Zum Inhalt springen

Product-Line Engineering for Metamodels

Aus SDQ-Institutsseminar
Vortragende(r) Henrik Kayser
Vortragstyp Masterarbeit
Betreuer(in) Fabian Eger
Termin Fr 13. November 2026, 11:30 (Raum 010 (Gebäude 50.34))
Vortragssprache Deutsch
Vortragsmodus in Präsenz
Kurzfassung Currently, reducing the size of a complex metamodel that has to be integrated into a Virtual Single Underlying Model (V-SUM) has to be done manually. To address this problem, we introduce the concepts of product line engineering into metamodels. We present a fully automatic approach that extracts a feature model from a given metamodel.

During feature extraction, we find an initial clustering of features using graph clustering techniques and refine them with heuristics derived from object-oriented modeling techniques. During model generation, we construct a feature tree using the dependencies between features. We further describe how CPRs can be used to extend the approach to multiple metamodels. Finally, we integrate the approach into the methodologist dashboard, where the feature model is used to configure and slice a metamodel. We observe that the heuristics generally increase the variability of the product line, albeit at the cost of higher complexity. Our feature models are comparable to real-world ones in most measures, except for the share of features involved in cross-tree constraints, which is considerably lower than in models of established product lines.We attribute this to our approach capturing only the syntactic dependencies of a metamodel, not the semantic dependencies of the modeled domain.