CIM to PIM Transformation - International Journal of Computer ...

Hamid Reza Sharifi et al ,Int.J.Computer Technology & Applications,Vol 3 (2), 791-796
ISSN:2229-6093
CIM to PIM Transformation: An Analytical Survey
Hamid Reza Sharifi
Department of Computer Engineering
Science and Research Branch,
Islamic Azad University
Tehran, Iran
hr.sharifi@srbiau.ac.ir
Mehran Mohsenzadeh
Department of Computer Engineering
Science and Research Branch,
Islamic Azad University
Tehran, Iran
mohsenzadeh@srbiau.ac.ir
Seyyed Mohsen Hashemi
Department of Computer Engineering
Science and Research Branch,
Islamic Azad University
Tehran, Iran
hashemi@srbiau.ac.ir
Abstract
Model transformation is one of the main issues and key
elements in Model Driven Architecture (MDA). Computation
Independent Model (CIM) to Platform Independent Model
(PIM) transformation is the first transformation in MDA and it
is very important in designing high quality software. Several
CIM to PIM transformation approaches have emerged, they
have not been objectively analysed yet. In this paper, we
provide a review of several CIM to PIM transformation
approaches, and present a criteria-based evaluation. The
results can be used for assessing and comparing CIM to PIM
transformation approaches.
Keywords: MDA; CIM; PIM; Transformation; Evaluation
Criteria.
1. Introduction
The MDA is an approach for software development defined
by the Object Management Group (OMG). MDA defines four
models in various levels of abstraction [1]. They are CIM, PIM,
PSM, and Code.
CIM is a model that describes a system from the
computation independent viewpoint. PIM is a model that
contains no information specific to the platform. Platform
Specific Model (PSM) is a model that includes information
about the specific technology that is used in the realization of it
on a specific platform. Code is a specification of the system in
the source code. MDA lifecycle begins from CIM and ends
with code. It is depicted in Fig. 1.
Figure 1. MDA life cycle.
MDA is a software development approach based on
creation of models and transformations between them. Model
transformation is one of the main issues and key elements in
MDA.
CIM to PIM transformation is the first transformation in
MDA and it is very important in designing high quality
software. The Benefits of MDA appear in this transformation.
Several CIM to PIM transformation approaches have
emerged, they have not been objectively analysed yet. This
research presents an analytical review and evaluation of these
approaches.
The main steps of the research are CIM to PIM
transformation approach selection, development of evaluation
criteria, and criteria-based evaluation of these approaches.
This paper is organized as follows: section 2 discusses
existing CIM to PIM transformation approaches. Section 3
discusses a conceptual framework for CIM to PIM
transformation. Evaluation of proposed approaches is
discussed in section 4. Section 5 discusses conclusion of the
paper.
791

Hamid Reza Sharifi et al ,Int.J.Computer Technology & Applications,Vol 3 (2), 791-796
ISSN:2229-6093
2. Review of CIM to PIM transformation
approaches
In this section, we discuss proposed approaches for
transformation from CIM to PIM.
In [2] proposed a disciplined approach for transformation
from CIM to PIM using feature-oriented and component-based
approach. In this paper, feature model used for representing
requirement in CIM and this model includes a set of features
and relationship between them. Also software architecture
used for representing PIM and it includes a set of components
and interaction between them. Also responsibility considered
as a connector between feature and component for facilitating
transformation. It is depicted in Fig. 2.
Figure 4. Proposed approach in [4].
In [5] presented an approach in which CIM is represented
by secure business process in BPMN [6] and transform with
the help of QVT [7] rules, checklists, and refinement rules into
UML use case which represents PIM. It is depicted in Fig. 5.
Figure 2. Proposed approach in [2].
In [3] presented a possible solution for CIM modeling and
then transform it to PIM using analytic method of
transformation. In this paper, CIM level representing by DFD
that it is used for business process modeling. PIM level
represented by UML diagrams includes use case diagram,
activity diagram, sequence diagram, and domain model. It is
depicted in Fig. 3.
Figure 3. Proposed approach in [3].
In [4] proposed a disciplined approach for transformation
of CIM into PIM. In this paper, CIM includes business process
model and requirement model. First, business process modeled
using an activity diagram then activity diagram details for
specifying system requirement. PIM represented by class
diagram. For this from the model of requirement elements the
system components are created. Finally, a set of business
archetypes help to transform the system components to the
PIM layer in details. It is depicted in Fig. 4.
Figure 5. Proposed approach in [5].
3. Conceptual framework of CIM to PIM
transformation
We have used method of paper [8] for creating conceptual
framework and extracting Evaluation Criteria from it.
We design a conceptual framework to extract Evaluation
Criteria for comparing proposed approaches. The conceptual
framework is composed of a "Static Model" describing
common concepts and their relationships, "Taxonomies" of
CIM and PIM for classifying, the types of CIM and PIM, and
"Transformation Approach". The comparison and evaluation
criteria are derived from this framework.
3.1 Static model
In this section, we formalize the concepts of CIM, PIM,
transformation, and traceability by means of a meta-model.
The meta-model is presented using the class diagram in Fig. 6.
792

