As software evolves, analysis and design models should be modified, correspondingly. In this scenario, one of the main problems is to detect which elements should be changed due to a given change. This paper presents an approach that applies data mining over a versioned UML repository in order to detect change patterns among model elements at different abstraction levels. These patterns are presented to the software engineer together with contextual information obtained from a change control system, so he can decide the software evolution task at hand.
Citation:
Cristine R. Dantas, Leonardo G. P. Murta, Cl?udia M. L. Werner, "Consistent Evolution of UML Models by Automatic Detection of Change Traces," iwpse, pp.14-147, Eighth International Workshop on Principles of Software Evolution (IWPSE'05), 2005