Detection of Incomplete Patterns Using FUJABA Principles

An approach for the detection of structural patterns in UML class diagrams is presented. It picks up some principles of the reverse engineering component of Fujaba, such as a hierarchical pattern definition and an alternating bottomup/top-down analysis. Furthermore it uses a fuzzy-like evaluation mechanism so that it is able to recognize not only entire patterns but also incomplete instances. The knowledge about incomplete instances, which obviously occur rather often during the developing process, assists developers not only while maintaining or reverse engineering existing software but already while designing and implementing new software.