Model Transformation Recommendations for

Service-Oriented Architectures

Dmitri Valeri Panfilenko

, Katsiaryna Hrom

, Brian Elvesæter

and Einar Landre

DFKI GmbH, Saarbrücken, Germany

T-Systems International GmbH, Saarbrücken, Germany

SINTEF, Oslo, Norway

Statoil, Stavanger, Norway

Keywords: SOA, MDA, Model Transformations, Recommendations, CIM, PIM, M2M, BPMN, SoaML.

Abstract: Services for service-oriented architectures can be modelled in different ways, including well-known existing

OMG standard SoaML and an IBM methodology SOMA. Involving domain expert stakeholders in the

system specification and development process plays an important role and is often inevitably combined with

model transformations between different levels of abstraction. Recommendations for those supporting users

during the modelling process along the chosen methodology can aid the development performance and thus

reduce model transformation efforts. This paper shows how bidirectional model transformations between

OMG MDA’s CIM and PIM levels can be enhanced through recommendations and which obstacles on the

way to a comprehensive framework for model-driven development are still to overcome.

1 INTRODUCTION

Nowadays it is impossible to think away from the

software systems in the enterprise workaday life

because they are facilitating a supportive role for the

enterprise management in the market competition.

The need for software solutions in the business

practice faces increasing complexity of these

solutions and permanently rising requirements for

performance, reliability and shorter technology

cycles. In addition, the changing requirements and

pressure for cost reduction count as well. The new

concepts and techniques for software engineering as

an actual problem are not to be undervalued.

The core principles of SOA – loose coupling,

atomicity and consistency (Erl, 2007) – are aspects

present in every system developed in this way, but

transferring the information from the higher levels of

abstraction as domain knowledge to the technical

system specification has always been an issue

researched upon (Kleppe, 2003); (Petrasch, 2006).

An MDA approach provides a basic scheme of

separation of concerns of domain experts and

technical specialists, giving them an opportunity of

working together and not having to get much

involved with concerns of other domains or

abstraction levels (research projects SHAPE:

www.shape-project.eu, REMICS: www.remics.eu).

The problem of transferring the information

between different abstraction levels is the

background of the present paper. We try to amend

the existing approaches for the model

transformations by engineering recommendations

for the end-users facilitating more automation in the

information propagation.

The paper flow proceeds as follows: section 2

describes the existing work in the areas of model

transformation and recommendation engineering.

Section 3 describes model transformation

application for engineering of service-oriented

systems. Section 4 presents a case study elaborated

in SHAPE project and evaluation of it. We discuss

strengths and weaknesses as well as implications

from the existing approaches to the model

transformation and recommendation engineering in

section 5. The section 6 provides conclusions and

sketches the future work plans in these areas.

2 RELATED WORK

The implementation of the MDA approach on the

CIM and PIM level is a field of interest for industry

and science. According to the literature survey of

248

Panﬁlenko D., Hrom K., Elvesæter B. and Landre E..

Model Transformation Recommendations for Service-Oriented Architectures.

DOI: 10.5220/0004445602480256

In Proceedings of the 15th International Conference on Enterprise Information Systems (ICEIS-2013), pages 248-256

ISBN: 978-989-8565-60-0

 2013 SCITEPRESS (Science and Technology Publications, Lda.)

existent MDA and MDA-based implementations the

most of them ignore the CIM level and therefore the

related model transformations (ATLAS, 2005);

(Jeary, 2008); (Petrasch, 2006). At the same time

CIMs and PIMs present business and analysis level

of software systems correspondingly and the models

of this level have a big impact on the software

development effort and the quality of software.

Quite recently many authors have mentioned that

automatic transformations from PIM to CIM are not

possible (Kahl, 2005); (Kleppe, 2003); (Koch,

2006). One of the last published practical researches

about CIM-to-PIM transformations showed that the

examined transformations could not be completely

automated “because it is needed both to add

information concerning the context and to make

decisions” (Lemrabet, 2010).

