Towards Outsource-ability Enabled BPMN

Mouna Rekik

1,3

, Khouloud Boukadi

1,3

and Hanene Ben-Abdallah

2,3

University of Sfax, Sfax, Tunisia

King Abdulaziz University, Jeddah, K.S.A.

Mir@cl Laboratory, Sfax, Tunisia

Keywords:

Business Process Outsourcing, Cloud Computing, BPMN Extension, Genetic Algorithm.

Abstract:

Business process outsourcing to the Cloud is increasingly being adopted as a strategy to save costs, improve

the business process performance, enhance the ﬂexibility in responding to costumers’ needs, etc. However, the

adoption of an outsourcing strategy faces several challenges like the enterprise data security, vendor-lock-in

and labor union. Weighing the pros and cons of outsourcing one’s business process is an arduous task. This

paper provides for assistance means: it extends the BPMN language to explicitly support the speciﬁcation of

outsourcing criteria, and it presents an automated approach to help decision makers identify those parts of

their business process that beneﬁt most from outsourcing to the Cloud.

1 INTRODUCTION

Face to the increasingly stringent business competi-

tion, small and medium size enterprises strive to ex-

cel in the marketplace by adopting different strategies

and solutions. Outsourcing their business processes

has been among the most widely adopted strategies

(Yang et al., 2007). Business Process Outsourcing

(BPO) is seen as a means to save costs, improve the

business process performance and make it more ﬂex-

ible, etc. These advantages along with others explain

the exponentially growing number of manufacturers

which are outsourcing substantial parts of their sup-

ply chain processes to outside contractors(Adesta and

Agusman, 2004).

In the recent years, the Cloud became the most

chosen outsourcing environment. Indeed, according

to the National Institute of standards and technology

(NIST) (Mell and Grance, 2011)), Cloud computing

is a model for enabling convenient, on-demand net-

work access to a shared pool of conﬁgurable com-

puting resources that can be rapidly provisioned and

released with minimal management effort or service

provider interaction. It is a pay-as-you-go model pro-

viding customers with the possibility of using its var-

ious offerings and pay only for the used resources.

When thinking about the suitability of BPO, busi-

ness experts must address several issues to properly

choose those parts of the business process to out-

source. This explains the fact that enterprises deci-

sion makers spend almost 80% of their time to de-

cide about the suitability to outsource theur business

processes (Yang et al., 2007). Indeed, BPO affects

and depends on several aspects of an enterprise: so-

cial, economic, legal, and IT (information technol-

ogy) aspects. To make a judicious outsourcing deci-

sion, business experts must have a clear speciﬁcation

of each of these aspects; we believe that such speci-

ﬁcation should be through a set of valued criteria in-

tegrated as much as possible with the business pro-

cess model. For example, it is easier to delimit the

social/economic/IT effects of an outsourcing decision

when the business expert has an explicit description

of the personnel in charge, costs and IT requirements

of each business activity. Currently available busi-

ness process modeling languages, and in particular

the standard BPMN ((OMG), 2011) ), do not provide

for most of these criteria; this is justiﬁed by the fact

that these languages’ primary concern is the business

modeling and not the outsourcing.

To overcome this expressive power shortage, this

paper proposes OutyBPMN, a lightweight BPMN ex-

tension for the speciﬁcation of outsource-ability char-

acteristics of business processes. Because our re-

search work deals with the Cloud as an outsourcing

environment, OutyBPMN extends BPMN with out-

sourcing concepts pertinent to the Cloud, e.g., cost

as cloud may be a pertinent solution to save over-

head cost, security to prevent outsourcing sensitive

data, etc. In a second contribution, this paper presents

Rekik M., Boukadi K. and Ben-Abdallah H..

Towards Outsource-ability Enabled BPMN.

