Empirical investigation into static and dynamic coupling metrics

Abstract

Recent research in the field of object-oriented software engineering has been focusing on the usage of software metrics for the assessment of a product's quality. Software design coupling metrics include some of the most widely applicable metrics used in the modern software industry. Although static coupling metrics are used for analyzing the structural aspects of the object-oriented software systems, their inability to capture the behavioral aspects is well known and hence has given way to the evolution of their runtime counterparts i.e. the dynamic coupling metrics. Most of the dynamic coupling metrics proposed until now have not yet been empirically validated due to the high evaluation cost involved in the runtime metric data collection, hence limiting their practical application. This paper, as a part of our ongoing research on empirical validation of the dynamic coupling metrics, presents an empirical investigation into the static and dynamic CBO (Coupling Between Objects) metrics using some open source real world java applications. A set of criteria for the selection of sample benchmark applications required for the dynamic metrics validation has also been devised. A scenario based approach in combination with the appropriate statistical techniques has been used in an attempt to correlate the static and dynamic coupling metrics.