Transforming Interactive Multi-objective Metamodel/Model Co-evolution into Mono-objective Search via Designer’s Preferences Extraction

Abstract

The simultaneous evolution of metamodels and models is called the meta-models/models co-evolution problem. While some Interactive/automated metamodel/model co-evolution techniques have been proposed using multi-objective search, designers still need to explore a large number of possible revised models. In this paper, we propose an approach to convert multi-objective search into a mono-objective one after interacting with the designer to identify a set of model changes based on his/her preferences. The first step consists of using a multi-objective search to generate different possible model edit operations by finding a trade-off between three objectives. Then, the designer may give feedback on some proposed solutions. The extracted preferences are used to transform the multi-objective search into a mono-objective one by generating an evaluation function based on the weights for the existing fitness functions that are automatically computed from the feedback. Thus, the designer will just interact with only one solution generated by the mono-objective search. We evaluated our approach on a set of metamodel/model co-evolution case studies and compared it to existing fully automated and interactive meta-model/model co-evolution techniques. The results show that the mono-objective search after the interaction with the users significantly improved the co-evolution changes for several widely used metamodels.