DOI: 10.5220/0005513500050014

In Proceedings of the 10th International Conference on Software Engineering and Applications (ICSOFT-EA-2015), pages 5-14

ISBN: 978-989-758-114-4

 2015 SCITEPRESS (Science and Technology Publications, Lda.)

an automated approach that uses OutyBPMN to iden-

tify the business process parts that beneﬁt most from

outsourcing to the Cloud. The identiﬁcation is in

fact a multi-objective optimization problem that can

be resolved through evolutionary algorithms such as

genetic algorithms, neuroevolution, genetic program-

ming, etc (Deb and Kalyanmoy, 2001). The herein

presented approach applies the penalty based genetic

algorithm (Hu et al., 2005) to select the activities can-

didate to outsourcing according to the criteria speci-

ﬁed in OutyBPMN. Initially randomly selected activ-

ities are generated as potential solutions. According

to their quality, only best ones are kept from gener-

ation to another allowing thus to keep solutions that

best meet speciﬁed criteria.

The remainder of this paper is structured as fol-

lows. Section 2 enumerates the criteria pertinent to

business process outsourcing to the Cloud. Section 3

presents how OutyBPMN extends BPMN to provide

for the explicit speciﬁcation of these criteria. Section

4 presents the proposed solution design whose evalu-

ation is discussed in Section 5. Section 6 summarizes

the work status and highlights its extensions.

2 BUSINESS PROCESS

OUTSOURCING CRITERIA

BPO is the process of delegating one or more ITinten-

sive business processes to third parties that may per-

form the business process more effectively and efﬁ-

ciently. Deciding whether to outsource ones business

process is not a trivial task. To assist with the out-

sourcing decision making, (Yang et al., 2007) suppose

that a set of potential determinants have been gath-

ered from experts. Indeed, enterprises expect from the

outsourcing to save costs, focus on core competence

and gain ﬂexibility in management. Other dimen-

sions may urge enterprises to outsource their business

processes such as speeding the time to market and

improving service quality (Li and Meissner, 2009).

Despite these advantages, some drawbacks may pre-

vent enterprises from outsourcing their business pro-

cesses. BPO inconveniences are related to security

issues, loss of management control, and vendor’s ser-

vice quality. In fact, cost saving is not always guar-

anteed owing to an inadequate business case or the

inability to predict all business requirements. More-

over, security concerns, loss of control and vendor

lock-in are among risks that may prohibit enterprises

from outsourcing their business processes. More de-

tailed risks are exposed in the research elaborated by

(Gewald and Rouse, 2012). A judicious outsourcing

decision needs to ﬁx a set of criteria that can help de-

cision maker in weighing the pros and cons of each

outsourcing solution. In the next sections, we pro-

vide a detailed deﬁnition for the most important set of

criteria that are widely considered when outsourcing

business processes to the Cloud.

2.1 Cost Saving

Almost all researchers and practitioners, agree on the

fact that cost saving is the most determinant factor that

attracts enterprises to outsource their related business

processes. In their analysis of data collected from

trade reports, (Rouse and Corbitt, 2004) showed that

outsourcing may yield a cost saving of 20%. (Yang

et al., 2007) argue that outsourcing is guided essen-

tially by overhead costs; that is, parts of a business

process to be outsourced are selected by ascertaining

ﬁrstly how much money they may save. To calcu-

late the cost of a business process, one should have

a knowledge about expenditure of setting up, execut-

ing and monitoring each of its a tasks/activities. An

activity’s cost is calculated essentially by investigat-

ing on expenditure related to its enactment cost (EC)

and realization cost (RC). The former corresponds to

the cost necessary for achieving essential steps in the

business process management life-cycle starting from

its design to the monitoring of its behavior. The latter

corresponds however, to the data transfer rate, trans-

action, or pre-payment for a period of time (Saeedi

et al., 2010).

2.2 Focus on Core Competence

Focusing on core competencies means that enter-

prises are spending ﬁnancial expenditure and business

efforts on activities expected to bring competitive ad-

vantage. As stated by Tom Peters, an expert in the out-

sourcing area, an enterprise should follow the rule of

”Do what you do best and outsource the rest” (Soiva,

2007). The activities which are typical for outsourc-

ing are those considered as noncore. More speciﬁ-

cally, the less strategic the activity is, the more likely

it can be outsourced. In this sense, business activities

can be categorized into three types according to their

strategic importance, listed from most to least criti-

cal:Core competence, critical noncore and noncritical

noncore. Although the general attempts when out-

sourcing is to delegate noncore activities/processes to

external provider, some enterprises are trying to out-

source also some of their related core business pro-

cesses (Theo Lynn et al., 2014).

