Towards the State of the Art of Extending Enterprise Modeling
Languages
Richard Braun
TU Dresden, Chair for Wirtschaftsinformatik, esp. System Development, 01062 Dresden, Germany
Keywords:
Enterprise Modeling, Modeling Extension, Meta Modeling, Extensibility, Meta Model Enhancement.
Abstract:
In the previous decade, more and more de facto standards of enterprise modeling languages (EML) evolved.
The establishment of EMLs leads naturally to an increasing number of EML extensions in order to integrate
requirements and needs from specific problems or domains in an EML. Thus, EML extensibility is proposed
as a relevant topic within both the field of meta modeling and enterprise modeling. We therefore conducted an
analysis of existing meta modeling languages and well known EML languages in order to derive the current
state of the art in terms of EML extensibility. In addition to that, classification schemes for extension pur-
poses and extension mechanisms are presented. Finally, topics for further research are proclaimed in order to
facilitate more research on language extensibility.
1 INTRODUCTION AND
MOTIVATION
Generally, Enterprise Modeling Languages (EML)
are of primary importance for conceptual and tech-
nical complexity management within enterprises
(Frank, 1999). For instance, EMLs are used within
enterprise architecture frameworks to facilitate the
management of different views (perspectives), as-
pects and several levels of abstraction in an integrated
manner (Frank, 2002; Braun and Winter, 2005).
EMLs can also provide the fundament for model-
driven engineering approaches (Atkinson and Kuhne,
2003). In the course of the last two decades, sev-
eral EMLs such as BPMN (Chinosi and Trombetta,
2012), ARIS (Scheer and N¨uttgens, 2000) or Archi-
Mate (Lankhorst et al., 2009) evolved. Also, the
general-purposemodeling language UML is often ap-
plied in the context of enterprise modeling.
As it is well known from software engineering
(e.g., in the field of ERP systems (Botta-Genoulaz
et al., 2005)), the prevalence of these standards leads
consequently to an increase of situations where EMLs
need to be customized or extended in order of satisfy
specific requirements coming from the peculiarities
of a specific industry, business or - more generally -
domain-specific problem. The increasing number of
EML extensions provides evidence for this assump-
tion (Braun and Esswein, 2014a; Pardillo, 2010).
Thus, it is necessary to address the issue of extend-
ing EMLs in detail due to the following reasons:
Domain-specific Configuration. There are numer-
ous reasons and scenarios that require the extension
of an EML. For instance, enhancing an EML with
analytical concepts (e.g., for performance measure-
ment), adding a new perspective to the EML (e.g., re-
source concepts in process models) or due to reasons
of interdisciplinarity (e.g., adapting process models in
manufacturing (Braun and Esswein, 2014b)). Further,
EMLs in the field of enterprise architecture manage-
ment are typically very abstract and under-specified,
wherefore domain-specific extension and customiza-
tion become necessary (Malavolta et al., 2013). This
situation also occurs when the general-purpose mod-
eling language UML needs to be extended for specific
platforms or domains (Pardillo, 2010). Extensibility
of EML is extremely relevant in the enterprise mod-
eling domain to the vast amount of stakeholders and
their perspectives to the enterprise (Atkinson et al.,
2013).
Managing Language Complexity and Method Plu-
ralisms. The design of universal languages, that
cover all possible purposes within enterprise mod-
eling or within one domain is barely shapeable and
only theoretically feasible, since different stakehold-
ers use different languages (Becker, 2014). Besides,
there is a traditional trade-off between language com-
plexity and language understanding (Malavolta et al.,
2013; zur Muehlen and Recker, 2008). Hence, it
394
Braun R..
Towards the State of the Art of Extending Enterprise Modeling Languages.
DOI: 10.5220/0005329703940402
In Proceedings of the 3rd International Conference on Model-Driven Engineering and Software Development (MODELSWARD-2015), pages 394-402
ISBN: 978-989-758-083-3
Copyright
c
2015 SCITEPRESS (Science and Technology Publications, Lda.)
seems to be more reasonable to proclaim the usage of
a small set of popular EMLs and extend them domain-
specifically (Loos et al., 2013). Thus, it might be pos-
sible to unify the EML language area and provide pre-
cise mechanisms for their extension instead of design-
ing “Yet Another Modeling Language”, which possi-
bly has even huge redundant overlaps with basal con-
cepts from standard EMLs.
Interoperability and Tool Support. There are also
several aspects from a technical point of view that
strengthen the relevance of EML extensibility. For
instance, modeling tools need to be extensible in or-
der to facilitate EML extensions in a standard man-
ner while ensuring a valid language core (see the pro-
posed plugin mechanism stated in (Atkinson et al.,
2013)). Also, the exchange of EML extensions be-
tween two modeling tools should be possible in or-
der to allow interoperability and data exchange. Thus,
not only machine-to-machine communication is facil-
itated, but also communication between human stake-
holders.