In the last few years there have been done many

research attempts aiming to apply MDA (or MDD)

paradigms to business processes and services. One

of the most similar works to our research is that one

from (Delgado, 2010) where MINERVA’s tool

support for service oriented development from

business processes is presented. This tool support

includes QVT automated model transformations

from BPMN to SoaML. De Castro et al. describe the

transformations between CIM and PIM models for

model-driven process (De Castro, 2011). In their

previous work (Rodríguez, 2007); (Rodríguez, 2008)

have proposed a CIM to PIM transformation

composed of QVT rules. (Touzi, 2009) designed

collaborative SOA according to MDA principles. As

a part of work the BPMN-to-UML transformations

were defined and implemented using ATL. In

(Roser, 2006) the model transformations from

business level ARIS models to platform-independent

ICT system models have been described.

The potential of the intelligent support and

recommendations for the enhancement of process

modelling has been recently recognized (Smirnov,

2009). However, the few implemented modelling

support tools have very restricted functionalities.

Existing work in the area of intelligent support and

recommendations for the process modelling can be

split up in 2 categories: 1) methods for process

analysis incl. model checking; 2) auto-completion of

models (e.g. auto-completion of model elements

identifiers, pattern recognition, auto-completion

based on the methods of information retrieval).The

most of implemented modelling support

functionalities are used for conformity checking of

models and based on auto completion of model

elements identifiers (e.g. process editor Signavio

(Fellmann, 2010)).

Another popular type of modelling support

presents the model syntax checking. In the

aforementioned Signavio the syntax checking has

been implemented for EPC, BPMN 1.2 /2.0. With

model checking different other model parameters

like performance, reliability, security, maintenance,

portability of models can be checked (Rech, 2009).

The different information criteria can be also

aggregated and used for the building of key figures

like the similarity of models or the coverage rate of

domain by one model (Fellmann, 2010).

More advanced modelling support systems

contain functionality for validation and verification

of process models as well as business process

optimization. There are a lot of examples in the

literature for the analysis of processes designed with

BPMN by means of Petri net, which present the

formal description of processes (e.g. (Desel, 2000)).

So, (Raedts, 2007) has used BPMN-to-Petri net

transformation for the validation and verification of

industrial models. (Dijkman, 2007) has described an

approach for the semantic correctness checking of

process models with the help of Petri net.

Further type of auto completion functionality for

modelling support systems is based on pattern

recognition. Nowadays there are empirical

researches dealing with the occurrence of patterns in

the process models, e.g. in (Lau, 2009); (Smirnov,

2009); (Thom, 2009). There are also tools which are

able to recognize patterns automatically and to

propose the variants of completion like for example

ProWAP (Thom, 2008).

The comprehensive concept of recommendation

based editor for the business process modelling has

been presented in (Hornung, 2008) and implemented

by (Koschmider, 2010). In contrast to the described

approaches this concept relates to the usual

recommendation systems from information retrieval.

This idea in the process modelling is that the

designed model is compared with known models and

the proposal for the extension or enhancement of

this model is generated.

The described approaches have mostly indirect

relation to service model transformation and

recommendations for them, but all of them can be

adapted to this field. Our proposal described in this

work is to bring all together: model-driven approach,

service orientation and intelligent (cross-level)

modelling support concept.

ModelTransformationRecommendationsforService-OrientedArchitectures

249

3 MODEL TRANSFORMATIONS

FOR SERVICE MODELS

3.1 CIM2PIM Transformation

for SOA

Below we shortly sketch what kinds of models are

defined on which abstraction level in the scope of

this paper and how we intend to transform the

models between those abstraction levels.

On the highest abstraction level CIM, business

models encompass business rules, processes,

services and other issues such as contracts involving

humans and organizations to achieve business goals.

These conform to the metamodel of CIMFlex

prototype. The middle layer – PIM – contains the

results of the proposal as transformation engines,

extended SOA models, the standardized UPMS

(SoaML).

The transformation engine should also support

visualization of services in business models (provide

transformation support both ways, top-down and

bottom-up between CIM and PIM levels). This

should provide a basis for service development for

different application domains that covers the life-

cycle of services from business goals and

requirements to platform specific models for various

platforms (Hahn, 2009).

Thus, the scope of this paper targets CIM

modelling and bidirectional consistent CIM-to-PIM

transformations.