ICSOFT-EA2015-10thInternationalConferenceonSoftwareEngineeringandApplications

Figure 1: OutyBPMN metamodel: BPMN2.0 extensions with outsourcing concerns.

2.3 Security Concerns

When outsourcing its business processes, an enter-

prise should be aware about risks that may lead to

undesirable outcome. Security concerns are the most

prominent factor that may prohibit enterprises from

outsourcing to external provider (Yang et al., 2007).

This concern is due to the fact that service providers

have control over the outsourced activities includ-

ing those dealing with the conﬁdential data of cos-

tumers. This risk increases when the service provider

lacks the means to encrypt data of outsourcing en-

terprises (Gewald and Rouse, 2012). Moreover, the

Internet-based connectivity between the outsourcing

enterprise and its service provider can present vulner-

abilities of unauthorized access to personal data, in-

trusion and hackers. To overcome this situation, an

outsourcing enterprise should think to keep conﬁden-

tial business data or activities dealing with high level

security data in premise. This way, it will be sure that

vital information and data are secured and protected

(Pathak and Joshi, 2009). It is worth noting that, in

spite of the existence of multitude solutions, security

issues still remain a serious problem when outsourc-

ing due to lack of trust in service providers.

2.4 Quality Improvement

Besides cost reduction, outsourcing relies on the hy-

pothesis that the service providers are capable of per-

forming the outsourced activities more efﬁciently and

effectively. The satisfaction of this hypothesis pre-

sumably leads to an increased customer satisfaction

(Gewald and Dibbern, 2009). Indeed, the majority

of outsourcing cases are elaborated after making sure

that the quality of the services provided by the ex-

ternal party is better compared to internal outcomes

(Yang et al., 2007). In addition, outsourcing enter-

prises may beneﬁt from external expertise as the ser-

vice supplier may have skills, platforms specialists

and technical staff for enacting business processes;

the outsourcing enterprise is often unable to provide

such human resources.

3 BPMN EXTENSION WITH

OUTSOURCING CONCERNS

Business processes are devoted to present the work-

ﬂow of activities within enterprises. In a service ori-

ented architecture, the business process is considered

as a set of logically interrelated services. In this con-

text each service is supposed to perform an activity

(Saeedi et al., 2010). Modeling a business process

is considered as a means to improve the way of the

business process operation. In general, the promi-

nence of modeling processes is due to the possibil-

ity it offers for sharing knowledge between enter-

prise stakeholder which allows to work harmoniously

towards goals (Eriksson and Penker, 1998). Sev-

eral number of process modeling languages and no-

tations have been emerged in the aim to assist en-

TowardsOutsource-abilityEnabledBPMN

terprises in the documentation and presentation of

their processes. However, BPMN (Business Pro-

cess Modeling Notation) ((OMG), 2011) is consid-

ered as the de facto standard (Rodr´ıguez et al., 2007)

approved by ISO/OSI (ISO 10303-203:1994, 1994).

BPMN is deﬁned by OMG in order to make the un-

derstanding of business processes easier from busi-

ness analysts to technical developers. Its elements

can be classiﬁed into ﬁve categories: Data, Flow

Objects, Connection Objects, Swimlanes and Arti-

facts; we refer readers to ((OMG), 2011) for more

details. BPMN2.0 introduces an extensibility mech-

anism for extending standard BPMN elements with

additional attributes and elements to specify a spe-

ciﬁc need. The BPMN2.0 extension consists essen-

tially on four different elements which are: Exten-

sion, ExtensionDeﬁnition, ExtensionAttributeDeﬁni-

tion, and ExtensionAttributeValue. The Extension-

Deﬁnition class deﬁnes additional attributes, however

ExtensionAttributeDeﬁnition presents the list of at-

tributes that can be attached to any BPMN element.

