InTra: A Pragmatic Approach of Using Rule-Based Model
Transformation to Reduce Complexity of UML and SysML Models
Philippe Barbie, Martin Becker and Andreas Sch
¨
afer
Fraunhofer IESE, Fraunhofer-Platz 1, 67663, Kaiserslautern, Germany
Keywords:
Model Transformation, Interaction-Based Transformation, Rule-Based-Modelling, Interaction-Patterns,
Complexity, MBSE, SysML, UML.
Abstract:
While UML and SysML are important concepts for digitally representing complex systems, we are still con-
fronted with the problems of comprehensibility and maintainability. Even models with comparatively few
elements can quickly become confusing and hard to read, due to their relationship density. This leads to
potential errors, both for the initial modelling and the later model evolution and evaluation phase. Driven
by the needs of a large-scale modelling project in the field of communication systems, we have researched
approaches to cope with the inherent model complexity by following a modelling approach, that is based on
interaction patterns. In this paper, we will give an introduction to the challenge of the consistently increasing
complexity of system models in the field of model-based systems engineering (MBSE). We will motivate our
need for rule-based modelling in order to handle a real industry use case and outline the problems of system
modeling, which could be solved by using a new rule-based modelling approach. InTra (Interaction-based
Transformation) is an approach to significantly reduce the complexity of a system model, by reducing the
number of connectors through the use of interaction rules. By doing so, we were able to create an abstracted
variant of the same system model, but with a highly reduced number of connectors used and thus an overall
reduced model complexity.
1 INTRODUCTION
The complexity of modern systems and the associ-
ated modelling effort is steadily increasing (Antinyan,
2020) (Warrilow, 2020) (Baduel, 2018) (St
¨
utzel,
2021). In a survey with 127 participants in 14
companies conducted in 2001, 84% of the partici-
pants answered the question ’What motivates your
company to introduce systems engineering?’ with
the reason of constantly increasing product com-
plexity (St
¨
utzel, 2021). It is inspiring to observe
that both David Long and Baduel et al. mention
the particular need for improvement in modelling
methodology to cope with the growing complexity
of future system models (Warrilow, 2020) (Baduel,
2018). Since the early 2000’s, a variety of dif-
ferent transformation approaches have been devel-
oped. While many of those approaches have their
focus on transforming system or software models
into executable code, there are also some languages
that concentrate on model-to-model transformation.
Some of the more popular approaches are, e.g.
PROGRES (RWTH Aachen, 2021), AGG (Taentzer,
2003), AToM
3
(McGill, 2021), GReAT (Balasub-
ramanian, 2006) (ISIS, 2021), GROOVE (Twente,
2021), BOTL (Braun, 2003) and Fujaba (Fujaba De-
velopment Group, 2021), even though some of them
are not developed any further (Kahani, 2019) (Sch
¨
urr,
2021).
In this paper we will investigate whether it is pos-
sible to use a model transformation approach to sig-
nificantly reduce the complexity of system models,
by transforming them into a rule-based representa-
tion. We developed a specific rule-based model trans-
formation approach called InTra (Interaction-based
Transformation), based on the problems introduced
by a real industrial communication system, with over
5,000 user roles and over 9,000 communication rela-
tionships. This approach has the goal to decrease a
system’s complexity, by reducing it to its basic struc-
ture and combining it with applicable relationship pat-
terns, thus decreasing the number of connectors in the
resulting model by an order of magnitude. Having a
large industry-related model, based on realistic data,
was especially useful when developing the approach
and testing its feasibility. To the best of our knowl-
edge, this is the first approach that uses model-to-
model transformation techniques to reduce a model’s
Barbie, P., Becker, M. and Schäfer, A.
InTra: A Pragmatic Approach of Using Rule-Based Model Transformation to Reduce Complexity of UML and SysML Models.
