Replication Package for "Towards a Non-invasive Robotic-supported Approach for Data Loss Detection in Android Applications"

This repository provides a static version of the dataset used for the study "Towards a Non-invasive Robotic-supported Approach for Data Loss Detection in Android Applications". Our research presents Robotic-supported Data Loss Detection (R-DLD), an evolution of our previous R-DLD framework that introduces a fully non-invasive testing for Android data loss detection. Our previous R-DLD required semi-invasive smartphone interactions while our new R-DLD operates without any system access (such as the ADB USB connection) or instrumentation: physical robotic actions replicate natural user interactions (touches, scrolls, and device rotations) through an industrial grade robotic arm. Our experimental validation demonstrates that R-DLD successfully reproduced 100% of known data loss cases across five benchmark applications. These results achieved the same detection capability as the original R-DLD while eliminating dependencies on USB-based debugging software or software instrumentation. The new R-DLD maintains backward compatibility with our previous version, which detected 341 unique data loss issues in 77 real-world applications. The developers have confirmed that 89.55% of our reported issues were real bugs. And they have subsequently fixed 35.82% of them. R-DLD has established a new standard for hardware-based, non-invasive mobile testing.