Recovering Workflows from Functional Tests

When enterprises outsource maintenance of IT systems to service providers, thorough knowledge acquisition is critical to the success of the engagement. Program comprehension contributes significantly to acquiring knowledge of the IT systems. It is a common practice to execute test scripts to identify critical scenarios in the system and then trace these as flows in the programs. Instead of executing test scripts, we propose the novel idea of mining workflows from test scripts to construct formal process models. The global view provided by the mined model can not only help transition teams gain high level understanding of the system but also help identify critical flows. We also suggest categorization of test cases using supervised learning to improve comprehension.