ExtensionAttributeValue contains attribute value. Fi-

nally the extension element imports the Extension-

Deﬁnition and its related attributes to a BPMN model

deﬁnition. Several works focus on the BPMN exten-

sion for different purposes. In fact, the extension al-

lows to give additional comprehension for business

process models. Moreover, adding new concepts to

business process modeling allows to switch the way to

use the BPMN from the contemplative manner to the

productive one by automating the analysis of the busi-

ness process or even its implementation (Bocciarelli

and D’Ambrogio, 2011). In this context, (Rodr´ıguez

et al., 2007) presents a new extension of BPMN to

incorporate security requirements to business process

diagrams. Additionally, the adding of the quality of

service into the business process modeling is con-

sidered in (Saeedi et al., 2010). Despite the multi-

tude researches dealing with business process exten-

sion, there is no work which deals with the extension

of BPMN for outsourcing concerns. In our elabo-

rated extension, we propose a generic concepts that

helps experts to decide the suitability of BPMN ele-

ments to be outsourced such as cost, business critical-

ity, security and speciﬁc concepts such as the location

side concept (in premise, in the cloud) and defaulting

service concept as we aim to enhance the web ser-

vice quality by enacting it in the Cloud environment.

This paper adopts BPMN2.0 to support the aforemen-

tioned outsourcing concerns. Figure 1 a) presents

the proposed class diagram corresponding to the meta

model used for illustrating the outsourcing concerns.

The outsourcing concerns class presents the Exten-

sionDeﬁnition class containing extension attributes

classes which are: Cost, BusinessCriticality, Secu-

rityLevel and DefaultingService. These latter classes

present the ExtensionAttributeDeﬁnitionillustrated in

details in ﬁgure 1 b). The extension proposed is in-

tended for adding new concepts to the activity ele-

ment as our aim is to identify then which activities are

the most suitable for outsourcing. The activity is the

generic term for work that enterprise performs. Table

1 presents the graphical representation of the exten-

sions over activity element and their corresponding

description. Only information about location side (in

premise/in cloud) are designed to be on the pool ele-

ment. In fact, we aim to group activities selected to

be outsourced into one BPMN element which is the

pool. The implementation of the proposal is done us-

ing Eclipse Modeler as a tool.

4 OUTSOURCED ACTIVITY

IDENTIFICATION

Our work aims to ﬁnd a good solution correspond-

ing to the appropriate set of activities suitable for

outsourcing respecting the enterprise experts prefer-

ences. These latter are speciﬁed in OutyBPMN. Our

solution search adopts penalty based genetic algo-

rithm. In this section, we ﬁrst present the problem

model. Secondly, we present our algorithm for the

identiﬁcation of activities to be outsourced. A report

on the algorithm performance is presented in Section

4.1 Problem Formulation

The input to our search algorithm is an OutyBPMN

model identifying in particular, the following activi-

ties:

• Expensive when executed in premise compared to

when executed in the cloud;

• Noncore;

• Handling and processing non sensitive data;

• Presenting a defaulting behavior when executed

as a service within the enterprise.

To look for an optimal set of activities best out-

sourced, we will use the following problem formu-

lation:

1. A={ A

,...,A

} is a set of activities com-

posing the business process.

2. The weights for outsourcing concerns, W

, W

and W

for Cost (C), Business Criticality (BC),

ICSOFT-EA2015-10thInternationalConferenceonSoftwareEngineeringandApplications

Table 1: Extensions graphics for BPMN elements.

Graphical extension Description

This icon means that experts should precise the business criticality,

namely the activity is core {high level of business criticality}, critical

noncore {medium level of business criticality} or noncritical noncore

{low level of business criticality}. To simplify for users, one can precise

the level of criticality by choosing one of proposed graphics presented

in the last row of this table.

IT and business experts should collaborate to identify whether the ac-

tivity handle conﬁdential data or not. The security graphic is used when

experts observe that the activity proceed sensitive data .