Language Evaluation. Extensibility of EMLs com-
prises also the opportunity of improving and evolving
an EML. For instance, commonly occurring exten-
sions can be seen as evidence for changing the EML
in some aspects. For example, Decker et al. (2007)
proposed a choreography extension of BPMN that
was widely integrated in BPMN 2.0 (Decker et al.,
2007).
All in all, we argue that extensibility of EMLs is
an importanttopic within enterprise modeling; mainly
due to reasons of separation of concern, interoperabil-
ity, avoidance of redundancy and finally better com-
munication at all. Within this article we aim to con-
sider the current state of the art in terms of extensi-
bility of EML in order to set up a foundation for fur-
ther research and the discussion of benefits and short-
comings within the research community. Therefore,
we investigate extension mechanisms of both exist-
ing EMLs and some meta modeling languages such as
MOF, MEMO and E3. Based on the theoretical anal-
ysis of both the literature and specifications of several
EMLs, we finally outline some classifications for ex-
tension mechanisms and extension techniques.
The remainder of this article is as follows: Sec-
tion 2 provides fundamental aspects and defines the
term extension. Afterwards, Section 3 examines meta
modeling languages regarding their explicit provision
of extension mechanisms. Consequently, Section 4
considers the extensibility of modeling languages.
Section 5 provides an classification of extension pur-
poses and mechanisms. The paper ends with an expli-
cation of proposed further research topics in this area
of discourse.
2 FUNDAMENTALS
2.1 Definition
Generally, the author follows the language-based
meta modeling definition (K¨uhne, 2006; Strahringer,
1998) and thus differentiates between four levels of
abstraction what is well known from OMG architec-
ture: The meta meta modeling level (M3), the meta
modeling level (M2), the model level (M1) and the
object level (M1). In detail, an EML is understood as
part of an enterprise modeling method (Greiffenberg,
2004). Within this research-in-progressarticle, we fo-
cus on the language grammar (syntax) and explicitly
on any kind of extensibility of both the abstract and
the concrete syntax (see Figure 1). As stated in Sec-
tion 1, we both consider several EMLs on level M2
and general extensibility possibilities on level M3.
Extended Modeling Method
Language Based
Meta Model
Procedure Model
Model
Extended Modeling
Language
Procedure
Domain
Immediate Model of Immediate Model of
Mediate Model of
Created in
Created by
Model of
Extension Concepts
Extension Procedure
Figure 1: Elements and instances of an extended model-
ing method (referring to (Greiffenberg, 2004)). The focus
of this research paper lies merely on syntactical aspects of
modeling languages (thick border lines).
Currently, there is a lack of a common EML exten-
sion definition. Weisem¨oller and Sch¨urr (2008) define
an extension in the context of MOF as follows: Let
m
1
, m
2
be meta models defined in MOF and p
1
, p
2
their outermost packages. M
1
is called an extension
of m
1
if m
1
!= m
2
and there exists a package p
ext
such
that p
2
is the result of merging p
ext
into p
1
.
However, that definitional approach is limited to
the package-related MOF. More general, an exten-
sion can be understood as an enhancement of an EML
with functionality according to the (domain-specific)
needs or requirements of language users. Generally,
an extension is neither useful nor functional on its
own and depends on the extended language (host lan-
guage). Further, a standard-conform extension is
understood as an extension that either follows the
well-defined extension mechanism of an EML or at
least do not contradict both syntax and semantics of
the language (Kopp et al., 2011). Thus, it is also pos-
sible to cover meta model changes as EML extension,
although a well-defined extension mechanism is miss-
ing.
TowardstheStateoftheArtofExtendingEnterpriseModelingLanguages
395
2.2 Related Work
To the best of our knowledge, only very research ar-
ticles explicitly address the issue of EML extensibil-
ity so far. Atkinson et al. (2013) emphasize the rel-
evance of the topic for enterprise modeling due to
the vast amount of domain-specific views and stake-
holders. Also, the authors discuss some pros and
cons of existing extension approaches (in-built, meta
model customization and model annotation). The au-
thors state the evident lack of suitable tool support
for all of these approaches since modeling tools only
(if indeed) provided hardwired, dedicated meta model
changeability. Thus, Atkinson et al. (2013) proclaim
the consideration of the following design principles
regarding EML extensibility: Separation of concerns;
low coupling between host language and extensions
as well as high cohesion within an extension. Con-
sequently, the authors introduce the OCA multi-level
modeling approach in the context of EML extensions.
This approach allows arbitrary ontological extensions
of a language while run-time. However, their research
is very engineering-driven and mainly focusses on
the concrete syntax and the definition of user-specfic
views (Atkinson et al., 2013).
Braun and Esswein (2014) conduct a systematic
review of existing BPMN extensions and provide a
framework for the analysis of extensions. Further,
their analysis include the consideration of the applied
methodology that leads to any proposed extension.
The authors state that only very few extensions are
conform to the BPMN standard and they also consti-
tute the need for an integrated methodologicalsupport
for language design and construction (Braun and Es-
swein, 2014a).
