A Meta-model based Automatic Conceptual Model-to-Model
Transformation Methodology
Tiexin Wang
1
, Sebastien Truptil
2
, Frederick Benaben
2
and Chuanqi Tao
1
1
College of Computer Science and Technology, Nanjing University of Aeronautics and Astronautics,
Jiangjun Road, Nanjing City, China
2
Centre Genie Industriel, IMT Mines Albi, Universite de Toulouse, Campus Jarlard, 81000 Albi, France
Keywords: Conceptual Dissimilarity, Automatic Model Transformation, Semantic Checking Measurements, Meta-
model based Transformation Process.
Abstract: Since model-based engineering theories and techniques becoming mature gradually, diverse engineering
domains have adopted the idea of employing modelling and model transformations to help simulate and
analyze domain specific problems. Consequently, substantial numbers of modelling techniques have been
developed. These modelling techniques define specific semantic and syntactic representations. Moreover,
models are normally built to represent systems from diverse domains. Both the conceptual dissimilarities
between modelling techniques and between diverse systems determine the particularity of models. In model
transformation process, distinguishing the conceptual difference from both semantic and syntactic aspects is
a time-consuming process relying mainly on manual effort. In order to remove the manual effort from
model transformation process, this paper proposes a generic automatic conceptual model-to-model
transformation methodology. This methodology employs semantic and syntactic checking measurements to
automatically detect the conceptual dissimilarities, and aims to solve both domain specific problems and
cross-domain problems. A refined meta-model based model transformation process is defined to better use
the two checking measurements.
1 INTRODUCTION
With the gradually mature of the theories and
techniques in model based engineering (MBE), more
and more engineering domains have adopted MBE
principles to solve domain problems. As two of the
key concepts in MBE modelling and model
transformation attract attention from both
academics and industrials.
Modelling means the activities of building
models. For different purposes, substantial numbers
of modelling techniques (e.g., UML, BPMN) have
been developed by employing specific semantic and
syntactic representations. A research of modelling is
presented in (Muller et al., 2012).
Models are built to represent systems, and model
transformations can simulate the interactions or
indicate the connections between systems.
Furthermore, for a specific system, concerning
different views, many models can be built to
represent it. Many defined model, table 1 shows four
definitions of model.
Table 1: Four definitions of model.
No.
Definitions
1
“Models provide abstractions of a physical
system that allow engineers to reason about that
system by ignoring extraneous details while
focusing on the relevant ones.” (Brown, 2004)
2
“A model is an abstraction of a (real or
language based) system allowing predictions or
inferences to be made.” (Kühne, 2006)
3
“A model of a system is a description or
specification of that system and its environment
for some certain purpose.” (OMG, 2006)
4
“Engineering models aim to reduce risk by
helping us better understand both a complex
problem and its potential solutions before
undertaking the expense and effort of a full
implementation.” (Selic, 2003)
A model is particular because it is built for a
specific purpose (e.g., describing a view of a
complex system) and by using a specific modelling
technique. Models can be divided into different
groups. As stated in (Fowler et al., 1999),
586
Wang, T., Truptil, S., Benaben, F. and Tao, C.
A Meta-model based Automatic Conceptual Model-to-Model Transformation Methodology .
DOI: 10.5220/0006718105860593
In Proceedings of the 6th International Conference on Model-Driven Engineering and Software Development (MODELSWARD 2018), pages 586-593
ISBN: 978-989-758-283-7
Copyright © 2018 by SCITEPRESS Science and Technology Publications, Lda. All rights reserved
depending on the level of precision, models are
divided into three levels namely Conceptual
Models, Specification Models and Implementation
Models. Another similar distinction proposed in
(Mellor, 2004), a model can be considered as a
Sketch, as a Blueprint, or as an Executable.
In order to build connections between models
in the same level and from different levels, model
transformation practices are required. However, in
model transformation practices, distinguishing the
conceptual difference between two models is a time-
consuming process which is mainly relied on manual
effort.
As stated in (Del Fabro and Valduriez, 2009),
in traditional model transformation practices there
are several weaknesses: low reusability, contain
repetitive tasks and involve huge manual effort, etc.
Due to the wide requirement and usage of model
transformation practices, it is unacceptable to do
model transformation manually. Thus, this paper
proposes a generic (domain-cross) automatic
conceptual model-to-model transformation
methodology (ACMTM), which is built on the base
of semantic and syntactic checking measurements
(S&S). S&S is used to automatically detect the
conceptual similarities and build mapping rules. A
refined meta-model based model transformation
process is defined to better combine S&S in.