DOI: 10.5220/0011615600003402
In Proceedings of the 11th International Conference on Model-Based Software and Systems Engineering (MODELSWARD 2023), pages 97-104
ISBN: 978-989-758-633-0; ISSN: 2184-4348
Copyright
c
2023 by SCITEPRESS Science and Technology Publications, Lda. Under CC license (CC BY-NC-ND 4.0)
97
Figure 1: Notation of a transformation rule for graph models, according to (Varr
´
o, 2007).
information complexity. In this case, model transfor-
mation is not used, to transform the model into a dif-
ferent state of development, but to transform between
different representations of the same model state.
The paper is structured as follows. Section 2 of-
fers a general overview of the field of graph-based
model transformation. Section 3 describes the in-
dustrial setting in which we developed and tested the
InTra approach. Section 4 elaborates the problems,
which led us into the development of a rule-based
model transformation approach. Section 5 explains
the InTra approach in general and why model trans-
formation is a necessary part of it. Section 6 will
present the basic functionality and application of In-
Tra, based on a simple family tree example model.
Section 7 explains InTra’s own model transformation
language, which is completely based on model ele-
ments itself. Section 8 describes the experiences and
results that were achieved in the given industrial set-
ting from Section 3. Section 9 reviews relevant ex-
isting work in the field, e.g. work that helped us to
derive necessary requirements for the development of
a model transformation approach. Finally Section 10
concludes the paper.
2 TRANSFORMATION OF
GRAPH-MODELS
While models exist in many different forms, such as
relational databases or text-based structures, in this
paper the scope will be centered around the transfor-
mation of graph models. Graphs are powerful data
structures that are able to represent interactions be-
tween system components in an easily comprehensi-
ble way. Essentially, any system that consists of com-
ponents with relations may be naturally described by
a graph in which the vertices represent the entities and
the edges represent the relations (Ghamarian, 2011).
A transformation rule for graph models usually con-
sists of two parts: Pattern-Matching and Rewriting
(Wieber, 2015).
Pattern Matching describes the structure of the pat-
tern, which is to be searched in the model. At all
places where a given subgraph is found that matches
the pattern, a transformation can take place. This first
section of a transformation rule is also called Left-
hand-side (LHS)’ (Wieber, 2015) (Czarnecki, 2006).
Rewriting, the second part of a transformation rule
describes the final state, which has to be realized on
all found pattern matching subgraphs. This part of
the transformation rule is also called Right-hand-side
(RHS)’ (Wieber, 2015) (Czarnecki, 2006).
In the example shown in Figure 1, taken from
(Varr
´
o, 2007), the LHS defines a class CS, which is
a child of a class CP, which has an attribute relation-
ship to attribute A. The RHS of the interaction rule
shows as a result of an applied model transformation,
a deleted relationship between class CS and attribute
A and the creation of a new relationship between the
child class CP and attribute A.
3 INDUSTRIAL CONTEXT
In a previous industry project, we had to model an ar-
chitecture representing a communication system with
over 5,000 user roles to capture individual require-
ments of those user roles and their communication.
The basic procedure was to model each user-role in
each scenario and map the communication between
those user roles via information exchange relation-
ships. Each of those information exchange relation-
ships contains at least one piece of information. Since
this is not comprehensibly representable without any
system structure, the user roles were grouped accord-
ing to their communication hierarchies. This included
further refinements in subgroups, as well as the map-
ping in superiors and subordinate user roles. Nev-
ertheless, there was an average of three information
exchanges per user-role in each communication sce-
nario. One of the technical requirements for the sys-
tem, which was set by the customer, was to use the
well established modelling tool Enterprise Architect
by Sparx Systems. As source of information, more
than 50 domain experts were interviewed and asked
what communication relationships exist in their spe-
cific working environment. The information collected
had to be modelled into the system context, attributed,
managed and delivered back for review. Even though
MODELSWARD 2023 - 11th International Conference on Model-Based Software and Systems Engineering
98
the information was available inside the model, it was
difficult and time-consuming to comprehend, verify,
and maintain the resulting system. As a result, even
subject matter experts had major problems in under-
standing the communication relationships they should
be already familiar with. Furthermore it was an es-
sential requirement, that the resulting model diagrams
could also be well understood by those who are not fa-
miliar with MBSE and system architectures. Unfortu-
nately, for reasons of confidentiality, the correspond-
ing system model and its contents cannot be made
available to the readership within this paper.
