Identifying components in object-oriented programs using dynamic analysis and clustering
Simon Allier, Houari A. Sahraoui, Salah Sadou · 2009
We propose an approach for component candidate identification as a first step towards the extraction of component-based architectures from object-oriented programs. Our approach uses as input dynamic call graphs, built from execution traces corresponding to use cases. This allows to better capture the functional dependencies between classes. The component identification is treated as a clustering problem. To this end, we use formal concept analysis and design heuristics.