In Model-Driven Engineering (MDE) design of modelling languages and model transformations is still an expert's task. Domain experts, i.e. the stakeholders of languages and transformations, would like to independently de ne and use their own MDE ecosystem, but can only support those activities. In this position paper we discuss this problem from a more general perspective, arguing for the need of emergent MDE, that is modelling languages and transformations shall be inferred from the usage it is done of modelling concepts. Moreover, the paper outlines a possible research agenda with corresponding challenges towards the goal of emergent MDE ecosystems.
metamodels and model transformations are created correspondingly. In other
words, the creation of an MDE framework is conceived as a bootstrap activity,
preliminary of any application design [
In this paper, we argue that the support for MDE should be reconsidered to enable also emergent approaches. More precisely, languages and transformations should incrementally inferred as the result of modelling needs and uses rather than being xed a-priori. Therefore, an ecosystem grows and is re ned in order to satisfy the modelling needs conveyed by users' usage. Emergent ecosystems would be naturally agile and exible, since they would include all and only the needed features, as captured and encoded through usage patterns. Admittedly, this paradigm shift does not come for free: in the remainder of this work, rst a discussion on the concepts of syntax, semantics, and ontology is proposed, and then a research agenda together with expected challenges and possible solution is illustrated. 2
In general, modelling activities have to deal with the interplay of syntax,
semantics, and ontology. The boundaries between the three are quite blurry. The
abstract syntax of a language, i.e. its metamodel, is supposed to be { indeed
{ only syntax. However, well-formedness constraints do include domain-speci c
information (e.g., types or multiplicities). With respect to semantics, it is usually
distinguished in structural and behavioural [
In software engineering ontological information is typically seen as a uniquely
de ned representation of a certain domain for reasoning purposes. On the
contrary, other elds of science as philosophy consider ontology as agent-based, that
is strictly related to the observer [12]. Going into more details, depending on the
amount of knowledge about a certain domain, di erent agents, or modelling
stakeholders in our case, would interpret the same information in di erent ways.
As an example, a sub-atomic physician would have a quite di erent perception
of a rainbow if compared to someone who is completely agnostic about physics.
This is also the cause of the intrinsic ambiguity of abstractions, because some
details relevant for the former observer would be totally irrelevant for the latter,
respectively [
By looking to the current MDE techniques, what is missing is a solid support of ontological concerns and the customization demands they introduce. Languages and transformations speci cation facilities should be aware of agentbased ontology and allow adaptability to the di erent degrees of knowledge a stakeholder might have. Therefore, ontology represents the main factor to be taken into account for customization purposes. This concept is instantiated quite naturally in the form of views, i.e. given a certain language di erent experts might need di erent information depending on their perspective. Moreover, consistency issues should be addressed as limited to the information available to each particular view.
An appropriate support for emergent MDE ecosystems should be able to infer languages and transformations as derivable from users' usage patterns, and hence relieve stakeholders from customization e orts. The ecosystem should be de ned in a user-speci c ontological layer, over the base syntax and semantics speci cations, and not as proper new languages and transformations [13]. 3
As discussed so far, in this paper we propose to tackle exibility and usability
issues of MDE ecosystems by adopting an induction strategy [
The base knowledge: i) MDE repositories, ii) composition operators, iii) common semantics.
An emergent approach relies on the existence of base/concrete knowledge that
can be used to create further abstractions. One option to store such
knowledge is to rely on a repository of existing languages and transformations, and
consequently to create a lattice of relationships between domain-speci c
languages [14]. Graph databases are a suitable support to manage these mechanisms
since they rely on nodes (portions of languages) and edges (relationships between
languages); moreover, they naturally allow pattern-based reasoning about the
information they store. In this scenario, the major problems would be due to
the creation of operators for (sub-)languages composition, and in particular for
semantics, that is not guaranteed to be compositional [15, 16]. In this respect,
interesting solutions enable re-use of metamodels and transformations portions
through adaptation [
The ontological layer: iv) ontology-based matching of modeling patterns, v) multilevel transformations, vi) base knowledge evolution and ontology layer co-evolution.