Moreover, Kopp et al. (2011)and Pardillo (2010)
examine BPEL extension respectively UML exten-
sions and profiles. Both approaches less focus on en-
terprise modeling and merely provide statistical anal-
ysis of published extensions.
3 EXTENSIBILITY IN META
MODELING LANGUAGES
Analyzing EML extensibility also requires the con-
sideration of existing meta meta modeling approaches
(languages defined on the M3 level), since these meta
languages might provide some advises or mecha-
nisms for extending languages (Frank, 2013).
3.1 MOF
Meta Object Facility (MOF) is the prevalent standard
for the definition of meta data and hence it is a popular
instrument for meta modeling, for example in the field
of model-driven engineering (K¨uhne, 2006). Conse-
quently, MOF is located on the topmost layer and its
primary responsibility is to define a language for the
specification of a meta model (OMG, 2014b). MOF
can be divided into an essential version and a com-
plete version of the language (EMOF and CMOF).
Capabilities of MOF are organized in four main pack-
ages: Common, Reflection, Identifier and Extension.
This package structure already indicates the relevance
of extensibility and MOF explicitly aims to provide
a simple definition of meta model extensions (OMG,
2014b, p. 5). Basically this is facilitated by package
merging and the shared usage of the Common Core
package (OMG, 2014b, p. 7).
However, MOF only provides very simple means
for extending model elements by providing name-
value-pairs within the Extension package (OMG,
2014b, p. 3). Thus, tags on Elements of the meta
model can be realized. Actually, this is the only
concept within the EMOF Extension package. The
CMOF Extension package merges the EMOF Exten-
sion package and adds the extension owner concept.
Next to this simple tagging approach, there seems
to be the possibility of a profile mechanism. This
is not directly explicated in the MOF specification,
but rather interpretable from the specification of UML
Infrastructure Library which states that its “profile
mechanism may be used with any meta model that
is created from MOF” (OMG, 2011c, p. 185). This
statement indicates, that each language defined by the
MOF can make use of the profile mechanism that is
commonly used in UML modeling (Pardillo, 2010).
The only condition for that is a required reference
of the profile to the respective meta model (OMG,
2011c, p. 176).
The mentioned statement arises the question, why
MOF does not define the profile mechanism on level
M3. The answer comes from the pretty strange ar-
chitecture of MOF: MOF merges packages that are
defined within the UML Infrastructure Library on
level M2. Strictly speaking, the Infrastructure Li-
brary acts both as a meta meta model (M3) and as
meta model (M2), what leads to a obvious conflict of
abstraction. Even when accepting these crude archi-
tecture, there still remains the problem that MOF ac-
tual does not have any valid syntactical access to the
Profile package. MOF only merges the packages In-
frastructureLibrary::Core::Basic and Infrastructure-
Library::Core::Constructs; but not the Profiles pack-
age. Also, it is confusing that the Profiles packages
is associated with the Core package (OMG, 2011c,
p. 173), although it is a separate package outside
MODELSWARD2015-3rdInternationalConferenceonModel-DrivenEngineeringandSoftwareDevelopment
396
the Core (OMG, 2011c, p. 12). Obviously, there is
some confusion and the syntax contradicts the above
mentioned indications that the profile mechanism can
be used by any meta model created from MOF (see
(OMG, 2011c, p. 185))!
At this point, both MOF and UML suffer from an
obscure differentiationbetween the abstraction layers.
It seems to be that the Infrastructure Library defines
the UML and MOF is only responsible for some el-
ementary concepts and tool support. Due to the fo-
cus on meta data definition and the mixture of meta
modeling and tool concepts, MOF provokes some
problems in the enterprise modeling domain (Frank,
2008).
All in all, we can conclude that MOF provides a
simple name-value-pair extension of each element on
M3 level. Further, there is confusion on the stated
profile mechanism. Based on the MOF meta model
we do not see any applicability of that mechanism
for MOF defined meta models at all. This is also
strengthened by the fact that none of the MOF based
languages uses profiles (see Section 4).
3.2 MEMO
MEMO (Multi-Perspective Enterprise Modeling) is a
framework for enterprise modeling and the integra-
tion of several views, perspectives and aspects of en-
terprises (Frank, 2002). The framework provides the
meta modeling language MEMO MML for the inte-
grated specification of domain-specific modeling lan-
guages (Frank, 2008, p. 22) and also provides a pro-
cedural description of required design and construc-
tion steps.
The MEMO framework does not provide a dedi-
cated extension mechanism, as its capabilities are de-
signed for the precise definition of domain-specific
languages, which spares extensibility (Frank, 2002,
p. 3), (Frank, 2008, p. 30). Consequently, MEMO
proclaims that each adaption of a language has to
be conducted on the M2 level (meta model) by lan-
guage engineers that are usually part of the domain-
specific context (Frank, 2013, p. 2). Nevertheless,
the MEMO author emphasizes the general necessity
of application-specific language extensibility (Frank,
2013, p. 15).
