Classificação de incompatibilidades cross-browser de layout: um estudo comparativo entre diferentes modelos

Abstract

The complexity of developing Web applications is increasing every day due to the large number of browsers available on the market. As a consequence, the number of inconsistencies between Web applications has also increased. When the same Web application is rendered in different browsers, the inconsistencies detected in the layout or behavior of the pages are known as (XBIs - Cross Browser Incompatibilities). The impact of XBIs on Web applications ranges from minor inconsistencies to critical flaws in the layout or functionality of a Web application. Thus, XBIs can directly affect the end user experience while browsing the Web (SILVA, 2005) application. To ensure the quality of web applications, testers and developers must manually inspect the applications in each specific browser, so that XBIs are identified and corrected before the system is deployed. Currently, there are several approaches in the literature for the identification and automatic correction of XBIs. These approaches have evolved with the aim of reducing false positives and negatives. This work proposes to compare some of the approaches, focusing on those that use the classification of XBIs of layout, through machine learning algorithms. There is still no work in the literature that makes this comparison, identifying its main advantages and disadvantages. This paper consists of an experiment that compares the results of the approaches and presents metrics that allow to affirm their effectiveness, aiming to bring important information as contributions to propose future works in relation to the evolution of the explored approaches. The result of the experiment is the F-Score metric. For this metric, the higher values imply greater efficiency in detecting incompatibilities between browsers, and the C5.0 10 iterations - Paes, Watanabe configuration obtained the best result in the experiment.