User Guided Automation for Testing Mobile Apps

Abstract—Mobile devices are gradually taking over traditional computers ’ dominance in human lives. With the ever-increasing shipment of mobile apps running on these devices, their quality issues become a severe challenge. Although automated testing techniques are being widely studied, they mostly fall short of handling mobile apps ’ complex interactions, e.g., a finger swipe or device shaking gesture, leading to inadequate testing. In this paper, we present a novel User Guided Automation (UGA) technique to address testing challenges incurred by such complex interactions. UGA exploits user insights to complement automated testing techniques by recording user-guided app exe-cutions, replaying apps to certain stop points, and systematically exploring state space from these stop points. We implemented our work as a prototype UGA tool on Android platform and evaluated it on seven real-world Android apps. Evaluation results show that UGA achieved 1.59–21.78 ⇥ improvement in terms of method coverage over state-of-the-art automated techniques in testing mobile apps. Index Terms—Android apps; Testing; Semi-automated; User trace