Enterprises nowadays operate in fast-changing environments and need to adapt dynamically to new circumstances. This impacts the way how enterprise information systems are analysed, designed and implemented. Conceptual modelling methods experience a constant evolution nowadays. The methods are re-constructed continuously to reflect the changing application/industrial domain. It is therefore required to provide additional design support to realise domain specific modelling methods and more precisely their underlying metamodel. The goal of this research project is to enable the knowledge representation of a metamodel to support its design process. Building upon the assumption that conceptual structures within a metamodel exist, a knowledge representation framework is proposed using conceptual graphs as the mathematical baseline. The proposal will be evaluated in a laboratory setting, applied on metamodel design challenges observed.
Enterprises operate today in fast changing environments. External factors influence and disrupt their business models; technology advances at a rapid pace and changes the way organisations offer their products and services; legal/regulatory requirements need to be reviewed continuously and respected.
All these factors influence the way information systems are designed, analysed and used. Enterprise modelling methods are an established means for the conceptual analysis, design and implementation of complex enterprise systems. They models realised using these methods provide value as a documentation ∗Supervisor: o. Univ.-Prof. Prof.h.c. Dr. Dimitris Karagiannis Copyright 2018 for this paper by its authors.
Copying permitted for private and academic purposes.
means and to support understanding by human beings[
In the field of conceptual modelling, we can distinguish between general purpose approaches, that aim to provide a standardised conceptual modelling foundation for a specific application field and domain-specific modelling methods that target specific interpretation and operation needs of an enterprise. The metamodel reflects these aspects at the core of any domain-specific modelling method: model-based operations define the concepts required to syntactically and semantically describe the organisation (or a subset of it), and enable model processing to create value.
Within this research project, domain-specific modelling methods are in
focus, specifically targeting the challenge of designing ”adequate” metamodel to
support dynamic (in the sense of quickly evolving) and complex (vertically and
horizontally integrated) environments. An initial understanding of ”adequate”
in relation with a domain-specific metamodels has been derived from [17, p. 5]
where usefulness defines scope of the abstraction performed in the design
process. Requirements derived during the design and runtime aspects influence the
design process. As a solution proposal, it is intended to extend the notion of
metamodel towards a conceptual structure. The term conceptual structure is
understood and closely linked to the work performed by Sowa in [
Problem Observed. The design process for metamodels as the core element of a modelling method is observed as a challenging issue: the task to derive a consistent metamodel that is adequate for a specific domain and application, enables model value and provides interpretation support for stakeholders involved, requires the metamodel engineer requires conceptual, technological and domainspecific knowledge to base design decisions upon and weight in and compromise on different interests.
Observing current practices in the field of metamodelling, these knowledge
challenges are introduced using the example of a development process observed
graphically shown in Fig. 1.
1. Understand requirements: in order to develop an adequate and useful
metamodel, the requirements for knowledge operations have to be captured and
understood. Knowledge on the application and industrial domain is required
to trigger the design process, e.g. simulation of manufacturing processes,
dependency analysis of community networks, verification of deadlocks in
automated business processes, integration of metamodel on different
formalisation levels into a consistent state for enterprise architecture management
in a given industrial domain.
2. Select a meta-modelling technique: based on these requirements, the
appropriate meta-modelling technique has to be selected. The technique provides
generic constructs to develop a metamodel, construction principle and usage
patterns and consequently enables specific functionalities e.g. code
generation, reasoning techniques on ontological metamodels, deduction in logic
based environments. The functionalities offered have to be mapped to the
requirement as a fundamental design decision that has to be taken.
Knowledge on specific techniques applicable is needed.
3. Design the metamodel: having accomplished these 2 prerequisites, the
metamodel is designed in conformance with the selected metamodelling approach.
The design techniques are typically adjustment, mapping or re-use patterns
and the resulting metamodel is strongly influenced by the expertise and
creativity of the metamodel engineer in the overall approach, design decisions
taken and specific usage techniques.
