An i* based Approach to Support Strategic Decision in Virtual
Enterprise
Clement Cherrey and Yan Liu
School of Software Engineering, Tongji University, 1239 Siping Road, Shanghai, China
Keywords: Virtual Enterprise, i*Framework, Goal-Oriented, Strategic Agent.
Abstract: Virtual Enterprise is becoming remarkable as the business environment for enterprise is more and more
dynamic. Existing approaches for describing and designing structure for those collaborations adopt Goal
oriented approach but tend to focus on functional goal for the system supplemented by an examination of
quality constraints. In trying to get a complete knowledge of the way to structure and organize the Virtual
Enterprise it is also necessary to have an understanding of the specific expectation, intent, and concerns of
the different partner involved in the Virtual Enterprise design process. This paper presents an Agent and
Goal oriented approach, iStarVE, taking i* framework as a basis, to model and evaluate the different views
of actors on the Virtual Enterprise and their potential impact on the design of its internal structure.
1 INTRODUCTION
Virtual Enterprise (VE) mostly results from an
evolution of the business context (Jähn et al., 2005).
When attempting to design the collaboration
structure between the stakeholders involved in the
VE, it can happen that many alternatives need to be
considered, each with different implications for the
parties that may have an interest in the sub-process.
Identify, evaluate, and classify alternatives regarding
to the impact on agent’s self-goals has only been
partially study yet.
The present paper intends to build a Goal-
Oriented approach, iStarVE, based on existing
frameworks designed for Requirement Engineering
and Computing system, in order to deal with these
problems.
Designing such approach faces multiple
challenges. The fact that this issue is situated on the
intersection of different areas of research makes the
design of consistent approach complex. Indeed we
need to consider research contribution about Virtual
Enterprise, Breeding Environment, Agent and Goal
Oriented approaches, in order to get a strategic view
on the agents involved in the process.
Therefore, iStarVE first contributes to extend
and apply goal-oriented approach to VE. It includes
extensions for i* framework to model VE
problematic. A second contribution is to support
strategic decisions about VE structure using agents’
point of views.
In the first part we introduce background notions
and identify specific issues for VE, and justify why a
goal-oriented approach is suitable. In the second part
we describe how it is possible to use i* framework
to model VE to facilitate strategic decision. The
third part illustrates iStarVE with a case study.
Finally we conclude, describing outcomes of
iStarVE, its limits and relevant further works.
2 RELATED WORKS
2.1 Virtual Enterprise (VE)
For the approach presented here, we take the
definition of VE formulate by (Thompson, 2008) as
a base. He pointed out that a VE is a voluntary and
dynamic community of SMEs that undertake to
work together for a set period of time and to
collectively seek opportunities to participate in
collaborative projects of mutual business interest.
Our approach focuses on issues appearing during
the early creation phase of Virtual Enterprise. Indeed
choosing architecture for the VE is equivalent to
allocate tasks, which must be performed, to the
different stakeholders (Kaisler et al., 2005). Along
that process we identified two main reasons of
dilemma to select the entity in charge of a task:
305
Cherrey C. and Liu Y..
An i* based Approach to Support Strategic Decision in Virtual Enterprise.
DOI: 10.5220/0004924003050310
In Proceedings of the 3rd International Conference on Operations Research and Enterprise Systems (ICORES-2014), pages 305-310
ISBN: 978-989-758-017-8
Copyright
c
2014 SCITEPRESS (Science and Technology Publications, Lda.)
• Several actors have the ability to perform the
task
• None of the partner already possess resources
and skills to complete the task
In the first one, several actors possess necessary
resources and expertise to be in charge of the task.
The second one can be seen as the opposite problem,
none of the stakeholder has adequate knowledge or
resources to implement the feature required. In both
scenarios the task allocation can not be performed
only considering actor abilities.
We also identified a third issue specific to
Extended Enterprise, which is a VE centralised with
a dominant actor. In this special kind of VE the
central actor may face difficulties to determine a
suitable status for his partners, and select a benefit
way of collaboration for both parties. This problem
is different from the tasks allocation issues, because
the tasks are clearly dispatched. This last point
equals to define boundary for the dominant actor.
The main idea developed by our approach is to
use and extend methods and theories coming from
Requirement Engineering to model those problems
at first, and then support strategic decision and
negotiation.
iStarVE must be implemented by a single entity
in order to improve its understanding and
positioning in the VE. The main idea, for the
company using iStarVE, is to assess the acceptance
of its partners for each alternative, in order to refine
its own strategy as one of the actors taking part in
the VE.
2.2 Goal Oriented Approach and Agent
Modelling
Goal Oriented Analysis occupied a central position
in Requirement Engineering Area. It can address the
issues we just underlined. In fact, a goal-oriented
approach start from high-level goals, then refines
them to obtain sub-goals and finally the tasks which
must be completed by the entities in the system (Van
Lamsweerde and Letier, 2004).