3.3 E3
The E3 meta modeling approach is not very ac-
quainted but provides a well-defined meta model-
ing language for the integrated definition of EMLs.
The E3 framework provides an integrated meta meta
model for the definition of EMLs by integrating con-
textual aspects (e.g., objects and its properties) and
presentational aspects (e.g., views and different pre-
sentations of a context object (Greiffenberg, 2004)).
The E3 frameworkintends to provide a generic frame-
work for language specification at all and integrates
all relevant aspects of language definition. In con-
trast to other approaches, it also proposes required
process steps within language design in order support
language engineers. However, extensibility of instan-
tiated languages is not considered. Rather it implies
direct adaptation on M2 level (meta model).
3.4 GME
The General Modeling Environment (GME) is a
toolkit for the creation of domain-specific modeling
languages and the integration of heterogeneous mod-
els (Ledeczi et al., 2001). The meta meta language
MetaGME is the kernel of the toolkit. MetaGME
bases on UML and OCL and provides generic con-
cepts for language definition such as Models, Parts
and so-called First Class Objects (Ledeczi et al.,
2001). GME provides minor guidance for language
creation but lacks in any consideration of language
extensibility. Moreover, it is up to the language engi-
neer to alter a meta model.
3.5 Gmodel
The objective of the Gmodel meta modeling lan-
guage is a compact, modular and extensible design
of domain-specific modeling languages that facili-
tate model integration and interoperability (Bettin and
Clark, 2010). The stated extensibility is enabled by
the Gmodel architecture that allows multi-level mod-
eling instead of the fixed four layered architecture of
OMG. Therefore, Gmodel proposes so-called Seman-
tic Identities on M3 level and also allows the integra-
tion of typed links on this level.
3.6 Conclusion
Reviewing the stated approaches reveals a lack of
suitable and more sophisticated mechanisms of ex-
tensibility on the level of meta modeling languages.
Only MOF and Gmodel consider extensibility to a mi-
nor degree: MOF provides fundamental concepts for
the definition of new attributes. Gmodel facilitates
vertical extensibility by multi-level modeling that is
additionally propose by Atkinson et al. (2013). Gen-
erally, the approaches leave language extensions open
to the language engineer and his alterations and cus-
tomizations of a meta model. However, missing guid-
ance and borders cause ad hoc changes depending
TowardstheStateoftheArtofExtendingEnterpriseModelingLanguages
397
on single engineers, which is only suitable if there is
no noteworthy dissemination of the language and no
need for comparison, integration or tool implementa-
tion.
4 EXTENSIBILITY IN EML
4.1 UML
Due to the dissemination of the Unified Model-
ing Language (UML), many extensions and adap-
tions of the language exist (Weisem¨oller and Sch¨urr,
2008). They are mainly used for domain-specific
view points, model transformation (MDA) and model
analysis (Selic, 2007). UML explicitly allows meta
model modification for custom purposes (heavy-
weight extension), but provides no detailed recom-
mendations on that. On the contrary, the lightweight
extension mechanism enables the definition of spe-
cific profiles within UML models on level M1. Pro-
files are limited extensions of referenced meta models
with the purpose of adapting the meta model to a spe-
cific platform or domain (OMG, 2011c, p. 183). Thus,
it can be seen as an “extension by addition” mech-
anism that generates UML dialects (OMG, 2011c,
p. 23). It isimportantto emphasize that profiles do not
change the UML meta model but rather its instances.
Actually, an intermediate meta level (M1.5)” is pro-
duced which enhances the vocabulary of UML with-
out changing the meta model. Therefore, several con-
cepts are included within UML meta model: Stereo-
types (extend meta classes of the UML), Tag Def-
initions (attributes of Stereotypes), Constraints and
some relationship types. Strict filtering options pro-
vide the chance to define tight DSMLs.
The concrete syntax of UML can be extended by
the annotation of specific icons to Stereotypes. Struc-
tured icons are not possible. While the UML spec-
ification does not provide procedural guidance for
the design of Profiles, there are several research pa-
pers addressing this issue (Selic, 2007; Lagarde et al.,
2008). UML profile mechanism is very popular due
its simple applicability and tool support. However, the
limited expressiveness of the approach might lead to
the definition of complex (and thus expansive) OCL
statements (Weisem¨oller and Sch¨urr, 2008).
4.2 Languages Defined with MOF
As stated in the previous section, MOF is a common
meta modeling language specified by the Object Man-
agement Group (OMG). Hence, we have analyzed a
range of business-oriented OMG languages. None
of the analyzed languages uses the MOF extension
mechanisms or re-uses the profile mechanism (see
Section 3.1)! The majority of the languages does not
provide extensibility and only BPMN, Essence and
KDM provide appropriate concepts (see Figure 2).
CMMN and DMN only emphasize forbidden exten-
sion operations within its models but do not pro-
vide a conceptual understanding of extensibility. Nei-
ther SBVR, VDML nor SysML consider extensibility.
