Mitigating Anonymity Challenges in Automated Testing and Debugging Systems
Silviu Andrica, George Candea · Infoscience (Ecole Polytechnique Fédérale de Lausanne) · 2013
Modern software often provides automated testing and bug reporting facilities that enable developers to improve the software after release. Alas, this comes at the cost of user anonymity: reported execution traces may identify users. We present a way to mitigate this inherent tension between developer utility and user anonymity: automatically transform execution traces in a way that preserves their utility for testing and debugging while, at the same time, providing k-anonymity to users, i.e., a guarantee that the trace can at most identify the user as being part of a group of k indistinguishable users. We evaluate this approach in the context of an automated testing and bug reporting system for smartphone applications. 1