3.2 Recommendations for Service

Model Transformations

The background for the idea of intelligent modelling

support system for transformations consists of 2

issues: 1) runtime of transformation for large models

and 2) necessity for model verification and

validation in time of their building. With the

intelligent modelling support the user can get the

information about how the changes made within one

modelling level affect the process model within

another modelling level, where the relation between

modelling levels is transformation (see Figure 1).

Change analysis can be used for round-trip

engineering, as repeating of MDA cycles with model

changes on different abstraction levels results in the

entanglement of the models for the user (Delgado,

2010).

The modelling support can cover a big spectrum

of aspects – from business level with process cost

and durability analysis as well as the appropriateness

of model changes up to technical level with the

analysis of data consistency in the database or the

changes of the methods in java class. A large

analysis coverage can give the user much important

information about models, but it is very risky to

provide the good level of usability of such analysis

tool at the same time (Fellmann, 2010). In our case

study one can see that, using a relatively small set of

analysis criteria, the analysis report can be very

large.

Figure 1: Implementation framework.

We examined all types of modelling support and

considering also the specifics of technical

implementation defined how different types of

support can be implemented for the service model

transformation in the MDA conceptually and on the

technical level. The tool architecture is presented on

the Figure 2. Transformation has been implemented

in ATL, as a BPMN model editor CIMFlex and as a

SoaML editor Modelio (www.modeliosoft.com) has

been used correspondingly, recommendations have

been implemented partly in ATL and in Java. We

have chosen Eclipse and the compatible plug-ins for

the implementation because the framework can be

simply extended and provide the certain flexibility

for developers (open source); moreover there are

many components for Eclipse which can be used for

the MDA.

Figure 2: The idea of modelling cross-level support.

To restrict the recommendation coverage we

have chosen for the implementation elements of 3

ICEIS2013-15thInternationalConferenceonEnterpriseInformationSystems

250

types of modelling support: 1) model checking

(compliance with specification), 2) analysis of

models by means of aggregated key figures and 3)

analysis of model/process changes based on pattern

recognition. The cross-level analysis can be done as

follows: the basis input CIM is transformed in PIM

and saved. Whenever the changes in CIM are done

and they should be analysed on the PIM level, the

new PIM is created, compared with the basis variant

and the information about changes is return to the

user.

The first idea here was to separate the

comparison and analysis logic especially as there

exist a lot of mature tools and approaches for the

comparison of models. For our implementation

framework the EMFT Compare looks like the most

suitable tool for the model comparison. It provides

generic support for any kind of metamodel in order

to compare and merge models in the EMF

framework. But because of many disadvantages like

complex handling of profiles using EMFT Compare,

no direct support of comparison of CIMFlex models,

etc. we left this idea for the future work. The other

existing tools for the model comparison operate on

the low technical level (XML) and cannot be used

for the implementation of our idea because we

cannot provide 1:1 mapping of model elements

repeating transformation between levels. The

comparison of models within the MDA should be

done on the level of model elements. To provide the

high level of usability for our modelling support tool

the most important details of models for the different

types of analysis were chosen as the base for model

analysis (usually it depends on the end user

demand). For the model analysis by means of

different key figures the model transformation has

already been used. Many implemented examples of

transformation of type model-to-measure can be

found in the ATL transformations Zoo or for

example in (Vépa, 2005). Such analysis is a time

consuming solution due to the metamodel for the

key figures and model-to-measure transformation.

All known approaches can be applied only for

existent models and thus not very efficient for the in

time modelling cross-level support. The approach

which we have developed has a technical

background and it is based on ATL Queries. Query

is a special type of routines to compute a primitive

value, such as a string or an integer, from source

models. The advantage of the usage of queries for

the cross-level model analysis is obvious: the query-

logic is separated from transformation logic; the

comparison of models is done on the model

elements level; it is not complicated to describe the

cross-level model changes only using the source

models. The last statement is based on the property

of queries to use the same elements of code as the

transformation; therefore, this approach minimises

the development effort and error rate of the analysis

results. The remaining issue in this case is the

interpretation of the analysis data and providing it

the user in intelligent way.

4 MODEL TRANSFORMATIONS

APPLICATION

4.1 Instantiation for BPMN2SoaML

To illustrate our concept we developed bidirectional

BPMN-to-SoaML transformation in ATL and

showed how the recommendations for the intelligent

modelling support for transformation can be

implemented using the selected tools.