IFML states some kind of black and white lists of el-
ements that can be extended or not.
0.2 Sprach-Übersicht
Syntax
Procedure
Abstract
Concrete
OMG languages
UML
!
"
"
BPMN
!
"
"
CMMN
DMW
#
SBVR
#
KDML
#
Essence
!
IFML
"
SysML
#
KDM
!
"
"
SMM
#
...
ArchiMate
"
ARIS
#
! (exists) " (partially) (does not exist)
Figure 2: Extensibility of abstract and concrete syntax
within MOF-based languages and their provision of me-
thodical support for extension design.
Essence introduces the Extension Element con-
cept and OCL based Extension Functions in order to
change attributes of elements (OMG, 2014a, p. 60-
65). In case of extension, the extended elements
remain oblivious of its modification. This “non-
destructive” kind indicates a run-time extension on
model level (OMG, 2014a, p. 97). SMM defines sim-
ple concepts (tag-value-pairs and textual annotations)
for the implementation of extension of every SMM
element on model level (OMG, 2012).
KDM (Knowledge Discovery Meta-Model) both
provides an extension of the language itself (level
M2) and its instances (level M1). For the first one,
a new package within KDM can be defined with
mandatorycomponents(Model, Abstract Element and
Abstract Relationship). For the latter, a dedicated
mechanism is specified, that provides elements for
the definition of extension classes, possible extension
value classes and simple key-value-pairs as annota-
tions (OMG, 2011b, p. 39). Although the architecture
and the semantics of the stated extension concepts are
quite similar to UML, KDM declares its own defini-
tion.
MODELSWARD2015-3rdInternationalConferenceonModel-DrivenEngineeringandSoftwareDevelopment
398
4.3 BPMN
BPMN (Business Process Model and Notation) is the
current de facto standard in the field of business pro-
cess modeling (Chinosi and Trombetta, 2012; OMG,
2011a). In contrast to many other languages, BPMN
provides an “extension by addition” mechanism con-
taining four class: An Extension Definition is a named
group of new attributes which can be referenced to
any BPMN element. An Extension Definition consists
of several Extension Attribute Definitions that define
the particular attributes. Values of these Extension At-
tribute Definitions can be defined by the Extension At-
tribute Value. The class Extension binds the entire ex-
tension definition and its attributes to a BPMN model
definition and makes it available and accessible.
Despite the well-defined syntax, there are some
shortcomings of the mechanism: For instance, the
XML definition of BPMN does not support exten-
sions completely. Also, each extension is related to
the generic Base Element of BPMN what hinders sep-
aration of concerns. Besides, BPMN lacks in provid-
ing methodical guidance for its application. A few au-
thors address this issue: Stroppi et al. (2011) proclaim
a MDA based approach for the transformation of con-
ceptual domain models into valid BPMN extension
methods (Stroppi et al., 2011). Braun and Schlieter
(2014) introduce a structured approach for the anal-
ysis of domain concepts that should be added to the
BPMN (Braun and Schlieter, 2014).
4.4 ArchiMate
ArchiMate is an EML for enterprise architectures
that divides an enterprise into layers, aspects, in-
ternal views and external views (van Haren, 2012).
In contrast to other EMLs from the architecture do-
main, ArchiMate provides two extension approaches.
First, the “profile specialization mechanism” enables
the annotation of attributes to concepts and relation-
ships. Thereby, a Profile is understood as a data
structure that is defined independently and connected
to concepts and relationships of ArchiMate (similar
to model weaving (Del Fabro and Valduriez, 2009)).
Thus, ArchiMate permits the integration of external
model elements. Moreover, ArchiMate explicates
the opportunity to specialize concepts and relation-
ships through inheritances between a general standard
element and a domain-specific element (similar to
UML Stereotypes). However, ArchiMate does neither
provide a well-defined meta model of all extension-
relevant concepts nor a procedure model for its appli-
cation.
4.5 ARIS
ARIS (Architecture of Integrated Information Sys-
tems) is multi-perspective enterprise modeling frame-
work that is mainly used in the field of business pro-
cess management and ERP management (Scheer and
N¨uttgens, 2000). However, there is no commonly ac-
cepted meta model of ARIS, as it is not standardized
by a larger institution. Consequently, ARIS does nei-
ther provide nor consider extensibility. Existing ARIS
extension first re-engineer the ARIS meta model and
then customize it ad hoc according to their require-
ments (Stein et al., 2008).
4.6 Conclusion
It can be summarized, that the minority of the ad-
dressed modeling languages provide an extension
mechanism. UML, BPMN, Essence, KDM and
ArchiMate provide concepts for the definition of new
language objects. However, only BPMN and KDM
allow heavyweight adaptations of the meta model in a
structured manner. The other approaches rather sup-
port the lightweight definition of domain-specific el-
ements on model level, without changing the meta
model. In contrast to abstract syntax, adaptations of
the concrete syntax is barely considered. Also precise
methodical guidance for the extension design process
is mostly missing.
