A new approach of component identification based on weighted connectivity strength metrics

Abstract

Once software is in production, maintenance works are inevitable e.g., bug fixes, requirement changes, etc. During the long-term software maintenance, documents become gradually inconsistent with the system and the source code becomes the most reliable source for design recovery. Currently there exist many approaches to identify component from source code, which adopt different component metrics and clustering algorithms. However, these approaches are not satisfying in the precision and efficiency. This study proposes a new component extraction approach based on 0 metrics and hierarchical clustering algorithm, which greatly improves the precision and efficiency. © 2008 Asian Network for Scientific Information.