Vous êtes ici

DISCOVERER: An Integrated Web-Based DSL Factory for Computational Science to Reason Over Wicked Problems

Equipe et encadrants
Département / Equipe: 
Site Web Equipe: 
http://diverse.irisa.fr/
Directeur de thèse
Benoit Combemale
Co-directeur(s), co-encadrant(s)
Olivier Barais
Jean-Raynald de Reuzy
Contact(s)
Sujet de thèse
Descriptif

A lot of problems that concern the daily life and a huge community can be seen as a wicked problem.  Wicked problems are those that have multiple interacting systems — social, ecological, and economic — a number of social and institutional uncertainties, and imperfect knowledge.   Wicked problems are not easy to solve because it is impossible to define and describe their full nature.  Additionally, wicked problems are continually evolving. As a result, managers are not facing problems that are independent of each other, but with dynamic situations that consist of complex systems of changing problems that interact with each other…

Decision making tools require lots of engineering effort. We propose to explore the use of domain specific abstractions grouped within a domain specific language to reason on these problems and their evolving models. Building these DSLs that ease the communication between domain experts, data scientists and citizens is a new challenge for the software language engineering community for several reasons:

  • First, data scientists come with their own background and technological stack. This must be correctly integrated in a DSL factory.
  • Wicked problem and computational science generally require lots of computing power and building interactive applications is still a challenge.
  • Alternatives to wicked problems come with associated uncertainty that must be considered in DSL implementation, including the semantics as well as the syntax to account for such uncertainty.
  • Working with citizens imposes to prevent to do any assumptions on the client environment. As a result, building the DSL and some applications require to get everything working on the Web.


The candidate will explore the design and implementation of DSLs for tradeoff analysis of wicked problems through what-if/for scenarios. In particular, she will extend the current state of the art to integrate data analysis and predictive model in the semantic domain of DSLs, and she will explore innovative facilities to relate inputs and outputs of such a semantic domain to support live modeling (live interactions between the inputs and outputs of the analysis).

The implementation of DSLs will be investigated in the context of a web-based architecture to support a broader engagement of the various stakeholders (incl. scientists, decision makers and the general public), and allow them to take informed decisions through domain specific views and indicators. Web-based architecture of DSL implementation raises many challenges such as, among others, dynamic adaptation of partitioning between the client and server, isolation of the IDE components as microservices in a serverless architecture, and graph-based manipulation of program representations.

Bibliographie

Jean-Michel Bruel, Benoit Combemale, Ileana Ober, Hélène Raynal. MDE in Practice for Computational Science. INTERNATIONAL CONFERENCE ON COMPUTATIONAL SCIENCE, Jun 2015, Reykjavík, Iceland. 2015. Cf. https://hal.inria.fr/hal-01141393

Benoit Combemale, Betty H.C. Cheng, Ana Moreira, Jean-Michel Bruel, Jeff Gray. Modeling for Sustainability. Modeling in Software Engineering 2016 (MiSE'16), 2016, Austin, United States. ACM, Modeling in Software Engineering 2016 (MiSE'16). https://hal.inria.fr/hal-01185800

Julien Richard-Foy, Olivier Barais, Jean-Marc Jézéquel. Efficient high-level abstractions for web programming. Jaakko Jarvi and Christian Kastner. Generative Programming: Concepts and Experiences, GPCE'13, Oct 2013, Indianapolis, United States. ACM, pp.53-60, 2013 Cf. https://hal.inria.fr/hal-00920786

Maxime Tricoire, Olivier Barais, Manuel Leduc, Johann Bourcier, François Fouquet, et al.. KevoreeJS: Enabling Dynamic Software Reconfigurations in the Browser. WICSA and CompArch 2016, Apr 2016, Venice, Italy. 2016. Cf. https://hal.inria.fr/hal-01354997

 

https://r-brain.io/ and https://shiny.rstudio.com/

https://www.eclipse.org/ice/

https://www.eclipse.org/che/

 

Début des travaux: 
2017
Mots clés: 
Sustainability systems, domain-specific language, live modeling, what-if/for scenarios
Lieu: 
IRISA - Campus universitaire de Beaulieu, Rennes