The BPMN and SoaML models belong to the

different abstraction levels, have different objects

and key aspects. Furthermore, the complexity level

of BPMN and SoaML is not identical. The study of

(Recker, 2009) has shown that the complexity of full

BPMN specification is higher as the complexity of

full UML specification. Hence, different restrictions

are inevitable for BPMN-to-SoaML transformations.

The description of mapping rules for BPMN-

elements into the corresponding UML/ SoaML

elements presented in (Hahn, 2010) as well as in

SHAPE project documentation. We enhanced and

specified the transformation rules more precisely

and adjusted them to the newer BPMN and

SoaML/UML specifications.

To illustrate our transformation and following

transformation support based on recommendations

we chose the business case “Voyage travel agency”

(presented at ECMFA’10 as a tutorial “Service

Modeling with SoaML”: http://www.ecmfa-

2010.org/index.php/tutorials),which includes 4 lanes

representing the client, the partner, the payment

centre and the travel agency itself. The client gets

information about the current travel offer and

chooses a travel. The client should provide personal

information to the travel agency and confirm the

order. Then the agency checks the client solvency

with the collaboration of the payment centre and if

the check is successful makes the reservations of

travel for the client.

The specific feature of our transformation is that

the CIM model built in BPMN contains 3 different

views: the structural, behavioural and data view.

This separation means that our BPMN-to-SoaML

ModelTransformationRecommendationsforService-OrientedArchitectures

251

transformation contains 3 different transformation

types in one and the result will be consolidated in

one resulting target model. The challenge of such

transformation was how to integrate all views in one

target model. The mapping rules from (Elvesæter,

Panfilenko, 2010); (Hahn, 2009); (Hahn, 2010)

include no integration aspects. Therefore, we have

defined view integration rules, which are collected

in the Table 1.

The target SoaML model contains also 3 views,

but indirectly in contrast to the source BPMN model.

These views can be seen in Modelio in 3 different

diagrams: structural model as the SoaML participant

diagram, behavioural model as the UML activity

diagram and the data model as UML class diagram.

Table 1: The integration rules for SoaML models.

View

SoaML/UML element superordinate

element

Struc-

tural

Model

ServiceArchitecture Model

Participant Model

ServiceContract Model

ServicePoint Participant

RequestPoint Participant

Beha-

vioural

Activity ServiceArchitecture

Activity Partition Activity

other behavioural

elements

Activity Partition

Data

All data elements Participant

All of the services from the BPMN model appear

in the SoaML model. The graphical correspondence

of the models is not evident on the level of services

because of incomplete compatibility of used tools,

but the correspondence of structure of models on the

internal XML level is satisfactory.

We don’t cite the transformation of BPMN

behavioural view into UML activity model in this

paper, because this transformation is trivial and was

described and implemented by the other authors (e.g.

(Lemrabet, 2010)). The only difference between our

and the other BPMN-to-SoaML transformations is

the mapping set and rules.

The second part of our transformations is the

inverse transformations. As mentioned above, a lot

of researchers are in doubt about the possibility of

PIM-to-CIM transformation. The challenge is that

whenever the business models are transformed into

service models, only the structural information is

transformed, but not dynamical. The bidirectionality

is only obtainable if the expression power of the

both languages is identical (Kleppe, 2003).

In the case of bidirectional transformations the

information is either lost (if the transformation

function is a surjective function) or added (if the

transformation function is an injective function)

after the transformation step; to define a bijective

model transformation function is almost impossible,

particularly for the model transformations used in

industry. There exists a method to solve this problem

by using a tracing-technique, but in general it is hard

solvable (Küster, 2004). In our research we have

determined that the loss of information happens

while changing the modelling level, but it can be

reduced to the minimum.

In our scenario there could be the following

information loss while transforming BPMN models

into SoaML models (CIM-to-PIM):

 the direction of associations from DataObjects to

Tasks in patterns Lane1→Lane2 und

Role1→Role2(table 7 in (Hahn, 2010) und 9 in

(Hahn, 2009) );

 the relationship between Roles and Tasks from

the pattern Role1→Role2;

 the type of Gateways;

 the source and target of Message Flows between

Pools (table 27 in (Hahn, 2009)).