In the context of this paper, IT experts should precise the cost of the

hardware supporting the activity. More speciﬁcally, experts should

be able to present in detail required expenditure of hardware mainte-

nance, monthly bills, and all corresponding costs for realizing an activ-

ity. These information will help next to compare between ﬁnancial cost

of executing the activity in premise and in the cloud. Techniques used

for calculating two prices is out of this paper scope. However, based

on the difference between two prices, experts can categorize the cost of

doing the activity internal {high,medium,low}.

As previously said, among potential determinants urging enterprises

to outsource their business processes, is the need to improve service

quality. As we focus on business processes running within Small and

Medium Enterprises (SME) that lack sufﬁcient IT expertise, cost and

required determinants for well realizing their business processes, we

decide to choose defaulting services most suitable for outsourcing. De-

faulting services are those have leading to the degradation of the busi-

ness process performance, more speciﬁcally, those preventing business

processes from attaining their goals. Discovering defaulting services

is done in our previous work. We use ”defaulting service” term and

”activities presenting a default behavior when executed as service” in-

terchangeably to refer the same thing.

Whenever the algorithm of selecting appropriate activities to be out-

sourced is done, experts should analyze results and precise on each ac-

tivity its location side {in premise, in the cloud}.

These graphics help experts to specify level of importance of the ex-

tended elements (business criticality, security, and cost). They refer

respectively to, high , medium and low level.

Defaulting Service (DS), and the number of activ-

ities composing the solution (L) respectively.

∑

k=1

= 1 (1)

3. S={S

,...,S

} is a set of security con-

straints imposed on data handled by correspond-

ing activities.

Our aim is to give a set A’ of activities that are most

suitable for outsourcing where A’⊂ A. We should no-

tice all the process may merit to be outsourced which

make A’=P where P is the entire business process al-

though it is not a preferable case. A’ is selected ac-

cording to the ﬁtness function (4).

We convert the optimization problem from a

multi-objective to single objective one by assigning

weights to each objective function composing the

function (2).

TowardsOutsource-abilityEnabledBPMN

Max F’=

∗ F

′

) +W

∗ F

′

) +W

∗ F

′

) +W

∗ F

′

)

(2)

Where

• F

′

) =

∑

i=1

C(a

)

• F

′

) =

∑

i=1

BC(a

)

• F

′

) =

∑

i=1

DS(a

)

• F

′

) =

(

1+ (

(N−N

′

)

′

−1)

) if 1 < N < N

′

1− (

(N−N

′

)

(L(P)−N

′

)

) if N

′

< N < L(P)

(3)

• a

is an activity that belongs to A’.

• N is the number of activities composing the solu-

tion A’.

• N’ is the preferred length of the solution A’. This

metric is added to F

′

to prevent foster solutions

having greater number of activities. The value of

this metric is automatically calculated by count-

ing the number of activities having a defaulting

behavior when executed as services and not deal-

ing with sensitive data.

• L(P) is the number of activities composing the

process.

4.2 Implementation of Genetic

Algorithm to Select Suitable

Activities for Outsourcing

Genetic algorithms (GAs) (Gewald and Rouse, 2012)

are evolutionary algorithms. Their main idea is to

simulate the evolution of population composed of di-

versiﬁed individuals. These individuals are subject

to operations such as recombination and mutation al-

lowing thus, by selecting best individuals, to enhance

the population quality. Choosing to work with GA is

due to the fact that it allows to ﬁnd solutions which

best meet different criteria. The genetic algorithm

presents solutions as individuals in a population that

varies each time its quality is enhanced. The individ-

uals composing a population vary when undergoing

a set of operations such as crossover and mutation.

Their quality is evaluated using a ﬁtness function.

4.2.1 Individuals Encoding

To ﬁnd out suitable solutions for the optimization

problem, appropriate encoding of individuals is nec-

essary. First of all, we should note that individuals

composing a population have variable lengths. Each

Figure 2: Individuals encoding.

gene, composing an individual, corresponds to an in-

teger referring an activity that belongs to the business

process. Figure 2 presents examples of individuals

encoding.

4.2.2 Infeasible Solutions

In our research case, some solutions are infeasible as

they violate security constraint. An individual may be

composed of one or more activities handling sensitive

data, we consider each activity dealing with sensitive

data as a constraint violation.

