A Framework for Detecting Compatibility-Issues-Proneness Apps Based on Multimodal Analysis in Android Platform

With the prosperity of Android, the compatibility issues in apps cause security flaws and bring damages to the user experience. Unfortunately, recent studies cannot help the developer to identify the apps that are compatibilityissue-proneness. It motivates us to propose an automated approach to identify these apps derived from multimodal learning. To this goal, we first present some potential modalities of apps based on previous insight and then leverage statistical methods to test the modalities. Finally, we use the selected modalities to identify these apps on a real dataset. Experimental results on apps demonstrate the effectiveness of our work.