Detecting Inconsistences in Multi-view UML Models

Abstract

Inconsistencies in conflicting multi-view UML models can be major obstacles to the quality and productivity of software development. In the current literature it can be observed that some tools were developed to support the detection of inconsistencies, but none of them are consolidated neither in the academic environment nor in the current market. In addition, many of these tools only evaluate syntactic inconsistencies, not considering semantic inconsistencies. Therefore, the tools available in the state-of-the-art are often unable to detect syntactic and semantic inconsistencies in conflicting multi-view UML models. To address this issue, we propose DIUML, a tool that includes: (i) detection of inconsistencies in multi-view UML models through design metrics, and (ii) detection of syntactic and semantic inconsistencies, indicating objects and affected classes, also qualifying the severities of each type of inconsistency. Our preliminary evaluation indicated that DIUML was able to detect inconsistencies in multi-view UML models with 337 elements from 10 different combinations of Class Diagrams and Sequence Diagrams.

Publication
International Journal of Computer Science and Software Engineering (IJCSSE). v. 5, n. 12, p. 259-263. 2016
Date
Links