Such refinement technique guarantees a perfect
traceability for the tasks identified. Therefore
including such approach to treat our problematic,
first allows identifying precisely the origin of the
task which has created the allocation dilemma
considered. Indeed this only requires to rollback the
refinement process until we obtain a suitable level of
detail for the goals description.
Moreover most Goal Oriented Approaches are
combined with analysis and models for system’s
agent. The definition of agent depends of the
framework considered. Nevertheless it is remarkable
that several frameworks already provided models for
Strategic Agent. Such description required to
distinguish the system’s goals from the internal
goals of the agents.
Moreover in order to get a rational representation
of systems, another distinction has been made
between main goal and soft-goal. Soft-goal are
mostly used to constraint the achievement criteria of
the main-goals, and bring consideration about
quality, efficiency, cost and so on. Soft-goals
represent a popular way to put strategic
considerations in the model. Nevertheless every soft-
goal does not refer to a unique level of strategy. Our
research leads to distinguish four level of view
which can be model by four types of soft-goals for
VE’s actors.
Figure 1: Categories of sot-goals.
Naturally we have selected i* framework which
focuses on strategic aspect for modelling systems.
The following section details possibilities to extend
it to model VE problematic.
3 i* FRAMEWORK OVERVIEW
3.1 Adoption of the i* Framework
The i* framework is originally designed to be
applied in computing area to support processes
involving computing treatment. It was created to
identify, evaluate and select process alternative (Yu,
1993).
Although it is possible to represent refinement
process, using graphic elements defined by the
framework, this is not the core of i*. Indeed i*
framework focuses on the “why” representation for
the process using strategic and intentional
representation of actor.
More precisely Yu defined two models to
represents the process. First, the Strategic
Dependency model (SD) describes dependency
relationships among organisational actors (Figure 2).
Considering VE problematic, it can provides a
synthetic representation for alternatives we needed
ICORES2014-InternationalConferenceonOperationsResearchandEnterpriseSystems
306
to compare.
Figure 2: Strategic Dependency diagram.
The i* framework also includes diagram to
represent the roles of the different actor and their
positions (Figure 3). An idea to build iStarVE
approach is to use i*’s role to aggregate the tasks for
each agent in a few number of role, in order to
simplify the representation of the problem.
Figure 3: Role-Position Diagram.
Another key idea used by iStarVE is that one
alternative can exactly correspond to one position
for each agent. Then, such “exhaustive” position can
be used to compare all alternatives using only one
diagram, and therefore provide a synthetic and
complete representation of the problem.
This idea has been combined with the second
model of i*, Strategic Rational model (RS), to
express rational attitudes representing internal goals,
tasks and resources for the agents. Indeed using
such “opened up” representation of agent, allow
focusing on internal strategic soft-goal (Yu, 2011). If
we combine it with “exhaustive” positions we
already introduced, we are able to create diagram to
evaluate the different alternative relatively to agents’
expectations (Figure 4 Case Study).
3.2 Modelling Process
Because iStarVE aims to support strategic decision
during VE creation phase, it must be implemented
by system architect and business expert to guarantee
a consistent analysis.
The approach includes several steps
progressively leading to assess the views of other
partners on the issue.
Figure 4: Approach’s Steps.
3.3 Strategic Decision Supporting
As defined by (Mariotti, 1996), enterprise
partnerships suggest a relationship between
companies and people who share common goals,
strive to achieve them together and do so in a spirit
of cooperation, collaboration and fairness.
Beyond that definition there are questions about
how to evaluate partnership, and do it relevantly in
the context of VE. Indeed a VE is unique as its
structure is dynamic, and can include asymmetric
partners with different sizes and level of implication
(Gajda, 2004). Therefore it is necessary to focus on
these two aspects for the evaluation.
The SD diagram (Figure2) can support an
evaluation about the interdependence of the partner
in front of business process. Studies about VE
suggest that more the business process is mature and
includes interdependent stakeholders, more the
durability of a VE is important.
Moreover an analysis of diagram representing
actors’ positions and their coverage in terms of role,
permit to assess the implication of actor in each
solution evaluated, and thus, draw conclusions about
the longevity of the partnership.
Finally, Actor-Position diagram is a key feature
provided by iStarVE, the acceptance of each
potential solution by the agents may be deduced
from it. But to obtain relevant evaluation we have to
combine it with assess about the importance of each
role and uncertainty about it.
The following formula is just an idea of how to
calculate the acceptance but we did not apply it in
iStarVE because it requires quantifying the impact
of the alternative on the soft-goal, as well as the
importance of the soft-goal.