4 PROBLEM STATEMENT
It can be observed that the technological barriers are
not posing the greatest problem in here, but rather
those of human cognitive perception. Mapping the
structure and interrelationships of complex systems
as a digital model is a major challenge. To be able
to comprehend, understand and maintain what is rep-
resented beyond that, is an even greater task. Of
course, UML and SysML offer a variety of methods
to decompose the system to reach lower complexity.
On the other hand, even using the best decomposi-
tion, you eventually reach the point where you need
to represent the relations between the system com-
ponents in the model. Normally this is done using
elements and connectors, but even with just a few el-
ements, this may mean a large number of information
flows on a single diagram. Yet, it is the understand-
ing of this connector-representation between system
elements, that is a major cognitive challenge for the
human brain (Koning, 2002).
In the study User preference of graph layout aes-
thetics: a UML study (Purchase, 2000) on the bet-
ter comprehensibility of UML diagrams, 93% of the
participants stated that crossing connectors in the di-
agram are an enormous obstacle to comprehensibil-
ity. 91% testified that bends in the connectors were
a major barrier. These were the two most frequently
mentioned comprehension barriers in the study.
In order to solve the problems mentioned above
and to counteract the challenge of increasing com-
plexity in system models, we will elaborate our ap-
proach named InTra (Interaction-based Transforma-
tion), which significantly reduces the number of con-
nectors in the model, but at the same time preserves
their information content in a comprehensible way.
5 InTra APPROACH
The approach introduced in this article is designed to
significantly reduce the complexity of a system model
with the help of individual interaction rules that work
on the basic system structure. In this way it is pos-
sible to reduce the number of connectors even during
the initial modelling phase, thus keeping the resulting
model complexity small, which in turn improves the
readability and maintainability. The term ’rule-based’
describes the idea of recognizing repeating interaction
patterns in the basic structure of the model, and the
definition of rules for those patterns, which describe
the interaction concept behind those relationships. By
including these rules, it is no longer necessary to store
the said relations by connectors in the model. The
creation of a rule is useful, if an interaction between
certain element types or constellations always runs in
the same way. Although in this rule-based form it is
a lot harder to evaluate the model with simulations or
programs, it is fully completed and understandable in
its rule-based abstracted form. This means, that all
information of the relations to be captured is already
mapped in the model, partly abstracted as interaction
rules to be interpreted. By expressing this information
as rules, the representation of the model is changed,
but its information content remains the same. Avoid-
ing the redundancy of information in this way also
reduces the risk of errors in the further processing of
the model, since changes only need to be made to one
rule instead of all the connectors that a single rule rep-
resents. Furthermore, the usage of interaction rules
makes the model easier to understand for domain ex-
perts. This is the case because domain-specific inter-
actions can be represented very efficiently and com-
prehensibly as interaction rules. Nevertheless, an ex-
perienced system architect should be responsible for
identifying the basic structure of the system, recog-
nizing repetitive patterns in the relationships between
model elements, defining and configuring interaction
rules for those patterns, and finally linking them to the
relevant parts of the system’s basic structure.
In an optional second phase of model transforma-
tion, the defined interaction rules are able to trans-
form the model into a rule-independent version, suit-
able for data usage through e.g. simulations, by us-
ing the automatic model transformation algorithm of
InTra. Therefore linked interaction rules are parsed
by the algorithm, to define active filters for the pattern
matching of the target model. Pattern matching will
then find valid interaction paths inside the model and
apply the defined rewriting effects between all start
and endpoints of those found paths. Even though the
rule-based version of the model is fully expressive, we
InTra: A Pragmatic Approach of Using Rule-Based Model Transformation to Reduce Complexity of UML and SysML Models
99
Figure 2: Extended family tree model with multiple incorporated relationship types.
need a way to (back-)convert the model to the tradi-
tional representation, as a backbone for our method-
ology. Furthermore, the transformation into indepen-
dent relations can help modelers to understand the
statement of a complex interaction rule and makes the
model readable again for simulations and other exter-
nal programs. For this reason we use model trans-
formation techniques to be able to switch between
rule-based and rule-independent representation of the
model.
