Identification d'une architecture a base de composants dans une application orientee objets a l'aide d'une analyse dynamique

A system is complex and particularly difficult to understand and to maintain when it is described with a large number of highly interdependent parties. An object-oriented application is often complex because it uses hundreds or thousands of classes with many different dependencies more or less explicit. The same application, using the component paradigm, contains a smaller number of loosely coupled parties, highly cohesive with clear inter-dependencies. Indeed, because the component paradigm provides a high-level representation, synthetic and well-organized structure of complex systems, it can provide a space of projection for object-oriented applications. Such projection facilitates the step of understanding a system prior to any activity of maintenance and/or evolution. In addition, it is possible to use this representation as a model to perform a complete restructuring of an operational object-oriented application into its equivalent operational component-based application. Thus, the new form of the application benefits from all the good properties associated with the component-oriented paradigm. The goal of my thesis is to propose a semi-automatic approach to identify a component-based architecture in an object-oriented application. This architecture should help in understanding the original application, but also simplifies the projection of the object-oriented application on a concrete component model. The identification of a component-based architecture is achieved in three main steps : i) obtaining data for the identification process. These data, which correspond to dependencies between classes, are obtained with a dynamic analysis of the target application. ii) identification of the components. Three methods were explored. The first uses the formal concept analysis, the second two meta-heuristics and the last a multiobjective meta-heuristic. iii) identification of the component-based architecture representing the target application. This is done by identifying the provided and required interfaces for each component. To validate this identification process, and the different choices made during its development, I realized several case studies. Finally, I show the feasibility of the projection of the identified component-based architecture on a specific component model.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.