4.2.3 Crossover and Mutation

We apply in our proposal the crossover and the mu-

tation operations. We adopt the classical one-point

crossover to generate each time two offsprings. Con-

cerning the mutation operator, we have use it in three

different ways:it can select randomly a position in the

individual and replace it by another activity, add an

activity, or delete existing one. An additional control

should be done in this level to prevent repeating the

same activity within an individual.

4.2.4 Fitness Function

An individual composed of activities requiring secur-

ing their data is considered as infeasible. To guaranty

that GA reaches an optimal or near-optimal solution,

these infeasible individuals should be kept as their

presence is essential in the building of solutions.

Thus, the idea is to give a penalty to ﬁtness values

relative to infeasible solutions. This leads to lower

ﬁtness values compared to those corresponding to

feasible solutions. Moreover, infeasible solutions

having more activities requiring security are more

harshly penalized. This process helps to reenforce the

presence of feasible solutions and the disappearance

of the infeasible ones from generation to another.

The equation (4) presents the deﬁnition of the ﬁtness

ICSOFT-EA2015-10thInternationalConferenceonSoftwareEngineeringandApplications

Figure 3: The business process used for the evaluation of the proposed GA.

function.

Fitness(X)=



′

∗ 0,5+ 0,5 if v(X) = 0

′

∗ 0,5− (0,1∗ v(X)/v

max

) otherwise

(4)

F’ is the objective function presented in section 4,

v(X) presents the total number of activities compos-

ing an individual requiring security, and v

max

stands

for the total number of activities requiring security

in the entire business process. The presented ﬁtness

guaranties that infeasible solutions have always less

ﬁtness values compared to feasible ones.

5 EXPERIMENT AND RESULT

We applied the penalty based genetic algorithm using

Java as a development language. Simulations were

conducted on a laptop computer with 2.5 GHz Intel

Core i7 CPU and 4GB RAM.

The business process illustrated in ﬁgure 3

presents an example of a case study that we analyzed.

As shown, the business process is composed of

13 activities. The suitable activities appropriate for

outsourcing is unknown rendering the research space

huge despite that experts deﬁne the preferred number

of activities to be outsourced. Preferred number of ac-

tivities to be outsourced in this example is 4 (the num-

ber of activities presenting defaulting behavior when

executed as services and not requiring security con-

straints). As shown, three activities require high level

of security which are: 2, 4, and 7. Individuals encom-

passing one or more of these three activities are penal-

ized but they still present solutions of our problem.

The evaluation of our penalty based GA is twofold:

we evaluated ﬁrstly the performance of our genetic

algorithm and then we veriﬁed the pertinence of the

results. A comparison between our algorithm and a

greedy algorithm was also elaborated in terms of per-

formance and the pertinence of results.

5.1 Experiment A

Table 2 presents parameters values we used to exper-

iment our algorithm. We evaluate the validity of our

penalty based genetic algorithm by comparing its time

cost in different computing scale.

Figure 4: The execution time consumed when increasing

the number of constraints.

We begin by observing the execution time con-

sumed by our algorithm when increasing the number

of constraints. Figure 4 shows that the number of con-

straints inﬂuences the execution time of the algorithm

which can be explained by the additional number of

treatment to be done.

Figure 5: The execution time consumed when increasing

the number of activities.

TowardsOutsource-abilityEnabledBPMN

The number of activities composing a business

process increases the time of execution as shown in

ﬁgure 5.

Table 2: Parameters used for our penalty based genetic al-

gorithm.

Parameters Description

Population size 50

Selection technique Tournament selection

Termination condition Number of genera-

tion=100

Crossover probability 0.5

Mutation probability 0.015

We compared our genetic algorithm with another

optimization algorithm namely the greedy algorithm

(Kodaganallur and Sen, 2010). Greedy algorithms are

known by their ability to ﬁnd quickly solutions. Gen-

erally, generated solutions are approximate, and opti-

mal ones are founded in few cases. The comparison

was done by testing the execution time of both algo-

rithms when applied to different number of business

processes. As shown in ﬁgure 6, the GA consumed a

bit more time to generate results compared to greedy

algorithms.