4. Enable metamodel operations: knowledge operations realise the value of
models, enabled by the corresponding metamodel. These operations are
established functionalities like composition (e.g. in case of sliced metamodel as
discussed in [
These knowledge-based challenges strongly influence the efficiency to establish an adequate design of a metamodel. The selection of a meta-modelling technique is currently driven by the metamodel engineers knowledge on a specific approach rather than the requirements elicited; during the process adjustments are performed to overcome this initial selection issue. It is assumed that these adjustment have an impact on the usefulness of the resulting metamodel. As a description framework of specific, existing metamodels is not available, a re-use of results is limited. This results in a re-implementation of similiar structures (using the same or a different technique).
The motivational example below showcases the observation. Different metamodelling techniques have been applied to represent the same metamodel requirement from a structural point of view: the sequence definition of a business process. The application of different approaches results in varying functionality based on the design and technique used.
Motivational Example: BPMN 2.0. The observation on the construction
of meta-models and its dependency on the meta-modelling approach is visually
shown in Fig. 5. The Business Process Model and Notation (BPMN) 2.0 [
The representation shows the aspect of sequence flows (highlighted in red/dashed
line) of the BPMN metamodel and how it has been designed using three different
meta-modelling techniques: a) the formal specification from the BPMN 2.0
standard document using class diagrams [24, p. 144], b) an ontology-based approach
to conceptualise the specification [25, p. 140] and c) a logic-based representation
using rules mapped to petri net constructs (introduced in [15, p. 51] extended in
[
It is assumed that the problem described using the motivational example is not only applicable for pre-existing metatmodels (and would result in an integration challenge of different realizations) but can be observed also for domainspecific metamodels and their characteristics that are constructed from scratch. Gap. Resulting from the problem observed above, the design or (continuous) adaptation/evolutions of metamodels is a tedious, inefficient and error-prone task that requires domain expertise on one hand, meta-modelling knowledge, an overview of existing results and creativity to perform the adjustment and interpretation of requirements to a new/evolved metamodel. It is proposed to close this gap by a) extending the notion of metamodels as conceptual structures as a means to define and describe metamodels independent of the technique used and collect these conceptual structures for re-use in the engineering process of metamodels. The adequateness, usefulness and purpose of the metamodel might influence the need for expressiveness of the conceptual structure for metamodels.
The toolbox of a metamodel engineer should include this collection and evolve
along industrial trends in a dynamic manner. Supporting agile techniques as
discussed in [
Research Objectives. Based on the identified gap, the research questions are formulated below. As a design based research methdology is followed (introduced in section 3. The objective of the research project is to develop a knowledge representation formalism as a conceptual structure for metamodels, representing the domain knowledge (syntax, notation, semantic, behaviour) encapsulated as elements in the structure. The design artefacts realized as the contribution are mapped to research objectives and questions depicted graphically in Fig. 3.
RQ1 aims to develop a framework for knowledge representation as conceptual structures that is adequate to describe metamodels. It is currently assumed that this knowledge representation can be derived from a) the definition of the term ”metamodel” in the scientific community, b) existing metamodels from academia and industry and their characteristics and construction principles, c) meta-modelling techniques and patterns that are currently applied, and d) knowledge operation as requirements on metamodels. Based on these preliminary input identification, the objective of the research questions is to identify a formalism that allows the description of metamodels using dimensions to be established as part of the research questions (e.g. syntactical, semantical, behavioural, operational, etc.).
The research questions targets the mechanisms required to a) describe and b) collect/support re-use metamodels as conceptual structure. This includes the analysis of description dimensions that enable e.g. their registration, retrieval and binding. Having the motivational example in mind, the second part of the research question targets the identification of patterns based on these established dimensions. Metamodels developed using modelling techniques of similar or varying expressiveness and functionality are reviewed to identify techniques to describe patterns and provide means for e.g. generalisation/abstraction of these patterns or fragments independent.
