Increasing the rigorousness of measures definition through a UML/OCL model based on the briand et al.'s framework

Abstract

The use of a formal definition of measures upon a metamodel assures that measures capture the software artifacts they intend for, improve repeatability and could facilitate the implementation of measures extraction tools. However, it does not assure that the measure captures the measurement concept it claims (like size, coupling, etc). For that purpose many formal frameworks had been defined. The well-known property-based framework proposed by Briand et al. defines the most important measurement concepts regardless the specific software artifacts to which these concepts are applied. In this article we define a UML/OCL model from the Briand's framework and we relate it with the formal definition of measures upon metamodels. We describe a set of well-formed properties that a measure should verify when capturing a measurement concept (which are derived from the model). We exemplify our approach through a thorough formal definition of UML statechart diagrams measures and its well-formed constraints of size measures. © 2013 Springer-Verlag Berlin Heidelberg.