A Novel Approach to Automated Design Pattern Detection

Nikolaos Tsantalis, Alexander Chatzigeorgiou, Spyros T. Halkidis, George Stephanides · 2005

The importance of the use of Design Patterns in order to build reusable and wellstructured software has been eminent since these patterns have been formalized. Thus, it became desirable to be able to detect which design patterns are present in a software system. Knowing this information it is possible to make an evaluation on different aspects of the system. Though, it is a very difficult task for a software engineer to pinpoint all the Design Patterns present in a system, without any assistance. Addressing this need, techniques for automated design pattern detection have appeared in the literature. Some are based on reverseengineering of already existing code while others can work already at design level by analyzing UML diagrams. Though, complexity is one of the characteristics of all the methods proposed until now. Furthermore, all of these techniques work only for a limited number of the GoF patterns. Our aim in this paper is to elaborate on a simple approach for automatic detection of design patterns that works by analyzing UML class diagrams. Our method can achieve the automated detection of all GoF patterns that do not require any code specific information to recognize them.

Read the paper · More papers on PaperTik