This is the question that we (Richard Paige and myself) try to answer in this a new “Expert Voice” article published in the Software and Systems Modeling Journal. Read the published manuscript, our personal open access version or keep reading for the summary and highlights.
In the column, we argue that the modeling field is rapidly evolving. As such, what constituted a good modeling research contribution ten years ago may not be the same today. We try to distill some insights of what we (and the community we aim to represent) consider today as key elements of a good research paper in the field of software and systems modeling. Such insights – which will need to evolve and adapt with time – will be useful not just for authors of new papers, but also for reviewers and editors.
You could argue whether we should try to pin down a definition of what constitutes a good contribution at all. As clearly articulated by Terence Tao in his reflection on “What is good mathematics?” by doing so we run the risk of arrogance and may fail to recognize exotic examples of good modeling. But there is also a risk in the opposite direction— to considering that all contributions to modeling are equally important, an idealist view that could confuse the members of our community, especially the more junior ones
So, we went ahead and identified six main types of modeling contributions:
- The contribution is a model
- The contribution is a language
- The contribution is a core modeling technique
- The contribution is a modeling application
- The contribution is a new modeling domain
- The contribution is a modeling tool <- “New” one! (stay tuned for upcoming news on this!)
and for each one, we try to give some hints on how a modeling contribution of that type should provide added value to the community (and therefore, how it should be evaluated). No surprise, we also state that many interesting papers could contribute to more than one category. And that there always good papers outside these categories (hopefully this reflection would quality as one of those!).
We do not forget either that all of the above is not enough if we want to increase the quality and diversity of modeling contributions. Given that modeling is an independent scientific discipline, we need to get out of our comfort zone and reach other communities that, we believe, could also be the source of great modeling contributions
Looking forward to your feedback 🦻🎙️🦻 as we hope to trigger a #community discussion 👫👫👫
FNR Pearl Chair. Head of the Software Engineering RDI Unit at LIST. Affiliate Professor at University of Luxembourg. More about me.
How about theories about modeling – e.g. cognitive theories about comprehension (a descriptive theory) or theories about the relation between “modeling” and “designing” or the relation between modeling and quality of software (i.e. in general the impact of modeling in some software engineering activity (empirical theories)).
To me, these examples would fit in the core modeling dimension (either as theory underlying some of the techniques or as validations of their usefulness / applicability).
Thanks for the thoughtful and though coverage of modeling Topic.
My company is embarking on a Journey of Open Software Model for the field of Automation. The main focus is IEC61499 a Distributed Event based software model. I have just started learning this new type of programming. As I learn the new event based style I would also like to learn more about modeling event based systems to improve my code quality. I hope this isn’t too much for a comment. If you can point me to some tutorials, books on Event based modeling and low cost or OpenSource modeling tools that would be very helpful.
Greatly appreciate you converting this topic over the years you have been totally focused. Thanks!
IEC61499 essentially defines mechatronics (or digital control) systems in terms of “functional blocks” (FBs) based on Execution Control Charts, which is a finite state automaton formalism defining the behavior of FBs. The UML-based SysML allows modeling “blocks” similar to FBs. FBs and SysML blocks can be connected via “ports” for allowing data flows and signal event flows between “blocks”.
UML/SysML limit the concept of events to specific uses in State Machine Diagrams and Activity Diagrams (“Call Event”,”Change Event”,”Signal Event”,”Time Event”). Unlike Object Event Modeling (OEM), UML/SysML do not support a general concept of events, which would include UI events and business events. For more on OEM, see https://dpmn.info/reading/OEMS
Hi Tim,
Thanks for stopping by!
On the event modeling part, have you checked OEM? https://modeling-languages.com/object-event-modeling-for-low-code-business-apps/
Regarding open source tools, OEM has one itself. You can also take a look at BESSER (for the pure low-code open source part) https://github.com/BESSER-PEARL
Thank you, the article is very nice and summarizes well the many attempts. It is time to create a body of knowledge.
The article should be enriched by one facet: The contribution is on science and art of modeling (what I call modelology – see my YouTube or my papers at ResearchGate or Academia). There are very few summaries of what modeling actually is and what it cannot do.
Something I also have to disagree with: there is good work in the past that everyone should know but only very few know about and which is therefore forgotten, even if you constantly reinvent these results and often just make them worse.
(https://www.youtube.com/@modelology-Modellkunde/videos)
How we could make sure this past knowledge is better ack and reused? I see that, often, as soon as we change terminology (e.g. from MDE to low-code) some people assume that any paper using the old terminology can be ignored, which obviously is stupid
This SOSYM contribution is really important. Hopefully, reviewers will adopt your take on this. For now, it is certainly not the case, even in the modelling community…
And regarding “modeling is an independent scientific discipline” – is it??? Specifically, does Scopus have a CiteScore category for modelling? To the best of my knowledge, the answer is no. Accordingly, our modelling publications cannot be properly assessed using the current metrics.
This observation is partiall correct. A new branch could be modelology as the art and science of models and modelling. See in Academia: “Modelology — the new Science, Life and Practice Discipline”. Scopus and CiteScore will come. Metrics have a delay. Everybody who whant to contribute is invited to participate in the MMM clubs (models–to_model_and_use–modelling). Best Bernhard Thalheim
https://www.academia.edu/122095271/Modelology_the_new_Science_Life_and_Practice_Discipline