Introducing Learning Mechanism for Class Responsibility Assignment Problem

Assigning responsibilities to classes is a vital task in object-oriented design, which has a great impact on the overall design of an application. However, this task is not easy for designers due to its complexity. Though many automated approaches have been developed to help designers to assign responsibilities to classes, none of them considers extracting the design knowledge (DK) about the relations between responsibilities in order to adapt designs better against design problems. To address the issue, we propose a novel Learning-based Genetic Algorithm (LGA) for the Class Responsibility Assignment (CRA) problem. In the proposed algorithm, a learning mechanism is introduced to extract DK about which responsibilities have a high probability to be assigned to the same class, and the extracted DK is employed to improve the design qualities of generated solutions. An experiment was conducted, which shows the effectiveness of the proposed approach.