims to ensure bothnormative and effective accessibility - guaranteeing the usability of the interface - by taking into account the context of its use (difficulties, strategies, needs, etc.).The first 2 studies analysed the activity of visually impaired users resulted in a model of the web navigation of these users. This model allows for designing interfaces followingthe holistic approach by taking into account their needs, especially the filtering of irrelevant and redundant information. In 3 experiments, visually impaired users performedtasks in two versions of websites: a version based on the holistic approach that gives access directly to content associated with the user’s action and a version using theexhaustive approach that transcribes exhaustively all the information present on the screen. The results show substantial benefit provided by the holistic version as regards thecognitive load and the usability of the interface according to 3 criteria: effectiveness, efficiency and satisfaction. Thus, this thesis provides cogent arguments promoting theholistic approach in order to guide the design of interfaces and the development of assistance solutions for screen reader users.

Made available in DSpace on 2016-06-02T19:05:50Z (GMT). No. of bitstreams: 1 3725.pdf: 10739075 bytes, checksum: 66873da98206d03f71730ae408582888 (MD5) Previous issue date: 2011-05-30 / Financiadora de Estudos e Projetos / Web 2.0 allowed users more interactivity with Web applications. The so-called Rich Internet Applications (RIAs) have transposed the boundaries of simple interfaces built only in Hypertext Markup Language (HTML). Through the adoption of technologies that enable the creation of more advanced interfaces, RIAs resemble the appearance and behavior of desktop applications. On the other hand, the demand for software in Ubiquitous Computing, in which access to applications occurs anywhere, at any time and from different devices, has raised new challenges for Software Engineering. One of these challenges is related to the adaptation of the contents of an application to the numerous devices that can access it in distinct contexts. Given the diversity of devices, access networks, environments and possible contexts, providing applications that meet the peculiarities of each access device, while keeping a consistent appearance and behavior in view of the changes occurring in the surrounding environment, has become a difficult task for software engineers. In applications that use rich interfaces in Web 2.0 for improving the interactivity, this task becomes even more complex due to the need of preserving the interaction aspects that afford users a richer experience with the application. This task can be facilitated using a software process that guides developers in building a ubiquitous application, considering the different contexts involved in its execution. Faced with these challenges, this work proposes a software process, named Model Driven RichUbi, aiming at supporting the development of rich interfaces for ubiquitous applications that adapt themselves when viewed on different types of devices. Based on the Model Driven Development and Domain-Specific Modeling conceptions, in the process are defined activities and artifacts that help in modeling and partial code generation of rich interfaces for different platforms. Besides, dynamic content adapters that refine the produced interface versions are also employed in the process, so that the developed interfaces can adapt to the peculiarities of the access device identified from the interaction context at runtime. The computational support focused on the Rich Interfaces Domain employed in the process is advantageous since it can be reused to simplify the development of adaptive rich interfaces for ubiquitous applications of several fields, which contributes to effort reduction and productivity increasing. / A Web 2.0 permitiu aos usuários maior interatividade com as aplicações Web. As chamadas Aplicações de Internet Ricas (RIAs Rich Internet Applications) transpuseram os limites das interfaces simples construídas apenas em Hypertext Markup Language (HTML). Através da adoção de tecnologias que permitem a criação de interfaces mais avançadas, as RIAs assemelham-se à aparência e comportamento das aplicações desktop. Por outro lado, a demanda por software na Computação Ubíqua, onde o acesso às aplicações ocorre de qualquer lugar, a qualquer hora e a partir de diferentes dispositivos, fez surgir novos desafios para a Engenharia de Software. Um desses desafios está relacionado com a adaptação das aplicações acessadas por diferentes dispositivos em contextos distintos. Dada a diversidade de dispositivos, redes de acesso, ambientes e contextos possíveis, prover aplicações que satisfaçam as peculiaridades de cada dispositivo de acesso, ao mesmo tempo em que mantêm um comportamento e aparência coerentes face às mudanças que ocorrem no ambiente ao redor, tornou-se uma difícil tarefa para os engenheiros de software. Nas aplicações que utilizam de interfaces ricas na Web 2.0, para melhorar a interatividade, essa tarefa torna-se mais complexa devido à necessidade de preservar os aspectos de interação que proporcionam aos usuários uma rica experiência com a aplicação. Tal tarefa pode ser facilitada usando um processo de software que oriente o desenvolvedor na construção de uma aplicação ubíqua, considerando os diferentes contextos em que se executa a aplicação. Diante desses desafios e visando a dar suporte ao desenvolvimento de interfaces ricas de aplicações ubíquas que se adaptam quando visualizadas em diferentes dispositivos, neste trabalho é proposto o processo denominado Model Driven RichUbi. Com base nas concepções de Desenvolvimento Dirigido a Modelos e Modelagem Específica de Domínio, são definidos atividades e artefatos que orientam a modelagem e geração parcial de código das interfaces ricas para diferentes contextos. No Model Driven RichUbi, também são utilizados adaptadores dinâmicos de conteúdo que refinam as versões produzidas das interfaces para se adequarem às peculiaridades do dispositivo de acesso identificadas do contexto da interação em tempo de execução. O apoio computacional no Domínio de Interfaces Ricas empregado no processo tem a vantagem de poder ser reutilizado no desenvolvimento de interfaces ricas adaptativas em aplicações ubíquas de diferentes domínios, colaborando para a redução de esforços e aumento da produtividade.