Varflix: A Flexible Approach for Variability Mining

Variability mining is a process to perform extractive adoption of Software Product Lines (SPLs). This process becomes necessary when manual variability management in existing products is no longer possible due to the size and complexity of the system and/or too many variants resulting from clone-and-own reuse. A key factor for satisfying variability mining results is the extensive involvement of experts of the domain or system under analysis. In existing approaches, this aspect is addressed to a limited extent and tools are highly automatic. Integrating checks for validation of the mining results is another aspect with little to no support in this context. In this paper, we present our work to design a framework for artifact-independent variability mining addressing these shortcomings called Varflix. The idea is to perform mining in multiple automatic steps but include user input after each step to adapt the result to user needs. This should not only improve the quality of the mining result, but also other aspects such as scalability or complexity. Varflix produces a variability model based on the input variants.