This paper is structured as follows. Section 2
presents the relevant theories to ACMTM. Section 3
shows an overview of ACMTM. A use case is
illustrated in Section 4. Finally, a conclusion draws
the advantages, potential improvement points and
future usage of ACMTM.
2 RELATED WORK
2.1 Model Transformation Definitions
Model transformation is a process, which contains a
sequence of activities operating on models. Many
propose the definitions about model transformation.
Table 2 shows three of them.
Model transformation is a process of generating
target models based on source models. The
transforming rules shall be built between same or
similar concepts that are from the two models,
respectively.
Table 2: Three definitions of model transformation.
No.
Definition
1
“model transformation is a program that
mutates one model into another” (Tratt, 2005)
2
“the process of converting a model into another
model of the same system” (Miller and
Mukerji, 2003)
3
“automatic generation of a target model from a
source model, according to a transformation
description” (Kleppe et al, 2003)
2.2 Model Transformation Category
Generally, model transformation can be divided into
three groups: Text-to-Model, Model-to-Model and
Model-to-Text. The content in models is presented
in abstract syntax, while the content in text is
presented in concrete syntax.
As defined in (Czarnecki and Helsen, 2003),
there are two main model transformation approaches:
model-to-code and model-to-model. For model-to-
code category, there are two kinds of approaches:
visitor-based” approaches and “template-based”
approaches. For model-to-model category, there are
five approaches: “direct-manipulation” approaches,
“relational” approaches, “graph-transformation-
based” approaches, “structure-driven” approaches
and “hybrid” approaches. In model-to-model
transformation category, there are also some other
approaches, such as: marking and pattern approach,
automatic transformation approach, meta-model
based transformation approach, model merging
approach, etc.
ACMTM belongs to model-to-model
transformation category. It is designed and
implemented as a hybrid approach which is also a
meta-model based.
2.3 Model Transformation Techniques
Focusing on model-to-model transformation
category, there are several well-known techniques.
Table 3 shows four of these techniques.
ATL and QVT are similar to each other on
architecture aspect. Both VIATRA2 and GReAT
focus mainly on graph models. Usually, specific
model transformation techniques can be only used
on models that are built by specific modelling
techniques. Also, model transformation techniques
integrate (or rely on) other techniques, such as
QVT OCL, VIATRA2 graph transformation
techniques, etc. Current model transformation
techniques lack the ability of automatically detecting
A Meta-model based Automatic Conceptual Model-to-Model Transformation Methodology
587
model transformation mappings, and require manual
effort to operate them.
Table 3: Model-to-model transformation techniques.
Name
Characteristic
Note
ATL
(Jouault et
al., 2008)
Hybrid (declarative
& imperative); three
layers architecture
self-executed
(provide both
transformation
language & toolkit)
QVT
(Omg,
2008)
Hybrid three kinds of
transformation
languages involved
based on MOF 2.0
(Omg, 2008)
integrated OCL
VIATRA2
(Varró and
Balogh,
2007)
Unidirectional
transformation
language; based
mainly on graph
transformation
techniques
operates on models
conformed to VPM
meta-modeling
approach
GReAT
(Karsai et
al., 2003)
Visual language
developed using
Generic Modeling
Environment
operates on models
conform to meta-
models specified in
UML
Based on these model transformation
techniques, numerous model transformation
practices have been developed, such as the work
stated in (De Castro et al., 2011; Fleurey et al., 2007;
García et al., 2013).
Comparing with the existing model
transformation techniques and practices, ACMTM
aims to be a generic, automatic conceptual
model-to-model transformation methodology. It
provides a theoretical framework and employs
semantic and syntactic checking measurements as
potential mappings detecting techniques.
3 ACMTM OVERVIEW
ACMTM employs S&S in a refined meta-model
based model transformation process. S&S is
illustrated first in this section. Then, the refined
transformation process is presented.
3.1 Semantic & Syntactic Comparisons
3.1.1 Use of S&S
In ACMTM, semantic checking and syntactic
checking measurements are combined as a single
function. This function is used between items on
meta-model level. Figure 1 shows the relation
between them and its usage in ACMTM.