Hamid Reza Sharifi et al ,Int.J.Computer Technology & Applications,Vol 3 (2), 791-796
ISSN:2229-6093
behaviour of the system by showing interactions among
objects, etc. As shown in Fig. 8, we classify the different
presentations of the structure aspect used in the proposed
approaches into three types: UML Structure Diagrams,
Domain Diagram and Architecture Concept. The behaviour
aspect is classified into one type: UML Behaviour Diagrams.
Figure 6. Static model.
As shown in Fig. 6, a CIM can be transformed into a PIM.
A CIM and PIM are composed of one or more Model Element.
Traceability Link is established between the model elements
of the CIM and the PIM.
3.2 Taxonomies
The taxonomy of CIM and PIM classify different their
aspects.
3.2.1 Taxonomy of CIM
In [9] is mentioned that Regarding CIM, there are two
basic streams of suggestions of what is to be represented by
CIM. One of the streams suggests that the business model is to
be represented at this level [10]. Another stream points to CIM
as a model, which represents system requirements [11]. Some
researchers position both models representing business
knowledge and system requirements at the CIM level [12].
In Fig. 7, we are depicted taxonomy of CIM.
Figure 8.Taxonomy of PIM.
3.3 Transformation approach
CIM is further transformed into a PIM .As shown in Fig. 9
transformation in proposed approaches is classified into Rule-
Based transformation. Rule-Based transformation uses a set of
predefined Transformation Rules.
A transformation rule is a description of how one or more
constructs in the source language can be transformed into one
or more constructs in the target language [14].
Figure 7. Taxonomy of CIM.
As shown in Fig. 7, we classify the different
representations of the Business Process Model aspect used in
the proposed approaches into three types: UML [13] Diagram
(Activity Diagram), Data Flow Diagram (DFD) and Business
Process Modeling Notation (BPMN). The Requirement Model
aspect is classified into two types: Use Case Model and
Feature Model.
3.2.2 Taxonomy of PIM
A complete PIM should describe two aspects of a system:
Structure and Behaviour. The structure (or static) aspect
emphasizes the static structure of the system using classes,
objects, attributes, operations, relationships, etc., while the
behaviour (or dynamic) aspect emphasizes the dynamic
Figure 9. Transformation approach.
4. Criteria-Based evaluation of CIM to PIM
transformation approaches
In this section, first we derive evaluation criteria from
conceptual framework in section 3 and then compare proposed
approaches.
4.1 Evaluation criteria
The evaluation criteria are used to evaluate each proposed
approach in terms of Coverage of CIM, Completeness of PIM,
Automation, Evaluation Methods, Traceability Support, and
793

Hamid Reza Sharifi et al ,Int.J.Computer Technology & Applications,Vol 3 (2), 791-796
ISSN:2229-6093
Tool Support. As shown in Fig. 10, for example, the
evaluation criterion “Completeness of PIM” is derived from
the Taxonomy of PIM. Notice that the Static Model of the
conceptual framework is not directly related to any evaluation
criteria; however, the taxonomies are all dependent on it. Last,
note that the criterion “Evaluation Methods” in proposed
approaches reports, and it is not therefore traced back to the
conceptual framework.
If a paper applies traceability methods between CIM and
PIM elements we label “Traceability Support” "yes" and
otherwise "no".
If there is a tool for an approach we label “Tool Support”
"yes", otherwise "no".
4.2 Discussion
In Table 1 we summarize the comparison of Evaluation
Criteria for proposed approaches in tabular form. In this
subsection we describe results from Table 1.
Figure 10. Mapping between the conceptual
framework and evaluation criteria.
4.1.1 Evaluation Criterion for CIM
We evaluate the CIM with respect to “Coverage of CIM”.
The evaluation criterion “Coverage of CIM” is derived from
the Taxonomy of CIM. “Coverage of CIM” indicates which
aspects of CIM covers by proposed approaches.
4.1.2 Evaluation Criterion for PIM
We need to evaluate the generated PIM with respect to its
completeness. If a generated PIM both describes the system
structure (e.g., class diagram) and behaviour (e.g., sequence
diagram, state machines, or activity diagrams), then we label
the generated PIM as complete. If a generated PIM describes
only one of these two aspects of a system (i.e., either the
structure or behaviour), then we label it as incomplete.
4.1.3 Evaluation Criterion for Transformation
We evaluate the transformation approaches proposed in the
approaches with respect to their automation. The automation
criterion evaluates whether a transformation is automatic,
semi-automatic, or manual. A transformation approach is
automatic if it has been fully implemented. A transformation is
semi-automatic if there is algorithm and transformation rule
for proposed approach and user interventions are required.
Last, some approaches are entirely manual.
4.1.4 Other Evaluation Criteria
We evaluate the approaches with respect to “Evaluation
Methods”, “Traceability Support”, and “Tool Support”.
If a paper applies other evaluation methods (besides case
studies) to evaluate its approach, we mention it in “Evaluation
Method” criterion.
CIM
We can see from Table 1, Two out of four approaches can
cover business model (e.g., DFD). One approach can cover
requirement model (e.g., Feature model). One approach is
capable of covering CIM, including both aspects.
PIM
We can see from Table 1, that two out of four approaches
can derive only structural model elements (e.g., objects,
classes, associations, components) from CIM. One approach
can generate behavioural features of a system (e.g., sequence
diagrams, state machines, and/or activity diagrams). One
approach is capable of generating PIM including both
structural and behavioural aspects of a system, which is
characterized as complete according to our evaluation criteria;
Transformation–automation
As shown in Table 1, two out of four approaches
require user intervention to semi-automatically perform
the transformation; two approaches require manual
transformations.
Evaluation Methods
All of the approaches have their transformation
approaches evaluated. Case study has been performed to
evaluate all the approaches.
Traceability support
All of the approaches do not propose any method for
traceability support.
Tool support
All of the approaches do not develop any tool for
supporting their approaches.
5. Conclusion
In this paper, we have surveyed several CIM to PIM
transformation approaches and have evaluated them using a set
of evaluation criteria. For this, we have designed a conceptual
framework and have derived Evaluation Criteria for
comparing proposed approaches. The results can be used for
794

