Customizable feature based design pattern recognition integrating multiple techniques

factory, Adapter, Builder, Command, Composite, Decorator, Factory method, Observer, Prototype, Singleton, State, Strategy, Template method, and Visitor patterns. The approach focuses on reducing search space, but it may return large number of false positives when certain roles are removed for detecting patterns. Arceli et al. [46] have presented a design pattern detection approach which is based on supervised classification and data mining techniques to extract behavioral design patterns. They used multi-label approach to create networks and performed their experiments using Chapter 3 Related Work 30 feed-forward neural networks and with back-propagation. The MARPLE (Matrix and Architecture Reconstruction Plug-In for Eclipse) tool is used which reconstruct software architecture and compute matrix that are used for pattern recovery. Joiner and neural network modules play key role in the recognition of design patterns. The approach has low precision and recall rates for Command, Strategy and Mediator due to difficulty in detecting these patterns.