A Meta-model for Dynamic Workflow Evolution
Berraouna Abdelkader
1
, Amirat Abdelkrim
1
and Meslati Djamel
2
1
University Mohammed Chérif Messaadia, 41000, Souk Ahras, Algeria
2
University Badji Mokhtar-Annaba, BP 12, 23000, Annaba, Algeria
Keywords: Meta-model, Workflow, Dynamic Evolution, Components.
Abstract Workflow systems such as they are today do not provide sufficient flexibility needed to deal with certain
situations that may arise during the execution, especially in critical environments such as medical systems
and banking systems. In this context, this paper treat the adaptive workflow based on Meta-Model approach
and components-based software approach. These provide granularity, flexibility and reliability needed for
effective and safe development of the workflow. The work is both in terms of the development process and
in terms of the support platform, which should ensure the adaptability and propose a generic component-
oriented framework to build, run and develop dynamic workflow templates. Our research objectives are: First,
the proposal of a proper workflow development approach that takes into account the evolution of these
systems appearance. Then the realization of a framework environment that takes advantage of the component
approach and the meta-model approach and its features.
1 INTRODUCTION
Companies in different fields have always shown
interest for information technology, and this interest is
growing. Thus was born an undeniable dependence of
computers in this area and it has become impossible
for a company to survive without the support of
adequate information technology.
One of these technologies is the "workflow" is
used to describe the automatic task execution cycle
time, validation methods, and to provide each player
the information necessary for the execution its task.
The workflow application fields are diverse as
financial services, telecommunications, the public
sector and even creative areas such as the service
industry and the media. With changing needs in this
wide range of areas, we have identified a gap in
research in the adaptation and flexibility of workflow.
(Kolar et al., 2013)
Given the level of competitiveness and the
environment in which businesses currently it is vital
to be able to change workflow systems either statically
or dynamically. However, workflow systems such as
they are today do not provide sufficient flexibility
needed to deal with certain situations that may arise
during the execution, especially in critical
environments such as health care systems and banking
systems. In this context, this paper deals with the
adaptive workflow based on both approaches, the
Meta-Model approach and the approach based
software components (Reichert and Dadam, 1997)
(Weske,1997) (López-Fernández et al,2015)(Tiwari
and Chakraborty ,2015). These provide granularity,
flexibility and reliability needed for effective and safe
development of the workflow. The work is both in
terms of the development process and in terms of the
support platform, which should ensure the adaptability
and propose a generic component-oriented framework
to build, run and develop dynamic workflow
templates. Our research objectives are:
First, we propose a workflow development
approach that takes into account the evolution of these
systems appearance. Then the realization of a
framework environment that takes advantage of the
component approach and the meta-model approach
and its characteristics. The motivations of this work
are many. In fact, dynamic evolve workflow systems
will help increase competitive businesses, reduce
costs has required the adaptation and evolution of
these systems and possibly improve the working
environment through active support, reliable and
adapted to change as needed. Thus it will be possible
for these companies to deal with different types of
change and crisis situations by adapting to new
contexts.
Abdelkader, B., Abdelkrim, A. and Djamel, M.
A Meta-model for Dynamic Workflow Evolution.
In Proceedings of the 1st International Conference on Complex Information Systems (COMPLEXIS 2016), pages 59-65
ISBN: 978-989-758-181-6
Copyright
c
2016 by SCITEPRESS – Science and Technology Publications, Lda. All rights reserved
59
The rest of the paper is organized as follows:
section 2 presents some of basic concepts and
terminology of the dynamic workflow. Section 3
outlines the similar work and highlights the
motivation of this work. Section 4 presents a summary
and discussion of existing approaches. Section 5
describes an introduction to the proposed framework
and will end with a conclusion and some perspectives.
2 CONCEPTS AND
TERMINOLOGY OF DYNAMIC
WORKFLOW
Before presenting the existing approaches for the
modeling of adaptive workflow we first present
several related concepts.
-Workflow: A workflow is a software tool dedicated
to the management tool processes. This tool define,
manages and executes processes by implementing
programs whose execution order is pre-defined in a
computer representation of the logic of these
procedures ( Workflow Management Coalition
Glossary 1996).
