Abstract

Usability is critical for any system, but insoftware it is one of the most important features. In fact,one of the main reasons for software failure is the systemlacking to achieve users specified goals and satisfaction. Forthis reason, usability evaluation is becoming an importantpart of software development. Software usability evaluationcan be costly in terms of time and human. Therefore,automation is promising way to augment existingapproaches especially if the evaluation is subjective wherethe usability concentrated about user's "opinion". Thispaper proposes to use opinion mining as an automatictechnique to evaluate subjective usability. Opinion miningis a research subtopic of data mining aiming toautomatically obtain useful opinioned knowledge insubjective texts. We propose a novel model to extractknowledge from opinions to improve subjective softwareusability. This is the first time opinion mining used insoftware usability. To evaluate our proposed model, a set ofexperiments was designed and conducted and we got anaverage accuracy of 85.41%. Also, we propose to usegraphics to visualize user's opinion in software and tocompare the usability of two software.