Model Fragment Reuse Driven by Requirements

Clone-and-Own is a common practice in families of software products, where parts from legacy products are reused in new devel opments. In industrial scenarios, CAO consumes high amounts of time and effort, not guaranteeing good results. We propose a novel approach, Computer Assisted CAO for Models (CACAO4M), that uses a Multi Objective Evolutionary Algorithm (MOEA) with two objectives (Model Fragment Similitude, and Model Fragment Understandability) to rank relevant model fragments for reuse. We evaluated our approach in the industrial domain of train control software. Our approach outperforms the results of a baseline that uses only the Model Fragment Similitude metric, which encourages us to further research in this direction.