On-Demand Cluster Analysis for Product Line Functional Requirements

We propose an on-demand clustering framework for analyzing the functional requirements in a product line. Our approach is novel in that the objects to be clustered capture the domain´s action themes at a primitive level, and the essential attributes are uncovered via semantic analysis. We provide automatic support to complement domain analysis by quickly identifying important entities and functionalities. A second contribution is our recognition of stakeholders´ different goals in cluster analysis, e.g., feature identification for users versus system decomposition for designers. We thus advance the literature by examining requirements clusters that overlap and those causing a minimal information loss, and by facilitating the discovery of product line variabilities. A proof-of-concept example is presented to show the applicability and usefulness of our approach.