RQ3: Do algorithms exist that are appropriate to suggest and propose knowledge operations in a domain context using the conceptual structure? The focus of this research question is to identify how knowledge operations can be proposed for specific conceptual structures. This builds on the assumption that specific operations require a conceptual structure to operate upon. It is planned to investigate available algorithms that suggest functionality for metamodels based on expressiveness of the conceptual structure and functional requirements not limited to interoperability, semantic lifting, reference alignment, mapping or consistency management. The assumption underpinning the research question is that a metamodel needs to fulfil certain requirements to be capable to perform a concrete knowledge operation to be applicable. 3
The research objective of this project is to develop artefacts to support the
engineering of domain-specific modelling methods, more precisely the design of
its metamodels. A novel and innovative framework that integrates the
conceptual structures as a building block will be proposed that allows a knowledge
representation of metamodels. The design-science based approach introduced
by Hevner and Chatterjee in 2010 in [
In line with the research questions, the research project aims to develop
an adequate conceptual structure to represent metamodels and corresponding
matching algorithms for knowledge operations. The design of this formalism
and algorithms is evaluated through a prototypical implementation of a
domainspecific language for metamodels. As a means to iteratively refine the design
artefacts, the concept established and prototype realised are continuously
evaluated in the context of the Open Models Initiative Laboratory (OMiLAB) [
The research approach for the project is structured according to the research methodology. These phases contribute individually to the research questions and are used during the Assess-Refine iterations. The five phases are executed in line with the methodology and will be iteratively run. This impacts individual design-evaluation cycles of each phase and also transitions in between phases. In general, phase 1 and 2 are contributing to the explication and requirements elicitation objective of the project, phase 3 to a refinement, whereas phase 4 performs the development and phase 5 the evaluation of the design artefacts.
(contributing to RQ1 and RQ2) During this phase, a literature review is conducted to establish an understanding of the terminology and definition applicable for metamodels and conceptual structures. The understanding of the term metamodel and its different interpretation will impact the design of the conceptual structure and vice-versa. The applicability of the research question is verified during this phase and further refined.
Preliminary Results. A preliminary review of literature on available definitions for the term meta-model aims to establish a common baseline. The results are presented below having motivated in an early stage of the research project the identification of the gap and research questions.
– Language-based Understanding of Metamodels: Strahinger investigates in [
This aspect of meta-modelling aims to compose metamodels from fragments
that support a specific subset of the modelling methods. The discussion of
the language elements required to describe method fragments, including the
ontological anchoring as defined in [
An observation at this stage shows that the research questions are in line
with the challenges observed in literature (e.g. [
(contributing to RQ1 and RQ2)
Concrete metamodels are identified and collected in this phase. This collection
acts as a basis for further analysis on the formalisation concepts and dimensions.
Fragmentation techniques as discussed in [
Preliminary Results. The collection phase is in progress and builds at this stage
on the metamodel developed at the OMiLAB and the literature review performed
in [
(contributing to RQ1 and RQ2)
The objective of this phase is to assess and evaluate requirements and the
resulting development/evaluation results in iterations. It is responsible to capture
intermediate results and feed the forward/back to the related phases.
Preliminary Results. Results from the design/evaluation iterations are not yet
available. Preliminary work has been performed in research projects that fed
into the design phases introduced above. The results achieved are published in
[
(contributing to RQ2 and RQ3)
The objective of this phase is to design the concept and implement a
prototypical realisation for conceptual structures.
Preliminary Results. As a preliminary result a procedural framework has been
introduced using abstract metamodelling building blocks. This framework acts
as a guiding element for the research to be conducted in the review and analysis
phases. It has been proposed in [
Each building block specified couples abstract model constructs, related model processing functionality and a description of the description of the abstract metamodel building block that enables the identification and integration of it in more complex scenarios that span multiple building blocks. The description should reflect the dimensions required for the conceptual structure. 2. Concept: the transformation from the approach stage is done through instantiation. The abstract blocks are made concrete using a selected metamodelling technique. 3. Implementation: represents the operationalisation level of the building blocks as the conceptual blocks are mapped to a concrete realisation technology and can be executed/run in a tooling infrastructure.
(contributing to RQ1, RQ2 and RQ3)
The evaluation of the concept and prototype is performed using identified
metamodels in an explorative laboratory setting. Feedback from the evaluation refines
the concept developed.
Preliminary Results. Evaluation results in a structured form are not yet available.
An initial indications for the applicability of the concept is discussed in [
This doctoral project is supervised by o. Univ.-Prof. Prof.h.c. Dr. Dimitris Karagiannis, head of the research group Knowledge Engineering and the Open Models Initiative Laboratory (OMiLAB) at the Faculty of Computer Science, University of Vienna.