A Model Driven Framework for Automatic Detection and Tracking Inconsistencies

Abstract

Software design model inconsistencies precipitate into flaws in system that can be avoided at the time of design of the system. Recent contributions in the software engineering domain confirmed this fact clearly. Obstructions in software development and delivery can lead to economic and time-to-market attributes of the software. The consequences of model inconsistencies will have ripple effect in three areas such as cost, development time and delivery. Though UML provides unified and common notations across the globe, the developers may use the models inappropriately leading to model inconsistencies. Many researches came into existence to handle model inconstancies. However, a comprehensive, flexible and extensible framework that caters to the needs of developers with diverse tool usage and skill set is desired. Towards this end, in our previous work, we proposed a framework named Extensible Real Time Software Design Inconsistency Checker (XRTSDIC) for checking software design inconsistencies. The framework was made flexible with placeholders to support different modeling tools, rule detectors and visualizations. In this paper we focus on providing more features for all placeholders to realize a truly flexible and extensible design inconsistency checker. This improved framework and the implemented prototype can help software engineers to build more consistent software design models that can avoid cost and budget overruns as the application can provide early detection of inconstancies. Moreover, the application can support certain degree of tolerance for inconsistencies and help software engineers to switch between UML tools, consistency checkers and visualization mechanisms. The empirical evaluation shows that our framework is flexible and reveals significant performance improvement over other state-of-the-art inconsistency checkers in terms of accuracy, speed and scalability.