RésuméThe Web is an essential way to search for information. In the professional world, a website reflects the corporate image and is an important and efficient means of communication in order to interact with customers. The most effective websites are those with a high level of usability (i.e. the effectiveness with which a user can achieve his goals). Therefore, research has studied ways to assess how websites should be structured to ensure their quality and usability. We review the scientific literature about the assessment of website quality and usability. We focus on an important characteristic of the usability: the website navigability (i.e. the ease with which users can locate and access relevant information). Most research has focused on assessing the quality of individual pages or of a site as a whole.
Based on two previous studies, we propose the use of a multi-level quality model, applied to the problem of assessing the website navigability. Our Multi-level Model computes a navigability score for each webpage (page-level model), an aggregated score for all the pages (composition model), and the website navigability score (site-level model). Our Multi-level Model combines these three “cascading models”. We contribute to the validation and comparison of these models. To test our models, we conducted an experiment in two phases, with 22 and then 24 subjects. We found that the multi-level model is better predictor of navigability. We investigated several possible composition models. We tested different weighting strategies based on known graph analysis algorithms. We found that a complex weighting strategy does not improve the performance of our models. Finally, we propose different ways to improve our models, according to the results of the experiment.
|la date de réponse||2011|
|Superviseur||Naji Habra (Promoteur)|