6 APPLICATION SHOWCASE
An illustrative example, although not productively us-
able, is a family tree as presented in Figure 2, which
shall serve as an example to visualize the InTra ap-
proach. A family tree is a good way to explain the
idea of InTra, because we already use the concepts
of rule-based modelling subconsciously when think-
ing about family trees. The shown tree model consists
of actors (elements), their gender (attributes), and the
relationships between family members (connectors).
As you can see, the model in Figure 2 shows sev-
eral family relations in only one diagram. Normally,
in a family tree we would expect exclusively Child
connectors, and we would still be able to read other
family relations from the model subconsciously. This
cognitive concept is very similar to rule-based InTra
approach. By using interaction rules to embed infor-
mation into a model, that can be abstracted from ex-
isting information, we are able to reduce the model to
its relevant basic system structure. In this case it is
sufficient to only keep the Child relations, as we are
used to from regular family trees, to be able to repre-
sent all other relationships with the help of rules.
For easier explanation we will concentrate on only
a small part of the given family tree, which can be
seen in Figure 3. Based on the here shown Child
relations, which are part of the model’s basic struc-
ture, we will build interaction rules that represent all
other visible connectors. For example, each male or
female source element of a Child relation is a fa-
ther or a mother of the target element of that rela-
tion, respectively. To replace this kind of relation
with an interaction-rule, such rule must be defined to
search for incoming Child connectors for each pos-
sible element and follow that connector exactly one
hop against its direction. Thereupon, the found par-
ent element must be distinguished between male (M)
or female (W), respecting the gender attribute (SEX ), in
order to then represent any Father or Mother rela-
tion in the model. Once these two rules are associated
with the family tree model, all Mother, and Father
connectors can be removed from the model without
deleting their information content.
Another, only slightly more complex example are
the grandparent relationships. For this purpose the
above described rules need to be modified only mini-
mally. Instead of following only one Child connector
from target to source, the rule must follow exactly two
Child connectors (two hops) in a row. The genders
MODELSWARD 2023 - 11th International Conference on Model-Based Software and Systems Engineering
100
Figure 3: Small excerpt of the family tree shown in Figure 2.
of the intermediate steps are irrelevant for the eval-
uation of the current rules. Thereupon, the distinc-
tion between female and male works exactly as for
the Mother and Father rule and replaces all relations
Grandmother and Grandfather respectively.
Table 1: Overview of the number of connectors replaced by
interaction rules in the family tree model.
Interaction-Rule Replaced connectors
Daughter Rule 26
Son Rule 24
Mother Rule 25
Father Rule 25
Sister Rule 34
Brother Rule 28
Grandmother Rule 33
Grandfather Rule 33
Granddaughter Rule 30
Grandson Rule 36
Aunt Rule 15
Uncle Rule 17
Niece Rule 21
Nephew Rule 11
Sum 358
By using this approach in the example model from
Figure 3, this model will result in only containing ac-
tors and Child connectors. Thus only 6 of the 16 con-
nectors shown in Figure 3 are still part of the model,
which is a reduction of 62.5%. However, the infor-
mation content of the other 10 connectors is not lost,
but merely abstracted in the form of interaction rules.
With increasing number of elements and connectors
in the model, interaction rules are a powerful tool to
reduce the overall complexity, since a single rule can
replace a large number of relations, and still remain
comprehensible and maintainable.
Applying the InTra approach in the same way to
the large family tree from Figure 2, using interac-
tion rules for all family relationships Daughter, Son,
Mother, Father, Sister, Brother, Grandmother,
Grandfather, Granddaughter, Grandson, Aunt,
Uncle, Niece and Nephew leads to even clearer re-
sults. Table 1 gives an overview about how many
connectors could be replaced by each of the 14 in-
teraction rules in the family tree example from Figure
2. A total of 358 connectors could be abstracted by
rules. This leaves 50 child relations, from which all
other connections can be derived through model trans-
formation. This corresponds to a reduction of 87%
of connectors in the family tree model from Figure
2. Nevertheless, this approach can be productively
applied to other system models with greater practical
relevance but similar reduction potential.
