Using Object-Level Run-Time Metrics to Study Coupling Between Objects

摘要

In this paper we present an investigation into the run-time\udbehaviour of objects in Java programs, using specially adapted\udcoupling metrics. We identify objects from the same class\udthat exhibit non-uniform coupling behaviour when measured\uddynamically.\udWe define a number of object level run-time metrics, based\udon the static Chidamber and Kemerer coupling between objects\ud(CBO) measure. These new metrics seek to quantify\udcoupling at different layers of granularity, that is at classclass\udand object-class level. We outline our method of collecting\udsuch metrics and present a study of the programs\udfrom the JOlden benchmark suite as an example of their\uduse.\udA number of statistical techniques, principally agglomerative\udhierarchical clustering analysis, are used to facilitate\udthe identification of such objects.