Modular Feature Models: Representation and Configuration

Within the realm of software product line engineering, feature modeling is one of the widely used techniques for modeling commonality as well as variability. Feature models incorporate the entire domain application configuration space, and are therefore developed collectively by teams of domain experts. In large scale industrial domains, feature models become too complex both in terms of maintenance and configuration. In order to make the maintenance and configuration of feature models feasible, we propose to modularize feature models based on the well-established Distributed Description Logics formalism. Modular feature models provide for an enhanced collaborative/ distributed feature model design, more efficient feature model evolution and better reusability of feature model structure. We also develop methods for the configuration and configuration verification of a modular feature model based on standard inference mechanisms. We describe and evaluate our proposed approach through a case study on an online electronic store application domain. ACM Classification: D.2.4, D.2.13