Figure 6: Comparison between the execution time con-

sumed corresponding to both algorithms.

5.2 Experiment B

To evaluate the pertinence of our algorithm, we inter-

viewed 5 business experts having knowledge on busi-

ness processes. We ﬁrst asked experts to rank the out-

sourcing concerns according to their preferences, the

results of this interview are shown in table 3.

According to the experts preferences, we at-

tributes to the weights explained in section 4.1, the

following values: W1=0.4 for (BC), W2=0.25 for

(DS), W3=0.25 for (C) and W4=0.1 for (L).

The extended BPMN illustrated in ﬁgure 3 is pre-

sented to the same experts to depict, according to

them, the most appropriate activities to be outsourced.

Table 3: Ranks of outsourcing concerns according to ex-

perts.

Rank Outsourcing

Concern

(OC)

% experts how se-

lect the correspond-

ing OC

1 BC 100%

2 DS & C 60%

3 L 100%

Among the thirteen activities, four experts chose the

activities 12 and 6 and one expert select in addition

to these activities, the activity 11. When we ap-

plied the GA, the individual having the best quality

according to its ﬁtness is composed of two activi-

ties: {6,12} where ﬁtness=0.8437. The greedy algo-

rithm generated an individualcomposed of 3 activities

{12,9,1}. Table 4 presents the precision and recall of

our penalty-based genetic algorithm and the same val-

ues corresponding to greedy algorithm.

Table 4: Evaluation of our GA.

precision

of GA

recall of

precision

of greedy

recall of

greedy

100% 66% 33% 33%

As shown in above table, the GA has a high value

of recall and precision which make it a relevant and

pertinent way to assist experts in the decision of busi-

ness process outsourcing. Moreover, despite that the

greedy algorithm take less time to generate results,

our GA generate better results if we refer to the com-

parison of the recall and precision of the two algo-

rithms. According to experiments, we can afﬁrm that

the proposed genetic algorithm is a pertinent way to

decide about the business process activities to be out-

sourced. However, in term of performance and more

speciﬁcally the response time, the genetic algorithm

may not be the most suitable one.

In the context of our research, when applying the

GA on the business process, the BPMN should be re-

designed so that, activities selected to be outsourced

are putted within a pool element designed to be en-

tirely in the cloud as presented in ﬁgure 7.

6 CONCLUSIONS

This paper presents a method to assist experts in the

fastidious task of BPO to the cloud decision. The

method offers a modeling language, OutyBPMN, an

extension of the BPMN to take into consideration

the outsourcing criteria. In addition it uses a penalty

based genetic algorithm to identify most appropriate

ICSOFT-EA2015-10thInternationalConferenceonSoftwareEngineeringandApplications

Figure 7: BPMN with deployment location data.

activities of a business process to be outsourced. In

the herein presented work, activity appropriateness

is determined based on its business criticality, cost,

its quality when executed and constrained by secu-

rity level of handled data. Based on our preliminary

experimental results, the proposed penalty based ge-

netic algorithm generates satisfactory results. Indeed,

the evaluation of the method presents its accuracy

and the similarity with experts preferences. We are

elaborating a deployment model for the execution of

the business process when part of it is located in the

cloud. Moreover, we are working on deﬁning a deci-

sion model to weight the importance of outsourcing

the selected pool against keeping it in premise. Ad-

ditional experimental evaluations are needed to adjust

the ﬁtness function and its related parameters to thor-

oughly examine the performance of the proposed al-

gorithm.

REFERENCES

Adesta, E. and Agusman, D. (2004). The evolution of

supply-chain management into extended enterprise.

In Engineering Management Conference, 2004. Pro-

ceedings. 2004 IEEE International, volume 3, pages

1298–1302 Vol.3.

Bocciarelli, P. and D’Ambrogio, A. (2011). A bpmn exten-

sion for modeling non functional properties of busi-

ness processes. In Proceedings of the 2011 Sympo-

sium on Theory of Modeling & Simulation: DEVS

Integrative M&S Symposium, TMS-DEVS ’11, pages

160–168, San Diego, CA, USA. Society for Computer

Simulation International.

