Condensation of Reverse Engineered UML Diagrams by Using the Semantic Web Technologies

Up-to-date software design documentation is valuable for maintenance engineers, testers and developers joining a project at a later stage; however, UML Models, e.g. class diagrams, are often poorly kept up-to-date during development and maintenance. Reverse engineering, therefore, has become a popular method to recover an up-todate design from the underlying source code. However current techniques yet produce a detailed representation of the underlying source code that would reduce the understandability. In order for the understandability to enhance, condensation of the reverse engineered diagrams has been proposed as a solution. However, current state-ofthe-art approaches in this area still demand a need for improving or alternative approaches. Consistently, in this paper we are putting forward a bridging idea of using the semantic web technologies for improving the condensation process. Two contributions are proposed that support each other. Firstly, the V-Ontmodel is suggested for updating software documentation over the software evolution. Secondly, a general architecture is proposed enabling us to reduce sophisticated analysis tasks for the condensation process of class diagrams to a few queries in SPARQL or its extensions like SPARQL-ML. To discuss the feasibility of the approach we focus on the condensation of reverse engineered class diagrams in the whole paper. Finally, an illustration example is presented in which helper classes are excluded from the class diagrams in the condensation process regarding the structural patterns extracted from the software metadata.