TRAVART: An Approach for Transforming Variability Models

Abstract

A large number of variability modeling approaches have been developed including feature modeling, decision modeling, and Orthogonal Variability Modeling (OVM). Multiple variants of each approach have been developed, i.e., there are many different types of feature models. The usefulness of variability modeling approaches has been demonstrated for both research and practice, e.g., to support domain analysis or configuration of products. However, industry still develops their own custom solutions to manage variability and academia frequently proposes new approaches. The growing plethora of modeling approaches, often only described in academic papers, makes it very difficult to find, understand, and eventually pick an existing approach for a specific context or (set of) system(s). In this paper, we present and evaluate TRAVART, an approach for transforming artifacts describing variability, such as variability models. TRAVART is based on generic transformation operations and implemented in specific transformation algorithms. Using TRAVART, researchers and practitioners can experiment with and compare different variability models and switch from one modeling approach to another. We present a feasibility study conducted with several feature models, decision models, and OVMs. Based on this study, we conclude that despite some limitations, our approach can correctly transform one model into the other and also supports a full roundtrip. TRAVART is a good basis to develop further transformations for other types of variability models and an important step towards unifying variability modeling.