Topic Families

Pattern detection and mining

• cross-project GoF pattern detection;
• distinguishing structurally similar patterns by semantic evidence;
• pattern evolution across versions;
• partial/degraded pattern instances;
• pattern mining from framework usage.

Smells and architecture anomalies

• context-sensitive smell thresholds;
• history-aware smell prioritization;
• architecture smell detection from dependency graphs;
• relationship between smells and change/defect risk;
• explainable smell detection.

Automated refactoring

• candidate selection;
• refactoring precondition checking;
• refactoring sequence planning;
• multi-objective quality improvement;
• testability-aware refactoring;
• semantic validation of automated transformations.

Search and learning

• genetic/multi-objective search;
• reinforcement-learning policies for refactoring;
• learned ranking of candidate refactorings;
• LLM-guided pattern recommendation;
• retrieval-augmented design assistance;
• hybrid static-analysis + LLM validation.

Empirical design research

• replication of a pattern/smell study;
• inter-rater agreement on design smells;
• pattern usage and change proneness;
• human evaluation of design recommendations;
• benchmark quality and data leakage analysis.