Deb, K. and Kalyanmoy, D. (2001). Multi-Objective Opti-

mization Using Evolutionary Algorithms. John Wiley

& Sons, Inc., New York, NY, USA.

Eriksson, H.-E. and Penker, M. (1998). Business Modeling

With UML: Business Patterns at Work. John Wiley &

Sons, Inc., New York, NY, USA, 1st edition.

Gewald, H. and Dibbern, J. (2009). Risks and beneﬁts of

business process outsourcing: A study of transaction

services in the german banking industry. Inf. Manage.,

46(4):249–257.

Gewald, H. and Rouse, A. (2012). Comparing business pro-

cess and it outsourcing risks–an exploratory study in

germany and australasia. In System Science (HICSS),

2012 45th Hawaii International Conference on, pages

275–284.

Hu, Y.-B., Wang, Y.-P., and Guo, F.-Y. (2005). A new

penalty based genetic algorithm for constrained opti-

mization problems. In Machine Learning and Cyber-

netics, 2005. Proceedings of 2005 International Con-

ference on, volume 5, pages 3025–3029 Vol. 5.

ISO 10303-203:1994 (1994). Information technology – ob-

ject management group business process model and

notation.

Kodaganallur, V. and Sen, A. (2010). Greedy by chance -

stochastic greedy algorithms. In Autonomic and Au-

tonomous Systems (ICAS), 2010 Sixth International

Conference on, pages 182–187.

Li, H. and Meissner, J. (2009). Improving quality in busi-

ness process outsourcing through technology. Work-

ing Papers MRG/0009, Department of Management

Science, Lancaster University.

Mell, P. M. and Grance, T. (2011). Sp 800-145. the nist deﬁ-

nition of cloud computing. Technical report, Gaithers-

burg, MD, United States.

(OMG), O. M. G. (2011). Business process model and no-

tation (bpmn) version 2.0. Technical report.

Pathak, R. and Joshi, S. (2009). Secured communication for

business process outsourcing using optimized arith-

metic cryptography protocol based on virtual parties.

In Ranka, S., Aluru, S., Buyya, R., Chung, Y.-C., Dua,

S., Grama, A., Gupta, S. K. S., Kumar, R., and Phoha,

V. V., editors, IC3, volume 40 of Communications in

Computer and Information Science, pages 205–215.

Springer.

Rodr´ıguez, A., Fern´andez-Medina, E., and Piattini, M.

(2007). A bpmn extension for the modeling of se-

curity requirements in business processes. IEICE -

Trans. Inf. Syst., E90-D(4):745–752.

Rouse, A. C. and Corbitt, B. J. (2004). It-supported business

TowardsOutsource-abilityEnabledBPMN

process outsourcing (bpo): The good, the bad and the

ugly. In PACIS, page 126. AISeL.

Saeedi, K., Zhao, L., and Sampaio, P. (2010). Extending

bpmn for supporting customer-facing service quality

requirements. In Web Services (ICWS), 2010 IEEE

International Conference on, pages 616–623.

Soiva, J. (2007). business process outsourcing case

study:How to develp Deloitte S2Gs shared service ac-

tivities in the Bercelona costumer response centre.

PhD thesis, , Lahti university of applied sciences.

Theo Lynn, N. O. C., John Mooney, M. H., David Corco-

ran, G. H., Lisa Van Der Werff, J. P. M., and Healy, P.

(2014). Towards a Framework for Deﬁning and Cate-

gorising Business Process-As-A-Service (BPaaS). In

21st International Product Development Management

Conference. Citations: Not Avail. ().

Yang, D.-H., Kim, S., Nam, C., and Min, J.-W. (2007).

Developing a decision model for business process

outsourcing. Computers and Operations Research,

34(12):3769 – 3778. Operations Research and Out-

sourcing.

ICSOFT-EA2015-10thInternationalConferenceonSoftwareEngineeringandApplications