Resumen de Using multiple feature models of domains and regulations to develop configuration systems
This dissertation aims to overcome some limitations of the existing approaches for modeling and creating configuration systems in the presence of multiple domains and standards: First, we propose an approach where the diverse domains and standards are modeled using: feature models for each domain, constraint sets representing the interactions among domains, and feature models for the standards. We have defined a systematic process to create the models, including steps to review them with different experts. These models can be reused and combined to support diverse product families. Second, we propose a set of operations for feature models: Considering that standards are crosscutting concerns that can be enforced or not depending on the customer requirements, we define operations for Conditional Intersection Merge, that enforces a standard only when it is selected, and for Partial Conditional Intersection Merge, that enforces only the part of a standard related to a domain. We define an operation for Combination that performs multiple operations to combine properly the diverse models. Third, we propose a strategy to derive Configuration Systems: on one hand, automated model transformations that takes feature models of a product family and the desired user interface to produce the corresponding configuration system. On the other hand, a set of components that processes the user decisions and updates the user interface consequently. In addition, we build a complete implementation of tools that: (1) help users to combine and analyze the feature models for domains and standards, (2) derive automatically configuration systems from those models, and (3) process decisions during the configuration. Finally, we have applied our approach in a joint research project between Siemens Colombia and Universidad de los Andes to model and create configuration systems for Electrical Transformers using our approach and tools.
