Abstract

Many academic, industrial, and government research units have unanimously acknowledged the importance of developing dependable software systems. At the same time they have also concurred on the difficulties and challenges to be surmounted in achieving the goal. The importance of domain analysis and linking domain models to software artifacts were also recognized by various researchers. However, no formal approach to domain analysis was attempted. The primary motivation for this thesis stems from this context. Component-based software engineering offers some attractive mechanisms to tackle the inherent complexity in developing dependable systems. Recently a formal approach has been put forth for such a development. This thesis provides a formal approach for domain analysis, and transforms the domain model to components desired by this development process.
Formal Concept Analysis (FCA) is a mathematical theory for identifying and classifying concepts. This thesis taps its potential to formally analyze the domain in a software development context. It turns out that the approach presented in this thesis cannot be fully automated; nevertheless several useful contributions have been made. These include (1) capturing formal concepts and defining them in FCA; (2) defining composition rules to categorize formal concepts and their trustworthy properties; (3) integrating partial formal context tables to build the concept lattice; (4) specifying and developing a model transformation approach to construct trustworthy OWL ontology; (5) implementing a model transformation technique to generate the TADL specification of the reusable component-based system. The proposed approach is applied to CoCoME, as a benchmark case study in the domain of component-based development.