Hide/Show Apps

A measurement framework for component oriented software systems

Salman, Nael
A measurement framework is presented for component oriented (CO) software systems. Fundamental concepts in component orientation are defined. The factors that influence CO systems’ structural complexity are identified. Metrics quantifying and characterizing these factors are defined. A set of properties that a CO complexity metric must satisfy are defined. Metrics are evaluated first using the set of properties defined in this thesis and also using the set of properties defined by Tian and Zelkowitz in [84]. Evaluation results revealed that metrics satisfy all properties in both sets. Empirical validation of metrics is performed using data collected from graduate students’ projects. Validation results revealed that CO complexity metrics can be used as predictors of development effort, Design effort, integration effort (characterizing system integrabiltiy), correction effort (characterizing system maintainability), function points count (characterizing system functionality), and programmer productivity. An Automated metrics collection tool is implemented and integrated with a dedicated CO modeling tool. The metrics collection tool automatically collects complexity metrics from system models and performs prediction estimations accordingly.