Hamid Reza Sharifi et al ,Int.J.Computer Technology & Applications,Vol 3 (2), 791-796
ISSN:2229-6093
assessing and comparing CIM to PIM transformation
approaches.
In the future work we want to propose a semi-automatic
approach for CIM to PIM transformation that generates a
complete PIM.
References
[1] J. Miller, J. Mukerji, “MDA guide version 1.0.1”, OMG, 2003.
[2] Wei. Zhang, Hong. Mei, Haiyan. Zhao, and Jie. Yang,
“Transformation from CIM to PIM: A Feature-Oriented
Component-Based approach”, in MoDELS, Montego Bay,
Jamaica, 2005.
[3] Martin. Kardoš, Matilda. Drozdová , “Analytical method of CIM
to PIM transformation in Model Driven Architecture (MDA)”,
JOURNAL OF INFORMATION AND ORGANIZATIONAL
SCIENCES, vol. 34, 2010, pp. 89-99.
[4] Samir. Kherraf, Éric. Lefebvre, and Witold. Suryn ,
“Transformation from CIM to PIM using patterns and
Archetypes”, 2008, pp. 338-346.
[5] Alfonso. Rodríguez, Eduardo. Fernández-Medina, and Mario.
Piattini, “Towards CIM to PIM transformation: from Secure
Business Processes defined in BPMN to Use-Cases”, Business
Process Management, vol. 4714, 2007, pp. 408-415.
[6] “BPMN: Business Process Modeling Notation specification”,
OMG, 2006.
[7] "QVT: Meta Object Facility (MOF) 2.0
Query/View/Transformation", OMG, 2008.
[8] T Yue, L Briand, and Y Labiche, "A Systematic Review of
Transformation Approaches between User Requirements and
Analysis Models", Requirements Engineering (Springer), 2011,
pp. 75-99.
[9] Marite. Kirikova, Anita. Finke, and Janis. Grundspenki, “What
is CIM: an information system perspective”, Advances in
Databases and Information Systems, vol. 5968, 2010, pp. 169-
176.
[10] Steffen. Becker, “Coupled model transformations”, in
Proceedings of the 7th international workshop on Software and
performance. Princeton, New Jersey, USA: ACM, 2008, pp.
103-114.
[11] Stefan. Biffl, Richard. Mordinyi, and Alexander. Schatten, “A
Model-Driven Architecture approach using explicit stakeholder
quality requirement models for building dependable information
systems”, in Proceedings of the 5th International Workshop on
Software Quality. Washington, DC, USA: IEEE Computer
Society, 2007, pp. 6-.
[12] Janis. Osis, Erika. Asnina, and Andrejs. Grave, “Computation
Independent Modeling within the MDA”, in IEEE International
Conference on Software-Science, Technology & Engineering,
Los Alamitos, 2007, pp. 22-34.
[13] OMG, “UML 2.1.1 superstructure specification”, 2007.
[14] A. Kleppe, J. Warmer, and W. Bast, MDA explained: the model
driven architecture: practice and promise, Addison Wesley,
2003.
795

Hamid Reza Sharifi et al ,Int.J.Computer Technology & Applications,Vol 3 (2), 791-796
ISSN:2229-6093
Table 1. Criteria based evaluation
Primary
Study
CIM
Rep. 1
Business
Model
CIM
Coverage
Requirement
Model
PIM
Rep.
PIM
Completeness
Structural
Behavioral
Automation
Traceability
Support
Evaluation
Methods
Tool
Support
Wei et al. [2]
Feature
Model
Yes
Software
Architecture Yes Semi-automatic No
Case Study
No
Kardoš et al.
[3]
DFD
Yes
Use Case
Diagram
Sequence
Diagram
Activity
Diagram
Domain
Model
Yes Yes manual No
Case Study
No
Kherraf et al.
[4]
Activity
Diagram
Use
Case
Diagram
Yes
Yes
Class
Diagram
Yes manual No
Case Study
No
Rodríguez et
al. [5]
BPMN
Yes
Use Case
Diagram Yes Semi-automatic No
Case Study
No
1 Representation
796