MULTI 2021

The 8th International Workshop on Multi-Level Modelling

MODELS 2021 - Fully virtual

October 12, 2021 (Tuesday)

Workshop date: October 12, 2021 (Tuesday) 10:00 - 13:00 and 16:00 - 19:00 (CET)

The yearly MULTI workshop is the premier event for researchers and practitioners working on multilevel modelling and multilevel software development.

Together with the Models Conference, MULTI 2021 will be fully virtual and will accommodate remote participation and presentation. All accepted papers presented online will be published by IEEE.

Multilevel language architectures represent a new object-oriented paradigm both for conceptual modelling and software engineering. In contrast to conventional approaches, they allow for an arbitrary number of classification levels and introduce other concepts that foster reuse and adaptability. While multilevel languages and tools have reached a considerable maturity, the field still offers numerous challenges. The MULTI workshop series is dedicated to bring together experts who develop and apply multilevel language technologies as well as those who focus on specific analysis and design methods or on economic aspects of this new paradigm.

Multilevel modelling is an emerging new modelling paradigm that offers exciting new perspectives not only for conceptual modelling, but also for the development of software systems that are integrated with models of themselves. Multilevel DSMLs allow for combining the benefits of economies of scale with the productivity enabled by concepts that were designed for very specific domains. Multilevel modelling has been used successfully in a wide range of projects.

The MULTI series is aimed at providing a platform for exchanging ideas and promoting the further development of multilevel languages, methods and tools. In particular, the goal is to encourage the community to delineate different approaches to multilevel modelling and define objective ways to evaluate their respective strengths and weaknesses.

Three kinds of papers are solicited: regular papers (10 pages), challenge papers (10 pages), and position papers (5 pages).

Topics for regular and position papers include, but are not limited to:

  • the nature of elements in a multi-level hierarchy and how best to represent them,
  • the importance and role of deep characterization mechanisms, including potency and its variants,
  • the structure and labelling of an MLM framework,
  • methods and techniques for discovering clabjects, specializations and classification relationships,
  • formal approaches to MLM,
  • fundamental aspects of MLM such as composition and decomposition,
  • experiences and challenges in providing tool support for MLM,
  • experiences and challenges in applying MLM to large and/or real-world problems,
  • model management languages (transformation, code generation etc.) in a multi-level setting,
  • criteria and approaches for comparing MLM approaches,
  • integration of modelling and programming languages in a multi-level setting,
  • definition of behavioral semantics in a multilevel setting,
  • design patterns addressing when and how to apply multi-level metamodelling,
  • case studies illustrating the usage of multi-level techniques to overcome difficulties occurring in classic modelling scenarios, and
  • multi-level modelling of AI systems (knowledge graphs, machine learning).

Authors submit their papers as PDF files via Easychair. Submissions must adhere to the IEEE formatting instructions. Challenge papers (see Challenge paper description) must be subtitled “A contribution to the MULTI Collaborative Comparison Challenge”. Accepted papers will be included in the joint workshop proceedings published by the IEEE.

Multi-level modeling addresses the modeling of subject domains that benefit from an explicit recognition of multiple levels of domain representation, such as software development, process modeling, capturing organizational roles, biological taxonomies, product hierarchies, and so on. Over the span of two decades many approaches for multi-level modeling have been proposed, all sharing the goal of extending traditional two-level approaches with constructs and concepts that naturally support multiple levels of domain representation, with the goal to increase model expressiveness while simultaneously reducing model complexity.

Numerous advances in multi-level modeling approaches and tools have, however, lead to a proliferation of available approaches, thus displaying a lack of consensus on what kinds of constructs and concepts provide the best support for multi-level modeling. In part, differences are owed to different application targets or different prioritizations of desirable model properties, yet not all existing differences can necessarily be motivated in this manner. Some differences at both foundational and realization levels may be perfectly justifiable while others may be reconcilable without diminishing effects.

The Collaborative Comparison Challenge aims towards increasing communication between multi-level modeling researchers by encouraging collaborations which may justify and thus clarify the need for existing differences, or, alternatively, lead towards homogenizing multi-level modeling. Previous challenges (the 2017 Bicycle Challenge and the 2019 Process Challenge) already invited researchers to demonstrate their approaches by addressing a set of requirements in a given domain and thus represented essential first steps towards the benchmarking of various approaches.

However, since these challenges only focused on a single approach respectively and did not specifically encourage the contrasting of approaches beyond regular related work discussions, their value in contrasting approaches and fostering a dialogue between researchers was limited. For this reason, the Collaborative Comparison Challenge specifically requires the application of two or more approaches to one domain example and mandates the discussion of commonalities and differences between the approaches in a joint paper authored by proponents of different multi-level modeling approaches. Commonalities and differences should be discussed as they manifest themselves in the treatment of the domain example but also at a more general level. Respective discussion subjects which authors may choose to elaborate on include, but are not limited to, fundamental concepts such as the nature of levels, cross-level relationships, classification vs generalization, deep characterization, the treatment of attributes and operations, and the use of structural and behavioral constraints. Discussions should seek to explore justifications for, and/or potential reconciliations of, fundamental differences rather than surface-level realization choices. An optional avenue towards contributing to the clarification of differences is the formalization of foundational concepts, thereby possibly discovering open questions and/or potential for unification.

