Year: 2017
Type: dissertation
Abstract: Most Software Product Lines are built from a set of existing products, that is re-engineered into reusable assets following feature location approaches. Traditional feature location approaches target program code, neglecting other software artifacts as models.\nWe present FLiMEA, an approach for Feature Location in Models by an Evolutionary Algorithm. FLiMEA capitalizes on experts domain knowledge to boost the feature location process and produces model fragments that properly capture the reusable units of the domain. FLiMEA performs a search (guided by a fitness function) over alternative feature realizations (generated through genetic operations). \nFeatures and their realizations must evolve over time. We propose Variable MetaModel (VMM), an approach based on variability modeling ideas applied at metamodel level to enable the co-evolution of the model fragments and the language used to create them. \nFLiMEA and VMM have been evaluated in our industrial partners, BSH and CAF. We explore different genetic operations and fitness functions so FLiMEA can be tailored to work under different industrial scenarios and we check that VMM is able to cope with the evolution of the features.
Author: Jaime Font Burdeus
Language: English
Subfield: Software
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W2787332107
