Business process families provide an over-arching representation of the possible business processes of a target domain. They are defined by capturing the similarities and differences among the possible business processes of the target domain. To realize a business process family into a concrete business process model, the variability points of the business process family need to be bounded. The decision on how to bind these variation points boils down to the stakeholders' requirements and needs. Given specific requirements from the stakeholders, the business process family can be configured. This paper formally introduces and empirically evaluates a framework called ConfBPFM that utilizes standard techniques for identifying stakeholders' quality requirements and employs a metaheuristic search algorithm (i.e., Genetic Algorithms) to optimally configure a business process family.

No abstract

No abstract

Abstract. Quality evaluation is a challenging task in monolithic software systems, and is even more complex when it comes to Service-Oriented Software Product Lines (SOSPL), as it needs to analyze the attributes of a family of SOA systems. In SOSPL, variability can be managed and planned at the architectural level to develop a software product with the same set of functionalities but different degrees of non-functional quality attribute satisfaction. Therefore, architectural quality evaluation becomes crucial due to the fact that it allows for the examination of whether or not the final product satisfies and guarantees all the ranges of quality requirements within the envisioned scope. This paper addresses the open research problem of aggregating QoS attribute ranges with respect to architectural variability. Previous solutions for quality aggregation do not consider architectural variability for composite services. Our approach introduces variability patterns that can possibly occur at the architectural level of a SOSPL. We propose an aggregation model for QoS computation which takes both variability and composition patterns into account.

No abstract