7 RULE DEFINITION
InTra was implemented as an addin for the modelling
tool Enterprise Architect and was programmed in C#.
With the help of this plug-in it is possible to use the
InTra: A Pragmatic Approach of Using Rule-Based Model Transformation to Reduce Complexity of UML and SysML Models
101
Figure 4: InTra RuleConfiguration: Mother.
proposed approach directly inside Enterprise Archi-
tect, without the need of additional tools or program-
ming knowledge.
As presented in Figure 4, interaction rules in In-
Tra are defined as model elements themselves, and
thus become part of the model in which they are used.
The shown rule consists of a RuleConfiguration,
which holds the rule’s Name and HopCount, a Pat-
tern Matching (Left-Hand-Side) area and a Rewrit-
ing (Right-Hand-Side) area. The names chosen for
these parts of a rule follow the common practice
of several authors, including (Varr
´
o, 2007), (Wieber,
2015) and (Kahani, 2019), among others. In con-
trast to the representation of LHS and RHS in Fig-
ure 1 from (Varr
´
o, 2007), InTra does not use a di-
rect subgraph representation for rule definition, but
rather defines LHS and RHS with the help of in-
teraction rules, by using diagram-based interaction
constraints for both vertices and edges of the target
model. The Pattern Matching area defines condi-
tions which need to be fulfilled by elements (defined
as WayPoint conditions) and connectors (defined as
LinkConfiguration conditions), respectively, to de-
fine a certain pattern of interaction and thus find a
set of valid interaction paths in the linked model seg-
ment. The Rewriting area describes the modifications
that are made to the model if a valid interaction path
is found and model transformation shall be executed.
Rewriting effects can be the creation or the removal
of certain kind of connectors between a start and end
WayPoint of a found interaction path. For the defi-
nition of filter conditions the element’s and connec-
tor’s attributes Name, Alias, EA-Type, Stereotype,
minimal Multiplicity, maximal Multiplicity,
Embedment and any kind of Tagged Value can be
analysed. Therefore the values of those attributes can
be compared by direct text comparison, regular ex-
pressions and numeric comparison operators like <,
<=, >, >=, == and ><, formulated as Tagged Value of
a WayPoint or a LinkConfiguration. To define a
Pattern Matching for possible interaction paths in the
model, the mentioned conditions are connected with
the core of the rule by the connectors From, Over, To
and RelatedBy. A valid interaction path must be-
gin with an element conforming to a From WayPoint,
be linked to other elements by connectors conform-
ing to a RelatedBy LinkConfiguration, end in a
element conforming to a To WayPoint and must be
connected to exactly (HopCount-1) elements in be-
tween, which are conforming to a Over WayPoint.
Multiple WayPoints or LinkConfigurations of the
same type will result in a logical OR for that spe-
cific condition pattern. While the rule shown in Fig-
ure 4 represents the most basic form of an inter-
action rule, a so called RuleConfiguration, there
is also the possibility to combine several of these
rather simple rules in a more complex rule, a so
called RuleChain. Thus, RuleConfigurations can
be thought of as atomic rule pieces that can be as-
sembled into a more complex RuleChain. Rules can
therefore be decomposed and offer high reusability.
An example RuleChain for the family model from
Figure 2 is the Sister rule shown in Figure 5. The
rule evaluation of a RuleChain always starts with a
RuleChainStart element. From there on, the con-
tained RuleConfigurations are executed one after
another (AppendRule), until a RuleChainEnd ele-
ment is reached. Each interaction path found by a
RuleConfiguration is thereby extended by the next
RuleConfiguration in the chain, by passing all To
WayPoints of one rule to the next one as possible
From WayPoints. The resulting interaction path of a
MODELSWARD 2023 - 11th International Conference on Model-Based Software and Systems Engineering
102
RuleChain thus consists of start elements of the first,
and end elements of the last rule of the chain. Thus,
it is certainly not a successive execution of transfor-
mation rules, but rather a narrowing down of the so-
lution space, based on conditions that sequentially
build upon each other. Each RuleChain defines its
own effects and reductions as LinkConfigurations,
which are linked to the endpoint of the chain anal-
ogous to a RuleConfiguration, using an Effect
or Reduction connector. In our Example from Fig-
ure 5 initially the Parent RuleConfiguration is
evaluated, which identifies the parents of a person.