- WorkFlow Engine: is a software service that
provides all or part of the execution environment of a
Workflow( Workflow Management Coalition –
Glossary 1996).
-Workflow Management System: is a complete
system that is used to define, manage and execute
processes by implementing programs whose
execution order is pre-defined in a computer
representation of the logic of these procedures
(Workflow Management Coalition – Glossary 1996).
- Workflow Service: is a software service that
consists of one or more workflow engines of the same
type used to define, manage and execute workflow
process (Workflow Management Coalition – Glossary
1996).
- Dynamism: is characteristic related to evolution of
the process model or to changes in the business
environment or business process reengineering
efforts. These changes are made in the time of
conception and involve a significant problem (Sadiq
et al., 2001) .In this characteristic the proposed
solution should reduce the number of situations needs
of change.
-Adaptability: Is the process capability to deal with
exceptional cases and a non-standard behavior. This
can be partially solved in time by adjusting the design
(Adapter accurately) the structure of the process
model (Kolar et al., 2013).
- Flexibility: is a characteristic of the process model
related to loose or partially defined model structures
specified in design-time. The full specification of the
process is finalized at the time of execution and may
be different for each process instance. Flexibility is the
main objective of this work (Kolar et al,2013).
3 EXISTING APPROACHES FOR
MODELING DYNAMIC
WORKFLOW
An example of a commercial workflow that allows
some dynamic adaptations is Tibco iProcess Suite
(version 10.5), which provides an orchestration
component "Orchestrator" which allows dynamic
allocation of sub-process variations in the execution
time. It requires a construction called dynamic event
to be modeled explicitly contains a number of sub-
processes listed as an "Array". When execution
reaches the dynamic event node, it will execute the
members of the «Array " on predefined conditional
basis, the table must be set statically before the process
is instantiated there is no possibility of modifications
execution.
Another trading system, CSR (Version 5.4) allows
an ad hoc manual execution adaptation such as
(reorganization of tasks, cancellation, repetition,
delay, termination).
The ADEPT2 prototype (Reichert et al ,2005)
supports the change process at runtime (ie, add, delete
and modify the sequence of tasks) to the model level
(dynamic evolution) and at Instance (ad hoc changes).
These modifications are made to a traditional
homogeneous model and must be achieved through
manual intervention of an administrator, summarized
in a high-level interaction. The system is also
responsible for the "jump forward" and "backward
jumps" in the process instances, but only by the
authorized person.
The system YAWL (Yet Another Workflow
Language) provides support for flexibility and
dynamic exception handling through the concept of
worklets, extensible directory of autonomous sub-
processes with associated selection rules this approach
provides direct a dynamic change and development
processes without intervention outside the system or
stop.
(Minoret al ,2014) Present a Case-based reasoning
(CBR) approach for automated workflow adaptation
through the reuse of experience.
This approach helps the expert in the execution of
these adaptations by an automated method. The
COMPLEXIS 2016 - 1st International Conference on Complex Information Systems
60
method uses the recorded case workflow adaptation in
adaptations of the past. The recorded changes can be
automatically transferred to a new workflow that is in
a similar situation changes, it uses CFCN the
workflow modeling language that was developed by
the University of Trier.
(Kolar et al., 2013) Proposes an approach based on
the constraints in the design process. This combined
approach is for business situations, including a rigid
structure to the process model causes a flexible
limitation.
It offers design process model that is suitable for
the modeling of ad hoc processes within BPMS
systems. The pattern may be used to define a process
structure in the declarative constraint- based manner.
In addition, it presents an application of the approach
in a real project, a BPMproject end to end for an
insurance company.
In (Tayade and Chavan, 2012), the authors have
taken an approach to identify the change as an ongoing
process, and integration of the processes of change in
the workflow process itself which is predefined and
can predefined and can be fixed in execution time.
The framework presented introduced the concept of a
flexible workflow process comprising a core and one
or more types or roles of flexibility in this kernel; the
following Figure shows briefly the basic structure of a
workflow modeling language:
Figure 1: The structure of a workflow modeling language.
The system includes four components as shown in
the following Figure :
Figure 2: FWMS Architecture.
