Rule-based handling of software quality and productivity models

Abstract

Each software system and each software project is unique. Modeling software quality or productivity therefore has to be product or project specific. A rule-based modeling technique is proposed, which uses weight functions to define factors of quality or productivity in terms of evaluation factors and which takes environment parameters to represent validity ranges. Objectives and applications are also defined by such rules. A third category of rules, namely interrelation rules, are used to define the ‘implementation’ of objectives in terms of quality factors and applications. Each set of rules might be viewed as an acyclic decomposition graph. Quality or productivity then is to be defined as the distance of an actual graph and a required graph.