S&S takes two words (strings) as inputs, and its
output is the matching possibility between the two
strings. For the syntactic checking part, it contains
two steps: predefined treatment (pretreatment) and
employing “Levenshtein distance” algorithm
(Hirschberg, 1997; Gilleland, 2009).
Figure 1: S&S illustration.
Predefined treatment” also contains two phases:
special forms detection and applying stemming
algorithms. Both the two phases aims to discover
special semantic relations (e.g., synonym and
antonym) between a pair of words. If the
pretreatment step fails in discovering such kinds of
semantic relations, then the second step employing
“Levenshtein distance” algorithm will be executed.
This algorithm calculates the syntactic similarity
between a pair of words. This syntactic similarity
stands by a value ranges between 0 and 1.
In order to detect the potential semantic relations
between two comparing words, a semantic thesaurus
ACMTM_ST, is created. Semantic relations
stands by a calculating (or assigned) value defined
within ACMTM context.
Equation (1) is defined to calculate the S&S
relation, between two words (strings). The S&S
relation is represented by a value which is the sum
of two aspects: semantic and syntactic.
S_SSV=SeV_weight*S_SeV+SyV_weight*S_SyV
(1)
“S_SSV” stands for the S&S value between a
pair of words. “S_SeV” stands for the semantic
MODELSWARD 2018 - 6th International Conference on Model-Driven Engineering and Software Development
588
value while “S_SyV” stands for the syntactic value.
Two coefficients: “SeV_weight” and “SyV_weight”
are defined. Their value range is 0 to 1, and the
sum of them is ‘1’. They are used to determine
which aspect is more important in determining S&S
value between a pair of words.
3.1.2 Syntactic Checking Measurements
Syntactic checking measurements focus on forms of
words (e.g., do-doing, student-students), formats of
concepts (date description in different cultures), and
units (Celsius, Fahrenheit and Kelvin measuring
temperature) used to describe subjects.
In the first checking phase, inspired by the
research work stated in (Benaben et al., 2013), a
profile is created and used to detect the different
formats and units standing for the same or similar
concepts. For words in different forms (also words
belong to the same semantic group: concerning the
stemming issue), a special algorithm word forms
detecting: WF_D is developed to detect these
situations. WF_D adopts parts of the porter
stemming (Porter, 1980) algorithms.
The second phase employs “Levenshtein
Distances” algorithm which is a string metric for
measuring the difference between two alphabet
sequences. Informally, the Levenshtein distance
between two words is the minimum number of
single-character edits (i.e. insertions, deletions or
substitutions) required to change one word into the
other.
Mathematically, the Levenshtein distance
between two strings: string a and string b with the
length │a│ and b│, respectively) is given by
“Lev
a,b
(│a│, │b│)”.In order to use this value,
equation (2) is defined.
S_SyV=1Lev
a,b
(│a│,│b│)/ max (│a│,│b│) (2)
The value of “S_SyV”, which first appears in
equation (1), shall always be within the range of 0 to
1. The higher of this value means the higher
syntactic similarity between two comparing words.
3.1.3 Semantic Checking Measurements
Semantic checking measurements focus on the
semantic meanings. Between a pair of words, one
syntactic similarity value always exists, while
several or no semantic relations (with different
semantic values) can exist.
To support semantic checking, ACMTM _ST,
which contains large amount of words, semantic
meanings and semantic relations, is particularly
created to support ACMTM. It adopts parts of the
content stored in “WordNet” (Fellbaum, 1998).
Figure 2 shows the structure of ACMTM _ST.
Three kinds of items are stored in ACMTM _ST.
Word Base: contains 147306 English words (i.e.,
nouns, verbs and adjectives).
Word-sense Base: contains 206941 senses that
owned by the words stored in “Word Base”.
Synset Base: contains 114038 synsets. A synset
contains a group of word senses, which own
synonym meanings; semantic relations are built
among different synsets.
Figure 2: ACMTM_ST structure.
The relation between word and word senses is
one-to-several, and the relation between word
senses to synset is several-to-one. Eleven kinds of
semantic relations (adopted from WordNet) are
maintained among synsets in ACMTM_ST. Table 4
shows these semantic relations and their values
pairs.
Table 4: Semantic relations maintained in ACMTM_ST.