All submission requirements and the domain example to use are available from the detailed MULTI 2021 challenge description.

Important dates

  • Paper Submission: 22nd 30th July 2021 (extended)
  • Authors Notification: 21st August 2021
  • Camera-ready Papers: 28th August 2021
  • Workshop: October 12, 2021 (Tuesday) 9:00 - 12:00 and 15:00 - 18:00 (London BST, UTC+01:00)

Steering Committee

  • Colin Atkinson (University of Mannheim, Germany)
  • Thomas Kühne (Victoria University of Wellington, New Zealand)
  • Juan de Lara (Universidad Autónoma de Madrid, Spain)

Workshop Organizers


Victorio Albani Carvalho

Federal Institute of Espírito Santo (IFES), Brazil


Gergely Mezei

Budapest University of Technology and Economics, Hungary


Bernd Neumayr

Johannes Kepler University Linz, Austria

Program Committee

  • Victorio Albani Carvalho, Federal Institute of Espírito Santo
  • Joao Paulo Almeida, Federal University of Espirito Santo
  • Colin Atkinson, University of Mannheim
  • Mira Balaban, Ben-Gurion University of the Negev
  • Tony Clark, Aston University
  • Marian Daun, University of Duisburg-Essen
  • Juan De Lara, Universidad Autonoma de Madrid
  • Cesar Gonzalez-Perez, Incipit CSIC
  • Georg Grossmann, University of South Australia
  • Jens Gulden, Utrecht University
  • Manfred Jeusfeld, University of Skoevde, School of Informatics
  • Monika Kaczmarek, University Duisburg Essen
  • Thomas Kuehne, Victoria University of Wellington
  • Fernando Macias, IMDEA Software Institute
  • Gergely Mezei, Budapest University of Technology and Economics
  • Bernd Neumayr, Johannes Kepler University Linz
  • Chris Partridge, University of Westminster, BORO Solutions
  • Andreas Prinz, University of Agder
  • Adrian Rutle, Western Norway University of Applied Sciences
  • Christoph G. Schuetz, Johannes Kepler University Linz
  • Markus Stumptner, University of South Australia
  • Manuel Wimmer, Johannes Kepler University Linz

Fully Virtual via Whova/Zoom - Room 5

Session 1 – MLM for MDE
Tuesday, 12 October, 10:00 – 11:15 (CET)
Chair: Bernd Neumayr

  • Welcome and Opening (Preface) (5min)
  • Keynote by Juan de Lara: Multi-level modelling for model-driven engineering (60 min) Slides
  • Discussion

Session 2 – Foundations
Tuesday, 12 October, 11:30 – 13:00 (CET)
Chair: Victorio Albani Carvalho

  • Joao Paulo Almeida, Victorio Albani Carvalho, Claudenir Fonseca and Giancarlo Guizzardi: A Note on Properties in Multi-Level Modeling (10 min + 5 min discussion) Slides
  • Daniel Toepel: Associations in Multi-Level-Modelling: Motivation, Conceptualization, Modeling Guidelines, and Implications for Model Management (15 min + 5 min discussion) Slides
  • Harald Koenig and Uwe Wolter: A General Methodology for Internalising Multi-Level Model Typing (15 min + 5 min discussion) Slides
  • Ulrich Frank: Prolegomena of a Multi-Level Modeling Method (15 min + 5 min discussion) Slides
  • Discussion

Session 3 – Applications
Tuesday, 12 October, 16:00 – 17:10 (CET)
Chair: Manfred Jeusfeld

  • Sybren de Kinderen, Monika Kaczmarek-Heß and Kristina Rosenthal: Towards an Empirical Perspective on Multi-Level Modeling and a Comparative Assessment to Conventional Meta Modeling (10 min + 5 min discussion) Slides
  • Sebastian Gruber, Bernd Neumayr, Michael Schrefl and Josef Niebauer: Towards Multi-level Modeling of Just-in-Time Adaptive Interventions (JITAIs) in Mobile Health (10 min + 5 min discussion) Slides
  • Maria Teresa Rossi, Martina Dal Molin, Ludovico Iovino, Martina De Sanctis and Manuel Wimmer: Leveraging Multi-Level Modeling for Multi-Domain Quality Assessment (15 min + 5 min discussion) Slides
  • Discussion: applications of MLM, current trends in modeling and how to apply MLM for them

Session 4 – Comparison and Unification
Tuesday, 12 October, 17:20 – 18:50 (CET)
Chair: Gergely Mezei

  • Manfred Jeusfeld and Ulrich Frank: Unifying multi-level modeling: A position paper (10 min + 5 min discussion) Slides
  • Introduction to the Collaborative Comparison Challenge (5 min) Slides
  • Sándor Bácsi, Arne Lange, Thomas Kuehne, Gergely Mezei and Colin Atkinson: Melanee and DMLA – A Contribution to the MULTI 2021 Collaborative Comparison Challenge (15 min + 5 min discussion)
  • Discussion: the challenge and other ways of unifying MLM Slides

Closing words (10 min)
Tuesday, 12 October, 18:50 – 19:00 (CET)

Screenshot from MULTI 2020
Screenshot from MULTI 2020