Ontology-Based Quality Attributes Prediction in Component-based Development

Despite the success that Component-Based Development (CBD) has achieved so far, component mismatch remains as a big obstacle for wider and smoother component reuse. Mismatch refers that the selected component does not satisfy the functional requirements, or that it fails the user's expectation in terms of the Quality Attributes (QAs) of the component-based system. This allows us the potential to predict the quality attributes of a software system by analysing the result of component retrieval. In this paper, applicable quality attributes for prediction are selected by investigating existing software quality attributes. A novel ontology-based approach was proposed to achieve precise component retrieval and quality attribute prediction. The approach contains three steps: the first is to develop a Quality Attributes Oriented Component Specification ontology model (QAOCS), where applicable QAs related knowledge of application domains were integrated into the ontology. The second is to establish an ontology-based QAs oriented component retrieval method to retrieve components according to the reuse requirements. The third is to predict the quality attributes of component-based system on the basis of the matching information. Based on these three steps, a prototype tool with an example component repository was built to verify and scale up the approach.