An automated approach to object oriented design pattern detection and extraction
Date
1997
Authors
Stier, Jochen
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Maintaining and re-engineering large software systems has proven to be extremely labor intensive, which is in part caused by a lack of knowledge about the system. Information gathering about design, structure and functionality often occupies a large portion of the maintenance efforts.
Object oriented design patterns describe common solutions to recurring problems in object oriented software systems. They are derived from the experience and expertise of partitioning software designers and developers. Literature about a pattern provides a vast pool of information ranging from pitfalls and advantages to detailed implementations.
Each pattern encapsulates much more information about a system than the individual software abstractions composing it. Detection and presentation of the design patterns aids and enhances the information gathering process of the re-engineering and maintenance task by providing documentation about system architecture, design and behavior. A pattern furthermore provides a means of system decomposition as well as common terminology for system components.
This thesis presents a strategy to identify, extract and visualize object oriented design patterns that are present within the source code of an existing software system. The approach combines techniques of reverse engineering, querying and visualization to provide a tool for automated design pattern detection and extraction.