Evaluation of reusable concern-oriented goal models

Abstract

A new unit of encapsulation called the concern is at the center of Concern-Orientation. Building on techniques for advanced Separation of Concerns, from Model-Driven Engineering, and from Software Product Lines, Concern-Orientation is a reuse paradigm that stipulates the use of three interfaces to enable broad, generic reuse: the variation, customization, and usage interfaces. Higher-level concerns reuse lower-level concerns, resulting in concern hierarchies where lower-level concern models are composed with higher-level concern models. As part of the variation interface, goal models are used to describe the impact of features of a concern on system qualities. Consequently, goal models of lower-level concerns must be combined with goal models of higher-level concerns to enable reasoning about system qualities in concern hierarchies. However, existing propagation-based reasoning mechanisms for goal models still assume a monolithic goal model, which is not appropriate for concern-oriented reuse. To address this issue, this paper presents novel modeling constructs to enable the reuse of lower-level goal models in the context of Concern-Orientation, extends existing propagation-based reasoning mechanisms of goal models for use in concern hierarchies, and reports on a proof-of-concept implementation of the novel modeling constructs and the extended reasoning mechanism.