(1)
According to (1), the acceptation of the n alternative
(A
n
) is a sum of the impact of the position in this
Ani*basedApproachtoSupportStrategicDecisioninVirtualEnterprise
307
alternative on the soft-goal (x
i
) weighted by the
importance of the soft-goal (y
i
). We can normalise
this result dividing by the some of the weights for
the soft-goals of the agent.
In addition, it is interesting to notice an i* social
modelling extension created by Alistair Sucliffe (see
Eric et al., 2011, p.669-691),
including diagrams to
model dominance among the partners. This
embodies another strategic view on VE actors,
which may be combined with iStarVE.
4 CASE STUDY
4.1 Context and Hypothesis
The Case Study is inspired from a Requirement
Analysis accomplished with KAOS to determine
tasks and feature for a potential collaboration. It had
been submitted to students by an enterprise which
did not have contact other partners at that time, so
the study initially represented a preliminary analysis
to determine the feasibility about a new VE.
For this case study we use this idea of VE as a
starting point. But we modified it, and formulated
the following hypothesis, for the case study:
• The minimum number of partner to launch the
VE have been identified and contacted
• A business process analysis to determine the
high level structure of the VE have been done
• The choice of the problematic correspond to the
implementation of a new feature which can
hardly been carry on by one actor alone
• The soft-goals identified refer to classic concern
of Actors in VE, and have been extract from
articles which provide a analysis of Agent in VE,
as it is detail in the following section
4.2 VE Considered and Problematic
A branch of a Car Rental firm and a local Gas
Station Company located in the same area decided to
create a new VE allowing to rent car near the city
centre. This service targets local customers without
car, who need cars for exceptional activities. In fact
the Gas Station Company became aware that they
disposed of vacancy area on small parking lot in
their station downtown. Therefore they want to
create a new car rental service. They plan to
implement an online reservation system and use
their staff members in the gas stations to deliver the
key to the client, and get back the vehicle after use.
So the main goal of the VE is to provide vehicle
with homogenous quality for rent in strategic place
inside or near the city centre, and to allow the client
to order and provide the entire guarantee needed
online.
The core of the new VE is an online reservation
system to support most of the interaction with the
client. The partners have to discuss who will be in
charge of the implementation.
4.3 Agent and Solutions Modelling
4.3.1 Alternatives Modelling
First it is necessary to model each alternative. So
first we use synthetic Strategy Dependency diagrams
to model those solutions. In our case Study they
corresponded to the following strategy of
implementation for the website:
Figure 5: Alternative 1.
Figure 6: Alternative 2.
These diagrams allow identifying the set of roles
potentially play by each agent in each alternative.
• Alternative 1: The Gas Station Company is only
a resource agent for the problem considered, and
the Car Rental company occupies a dominant
position in the organization
• Alternative 2: Both Companies are in charge of a
ICORES2014-InternationalConferenceonOperationsResearchandEnterpriseSystems
308
part of the online system.
• Alternative 3 (Figure 2): Both play the role of
resources agents, and another organization
implements the web portal.
4.3.2 Agent Expectations Modelling
The second part of the modelling process focuses on
the agents soft-goals. The study at the origin of the
case study did not include consideration about the
expectation and concerns of the agents, so the soft-
goals listed for the agents involved came from other
study on VE and partnership in general, such as
(Zota and Fratila, 2013) and (Martinez et al., 2001).
They correspond to classical agent concerns in inter-
enterprise relationship, and we choose to only
attribute less than three soft-goals by actor in order
to facilitate the comprehension.
Moreover the quantification for the importance
of the role and their uncertainty has been design in
order to be realistic. This is acceptable as the present
case study is presented first to illustrate a potential
application of the approach, and as the deductions
about the VE considered do not have any value for
itself.
Respectively the soft goal identified represents
the following concerns for the agents:
• Exclusivity (Car Rental Co.): occupy a dominant
position in the VE to block association of the
Gas Station with other partners
• Improvement of its reservation system (Car
Rental Co.)
• High profitability (Car Rental Co.)
• Visibility (Gas Station Co.)
• Investment in the VE limited (Gas Station Co.)
Finally, we can model these soft-goals using an
actor-position diagram (Figure 7).
Figure 7: Diagram Position/Soft-goal final.
4.4 Case Study Outcomes
Applying our approach to the present case study
highlights potential issues related to the choices
made to construct the approach. Indeed the creation
of the case study itself, based on hypothesis to
define concerns for the stakeholders underline that a
central question is the definition of a strategic view
for the different stakeholders.
Furthermore, it appears that several kinds of
diagrams are not relevant all the time, especially
diagram describing roles and position. We can skip
it because in that case it is not necessary to
aggregate the tasks in role as the number of tasks
considerate is already limited.
In addition is seems complex to define explicit
name for the position of each actor as the position
must summarize the complete set of task of an actor
for one alternative.
