Cross product line analysis

Abstract

Due to increase in market competition and merger and acquisition of companies, different software product lines (SPLs) may exist under the same roof. These SPLs may be developed applying different domain analysis processes, but are likely not disjoint. Cross product line analysis aims to examine the common and variable aspects of different SPLs for improving maintenance and future development of related SPLs. Currently different SPL artifacts, or more accurately feature models, are compared, matched, and merged for supporting scalability, increasing modularity and reuse, synchronizing feature model versions, and modeling multiple SPLs for software supply chains. However, in all these cases the focus is on creating valid merged models from the input feature models. Furthermore, the terminology used in all the input feature models is assumed to be the same, namely similar features are named the same. As a result these methods cannot be simply applied to feature models that represent different SPLs. In this work we offer adapting similarity metrics and text clustering techniques in order to enable cross product line analysis. This way analysis of feature models that use different terminologies in the same domain can be done in order to improve the management of the involved SPLs. Preliminary results reveal that the suggested method helps systematically analyze the commonality and variability between related SPLs, potentially suggesting improvements to existing SPLs and to the maintenance of sets of SPLs.