Layout cross-browser incompatibility detection using machine learning and DOM segmentation
Fagner Christian Paes, Willian Massami Watanabe · 2018
Cross-Browser Incompatibilities, namely XBIs, are differences in the behavior of web applications as they are rendered in distinct browser implementations. Web applications can be rendered in a wide variety of configuration environments, varying their browser implementation (eg. Google Chrome, Microsoft Internet Explorer and Mozilla Firefox). Even though there are technical specifications for building these environments, the way web applications are rendered in distinct browsers are not always consistent, depending on the technological resources and properties which are used. Currently, in the web engineering process, testers and developers must manually inspect their web application in each specific browser, so that XBIs are identified and fixed before deploying the system. This research reports the development of a Layout XBI detection approach based on the use of Machine Learning and DOM Segmentation. The approach segments a single web application in multiple DOM elements which compose the web application. The task of Layout XBI detection was modeled as a supervised learning classification problem using as features: differences in position, size and screenshot comparison of each DOM element of a web application. We validated our approach in an experiment which investigated the efficacy of the classification model. The experiment used 66 web applications which contained 5081 DOM elements rendered in three different browsers (Google Chrome, Mozilla Firefox, Internet Explorer). The experiment reported significant accuracy results according to F-measure analysis having reached 0.91.