Usage of dynamic decision models as an agile approach to COTS taxonomies construction

The rapid expansion of the COTS domain causes a "snowball" effect to the growth of information about and used by components. One of the proofed methods to organize information is taxonomies. Although there exist many upper and domain-specific taxonomies, the need to create or extend a taxonomy often arises. However, the effort needed for the construction of a successful COTS taxonomy could be an obstacle because it requires expertise and skills different to the ones needed for COTS development. This and many other problems can drive COTS developers to abandon the taxonomies. This paper proposes an agile approach for COTS taxonomies development based on our work in the MERCED project. The method introduces work separation and step-by-step development in the process of taxonomy construction. First the approach allows a quick start using a high-level taxonomy, created without spending too much effort for its development. Then, in parallel with taxonomy use, the approach allows its extension using dynamic decision models. These dynamic decision models simplify not only the taxonomy development, but also many other activities as quality control, documentation and configuration control. In this paper, we also present a real-world example showing how the method operates within the domain of real-time embedded software components.