Qualitative Simulation and the Study of Software Evolution

Empirical studies of software evolution will benefit from simulation techniques such as qualitative simulation, which addresses the sparseness of the empirical data and the imprecise knowledge about the relationships between the attributes being modeled. This paper introduces the technique and outlines qualitative simulation results of various high level models of the software evolution. It also shows how empirical data can be abstracted into a qualitative form, and the results compared to the qualitative simulation traces. The study concludes that the qualitative patterns observed in several different software systems studied display remarkable similarities and that similar patterns are produced by the models under specific external conditions. Results provide support for the models as planning aids and suggest that qualitative abstraction and simulation have great potential for further research in this area. 1