Aspect-oriented software development has become an important development and maintenance approach to software engineering across requirements, design and implementation phases. However, discovering early aspects from requirements for a better integration of crosscutting concerns into a target system is still not well addressed in the existing works. In this paper, we propose a Goal-driven Early Aspect approach (called GEA) to discovering early aspects by means of a clustering algorithm in which relationships among goals and use cases are utilized to explore similarity degrees of clustering goals, and total interaction degrees are devised to check the validity of the formation of each cluster. Introducing early aspects not only enhances the goal-driven requirements modeling to manage crosscutting concerns, but also provides modularity insights into the analysis and design of software development. Moreover, relationships among goals represented numerically are more informative to discover early aspects and more easily to be processed computationally than qualitative terms. The proposed approach is illustrated by using two problem domains: a meeting scheduler system and a course enrollment system. An experiment is also conducted to evaluate the benefits of the proposed approach with Mann-Whitney U-test to show that the difference between with GEA and without GEA is statistically significant.
1 June 2010 in International Journal of Intelligent Systems
Design patterns provide solutions to recurring problems in software design and development, promoting scalability, readability, and maintainability. While past research focused on the utilization of the design patterns and performance, there is limited insight into their impact on program evolution. Dependency signifies relationships between program elements, reflecting a program’s structure and interaction. High dependencies indicate complexity and potential flaws, hampering system quality and maintenance. This paper presents how design patterns influence software evolution by analyzing dependencies using the Abstract Syntax Tree (AST) to examine dependency patterns during evolution. We employed three widely adopted design patterns from the Gang of Four (GoF) as experimental examples. The results show that design patterns effectively reduce dependencies, lowering system complexity and enhancing quality.