A Metrics-Based Evolutionary Approach for the Component Selection Problem

Component-Based Software Engineering is concerned with the assembly of pre-existing software components that leads to software systems that respond to client specific requirements. Component selection and component systems assembly have become two of the key issues involved in this process. There may be different alternative components that can be selected (to obtain a configuration), each coming at their own set of offered requirements. Our goal is to select the best existing component alternatives. In order to attain this goal, the evaluation of the components is a necessity. As a result in this direction, software metrics are very useful being a mean to quantify those attributes considered important for the system that will built. We formulate the problem as multiobjective, considering different metrics values. The approach is an evolutionary computation technique. The experiments and comparisons (considering different criteria and various number of individuals and iterations) shows the importance of multicriteria-based solution.