Work Node Manager
This is the basic component of the work process all
working nodes are stored in the nodes of the database
with the appropriate classification.
Work Process Manager
Manages the workflow templates and instances,
including creating, editing, deleting and querying
workflow.
User's Task Manager
This component manages users there's including the
verification of the user connections and
disconnections.
Work Process Executor
This component represents the system core, includes
eight main components - runtime components, Service
task, Repository Service, Management service,
Service of history, identity service, DAOFactory core.
The meta-model approach is used (explicitly or
implicitly) to identify structures and types of
workflow model of constituent components. A set of
primitives is typically defined in which change
operations can be performed a certain model or
bodies. Examples of this approach include
ADEPTflex (Reichert and Dadam ,1997) WASA
(Weske ,1997) and WIDE (Casati et al,1997). The
syntactic correctness and modifications can be made
by users.
A meta-model describes the structure of models and
allows reasoning on models like the Knowledge of
first Level.
Some object-oriented languages such as Java,
describe the same way the classes, meta-classes and
objects. In software engineering, it is with the concept
of UML meta-model was developed. The meta-model
describes the models: classes, objects, attributes,
relationships and self-described, also means the model
of the model. It can be defined as the representation of
a particular view on model.
In the Components-based approach the modeling
goal is the construction of more complex models,
through reusing consistent and functional models.
Each of them is a process component. When modeling
components we should consider the following issues:
Building an information base on reusable
components
Determine the function of each component in
the new process model
Dynamic composition of heterogeneous
components.
A Meta-model for Dynamic Workflow Evolution
61
4 SUMMARY AND DISCUSSION
OF EXISTING APPROACHES
We studied several approaches proposed in the
literature about workflow flexibility, we can mention
some features on the dynamic evolution of workflow:
The Factors of Evolution in Workflow
Management Systems (Hanet al., 1998)
- Change in the environment.
- Technical progress.
Ad-hoc Diversions During Workflow Execution
(Hanet al ,1998)
- Refinement of dynamic workflow model.
- User intervention.
- An unpredictable event.
- System failure.
Classification of Adaptive Workflow Adaptations
Levels (Hanet al ,1998)
- Adaptation to the context level domain.
- Adaptation in process Level
- Adaptation at Resource Level
- Adaptation to the infrastructure level.
Generally, there are two types of changes:
1 ad-hoc changes
2 Evolutionary change.
Ad-hoc changes are handled case by case. To
provide customer-specific solutions or handle rare
events, the process is suitable for one or more cases.
Evolutionary change is often the result of
reengineering efforts. The process is modified to
improve customer responsiveness and improve
efficiency.
Dynamic workflow refers to the extension of the
workflow process so that when the change occurs, the
process model may be modified or completed, rather
than build a new model. The change can be considered
(which concern only the current instance) or may need
to be applied either temporarily or permanently. The
adaptation is done on two levels.
- level process model
- level instances
In instances running must be managed when the
process model from which it was instantiated it (the
instance should be canceled, restarted) so the
workflow must provide support for changes to the
process model job but incremental and ad hoc changes
can be arranged as required.
From several criteria identified for comparison, we
present a subset of them: the nature of flexibility,
formalism and flexibility techniques. According to
(
Workflow Reference Model, 1995) WFMS can be
characterized as a support for three areas functional:
- The Build-time Functions involved in the
definition and possibly modeling, workflow
process and its constituent activities.
- Run-time Functions involved in the control of
workflow processes in an operational environment
and scheduling to handle various activities as part
of each process.
- Run-time Interactions with human users and its
implementation tools available for the various
stages of activity.
The study of literature allowed us to distinguish two
main types of flexibility based on the ability to cope
with changes that may be incorporated into the process
definitions when building or running time.
- Posterior Flexibility by adapting the process
definition or one of its instances during execution.
The approaches that offer this kind of flexibility
are based on prescriptive modeling formalisms. It
could be considered that the definition of processes
and the resulting are not really flexible, but rather to
adapt or evolve. In fact, these approaches can not
anticipate the ability to change at time of construction,
it is the most common situation found in the literature.
Prescriptive modeling formalisms are well suited
to specify business processes that require a high
degree of control and prediction and for which the
need for change is an exception such as production
workflow.
