No abstract

In model-driven software engineering, models are used in all phases of the development process. These models may get broken due to various editions throughout their life-cycle. There are already approaches that provide an automatic repair of models, however, the same issues might not have the same solutions in all contexts due to different user preferences and business policies. Personalization would enhance the usability of automatic repairs in different contexts, and by reusing the experience from previous repairs we would avoid duplicated calculations when facing similar issues. By using reinforcement learning we have achieved the repair of broken models allowing both automation and personalization of results. In this paper, we propose transfer learning to reuse the experience learned from each model repair. We have validated our approach by repairing models using different sets of personalization preferences and studying how the repair time improved when reusing the experience from each repair.

In the last few years, telerehabilitation and telecare have become important topics in healthcare since they enable people to remain independent in their own homes by providing person-centered technologies to support the individual. These technologies allows elderly people to be assisted in their home, instead of traveling to a clinic, providing them wellbeing and personalized health care. The literature shows a great number of interesting proposals to address telerehabilitation and telecare scenarios, which may be mainly categorized into two broad groups, namely wearable devices and context-aware systems. However, we believe that these apparently different scenarios may be addressed by a single context-aware approach, concretely a vision-based system that can operate automatically in a non-intrusive way for the elderly, and this is the goal of this paper. We present a general approach based on 3D cameras and neural network algorithms that offers an efficient solution for two different scenarios of telerehabilitation and telecare for elderly people. Our empirical analysis reveals the effectiveness and accuracy of the algorithms presented in our approach and provides more than promising results when the neural network parameters are properly adjusted.

Metamodels play a crucial role in any modeling environment as they formalize the modeling constructs underpinning the definition of conforming artifacts, including models, model transformations, code generators, and editors. Understanding the structural characteristics and the quality of the metamodels that are available in public repositories before their reuse is a critical task that demands the adoption of different tools, which might not be easy to adopt. Even the selection of metamodels to be used for experimenting with new tools is not straightforward as it involves exploring various sources of information and dig in each metamodel to check its appropriateness for the evaluation of the tool under development. In this paper, we present a dataset of metamodels, which has been collected for experimenting with different approaches conceived by the authors. The dataset has been automatically curated using a toolchain, which has been redesigned post-ante the definition of the proposed approaches to foster its future reuse. CCS CONCEPTS • Software and its engineering → Model-driven software engineering; Software system models; Software system structures; Software organization and properties.

No abstract

In model-driven software engineering, models are used in all phases of the development process. These models may get broken due to various editions during the modeling process. To repair broken models we have developed PARMOREL, an extensible framework that uses reinforcement learning techniques. So far, we have used our version of the Markov Decision Process (MDP) adapted to the model repair problem and the Q-learning algorithm. In this paper, we revisit our MDP definition, addressing its weaknesses, and proposing a new one. After comparing the results of both MDPs using Q-Learning to repair a sample model, we proceed to compare the performance of Q-Learning with other reinforcement learning algorithms using the new MDP. We compare Q-Learning with four algorithms: Q(λ), Monte Carlo, SARSA and SARSA (λ), and perform a comparative study by repairing a set of broken models. Our results indicate that the new MDP definition and the Q(λ) algorithm can repair with faster performance. CCS CONCEPTS • Software and its engineering → Model-driven software engineering; • Theory of computation → Reinforcement learning.

Models are core artifacts of modern software engineering processes, and they are subject to evolution throughout their life cycle due to maintenance and to comply with new requirements as any other software artifact. Smells in modeling are indicators that something may be wrong within the model design. Removing the smells using refactoring usually has a positive effect on the general quality of the model. However, it could have a negative impact in some cases since it could destroy the quality wanted by stakeholders. PARMOREL is a framework that, using reinforcement learning, can automatically refactor models to comply with user preferences. The work presented in this paper extends PARMOREL to support smells detection and selective refactoring based on quality characteristics to assure only the refactoring with a positive impact is applied. We evaluated the approach on a large available public dataset to show that PARMOREL can decide which smells should be refactored to maintain and, even improve, the quality characteristics selected by the user.

Artificial intelligence has already proven to be a powerful tool to automate and improve how we deal with software development processes. The application of artificial intelligence to model-driven engineering projects is becoming more and more popular; however, within the model repair field, the use of this technique remains mostly an open challenge. In this paper, we explore some existing approaches in the field of AI-powered model repair. From the existing approaches in this field, we identify a series of challenges which the community needs to overcome. In addition, we present a number of research opportunities by taking inspiration from other fields which have successfully used artificial intelligence, such as code repair. Moreover, we discuss the connection between the existing approaches and the opportunities with the identified challenges. Finally, we present the outcomes of our experience of applying artificial intelligence to model repair.