Intelligent Mining of Association Rules Based on Nanopatterns for Code Smells Detection

Table 2

Outline of code smell detection approaches.

Approaches	Outline
Search-based	Solves optimization problems to find the best possible subset of solutions
Metric-based	Creates a rule based on metrics and respective thresholds
Symptom-based	Describes symptoms as class roles and structures that are transformed into detection algorithms
Visualization-based	Semiautomated process of visually representing data with metrics using visual metaphors
Probabilistic	Related to the degree of uncertainty of a class that indicates an occurrence of code smell
Cooperative-based	Improves performance and accuracy in detecting code smells by executing activities cooperatively
Manual	The human-centric process that requires a great human effort, extensive analysis, and interpretation effort from software maintainers to find design fragments that correspond to code smells