The loss of information is the cause of incomplete

transformation definition, i.e. not every element

from the CIM has a corresponding element on the

PIM level. On the conceptual level we don’t need to

take into account all of the details defining the

transformation rules, but in implementation stage

these details can be important and must be included

in the set of transformation rules. In our work we

determined that using the appropriate traceability

mechanism it is possible to achieve the complete

restorable mapping.

We used the ATLAS transformation language,

which does not support the extern traceability

directly. One method of ensuring the traceability

support in ATL was provided by (Jouault, 2005).

The traceability information is considered as a

special model and is described by means of

traceability metamodel. This method was

implemented as a special program – TracerAdder. It

allows automating the extern traceability partially,

because the program works in a special refining

mode which is not fully supported by ATL yet.

Nevertheless, this method has a big potential

because the tracing and transformation logic are

independent from each other as well as no additional

language constructs or ATL engine modifications

are needed, therefore its implementation is only the

matter of time.

ICEIS2013-15thInternationalConferenceonEnterpriseInformationSystems

252

The similar technique can be implemented using

profiles. This method was developed for UML-to-

UML transformations (Vanhooff, 2005), but can be

also adjusted for the other transformations.

In our concrete case we detected also another

type of information loss which deals with the

concept of “3 views”-representation of models

described above. In BPMN models these views are

integrated, whereas in UML/SoaML models they are

completely independent. A lot of model elements

can be assigned to different views, and therefore

they are contained in the resulting SoaML models

twice or even threefold (e.g. DataObject

“ClientDefinition” appears in all three views), i.e.

the target models of BPMN-to-SoaML

transformation contain much more objects as the

source models. This should be taken into account by

the inverse SoaML-to-BPMN transformations.

Because there is no relationship between such

elements having “the same origin” in the SoaML, we

need to think about the enhanced tracing technique

which should be applied to all elements of models.

To sum up, the problem of the information loss

by the bi-directional CIM-to-PIM transformations is

solvable with one of the described traceability

methods, but it still requires the adjustment of the

whole concept and architecture.

4.2 SHAPE Project Case Study

Evaluation method in this paper refers to the

evaluation conducted within the scope of the

SHAPE project (Elvesæter, 2010). Evaluation

should be split into two parts due to the nature of

software development in companies like Saarstahl

(www.saarstahl.com), having their own IT-

department responsible for business process

modelling. Firstly, a system architect asserts whether

the solution proposed works and if it helps.

Secondly, management has to assert the proposed

solution fulfils the company guidelines for ROI. The

evaluation conducted in SHAPE project is a subject

to time and resource constraint and thus restricts the

complete evaluation volume to performance

measures definition, which in turn provides basis for

answering the questions whether the solution works

and whether it helps the development.

The full description of the Saarstahl use case

realisation can be seen in (Elvesæter, 2010), whereas

the scope of this paper is on the CIM level

modelling and CIM-to-PIM transformations with

recommendations for the latter. As for these the

evaluation report states that on the CIM level

CIMFlex allows modelling of business processes in

an abstract manner. As for CIM-to-PIM

transformations, these generate a skeleton SoaML

model that needs further manual refinements. The

recommendations for the model transformations are

the research point, especially given attention in this

paper. Using these techniques Saarstahl was able to

model its business processes on different abstraction

levels, whereas the produced source code and

system reflect the business models on the top

abstraction level, which in turn proves that the

proposed technology works. Another part of the use

case was to wrap the existing legacy systems behind

the web services, which also has been successfully

evaluated. The modelling as described and the

wrapping of the legacy systems lead to increased

interoperability of the complete IT landscape at

Saarstahl, which proves the consistency of the

SHAPE technology. The complete list of

performance measures that provide basis for

answering the question about ROI can be seen in

(Elvesæter, 2010).

