Treffer: Runtime Analysis of Object-Oriented Designs for Predefined Design Pattern Detection: A Conceptual Validation Study.

The process of Automatic Design Pattern Recovery (ADPR) aims to identify and document design patterns within software systems by analyzing code for recurring signatures indicative of predefined patterns. Despite the development of numerous techniques to address this challenge, the varying implementations of design patterns across different programming languages, combined with the lack of contextual analysis, hinder the effectiveness of these methods. Building on prior research, this paper introduces an innovative approach to ADPR that integrates structural and behavioral analysis of source code. This approach leverages the characteristic that design patterns tend to exhibit predictable behaviors as software architecture scales or when the behavior of its constituent objects is filtered. The core contribution of this work is the introduction of Behavio-Structural Sequences (BSS), which encapsulate key features for identifying the intrinsic nature of design patterns. These sequences are filtered based on architectural and behavioral properties, transformed into text strings, and classified using the Levenshtein distance metric. The proposed classifier is evaluated on a diverse set of source codes, achieving an 89% accuracy rate with a 0.09 miss rate. [ABSTRACT FROM AUTHOR]