Semantic evolution analysis of feature models.

During the development process, feature models change continuously. Analyzing the semantic differences between consecutive feature model versions is important throughout the entire development process to detect unintended changes of the modeled product line. Previous work introduced a semantic differencing technique for feature models based on a closed-world assumption, which reveals the differences between two feature models when allowing products to only contain features used in the models. However, this does not reflect the stepwise refinement of feature models in early development stages. Therefore, we introduce an open-world semantics, an automatic method for semantic differencing of feature models with respect to the novel semantics, and formally relate the open- and closed-world semantics. We formally prove our results, including the relation between the different semantics as well as the soundness and completeness of the semantic differencing procedure. In conjunction with previous work, the results enable effective semantic feature model evolution analyses throughout the entire development process.