5 DISCUSSION AND

IMPLICATIONS

Of course, the described work is a trial version of

BPMN-to-SoaML transformation to evaluate our

concept of intelligent cross-level modelling support

and review the gaps in it. The described

transformations are implemented with the pre-

defined tools and big restrictions to be solved in the

future:

1) The set of mapping rules does not cover all

elements of metamodels which are the basis of the

used modelling editors. The consequence is that the

source models should be manually checked and

restricted to include only the elements described in

transformation rules. It causes the additional

difficulties in testing and validation of

transformations particularly for the models with a lot

of elements.

2) The mapping rules are usually defined on the

conceptual level and do not contain a specific

technical information. This requires extending the

set of transformation rules with additional ones,

which complicates the understanding of the

transformations. We defined additional rules (see

Table 1) to achieve the unique structural

representation of PIMs in our transformation,

because the initial set of mapping rules did not

includ the integration aspect of different view in

SoaML models.

ModelTransformationRecommendationsforService-OrientedArchitectures

253

3) The transformation rules of type pattern-to-

element require a lot of implementation effort. We

used the linear search approach to identify the

patterns and tested our transformation on the models

containing not more as 120 elements. Probably it can

be necessary in the future to think of a special

optimisation algorithm for larger models.

4) Further problem with patterns consists in the

interpretation of patterns combination (Elvesæter,

Panfilenko, 2010), (Hahn, 2009), (Hahn, 2010). We

detected only some combinations which require

more precise description, but it can be identified

more. So it is necessary to include the interpretation

of patterns combinations in the mapping rules or

define the patterns more exactly.

5) The implemented model transformation and

generation of the recommendations for the

transformations are only partly automated due to the

existing tools not achieving such level of maturity

yet. In our case the testing tool, the validation and

verification environments are missing. It is possible,

though, to find more appropriate technologies to

automate the desired concept.

6) The implemented recommendations cover only

some aspects of possible intelligent cross-level

modelling support. It is possible to extend the

number of recommendations by including other

types of model analysis or extending the existent

recommendations for the bigger set of model

elements.

7) The transformations and recommendations were

implemented with big restrictions, which are the

cause of compatibility problems between different

tools we used in our research. These problems are

solvable by means of different techniques indeed,

but make the transformations very complicated to

understand, to modify and to reuse.

6 CONCLUSIONS AND FUTURE

WORK

MDA is a powerful concept which can be used in

many fields like process analysis, process

optimisation, model validation, model driven testing,

etc. However, there are a lot of problems (e.g. model

transformation) on the way of the MDA

implementation mentioned in the literature and

verified in our work as well.

The technology for the implementation of model

transformations should be chosen very carefully. All

known transformation languages have such level of

maturity at the moment that they cannot satisfy all

requirements for the model transformations in the

MDA. Further aspect, which makes the development

of model transformations time-consuming, is that

there is no integrated implementation framework for

the transformation, i.e. validation, testing and

debugging should be done manually. Moreover, all

of the existent model editors do not support the

automatic arrangement and alignment of the

graphical elements for the exported models.

Our concept, including the implementation

technique, allows for strict distinction between the

transformation and analysis logics, but they use the

same functions at the same time, which guarantees

the synchronisation of model transformations and

the analysis of transformed models. This way does

not require any complicated modifications in other

tools. However, the recommendations coverage

should be still defined. Once all necessary

formalisms are defined, the model analysis can be

easily applied on different modelling levels as well

as in the cross-level field.

The unique architecture for the transformation

and cross-level recommendations in the MDA

should still be elaborated. The concept of model

validation should be implemented; a big advantage

would be an appropriate traceability mechanism for

the transformation as well as an automated test

environment. The smooth interaction of all

necessary components still requires a lot of work on

the both conceptual and technical levels.

ACKNOWLEDGEMENTS

This research was co-funded by the European Union

in the frame of the SHAPE FP7 project (ICT- 2007-

216408). The authors would like to express their

acknowledgments to SHAPE colleagues.

REFERENCES

ATLAS Group (INRIA and LINA) 2005: “ATL

Transformation Examples. The KM3 to Metric ATL

transformation”, Version 0.1.

De Castro V., Marcos E., Vara J. M., 2011: “Applying

CIM-to-PIM model transformations for the service-

oriented development of information systems,” in

Information and Software Technology, Volume 53,

Issue 1, pp 87-105.

