Model-based testing of end-user collaboration intensive systems

Collaboration intensive systems like social networks support the interaction of multiple end-users playing different roles such as "friend" or "post owner". To ensure that end-users achieve the intended type of collaboration, systematic testing can be an effective means. However, manually creating effective test cases is cumbersome and error prone as the amount of end-users interactions to test grows exponentially with the number of involved end-users and roles. In this paper, we present a novel approach for automatic test case generation via modeling user collaborations as synchronized, non-deterministic Finite State Machines. As our preliminary evaluation shows, such collaboration models are effective and efficient: compared to collaboration-unaware alternatives, we generated test cases which achieve higher code coverage and trigger more errors, up to 10X faster.