Feature-based software design pattern detection

Software design patterns are standard solutions to common problems in software and architecture. Knowing that a particular module implements pattern is shortcut comprehension. Manually detecting time consuming challenging task, therefore, researchers have proposed automatic detection techniques. However, these techniques show low performance for certain patterns. In this work, we introduce approach, DPDF improves the over state-of-the-art by using code features with machine learning classifiers automatically train detector. creates semantic representation of Java source call graph, applies Word2Vec algorithm on construct word-space geometric model code. then builds classifier trained labelled dataset identifies 80% Precision 79% Recall. Additionally, compared two existing namely FeatureMaps & MARPLE-DPD. Empirical results demonstrate our approach outperforms approaches approximately 35% 15% respectively terms Precision. The run-time also supports practical applicability classifier.