Delgado A., García-Rodríguez de Guzmán I., Ruiz F.,

Piattini M. 2010: “Tool support for Service Oriented

development from Business Processes,” in MOSE2010

in conjunction with the TOOLS2010, Málaga, Spain.

ICEIS2013-15thInternationalConferenceonEnterpriseInformationSystems

254

Desel, J.; Erwin, T. 2000: “Modeling, Simulation and

Analysis of Business Processes,” in: van der Aalst, W.

et al. (eds.): Business Process Management, LNCS

1806. Berlin Heidelberg: Springer, pp. 129-141.

Dijkman, R.M.; Dumas, M.; Ouyang, C. 2007: “Formal

Semantics and Analysis of BPMN Process Models

using Petri Nets,” Technical Report 7115, Queensland

University of Technology, Brisbane.

http://eprints.qut.edu.au/archive/00007115/.

Elvesæter, B.; Landre, E.; Hahn, C.; Jacobi, S. 2010: Case

Study Execution and Validation”. SHAPE-Project,

Deliverable D1.4, version 1.0/

Elvesæter, B.; Panfilenko, D.; Jacobi, S.; Hahn, C. 2010:

“Aligning Business and IT Models in Service-

Oriented Architectures using BPMN and SoaML,” in:

MDI10, Bézivin, J.; Soley R.M.; Vallecillo A.(eds.),

ACM/IEEE MoDELS10, Oslo, Norway. New York:

ACM Press, pp. 61-68.

Thomas Erl. 2007. Soa: Principles of Service Design (First

ed.). Prentice Hall Press, NJ, USA.

Fellmann, M.; Zarvic, N.; Thomas, O. 2010:

„Unterstützung von Modellierungsleistungen durch

semantische Technologien,“ in:

Dienstleistungsmodellierung 2010. Interdisziplinäre

Konzepte und Anwendungsszenarien, Thomas, O.;

Nüttgens, M. (eds.). Heidelberg: Springer, pp. 70-80.

Hahn, C.; Panfilenko, D. 2009: “Model transformations

from business models to UPMSHA – Initial version”.

SHAPE-Projekt, Deliverable D5.2, Version 0.5.

01.12.2009. p. 7-10.

Hahn, C.; Panfilenko, D.; Fischer, K. 2010: A Model-

Driven Approach to Close the Gap between Business

Requirements and Agent-Based Execution.

Hornung, T., Koschmider, A., Lausen, G. 2008:

“Recommendation based process modelling support:

Method and user experience,” in: ER 2008. LNCS

5231, Li, Q., Spaccapietra, S., Yu, E., Olivé, A. (eds.).

Heidelberg: Springer.

Jeary, S.; Fouad, A.; Phalp, K. 2008. “Extending the

Model Driven Architecture with a preCIM level,” in

TOOLS EUROPE Workshop, 1st International

Workshop on Business Support and MDA (MDABIZ),

Zurich, p.2.

Jouault, F. 2005: Loosely Coupled Traceability for ATL.

http://www.sciences.univ-nantes.fr/lina/atl/.

Kahl, T.; Kupsch, F. 2005. „Transformation und Mapping

von Prozessmodellen in verteilten Umgebungen mit

der ereignisgesteuerten Prozesskette,“ in: EPK 2005 -

Geschäftsprozessmanagement mit Ereignisgesteuerten

Prozessketten - 4. Workshop der Gesellschaft für

Informatik e.V. (GI) - und Treffen ihres Arbeitkreises

„Geschäftsprozessmanagement mit

Ereignisgesteuerten Prozessketten“, Nüttgens, M.;

Rump, F. J. (eds.), Hamburg, p. 3.

Kleppe, A.; Warmer, J.; Bast, W. 2003.„MDA Explained,

The Model Driven Architecture: Practice and

Promise”, Addison-Wesley.

Koch, N. 2006. “Transformation Techniques in the Model-

Driven Development Process of UWE,” in: ICWE'06

Workshops. Palo Alto, CA.

Koschmider, A.; Oberweis, A. 2010: “Designing Business

Processes with a Recommendation-Based Editor,” in:

Handbook on Business Process Management, Band 1.

Introduction, Methods, and Information Systems,

vomBrocke, J.; Rosemann, M. (eds.). Berlin

Heidelberg: Springer, pp. 299-313.