Semantic relation
S_SeV
Example
synonym
0.9
shut & close
hyponym
0.6
person-creator
hypernym
0.8
creator-person
similar-to
0.85
perfect & ideal
partmeronym
0.7
tire & car
partholonym
0.55
car & tire
membermeronym
0.65
car & traffic jam
memberholonym
0.45
traffic jam & car
Antonym
0.1
good & bad
iterative hyponym
0.6
n
person-creator-maker
iterative hypernym
0.8
n
maker-creator-person
A Meta-model based Automatic Conceptual Model-to-Model Transformation Methodology
589
The “S_SeV” (first introduced in equation (1))
stands for the semantic similarity between a pair of
comparing words. The higher of this value means
the closer of the two words in semantic aspect. All
these “S_SeV” values are assigned directly (based
on experience).
Both the calculating rules for “S_SyV” and
“S_SeV” are illustrated. The “S_SSV” between any
pair of comparing words is computable.
On the basis of semantic and syntactic checking
measurements, a S_SSV value can be calculated.
This S_SSV value means the possibility of
matching two words. The determination mechanism
is shown in Figure 3.
Figure 3: Matching pair chosen mechanism.
According to the range of S_SSV, three regions
are divided. If two words have a S_SSV in Region 1,
the two words have a high matching possibility.
While this value in Region 2, the two words have a
medium match degree. If this value is in region 3, no
matching can be made between the two words.
3.2 ACMTM Theories & Process
The S&S illustrated above are used between word
pairs, while ACMTM focuses on transforming
models. So, a refined meta-model based model
transformation process is created.
3.2.1 ACMTM-MMM
Meta-model is a special kind of model which defines
the rules of building models. Meta-models can exist
in several levels.
In a model transformation process, a model is
regarded as two parts: shared part (transformable)
and specific part (non-transformable). Both shared
and specific part on model layer can be traced on
meta-model layer as shared and special concepts. In
this way, identifying the shared part on model layer
becomes detecting the shared concepts on meta-
model layer. In ACMTM, the mechanism of
applying S&S is defined in a meta-meta-model.
There are several meta-modelling architectures,
two of them are: “MOF: Meta-Object Facility”
(Omg, 2008) and “ISO/IEC 24744” (Henderson-
Sellers and Gonzalez-Perez, 2008). These are
general-purpose architectures. For supporting
particularly to model transformation field, a specific
meta-meta-model ACMTM-MMM is created.
As shown in Figure 4, there are nine core
elements in this meta-meta-model. Model stand
for all the model instances. Model is made of
Element”, which has two inheritances: Node
(concepts) and Edge (relations). Element is self-
contained. “Node” are linked by “Edge” based on
their roles”. Element has a group of Property”,
Property can identify and explain the “Element”.
Semantic Relation and Syntactic Relation
exist between different kinds of items (i.e. between
elements pairs, between propertys pairs, between
models pairs and between environments pairs).
Potential model transformation mappings are built
based on them.
Figure 4: The structure of ACMTM-MMM.
3.2.2 Iterative Transformation Process
Model transformation is regarded as an iterative
process in ACMTM. Between the original source
model and final target model, several intermediate
models can be generated. The target model of former
iteration becomes the source model of latter iteration.
In each iteration phase, the specific part of
source model shall be stored in ontology named
ACMTM_O. Also, the specific part of target
model shall be enriched by additional knowledge
stored in ACMTM_O.
As shown in Figure 5, the content stored in
ACMTM_O comes from both the specific part of
source models and other knowledge base (e.g.,
domain ontologies).
MODELSWARD 2018 - 6th International Conference on Model-Driven Engineering and Software Development
590
Figure 5: Iterative transformation process illustration.
3.2.3 Four Matching Steps
To apply S&S to build potential mappings between
meta-models, four matching steps are divided. These
four matching steps aim to solve the inherent
granularity issue in model transformation domain:
M-to-N matching and cross-level (element-property)
matching. Figure 6 shows an overview of the four
matching steps.
Figure 6: Four matching steps.
The first step matching on element level:
aims to build mappings between element’s pairs
(considering elements’ names and their property
groups) and between property’s pairs (considering
properties’ names and types) which are within the
matched element pairs. Two equations: (3) and (4)
are defined to do this matching step.
Ele_SSV=name_weight*S_SSV+
property_weight*( 

)/x
(3)
P_SSV=pn_weight*S_SSV+pt_weight*Id_type (4)
The second step hybrid matching focuses on
properties (property-to-property matching), which
are unmatched after executing the first matching
step. Equation (5) is defined for this matching step.
