Personalized and Automatic Model Repairing using Reinforcement Learning

When performing modeling activities, the chances of breaking a model increase together with the size of development teams and number of changes in software specifications. Model repair research mostly proposes two different solutions to this issue: fully automatic, non-interactive model repairing tools or support systems where the repairing choice is left to the developer's criteria. In this paper, we propose the use of reinforcement learning algorithms to achieve the repair of broken models allowing both automation and personalization. We validate our proposal by repairing a large set of broken models randomly generated with a mutation tool.