Then, these elements are passed as potential From
WayPoints of the Daughter RuleConfiguration.
Evaluating this rule finds all daughters of a person,
and in this particular case all daughters of a person
who is parent of a different person. So, in summary,
the two rules in succession find all daughters of all
parents of a person, which corresponds to all sisters
of that person. Some attentive readers may wonder
at this point what happens if the source element itself
is female. If the source element itself is female, it is
correctly not identified as a sister of herself, because
connectors that have already been visited by one of
the RuleConfigurations are not visited again dur-
ing Pattern Matching of a RuleChain. For this rea-
son, it is also impossible to get caught in cycles dur-
ing the model transformation. The fact that interac-
tion rules themselves are also defined as model el-
ements has the additional advantage that interaction
rules themselves can also be transformed via other In-
Tra rules. Due to this property InTra rules can be seen
as rules of higher order (Mens, 2006).
Figure 5: InTra RuleChain: Sister.
8 INDUSTRIAL APPLICATION
RESULTS
In the communication model introduced in Section 3,
the InTra approach could be successfully applied to
abstract the communication relationships in multiple
views of the model, thus highly reducing the com-
plexity of the system. This has proven to be extremely
helpful, especially with regard to the readability and
maintainability of the model. By building 128 in-
teraction rules and creating 255 links between rules
and parts of the model, the number of information ex-
changes in the model could be reduced from 9,355
to 2,615. This is a reduction of 72 %. Therefore, it
could be proven, that the InTra approach is able to
be applied lucratively in realistic, productive systems
and model environments. Furthermore, it was possi-
ble to explain the contents of the model to the domain
experts more easily with the help of interaction rules.
Subsequently, we received feedback from the future
model users, that they perceived the new approach to
be comprehensible and helpful.
9 RELATED WORK
In preparation for the development of the InTra ap-
proach, the current state of the art on the topic was ex-
amined, whereby it was necessary to find out, which
generic requirements a transformation approach must
fulfill in order to be considered suitable. These re-
quirements were identified based on the taxonomies
and comparative characteristics of the contributions
of (Czarnecki, 2006), (Ghamarian, 2011), (Kahani,
2019) and (Mens, 2006). All listed works offered the
possibility of identifying important criteria regarding
the requirements for rule-creation, model transforma-
tion and general usability of the approach, but espe-
cially (Czarnecki, 2006) and (Czarnecki, 2002) pro-
vided a majority of useful properties and were able
to give a good view on existing model transformation
approaches. They also provided their findings based
on a feature diagram according to (Kang, 1990) in a
structured and comprehensible way. (Kahani, 2019)
and (Varr
´
o, 2007) give an excellent introduction into
the field of graph-based model transformation. Both
(Valiente, 1997) and (McCreesh, 2020) are dealing
with the challenge of the NP-complete complexity of
the subgraph-isomorphism problem, which describes
the usual run-time limit for the Pattern Matching in
traditional model transformation approaches.
InTra: A Pragmatic Approach of Using Rule-Based Model Transformation to Reduce Complexity of UML and SysML Models
103
10 CONCLUSION
In this paper we investigated the usage of model
transformation techniques to reduce the complexity
of large system models, to improve the usability
and comprehensibility. We elaborated on the com-
mon problems of system modeling, gave an overview
about the field of graph-based model transformation,
introduced our own rule-based model transformation
approach InTra, described how rules can be defined
and applied and gave an example of the results that
can be expected by using the approach in large system
models. We also highlighted the advantages of the
approach and how it can be used to tackle the prob-
lems that may arise in large models. We were able to
successfully use the approach in the industrial setting
described in Section 3, and presented our results and
experiences in Section 8. As application showcase we
presented a family tree model in which arbitrary fam-
ily relations could be derived using the existing child
relations and the given basic structure of the model.
