Towards an automatic evaluation of web applications

Evaluating the usability of computer applications using traditional laboratory-based tests is costly and time consuming. For modern Web applications, usually developed, tested and deployed in "Internet Time", this approach is simply not feasible. A more effective way to evaluate the usability of Web applications consists in gathering information from the user's interactions and automatically processing this data in order to detect usability problems in the execution of predefined tasks. The reported solutions based on this approach usually fail on providing efficient tools for the definition of tasks, specially in large and dynamic Web applications. In order to tackle this problem, we developed USABILICS, a system targeted for the automatic remote evaluation of usability based on an interface model. The proposed model allows the definition of tasks using a simple and intuitive approach, which can be applied to large and dynamic Web applications. USABILICS analyzes the execution of tasks by calculating the similarity among sequence of events produced by users and those previously captured by evaluators. Based on this analysis, USABILICS provides an usability index, as well as recommendations for solving usability problems detected on the execution of each task.