Using a Configurator for Predictable Component Composition

Predicting the properties of a component composition has been studied in component-based engineering. However, most studies do not address how to find a component composition that satisfies given functional and nonfunctional requirements. This paper proposes that configurable products and configurators can be used for solving this task. The proposed solution is based on the research on traditional, mechanical configurable products. Due to the computational complexity, the solution should utilise existing techniques from the field of artificial intelligence. The applicability of the approach is demonstrated with KumbangSec, which is a conceptualisation, a language and a configurator tool for deriving component compositions with given functional and security requirements. KumbangSec configurator utilises existing inference engine models to ensure efficiency.