Assessing and interpreting object-oriented software complexity with structured and independent metrics

Abstract

Object-oriented software complexity is difficult to assess due to its manifold influences from cognition science or algorithmic complexity theory. A practical process for a structured complexity assessment is presented in this paper. It starts with considerations for measurement and data preparation. Using mathematical transformation techniques, independent complexity metrics are gained. With these results, complexity aspects of a software system can be defined. This makes a complexity comparison through system classes possible, which helps getting an overview on large systems. These process steps are then illustrated with an industrial example.