Empirical Validation of Component-based Software Systems Generation and Evaluation Approaches

Abstract

Component-based software development needs to formalize a process of generation, evaluation andselection of Composite COTS-based Software Systems (CCSS), enabling software architects to make earlydecisions; the Azimut approach and its associated software tool were proposed to tackle this problem.This article presents an experimental study conduced to compare Azimut approach with a SystematizedAd-Hoc approach, regarding generated solutions quality, cost and e®ort. Results suggest that: (1) Azimutgenerate better quality solutions at lower cost, but not statistically signi¯cant, and (2) there is strongevidence showing that the e®ort required is higher than for Systematized Ad-Hoc approach; re-samplingmethods (Bootstrap and Jackknife) were applied to reinforce these conclusions. Also this study serves asa framework for validating approaches, process and tools for generating and evaluating component-basedsoftware systems.