An Efficient Algorithm for Feature-Model Slicing

Feature models are a well-known concept to represent variability in software product lines. A feature model defines all features of a product line and their corresponding interdependencies. During software product line engineering, there arise situations that require the removal of certain features from a feature model such as feature-model evolution, information hiding, and feature-model analyses. However, crude deletion of features in a model typically has undesirable effects on interdependencies of the remaining features. Moreover, current algorithms for dependency-preserving feature removal (known as feature-model slicing) do not perform well when removing a high number of features from large feature models. Therefore, we propose an efficient algorithm for feature-model slicing based on logical resolution and CNF minimization.