Usability problems discovery based on the automatic detection of usability smells

Abstract

Web applications usage has gone through a big and fast growth in the past few years, becoming part of our daily lives. Large companies use these applications to provide their services. So, it is necessary to ensure the development of high-quality applications. One of the main attributes of a Web application that directly determines its success or failure is their usability, and several methods were developed to evaluate it. The most popular method of usability evaluation is the laboratory testing because it can directly evaluate users' reactions while interacting with the final application. However, performing this test entails high cost and complexity. So, aiming to simplify the execution of this evaluation and to ease the discovery of usability problems, this work proposes an approach to detect indicators of problems (usability smells), based on the capture of the user interaction in the production environment and automatic analysis of this interaction. A study carried out with the approach showed that smells indicated by it were able to aid the detection and correction of usability problems in a real application.