The ontological layer is meant to support user-tailored customizations, and is
incrementally extended and re ned depending on the patterns traversed by a
certain user. In particular, users can list a personal set of concepts and
relationships that have to be matched in the repository. Once the set is considered
satisfactory, a user has the option of renaming elements and relationships
between them, and to de ne a custom concrete syntax. As previously discussed,
these aspects pertain to the user ontology, i.e. the way in which a certain
observer perceives the reality, and therefore do not change the base knowledge.
Technically, these layers can be implemented by means of multi-level modelling
solutions, which support multiple instantiation layers: existing works already
deal with creating ontological layers of metamodels, including the de nition of
concrete syntaxes [
Another major concern is represented by possible changes in the base
knowledge and their propagation to existing ontological layers [12]. In fact, there shall
be a way to assess if changes are breaking a certain ontological layer, both in
terms of model entities and transformations. Moreover, it should be possible to
understand if the chosen concrete syntax might be a ected by the changes. This
issue shares a lot of open research problems with metamodel evolution and the
co-evolution of related artefacts [
This position paper discusses the problems related to MDE ecosystems usability
and exibility and proposes a research direction aiming at realizing emergent
customization mechanisms. We advocate the de nition of a clearer distinction
between syntax, semantics, and ontological aspects of modelling and illustrate a
research agenda towards emergent MDE. It is interesting to notice that most of
the intricacies discussed in the research agenda raise also in the development of
cyber-physical systems (CPSs): as a matter of fact, their modelling demands the
integration of multi-paradigm modelling techniques, di erent levels of
abstractions, and disparate semantics. Available solutions tend to create tool chains
in which a shared semantics, typically a model of computation, is exploited as
a common denominator devoted to consistency management and synthesis of
the resulting integration. The repository mechanism mentioned in this work can
be conceived as a generalization of modelling solutions adopted for CPSs, and
the adoption of models of computation as shared semantics is an interesting
investigation direction [
The author would like to thank Federico Ciccozzi and Alfonso Pierantonio for the interesting discussions on the topics covered in this work. 12. Ermolayev, V., Davidovsky, M.: Agent-based ontology alignment: Basics, applications, theoretical foundations, and demonstration. In: Proceedings of the 2Nd International Conference on Web Intelligence, Mining and Semantics. WIMS '12, New York, NY, USA, ACM (2012) 3:1{3:12 13. Izquierdo, J.L.C., Cabot, J., Lopez-Fernandez, J.J., Cuadrado, J.S., Guerra, E., Lara, J.: Engaging End-Users in the Collaborative Development of Domain-Speci c Modelling Languages. In: Cooperative Design, Visualization, and Engineering: 10th International Conference, CDVE 2013, Alcudia, Mallorca, Spain, September 22-25, 2013. Proceedings. Springer Berlin Heidelberg, Berlin, Heidelberg (2013) 101{110 14. Hebig, R., Seibel, A., Giese, H.: On the Uni cation of Megamodels. In Amaral, V., Vangheluwe, H., Hardebolle, C., Lengyel, L., Magaria, T., Padberg, J., Taentzer, G., eds.: Proceedings of the 4th International Workshop on Multi-Paradigm Modeling (MPM 2010). Volume 42 of Electronic Communications of the EASST. (2011) 15. Jezequel, J.M., Combemale, B., Barais, O., Monperrus, M., Fouquet, F.: Mashup of metalanguages and its implementation in the kermeta language workbench. Softw.
Syst. Model. 14(2) (May 2015) 905{920 16. Vallecillo, A.: On the combination of domain speci c modeling languages. In: Proceedings of the 6th European Conference on Modelling Foundations and Applications. ECMFA'10, Berlin, Heidelberg, Springer-Verlag (2010) 305{320 17. Basciani, F., Di Ruscio, D., Iovino, L., Pierantonio, A.: Automated Chaining of Model Transformations with Incompatible Metamodels. In: Model-Driven Engineering Languages and Systems: 17th International Conference, MODELS 2014, Valencia, Spain, September 28 { October 3, 2014. Proceedings. Springer International Publishing (2014) 602{618 18. Bryant, B.R., Gray, J., Mernik, M., Clarke, P.J., France, R.B., Karsai, G.: Challenges and directions in formalizing the semantics of modeling languages. Comput. Sci. Inf. Syst. 8(2) (2011) 225{253