Design pattern mining using distributed learning automata and DNA sequence alignment

Area of Science:

• Software Engineering
• Object-Oriented Programming
• Pattern Mining

Background:

• Design patterns offer reusable solutions for common software engineering problems.
• Identifying and understanding relationships between design patterns is crucial for effective software development.

Purpose of the Study:

• To introduce a new method and tool, DLA-DNA, for mining design patterns and their relationships.
• To evaluate the effectiveness of DLA-DNA in terms of precision and recall compared to existing tools.

Main Methods:

• Mining structural design patterns from object-oriented source code.
• Extracting strong and weak relationships between identified design patterns.
• Utilizing distributed learning automata (DLA) and deoxyribonucleic acid (DNA) sequence alignment principles.

Main Results:

• The DLA-DNA method achieves high precision and recall in design pattern detection.
• DLA-DNA outperforms Pinot, PTIDEJ, and DPJF in identifying design patterns and their relationships.
• The proposed method shows average improvements of 20% and 9.6% in precision and recall over Pinot, respectively.

Conclusions:

• The proposed method effectively identifies elemental and actual design patterns in software.
• DLA-DNA provides a robust approach for analyzing design patterns and their interdependencies.
• This method enhances code analysis by enabling determination of object and component dependency rates.