- Flexibility Priori or by Selection based on
modeling formalisms that can offer the ability to cope
with environmental change without changing process
definitions. This means that this capacity should be
integrated into the process definitions during
construction. The process definition should be
specified in a sufficiently flexible to adapt with the
environment without breaking execution.
Consequently, the workflow service should be able to
run the specification "incomplete" process definitions.
5 PROPOSED FRAMEWORK
We focused specifically on workflow modeling
approaches based components and meta-model, the
control and data integration are implemented from a
model based components. Then, the process is seen as
a set of inter operative components. Data
dependencies are reduced by encapsulating the
components behind well-defined interfaces. This
allows dynamically changing workflow systems, and
proposing a generic component-oriented framework
to build, run and develop dynamic workflow
templates.
COMPLEXIS 2016 - 1st International Conference on Complex Information Systems
62
The idea is to see the workflow as an open process
and its implementation is not finished.
We used the OPC
(Open Process Component)
(Sadiq et al., 2001) mechanism to adopt
implementation, this mechanism allows:
Dynamic Evolution of Software Developed.
Flexibility
The development of a basic component
Provides a meta-layer model for software
Enables configuration management
Provides meta-process support
Figure 3: OPC Framework.
The OPC (Gary et al., 1997) provides an environment
for production processes based components.
In this approach, the process models are constructed
as a set of components that interact through interfaces
well defined. It also offers an infrastructure for
interoperability and reuse components of
heterogeneous processes, built on three levels of
abstraction.
1. The meta-model level that identifies the
fundamental entities of the process and their
relationships.
2. The finite-state machine use for
representation of
the behavior of the process entities
3. An object-oriented framework related to the
process representation formalism and its area of
application.
It provides support for dynamic processes, a
component is defined with the model and the
instantiation information and allows the component
manage its own Evolution, a Component has generic
information (in the framework) and more
specialization (adaptation) in the instantiation.
The first step in the development of our framework
is to design a meta-model (meta-model level) that can
capture all the features and functionality of the various
workflows in the trade business domain
The proposed meta-model is shown in Figure 4In this
meta-model the class Activity is the central part. A
process contains several FlowNode that can be
Avtivity Group,Simple Activity or Control Node. The
class FlowNode has a reflexive association called seq
to define the sequence relationship.
An activity have tow author relation with Resource
class and Actor class.
The class Activity has an attribute state assigned to
the enumeration type State that indicates in which
execution state the activity is currently, this
enumeration has five stats (waiting, running , done
,skipped and failed).
To ensure the control of different activity we have the
class Control that contains three type of control Join,
Fork and Sequence.
The activities in which decisions have to be made are
modeled explicitly with the class Decision. The
conditions for selecting the right path is modeled by
class Guard which is connected to the subsequent
activities. At runtime the user can select the desired
Guard with the select() operation, the Decision
activity can either be an XOR that selects exactly one
path or an OR that can select more than one or AND
that select the both.
Figure 4: Workflow meta-model.
A Meta-model for Dynamic Workflow Evolution
63
6 EXAMPLE PROCESS
MODELING WITH
In this section the metamodel of the previous section
is used to create workflow models. In the first
subsection the workflow is described in natural
language. In next subsection the described workflow
is modeled in GMF tool with the metamodel elements
introduced in this model Resources and data are not
considered in the model because they are not part of
the metamodel. Only activities and control flow
aspects are modeled.
6.1 Process Description
The following example describes globally the
workflow for making an order.
We identify three actors in this system, namely the
customer, accounting service and delivery service.
The action takes place as follows:
The client begins by make an order, giving rise to the
development of a Quote by the Accounting Service
.making quotation activity itself is decomposed into
two sub-avtivity check availability of the ordered
products and calculate the order price.
Quote is the name of a class of system, data stream
are typed the customer can then change its order (back
to the initial activity to order), to cancel (go to final
state) or validate the Quote.
If the costumer validate Quote two actions can be
undertaken in parallel:
- the preparation of the order by the delivery
service.
- invoice processing and payment.
