An empirical investigation in quantifying inconsistency and incompleteness of UML designs

The UML is becoming the de-facto notation for software engineering projects. It is a common hypothesis that incompleteness and inconsistency allowed by UML are a source for problems in the software development process. However, it appears that many adequate software systems are built with the use of UML. This raises the question as to what degree inconsistency and incompleteness in UML designs impact software engineering projects. For instance, are there typically many or few inconsistencies in a design? Which type of inconsistency occurs most often? Do the types of inconsistencies that are present in a design change in the course of the design process? If so, how? To investigate these questions, we have developed a number of techniques for analyzing UML designs. In this paper, we present the results of our study on inconsistency and incompleteness in large industrial systems