MDE Intelligence
Co-located with MODELS. October 2, 2023. Västerås, Sweden.
Artificial Intelligence (AI) has become part of everyone's life. It is used by companies to exploit the information they collect to improve the products and/or services they offer and, wanted or unwanted, it is present in almost every device around us. Lately, AI is also impacting all aspects of the system and software development lifecycle, from their upfront specification to their design, testing, deployment and maintenance, with the main goal of helping engineers produce systems and software faster and with better quality while being able to handle ever more complex systems and software.
There is no doubt that MDE has been a means to tame until now part of this complexity. However, its adoption by industry still relies on their capacity to manage the underlying methodological changes including among other things the adoption of new tools. To go one step further, we believe there is a clear need for AI-empowered MDE, which will push the limits of "classic" MDE and provide the right techniques to develop the next generation of highly complex model-based system and software systems engineers will have to design tomorrow.
This workshop provides a forum to discuss, study and explore the opportunities and challenges raised by the integration of AI and MDE.
We would like to address topics such as how to choose, evaluate and adapt AI techniques to Model-Driven Engineering as a way to improve current system and software modeling and generation processes in order to increase the benefits and reduce the costs of adopting MDE. We believe that AI artifacts will empower the MDE tools and boost hence the advantages, and then adoption, of MDE at industry level.
At the same time, AI is software (and complex software, in fact), we also believe that such AI-powered MDE approach will also benefit the design of AI artifacts themselves and specially to face the challenge of designing "trustable" AI software.
Last but not least, although AI is the most popular branch of computer science to create and simulate intelligence, we also believe that any kind of technique that provides human cognitive capabilities and helps creating "intelligent" software are also in the scope of this workshop. An example would be the knowledge representation techniques and ontologies that can be useful on its own or support other kinds of AI techniques.
Model-driven engineering (MDE) and artificial intelligence (AI) are two separate fields in computer science, which can clearly benefit from cross-pollination and collaboration. There are at least two ways in which such integration—which we call MDE Intelligence—can manifest:
Topics of interest for the workshop include, but are not limited to:
The 5th edition of the MDE Intelligence workshop will be co-located with MODELS 2023 and aims to discuss current work and challenges at the intersection of MDE and AI.
During the workshop, there will be a session of lightning talks around the topics that fall under the scope of the workshop.
We believe that lightning talks are a great opportunity for presenters to promote their work, to receive timely and helpful feedback and to find new collaborations. At the same, these talks will be beneficial for the workshop participants as they may broaden their knowledge and they will be able to actively participate and engage in the discussion around the presented topics.
If you are interested in presenting a lightning talk, please follow the instructions below.
Presenters will have 1-2 minutes to communicate their ideas and they can choose to use up to one slide.
We encourage the submission of proposals around the following topics:
Proposals must be submitted using the submission form. The deadline for submitting proposals is September 28, 2023.
8:30 - 10:00 | Session I: chaired by Lola Burgueño and Manuel Wimmer |
---|---|
|
|
10:00-10:30 | Coffee break |
10:30 - 12:00 | Session II: chaired by Bentley James Oakes |
|
|
12:00-13:30 | Lunch break |
13:30 - 15:00 | Session III: chaired by Sébastien Mosser |
|
|
15:00-15:30 | Coffee break |
15:30 - 17:00 | Session IV: Lightning Talks and Discussion chaired by Dominik Bork |
|
At Method Grid, we have embraced the use of generative AI tools in our software development workflow. While generative LLMs can be a productivity boost for developers, they can also at times be a distraction, struggling to generate code that conforms to our domain-specific models and requirements. This has raised the question: can we integrate software models into this AI code generation workflow to generate code that more aptly adheres to Method Grid’s domain-specific language and requirements. Through a comparative analysis, we highlight the discernible differences in code generated by LLMs with and without the aid of UML diagrams and, using saliency maps, show which regions of the UML diagrams are used by LLMs during the code generation process. Finally, we discuss how such mappings of UML diagram features to generated code tokens could in the future be used to track and iteratively generate code that conforms to an evolving software model.
If you have questions, contact us by email at: mdeintelligence2023@easychair.org