Küster, J. M.; Sendall, S.; Wahler, M. 2004: “Comparing

two model transformation Approaches,” in: UML

2004, Workshop OCL and Model Driven Engineering,

Proceedings, Lisbon, Portugal.

Lau, J. M.; Iochpe, C.; Thom, L. et al. 2009: “Discovery

and Analysis of Activity Pattern Cooccurrences in

Business Process Models,” in: Proc. 11th

International Conference on Enterprise Information

Systems (ICEIS'09), Volume on Information Systems

Analysis and Specification, Milan, Italy, LNBIP 24,

Filipe, J; Cordeiro, J. (eds.). Berlin Heidelberg:

Springer, pp. 83-88.

Lemrabet, Y.;Touzi, J.; Clin, D.; Bigand, M.; Bourey, J.-

P., 2010. “Mapping of BPMN models into UML

models using SoaML profile,” in: MOSIM’10,

Hammamet, Tunisia.

Petrasch, R.; Meimberg, O. 2006. „Model

DrivenArchitecture. Eine praxisorientierte Einführung

in die MDA,“ Heidelberg: dpunkt.verlag.

Raedts, I.; Petkovic, M.; Usenko, Y. et al. 2007:

“Transformation of BPMN models for behaviour

analysis,” in: Proceedings of MSVVEIS, Augusto, J.;

Barjis, J.; Ultes-Nitsche, U. (eds.). INSTICC Press, pp.

126-137.

Rech, J.; Schmitt, M. 2009: “Embedding Information

about Defects, Decisions, Context, Quality, and

Traceability in CIM- and PIM-level Software

Models,” in: Information and Software Technology

(IST), p. 2.

Recker, J. C.; zurMuehlen, M.; Siau, K. et al. 2009:

“Measuring method complexity: UML versus

BPMN,” in: Proc. AMCIS09, San Francisco,

California. p. 7.

Rodríguez, A.; Fernández-Medina, E.; Piattini, M. 2007:

“Towards CIM to PIM Transformation: From Secure

Business Processes Defined in BPMN to Use-Cases,”

in: 5th Int. Conf. on BPM, LNCS 4714, pp. 408-415.

Rodríguez, A.; Fernández-Medina, E.; Piattini, M. 2008:

"CIM to PIM Transformation: A Reality,” in Research

and Practical Issues of Enterprise Information

Systems II IFIP International Federation for

Information Processing, Volume 255, pp. 1239-1249.

Roser,S., Bauer,B., Muller,J. 2006: “Model- and

Architecture-Driven Development in the Context of

Cross-Enterprise Business Process Engineering,” in:

SCC’06.

Smirnov, S.; Weidlich, M.; Mendling, J. 2009: “Action

Patterns in Business Process Models,” in: Service-

Oriented Computing. LNCS 5900, Baresi, L.; Chi, C.-

H.; Suzuki, J. (eds.). Heidelberg: Springer , pp. 115-

129.

Thom, L.; Reichert, M.; Chiao, C. M. et al. 2008:

“Applying Activity Patterns for Developing an

Intelligent Process Modeling Tool,” in: ICEIS'08,

ModelTransformationRecommendationsforService-OrientedArchitectures

255

Barcelona, Spain.

Thom, L. H.; Iochpe, C.; Reichert, M. et al. 2009:

“Activity patterns in process-aware information

systems: Basic concepts and empirical evidence,” in:

IJPIM, Vol. 4, No. 2, pp. 93-110.

Touzi J., Benaben F., Pingaud H., Lorré J.P. 2009: “A

model-driven approach for collaborative service-

oriented architecture design,” in: International Journal

of Production Economics, Volume 121, Issue 1, pp. 5-

20.

Vépa, E. 2005: “ATL Transformation Examples. UML2 to

Measure,” Version 0.1. http://www.eclipse.org/m2m/

atl/atlTransformations/.

Vanhooff, B.; Berbers, Y. 2005: “Supporting Modular

Transformation Units with Precise Transformation

Traceability Metadata,” in: ECMDA-TW 2005,

Proceedings. Nuremberg, p. 20.

ICEIS2013-15thInternationalConferenceonEnterpriseInformationSystems

256