We were able to show that the approach in this ex-
ample was well suited to reduce the system complex-
ity without affecting the information content of the
model. The InTra rule-based modelling approach has
the potential to make the growing complexity of to-
day’s system models more manageable, comprehen-
sible and maintainable through the use of rules.
However, it is primarily necessary to identify fur-
ther real industry scenarios, similar to the one from
Section 3, in order to further test the approach for
its abilities in real world scenarios. Potential cases
should primarily have a clear basic system structure
and reoccurring relation patterns. Likewise, the mod-
els should be quite large, in order to justify a rule-
based reduction of connectors. Case studies based on
further practical application could be used to inves-
tigate and enhance the InTra approach. In addition,
it is necessary to investigate with further empirical
studies, if a reduction in the number of connectors
actually leads to a reduction in complexity in terms
of human perception, and whether interaction rules
are truly more comprehensible for the viewer than
the sum of connectors they replace. Nevertheless, it
should be considered that domain experts and sys-
tem architects will perceive models in different ways,
which could lead to diverging results.
REFERENCES
Antinyan, V. (2020). Revealing the Complexity of Automo-
tive Software. Volvo Car Group.
Baduel, R. (2018). SysML Models Verification and Valida-
tion in an Industrial Context: Challenges and Experi-
mentation.
Balasubramanian, D. (2006). The Graph Rewriting and
Transformation Language: GReAT.
Braun, P. (2003). BOTL The Bidirectional Object Oriented
Transformation Language.
Czarnecki, K. (2002). Classification of Model Transforma-
tion Approaches.
Czarnecki, K. (2006). Feature-Based Survey of Model
Transformation Approaches.
Fujaba Development Group (2021). Fujaba Tool Website.
https://web.cs.upb.de/archive/fujaba/. Website Uni-
versity of Paderborn.
Ghamarian, A. H. (2011). Modelling and analysis using
GROOVE.
ISIS (2021). GReAT Tool Website. https://www.isis.
vanderbilt.edu/tools/GReAT.
Kahani, N. (2019). Survey and classification of model trans-
formation tools.
Kang, K. C. (1990). Feature-Oriented Domain Analysis
(FODA) Feasibility Study.
Koning, H. (2002). Practical Guidelines for the Readability
of IT-architecture Diagrams.
McCreesh, C. (2020). The Glasgow Subgraph Solver: Us-
ing Constraint Programming to Tackle Hard Subgraph
Isomorphism Problem Variants.
McGill (2021). AToM3 Introduction Website. McGill Uni-
versity.
Mens, T. (2006). A Taxonomy of Model Transformation.
Purchase, H. C. (2000). User Preference of Graph Layout
Aesthetics: A UML Study.
RWTH Aachen (2021). PROGRES Tool Website. http://
www-i3.informatik.rwth-aachen.de.
Sch
¨
urr, A. (2021). PROGRES - A Graph Transfor-
mation Programming Environment. https://www.es.
tu-darmstadt.de/.
St
¨
utzel, B. (2021). Systems Engineering in Deutschland -
Die deutsche Unternehmenslandschaft im Vergleich.
Studie 2021 — Publisher: Prozesswerk GmbH.
Taentzer, G. (2003). AGG: A Graph Transformation Envi-
ronment for Modeling and Validation of Software.
Twente (2021). GROOVE Tool Website. https://groove.ewi.
utwente.nl/about. Twente University.
Valiente, G. (1997). An algorithm for graph pattern-
matching.
Varr
´
o, D. (2007). The model transformation language of the
VIATRA2 framework.
Warrilow, R. (2020). Lost in Translation? – How MBSE is
Evolving to address Today’s Complexity Challenges.
Wieber, M. S. (2015). Qualit
¨
atssicherung von Modelltrans-
formationen -
¨
Uber das dynamische Testen program-
mierter Graphersetzungssysteme.
MODELSWARD 2023 - 11th International Conference on Model-Based Software and Systems Engineering
104