Intelligent Engineering Systems through Artificial Neural Networks, Volume 16
12 Multi-Objecttve Genettc Algorithms in Object-Oriented Conceptual Software Design
Download citation file:
- Ris (Zotero)
- Reference Manager
The conceptual design of object-oriented software is difficult to learn and perform yet has a crucial impact on subsequent downstream software development. In an attempt to support the human designer during conceptual object-oriented software design, a multi-objective genetic algorithm has been developed to search and explore the design space. Two case studies are investigated using class cohesion and size as multi-objective fitness functions, and the generated solutions are compared with those from manual designs. While cohesion values are broadly similar, the genetic algorithm also produces a variety of interesting design variants of equivalent fitness that have not been identified by manual design. These promising results, when combined with favorable performance times, suggest that the multi-objective genetic algorithms offer potential as the basis of computational tool support for interactive human / machine search and exploration of the conceptual object-oriented design space.