The Accounting Service create invoices and send it
to customer (the object invoices transmitted in the
state sent) this makes the payment and sends the
invoice (which is present in the regulated state).
When the order is ready and confirmed the
payment of the client (by appointment
synchronization of these two activities) the order is
delivered to the client and the treatment is finished.
6.2 Workflow in GMF Tool
The proposed modeling of workflow is shown in
figure 5
Figure 5: Workflow model.
7 CONCLUSIONS
Workflow management systems must provide high
support flexibility with easy and simple ways. This
change can manifest in different forms in the
workflow process, in this article we presented
different approaches for the modeling and the
development of flexible and adaptive workflow with
a comparison and synthesis.
We also presented research aims to develop a
generic component-oriented framework to ensure the
dynamic development prospects as workflow.
In further work we aim to develop and implement this
framework to ensure the dynamic development
workflow
REFERENCES
Kolar, J., Dockal, L., & Pitner, T. 2013” A Dynamic
Approach to Process Design: A Pattern for Extending
the Flexibility of Process Models. In The Practice of
Enterprise Modeling “(pp. 176-190). Springer Berlin
Heidelberg.
Reichert, M., Rinderle S., Kreher, U., Dadam P. 2005
Adaptive Process Management with ADEPT2”.
International Conference on Data Engineering, Demo
Session, Tokyo, Japan, p. 1113±1114.
COMPLEXIS 2016 - 1st International Conference on Complex Information Systems
64
Adams, M., ter Hofstede, A., Edmond, D., van der Aalst,
W., 2006 Worklets: A Service-Oriented Implementation
of Dynamic Flexibility in Workflows”. International
Conference on Cooperative Information Systems,
Montpellier, France, pp.450-456.
Minor, M., Bergmann, R., & Görg, S., 2014“Case-based
adaptation of workflows”. Information Systems, 40,
142-152.
Tayade, S. M., & Chavan, V. 2012 “Roles of Flexibility in
Workflow Engineering”. International Journal of
Modeling and Optimization, 2(5), 644-647.
Han, Y., Sheth, A., & Bussler, C, 1998.” A taxonomy of
adaptive workflow management”. In Workshop of the
1998 ACM Conference on Computer Supported
Cooperative Work. pp 150-156.
Workflow Reference Model 1995, WfMC-TC-1003 v1.1,
V 1.1.
Sadiq, S., Sadiq, W., Orlowska, M. 2001 “Pockets of
flexibility in workflow specification”. In: Kunii, H.S.,
Jajodia, S., Sølvberg, A. (eds.) ER 2001. LNCS, vol.
2224, pp. 513–526. Springer, Heidelberg.
Reichert M., Dadam P.1997 "A Framework for Dynamic
Changes in Workflow Management Systems”
Proceedings of 8th International Workshop on
Database and Expert Systems Applications, France,
September, 42-48.
Weske M. 1998 "Flexible Modeling and Execution of
Workflow Activities", Submitted to 31st Hawaii
International Conference on System Sciences (HICSS-
31)pp 201-207.
Casati F., Ceri S., Pernici B., Pozzi G. 1996 "Workflow
Evolution", Proceedings of 15th International
Conference on Conceptual Modeling (ER’96),
Germany, October, pp.438-455.
Workflow Management Coalition – Glossary 1996.
http://www.aiai.ed.ac.uk/project/wfmc/ARCHIVE
/DOCS/glossary/glossary.html.
Workflow Reference Model ,1995, WfMC-TC-1003 v1.1,
V 1. January.
Gary, K., Lindquist, T., and Koehnemann, H. 1997
Compo-nent-based Process Modeling”. Technical
Report TR-97-022, Computer Science Department,
Arizona StateUniversity. May.
López-Fernández, J. J., Cuadrado, J. S., Guerra, E., & de
Lara, J. 2015. Example-driven meta-model
development. Software & Systems Modeling, 14(4),
1323-1347.
Tiwari, A., & Chakraborty, P. S. 2015. Software
Component Quality Characteristics Model for
Component Based Software Engineering. In:
Computational Intelligence & Communication
Technology (CICT), 2015 IEEE International
Conference on pp. 47-51.
A Meta-model for Dynamic Workflow Evolution
65