We can also notice that the initial task
decomposition is critical as it conditions the creation
and the modelling of the alternatives. For example
here we identify two principal features for an online
reservation, which is the treatment of the customer
requests and the verification of client information.
But another tasks’ decomposition is maybe possible
and lead to considerate different possibilities.
Nevertheless this decomposition is supposed already
done, by business experts, when iStarVE is
implemented.
5 CONCLUSIONS AND FUTURE
WORK
5.1 The Outcomes of iStarVE
First iStarVE includes factual representations and do
not include any assessment about personal goals of
the different parties. This could be used to support
discussion between the partners.
Nevertheless the diagram representing internal
soft-goals for each agent and the evaluation extract
from the diagram must not be shared, as they permit
to identified weakness of other agents, and thus,
refine negotiation strategy. The actor-position
diagram is a part of it. Sharing assessments about
relative importance of goals for other actors can lead
to a switch in their strategies.
As it is the case in RE the analysis of the actors
leading to identify soft-goals is the more critical part
in the process. It is directly linked to the knowledge
of the entity which conducts the study about its
partners.
Ani*basedApproachtoSupportStrategicDecisioninVirtualEnterprise
309
Moreover the quantification of the importance is
based on assessment, and has only been considered
to prioritize the different soft-goals. The early
identification of those limits pushes us to the
introduction of uncertainty factor, but all those
parameters are not yet well formalized.
5.2 Further Work
A first step for further works would be to define
uncertainty and importance indicators, with detailed
description of each level, and then refine the formula
assessing acceptance.
A second point would be to establish more
precise characterizations for agents and VE’s
structures. Definitely, typing architecture, with
precise description of their specificities, would allow
reusing analysis of previous cases, and build little by
little a bank of classic sub-structure includes in VE.
Also, typing actors according to their position in the
VE would provide a mean to check all along the
design process if the way the structure is evolving
match the initial idea for role distribution among the
partners.
Finally we believe that it could be relevant to put
forward the idea of evaluation focused on Agent,
and consider each stakeholder as a user of the VE,
who tries to use it to complete his own goals, and so,
conduct strictly user-oriented evaluation (Mourouzis
et al., 2006).
ACKNOWLEDGEMENTS
This work was financially supported by the Science
and Technology Commission of Shanghai
Municipality (12dz1507000) and the Fundamental
Research Funds for the Central Universities
(2100219030).
REFERENCES
Jähn, H., Fischer, M., & Zimmermann, M., 2005. An
approach for the ascertainment of profit shares for
network participants. In Collaborative Networks and
Their Breeding Environments (pp. 257-264). Springer
Kaisler, S. H., Armour, F., & Valivullah, M., 2005.
Enterprise architecting: Critical problems. In System
Sciences, 2005. HICSS'05. Proceedings of the 38th
Annual Hawaii International Conference on (pp.
224b-224b). IEEE.
Gajda, R., 2004. Utilizing collaboration theory to evaluate
strategic alliances. American Journal of Evaluation,
25(1), 65-77.
Van Lamsweerde, A., & Letier, E., 2004. From object
orientation to goal orientation: A paradigm shift for
requirements engineering. In Radical Innovations of
Software and Systems Engineering in the Future (pp.
325-340). Springer Berlin Heidelberg
Yu, E., 2011. Modelling strategic relationships for process
reengineering. Social Modelling for Requirements
Engineering, 11, 2011.
Thompson, K. (2008). The Networked Enterprise.
Competing for the Future Through Virtual Enterprise
Networks. Meghan Kiffer, Tampa.
Mariotti, J. L., 1996. The Power of Partnerships: The Next
Step beyond TQM, Reengineering, and Lean
Production. Blackwell Business.
Yu, E. S., 1993. Modelling organizations for information
systems requirements engineering. In Requirements
Engineering, 1993, Proceedings of IEEE International
Symposium on (pp. 34-41). IEEE.
Martinez, M. T., Fouletier, P., Park, K. H., & Favrel, J.,
2001. Virtual enterprise–organisation, evolution and
control. International Journal of Production
Economics, 74(1), 225-238.
Eric, S. K., Giorgini, P., Maiden, N., & Mylopoulos, J.
(Eds.)., 2011. Social modelling for requirements
engineering. (pp 669-691).Mit Press.
Daniel Zota, R., & Fratila, A., 2013. Toward the selection
of an enterprise architecture model for a cloud
environment. In Roedunet International Conference
(RoEduNet), 2013 11th (pp. 1-6). IEEE.
Mourouzis, A., Antona, M., Boutsakis, E., & Stephanidis,
C., 2006. A user-orientation evaluation framework:
Assessing accessibility throughout the user experience
lifecycle. In Computers Helping People with Special
Needs (pp. 421-428). Springer Berlin Heidelberg.
ICORES2014-InternationalConferenceonOperationsResearchandEnterpriseSystems
310
