Automated visual classification of DOM?based presentation failure reports for responsive web pages

Since it is common for the users of a web page to access through wide variety devices—including desktops, laptops, tablets and phones—web developers rely on responsive design (RWD) principles frameworks create sites that are useful all devices. A correctly implemented adjusts its layout according viewport width device in use, thereby ensuring suitably features content. use complex RWD often leads pages with hard?to?detect failures (RLFs), employ testing tools generate reports potential RLFs. pages, like ReDeCheck, analyse representation called Document Object Model (DOM), they may inadvertently flag concerns not human visible, requiring manually confirm classify each RLF as true positive (TP), false (FP), or non?observable issue (NOI)—a process time consuming error prone. The conference version this paper presented Viser, tool automatically classified three types RLFs reported by ReDeCheck. Viser was designed two ReDeCheck's DOM?based analysis could surface, introduces Verve, classifies Along manipulating opacity HTML elements page, does Verve also uses histogram?based image comparison pages. Incorporating both 25 used prior experiments 20 new previously considered, paper's empirical study reveals Verve's classification five frequently agrees classifications produced humans. reveal took average about 4?s any among 469 demonstrates classifying an TP, FP, NOI publicly available tool, less subjective prone than same manual done developer, we argue well?suited supporting