Entropy as a Measure of Object-Oriented Design Quality

In this paper, object-oriented designs are approached from an information theoretic point of view and entropy is proposed as a design quality metric. One of the primary aims of object-orientation is the flexibility and the ease in extending a system’s functionality, with limited alterations to existing modules. This feature is evaluated defining an appropriate probability space according to the number of unary associations enabling the definition of an entropy metric. The entropy of the next generation of an object-oriented system with enhanced functionality remains close to its previous level in case the added functionality affects a limited number of existing classes; on the other hand, a poorly designed system increases entropy drastically. In this way, not only a given system is evaluated but it is also possible to assess the degradation of a system and its “distance” from the original design.