On Language Levels for Feature Modeling Notations

Abstract

Configuration is a key enabling technology for the engineering of systems and software as wells as physical goods. A selection of configuration options (aka. features) is often enough to automatically generate a product tailored to the needs of a customer. It is common that not all combinations of features are possible in a given domain. Feature modeling is the de-facto standard for specifying features and their valid combinations. However, a pivotal hurdle for practitioners, researchers, and teachers in applying feature modeling is that there are hundreds of tools and languages available. While there have been first attempts to define a standard feature modeling language, they still struggle with finding an appropriate level of expressiveness. If the expressiveness is too high, the language will not be adopted, as it is too much effort to support all language constructs. If the expressiveness is too low, the language will not be adopted, as many interesting domains cannot be modeled in such a language. Towards a standard feature modeling notation, we propose the use of language levels with different expressiveness each and discuss criteria to be used to define such language levels. We aim to raise the awareness on the expressiveness and eventually contribute to a standard feature modeling notation.