5 CLASSIFICATION
5.1 Extension Purposes
The aspect “extension purpose” covers the objective
that is related to the aimed extension and reflects the
purpose that needs to be fulfilled. So far, only Atkin-
son et al. (2013) addresses this issue and proclaim
“enhancement” and “augmentation”. Based on a lit-
erature review and the consideration of extension re-
views (Braun and Esswein, 2014a; Pardillo, 2010),
we have extended their purpose setting as depicted in
Figure 3.
Figure 3: Consolidated extension purposes.
TowardstheStateoftheArtofExtendingEnterpriseModelingLanguages
399
Enhancement stands for the extension of a host
language with elements from the same domain (e.g.,
new gateway types within BPMN). Thus, it can be
seen as a horizontal EML extension (Atkinson et al.,
2013). On the other hand, augmentation stands for
the vertical extension of a host language with con-
cepts, attributes or rules from an other domain (Atkin-
son et al., 2013). Both purposes can be integrated
(Enhancement and Augmentation).
Specialization covers all purposes that aim to
characterize an EML for a specific domain, with a
very limited number of new concepts. Moreover, the
abstract syntax remains more or less the same, but
the semantical aspects change. Thus, Re-Definition
stands for the alteration of semantics of the host
EML. For instance, by re-defining the interpretation
of some EML concepts in the context of a specific do-
main. Customization stands for specifying the con-
crete syntax or the semantics of EML concepts in pre-
defined boundaries (e.g., by integrating specific ter-
minology of a domain or platform). Accordingly, the
elements are only specified and not added. Thus, el-
ements of the host EML remain under-specified and
are intended to be adapted to a specific need.
Moreover, Model Usage encompasses all exten-
sions that support some kind of working on EML
models. We divide this purpose type into operations
and usability. Operations covers all helper elements
that are needed for model analysis or model transfor-
mation. Hence, it is a special type of augmentation (to
the effect, that the considered domain is understood as
an analytical area). Besides, Usability reflects all el-
ements that are needed for the support of the model
user (e.g., additional annotation elements).
5.2 Extension Mechanisms
An extension mechanism is understood as either an
explicit mechanism of an EML for the extension of
this language or a more general approach for exten-
sion. We have evolved the following mechanisms
based on the review of both the literature (Atkinson
et al., 2013; Braun and Esswein, 2014a) and exist-
ing EML specifications. Figure 4 presents all consol-
idated mechanisms.
In-Built
Profiles
Extension
Framework
Under
Specification
Meta Model
Customization
Model
Annotation
Multi Level
Modeling
Figure 4: Consolidated extension mechanisms.
The In-Built group covers all approaches where
the EML meta model contains an inherent mecha-
nism for language extension on the meta level M2.
The group can be divided into three sub mechanisms:
Profiles, Meta Model Customization and Under Spec-
ification. A Profile constitutes well-defined language
elements defined on level M2 that can be applied for
the specification of new elements on level M1 (e.g.,
UML profiles). In this case, the meta model is not
altered at all. Rather, some kind of an intermediate
model level is built that allows the instance-specific
definition of new elements in level M1. The Exten-
sion Framework type stands for a similar type, but
in this case the proposed extension elements are less
strict defined. It provides more a framework for cus-
tom extension, but its implementation remains open.
For instance, within BPMN it is not clearly specified
whether an extension should be implemented in the
style of a profile or as a meta model alteration. Fi-
nally, Under Specification stands for specifications,
whose elements are intentionally more generic and
provide space for custom specification (e.g., Archi-
Mate or KDM). This could also affect only single el-
ements (e.g., Lanes in BPMN).
Meta Model Customization can either be ex-
plicitly allowed by an EML or applied in an ad hoc
manner individually. By reviewing existing meta
model specifications, we found that actually no lan-
guage supports a structured meta model customiza-
tion. Rather, it remains open to the language engi-
neer. Indeed, some languages indicate what elements
can be extended or not, but there is no appropriate
procedure for meta model customizations. Of course,
altering a meta model always depends on the applied
meta meta modeling language (see Section 3.1).
Model Annotation references the connection and
integration of external models with the host language
meta model. Such an integration can be realized by
techniques like model weaving (Del Fabro and Val-
duriez, 2009). So far, none of the examined languages
supports this concept explicitly. Only BPMN pro-
vides similar concepts by its External Relationship
elements, which supports the integration of external
domain models (however, on level M1 (OMG, 2011a,
p. 62)).
Last but not least, Atkinson et al. (2013) intro-
duced the novel approach of Multi Level Modeling
in the context of EML extensions. In contrast to the
rigid four layered OMG architecture, the authors pro-
pose a flexible multi level architecture based on the
orthogonal classification architecture (Atkinson et al.,
2009). Although the approach is not very diffused yet,
multi level modeling is promising regarding to exten-
sibility since it could solve abstraction difficulties.
