Poster: Repair Cross Browser Layout Issues by Combining Learning and Search-based technique

With the increasing number of browsers and platforms on which the applications can be executed, cross-browser incompatibilities (XBIs) are becoming a serious problem for organizations to develop web-based software. In order to eliminate cross browser issues, existing work tries to repair layout XBIs using search-based technique. However, the designed fitness function is too strict, and may miss the chance to repair some layout XBIs. For each reported layout XBI, it will search all possible candidate fixes, and cannot reuse existing repair solutions for similar issues. This paper proposes to combine learning and search based technique to improve the state-of-the-art of automated repair of layout XBIs. By extracting the characteristics of successfully repaired layout XBIs and learn a decision tree, our approach will accelerate the search process by directly applying the recommended solution to fix similar XBIs. Moreover, by redesigning search-based repair process, our technique can improve the chance to repair layout XBIs. The initial evaluation shows that the proposed approach is effective.