HM_SSV=en_weight*S_SSV+pl_weight*P_SSV (5)
The third step cross-level matching concerns
making mappings between properties and elements.
This step focuses on the unmatched elements and
properties after executing the two former matching
steps. S&S are applied between elements names
and properties names. Equation (6) is defined to
work for this step.
CM_SSV=sem_weight*S_SeV+syn_weigh*S_SyV (6)
Ele_SSV stands for the semantic and syntactic
value between an elements pair, while P_SSV
stands for this value between propertys pairs.
HM_SSV stands for the value of hybrid matching
and CM_SSV for cross-level matching value. All
of the four values are the sum of two variables. In
each of the equations, two impact factors (e.g.
name_weight & property_weight), the sum of them
is 1, are defined to determine which of the two
variables plays a more important role in deciding the
final equation value.
All the three matching steps aim to define
mappings within the shared part. For the specific
parts, the fourth step “auxiliary matching” can be
used.
Auxiliary matching focuses on enriching the
specific parts of target models by extracting
additional knowledge from ACMTM_O. It reuses
the three former matching steps to detect
potential model transformation mappings, while
taking ACMTM_O as the source meta-model.
4 USE CASE
To explain and test the working mechanism of
ACMTM, a simple use case is illustrated in this
section. This use case concerns the process of
comparing two Elements. The two elements are
shown in Figure 7.
A Meta-model based Automatic Conceptual Model-to-Model Transformation Methodology
591
Figure 7: Iterative transformation process illustration.
Two elements student (with five properties)
and person (with seven properties) are taken as
inputs. The outputs are potential mappings between
them. Before executing the detecting process,
concrete values are assigned to the parameters used
in equation (1), (3), and (4). Table 5 shows the
assigning value pairs.
Table 5: Assigning values to parameters.
No.
Equation
Parameter
value
1
SeV_weight, SyV_weight
0.9, 0.1
3
name_weight,
property_weight
0.5, 0.5
4
pn_weight, pt_weight
0.8, 0.2
Taking the calculation process of Ele_SSV”
between two elements: student and person as an
example; equation (3) is used to do this step. Figure
8 is the screenshot of calculating the S_SSV value
between elements names: “student” and “person”.
Figure 8: S&S comparisons between elements names.
The word “student” has two semantic meanings,
and the word “person” has three semantic meanings.
The semantic relation between the two words is
“iterative hypernym”, and the semantic value
between them is 0.64”. The syntactic similarity
value between them is: 0.1428. In this use case,
semantic relation is assumed more important than
syntactic relation, so two coefficients:
“SeV_weight” and “SyV_weight” in equation (1)
are assigned with values as 0.9 and 0.1, respectively.
The final S&S value between the two words is:
0.5903.
The S&S comparisons between the two elements
properties groups are calculated by using equation
(4). Table 6 is created to store these comparison
values. Between each pair of properties, a P_SSV
can be calculated. The two parameters pn_weight
and pt_weight are assigned with values 0.8 and
0.2. This means property name is more important
than property type when making mappings.
When calculating Ele_SSV, the two parameters
in Equation (3) are assigned as 0.5 and 0.5. This
means element name and property group have the
same weight in deciding element matching pairs.
The “Ele_SSV” calculated between student
and person is: 0.695. According to the matching
pair chosen mechanism, there is a medium potential
mapping exist between them.
5 CONCLUSION
This paper presents an automatic conceptual model-
to-model transformation methodology: ACMTM.
Comparing with the existing model transformation
methodologies, two main characteristics of ACMTM
are: generic and automatic.
ACMTM combines semantic and syntactic
checking measurements into a refined meta-model
based model transformation process. Also, ACMTM
takes model transformation as an iterative process
and four matching steps are divided within each
iteration phase. To better use S&S, five equations
have been defined to use in different matching steps.
Some potential improvements in ACMTM are as
follows.
A validation and evaluation process of the
automatic generated model transformation
mappings is required.
Strengthen semantic checking measurements by
extending ACMTM_ST with more content from
specific domains (e.g., ontology).
A better way to assign values to coefficients
defined in equations (3), (4), (5) and (6) (e.g.,
mathematical, statistical analysis).