MODELSWARD2015-3rdInternationalConferenceonModel-DrivenEngineeringandSoftwareDevelopment
400
6 CONCLUSION AND FURTHER
RESEARCH
Within this article, the topic of language extensibil-
ity in the field of enterprise modeling languages was
introduced by a review of existing extension mech-
anisms in both meta modeling languages and some
EMLs. Based on the examination of these languages
and a review of the (scarce) literature on that topic,
a classification of extension purposes and extension
mechanisms was proposed. Since this article repre-
sent research-in-process work, there are several as-
pects for further investigation:
Syntax. Currently, there is a lack of clear and pre-
cise definitions of extension mechanisms in existing
EML specifications. Although several approaches ex-
ist, actually each of them has some inaccuracies or in-
consistencies. Perhaps, this issue can be improved by
the definition of a common extension reference model
that could both support the revision of meta mod-
els concerning extensibility and also facilitate the ex-
change of model data. Such a reference model might
also help to establish a common accepted understand-
ing of EML extensibility and its elements. Therefore,
the integrated E3 meta model of Greiffenberg (2004)
seems to be a feasible starting point. Further, it is
also important to examine the consequences of EML
extension to modeling languages within software en-
gineering in order to enable a tight business IT align-
ment.
Tool Support. The design of EML extension pat-
terns is promising in order to integrate extension pur-
poses and mechanisms and provide a productive base
for reuse in extension design. Besides, syntactical is-
sues are always related to a range of technical aspects
such as flexible tool implementation, interoperabil-
ity or possible run-time model alterations (Atkinson
et al., 2013). Working on meta models also arises the
necessity of suitable revision management (Esswein
and Weller, 2007). Especially, when meta model el-
ements are removed (meta model reducing) or over-
written - both aspects are barely examined so far.
Method. Although procedure models are impor-
tant components of modeling methods (see Figure 1),
there is a severe lack of methodological support for
the design of valid and sense making extensions. For
instance, such procedure models should support the
domain analysis phase in order to identify an ex-
tension need. Also, they could recommend suit-
able extension patterns. Perhaps, preliminary stud-
ies in the field of situational method engineering can
support the design of appropriate extension methods
(Brinkkemper et al., 1999).
Acknowledgement. This research was funded by the
German Research Foundation (DFG) within the re-
search project “SFB Transregio 96”. The author is
grateful for the funding and the research opportuni-
ties.
REFERENCES
Atkinson, C., Gerbig, R., and Fritzsche, M. (2013). Model-
ing language extension in the enterprise systems do-
main. In 17th IEEE International Enterprise Dis-
tributed Object Computing Conference, pages 49–58.
Atkinson, C., Gutheil, M., and Kennel, B. (2009). A flex-
ible infrastructure for multilevel language engineer-
ing. IEEE Transactions on Software Engineering,
35(6):742–755.
Atkinson, C. and Kuhne, T. (2003). Model-driven devel-
opment: a metamodeling foundation. IEEE Software,
20(5):36–41.
Becker, J. (2014). Interview with reinhard sch¨utte on “man-
aging large-scale bpm projects”. Business & Informa-
tion Systems Engineering, pages 1–3.
Bettin, J. and Clark, T. (2010). Advanced modelling made
simple with the gmodel metalanguage. In Proceedings
of the First International Workshop on Model-Driven
Interoperability, pages 79–88. ACM.
Botta-Genoulaz, V., Millet, P.-A., and Grabot, B. (2005). A
survey on the recent research literature on erp systems.
Computers in Industry, 56(6):510–522.
Braun, C. and Winter, R. (2005). A comprehensive enter-
prise architecture metamodel and its implementation
using a metamodeling platform. Proceedings of the
Workshop Enterprise Modelling and Information Sys-
tems Architectures, pages 24–25.
Braun, R. and Esswein, W. (2014a). Classification of
domain-specific bpmn extensions. Lecture Notes of
Business Information Processing, 147:42–57.
Braun, R. and Esswein, W. (2014b). Extending bpmn for
modeling resource aspects in the domain of machine
tools. Advanced Materials and Information Technol-
ogy Processing, 87:449.
Braun, R. and Schlieter, H. (2014). Requirements-based
development of bpmn extensions: The case of clinical
pathways. In IEEE 1st International Workshop on the
Interrelations between Requirements Engineering and
Business Process Management, pages 39–44.
Brinkkemper, S., Saeki, M., and Harmsen, F. (1999). Meta-
modelling based assembly techniques for situational
method engineering. Information Systems, 24(3):209–
228.
Chinosi, M. and Trombetta, A. (2012). Bpmn: An intro-
duction to the standard. Computer Standards & Inter-
faces, 34(1):124–134.
Decker, G., Kopp, O., Leymann, F., and Weske, M. (2007).
Bpel4chor: Extending bpel for modeling choreogra-
phies. In IEEE International Conference on Web Ser-
vices, pages 296–303.
