Measures for quality evaluation of feature models

Abstract

In Software Product Lines (SPL), quality evaluation is a critical factor, because an error in a SPL can spread to various end products. However, it is often proved impractical to ensure the quality of all products of a given SPL both for economic reasons and the effort needed due to their large number. In this context, a strategy that can be used is to make quality assessments on the initial phases of the SPL development. This approach avoids having errors that could be propagated to the next SPL phases. So, taking into account the feature model, which is one of the most important artifacts in a SPL since its quality directly affects the quality of the SPL end products, to assure the quality of the feature model is one of the current strategies to assess the quality of a SPL. In this sense, one way to evaluate the feature model is to use measures, which could be associated with the feature model quality characteristics and their quality attributes. This paper presents a measures catalog, which can be used to support the quality evaluation of the feature model. In order to identify these measures, a systematic mapping is conducted and to validate the measures catalog, we perform a peer review with experts in software quality and SPL. Besides that, to evaluate the use of the proposed catalog, we apply the measures in three feature models in the domain of mobile applications. The results show that the proposed measures catalog can be effectively deployed to support the quality evaluation of the feature models.