MODELSWARD 2018 - 6th International Conference on Model-Driven Engineering and Software Development
592
Table 6: S&S comparisons for property groups.
person \ student
id
surname
age
phone
address
id
1
-
-
-
-
name
0.2
0.6777
0.04
0.016
0.011
age
0
0.0229
1
-
-
address
0.21
0.2
0.02
0.011
0.8
sex
0.2
0.2114
0
0
0.011
teacher
0
0
0.02
0.011
0.2
ACKNOWLEDGEMENTS
The authors would like to acknowledge the financial
support from European Commission C2Net project
(H2020-FoF-1-2014/636909), Chinese Scholarship
Council, National Natural Science Foundation of
China (61502231) and Natural Science Foundation
of Jiangsu Province (BK20150753).
REFERENCES
Benaben, F., Boissel-Dallier, N., Pingaud, H., Lorre, J. P.
2013. Semantic issues in model-driven management of
information system interoperability. International
Journal of Computer Integrated Manufacturing,26(11),
1042-1053.
Brown, A.W., 2004. Model driven architecture: principles
and practice. SoSyM 3(3), 314327.
Czarnecki K, Helsen S., 2003. Classification of model
transformation approaches[C]//Proceedings of the 2nd
OOPSLA Workshop on Generative Techniques in the
Context of the Model Driven Architecture, 45(3): 1-
17.
De Castro V, Marcos E, Vara J M., 2011. Applying CIM-
to-PIM model transformations for the service-oriented
development of information systems [J]. Information
and Software Technology, 53(1): 87-105.
Del Fabro, M. D., & Valduriez, P., 2009. Towards the
efficient development of model transformations using
model weaving and matching transformations.
Software & Systems Modeling, 8(3), 305-324.
Fellbaum, C., 1998. WordNet. Blackwell Publishing Ltd.
Fleurey, F., Baudry, B., France, R., Ghosh, S., 2007. A
generic approach for automatic model composition. In
Models in Software Engineering (pp. 7-15). Springer
Berlin Heidelberg.
Fowler, M., Scott, K., Booch, G., 1999. UML distilled,
Object Oriented series, 179 p.
García, J., Diaz, O., Azanza, M., 2013. Model
transformation co-evolution: A semi-automatic
approach. Software Language Engineering,7745, 144-
163.
Gilleland, M., 2009. Levenshtein distance, in three flavors.
Merriam Park Software: http://
www.merriampark.com/ld.htm.
Henderson-Sellers B, Gonzalez-Perez C., 2008.
Standardizing methodology metamodelling and
notation: an ISO exemplar [M]. Springer Berlin
Heidelberg.
Hirschberg, D., 1997. Serial computations of Levenshtein
distances.
Jouault, F., Allilaire, F., Bézivin, J., & Kurtev, I., 2008.
ATL: A model transformation tool. Science of
computer programming, 72(1), 31-39.
Karsai, G., Agrawal, A., Shi, F., & Sprinkle, J., 2003. On
the use of graph transformation in the formal
specification of model interpreters. J. UCS, 9(11),
1296-1321.
Kleppe, A. G., Warmer, J. B., Bast, W., 2003. MDA
explained: the model driven architecture: practice and
promise. Addison-Wesley Professional.
Kühne, T., 2006. Matters of (meta-) modeling. SoSyM,
5(4).
Mellor, S. J., 2004. MDA distilled: principles of model-
driven architecture. Addison-Wesley Professional.
Miller, J., Mukerji, J., 2003. MDA Guide Version 1.0. 1
Muller, P. A., Fondement, F., Baudry, B., & Combemale,
B., 2012. Modeling modeling modeling. Software &
Systems Modeling, 11(3), 347-359.
OMG 2006: Model Driven Architecture. http://
www.omg.org/mda/.
Omg, 2008. Meta object facility (mof) 2.0
query/view/transformation specification. Final
Adopted Specification.
Porter, M. F., 1980. An algorithm for suffix stripping.
Program, 14(3), 130-137.
Selic, B., 2003. The pragmatics of model-driven
development. IEEE Softw. 20(5), 1925.
Tratt, L., 2005. Model transformations and tool
integration. Software & Systems Modeling, 4(2), 112-
122.
Varró, D., & Balogh, A., 2007. The model transformation
language of the VIATRA2 framework. Science of
Computer Programming, 68(3), 214-234.
A Meta-model based Automatic Conceptual Model-to-Model Transformation Methodology
593