TowardstheStateoftheArtofExtendingEnterpriseModelingLanguages
401
Del Fabro, M. D. and Valduriez, P. (2009). Towards the
efficient development of model transformations using
model weaving and matching transformations. Soft-
ware & Systems Modeling, 8(3):305–324.
Esswein, W. and Weller, J. (2007). Method modifications in
a configuration management environment. Proceed-
ings of the Fifteenth European Conference on Infor-
mation Systems, pages 2002–2013.
Frank, U. (1999). Conceptual modelling as the core of the
information systems discipline-perspectives and epis-
temological challenges. AMCIS 1999 Proceedings,
page 240.
Frank, U. (2002). Multi-perspective enterprise model-
ing (memo) conceptual framework and modeling lan-
guages. In Proceedings of the 35th Annual Hawaii
International Conference on System Sciences, pages
1258–1267.
Frank, U. (2008). The memo meta modelling language
(mml) and language architecture. ICB Research Re-
port 24, Universit¨at Duisburg-Essen.
Frank, U. (2013). Domain-specific modeling languages: re-
quirements analysis and design guidelines. In Domain
Engineering, pages 133–157. Springer.
Greiffenberg, S. (2004). Methodenentwicklung in
Wirtschaft und Verwaltung. Kovac.
Kopp, O., G¨orlach, K., Karastoyanova, D., Leymann, F.,
Reiter, M., Schumm, D., Sonntag, M., Strauch, S.,
Unger, T., Wieland, M., et al. (2011). A classification
of bpel extensions. Journal of Systems Integration,
2(4):3–28.
K¨uhne, T. (2006). Matters of (meta-) modeling. Software
and Systems Modeling, 5(4):369–385.
Lagarde, F., Espinoza, H., Terrier, F., Andr´e, C., and
G´erard, S. (2008). Leveraging patterns on domain
models to improve uml profile definition. In Fun-
damental Approaches to Software Engineering, pages
116–130. Springer.
Lankhorst, M. M., Proper, H. A., and Jonkers, H. (2009).
The architecture of the archimate language. In En-
terprise, Business-Process and Information Systems
Modeling, pages 367–380. Springer.
Ledeczi, A., Maroti, M., Bakay, A., Karsai, G., Garrett,
J., Thomason, C., Nordstrom, G., Sprinkle, J., and
Volgyesi, P. (2001). The generic modeling environ-
ment. In Workshop on Intelligent Signal Processing,
Budapest, Hungary, volume 17.
Loos, P., Mettler, T., Winter, R., Goeken, M., Frank, U., and
Winter, A. (2013). Methodological pluralism in busi-
ness and information systems engineering? Business
& Information Systems Engineering, 5(6):453–460.
Malavolta, I., Lago, P., Muccini, H., Pelliccione, P., and
Tang, A. (2013). What industry needs from architec-
tural languages: A survey. IEEE Transactions on Soft-
ware Engineering, 39(6):869–891.
OMG (2011a). Business Process Model and Notation, Ver-
sion 2.0.
OMG (2011b). Knowledge Discovery Meta-Model, Version
1.3.
OMG (2011c). Unified Modeling Language, Infrastructure,
Version 2.4.1. OMG.
OMG (2012). Structured Metrics Metamodel, Version 1.0.
OMG (2014a). Essence - Kernel and Language for Software
Engineering Methods, Beta 2.
OMG (2014b). Meta Object Facility (MOF) Core Specifi-
cation, Version 2.4.2.
Pardillo, J. (2010). A systematic review on the definition of
uml profiles. In Model Driven Engineering Languages
and Systems, pages 407–422. Springer.
Scheer, A.-W. and N¨uttgens, M. (2000). ARIS architecture
and reference models for business process manage-
ment. Springer.
Selic, B. (2007). A systematic approach to domain-
specific language design using uml. In 10th IEEE
International Symposium on Object and Component-
Oriented Real-Time Distributed Computing, pages 2–
9.
Stein, D.-W. F. S., Lauer, D.-I. F. J., and Ivanov, K.
(2008). Aris method extension for business-driven
soa. Wirtschaftsinformatik, 50(6):436–444.
Strahringer, S. (1998). Ein sprachbasierter metamodellbe-
griff und seine verallgemeinerung durch das konzept
des metaisierungsprinzips. In CEUR Workshop Pro-
ceedings Modellierung.
Stroppi, L. J. R., Chiotti, O., and Villarreal, P. D. (2011).
Extending bpmn 2.0: Method and tool support. In
Business Process Model and Notation, pages 59–73.
Springer.
van Haren (2012). Archimate 2.0 specification.
Weisem¨oller, I. and Sch¨urr, A. (2008). A comparison of
standard compliant ways to define domain specific
languages. In Models in Software Engineering, pages
47–58. Springer.
zur Muehlen, M. and Recker, J. (2008). How much lan-
guage is enough? theoretical and practical use of the
business process modeling notation. In Advanced in-
formation systems engineering, pages 465–479.
MODELSWARD2015-3rdInternationalConferenceonModel-DrivenEngineeringandSoftwareDevelopment
402