IERI Procedia 4 ( 2013 ) 99 – 109

Available online at www.sciencedirect.com

2212-6678 © 2013 The Authors. Published by Elsevier B.V.Selection and peer review under responsibility of Information Engineering Research Institutedoi: 10.1016/j.ieri.2013.11.016

ScienceDirect

20

Conv

Abstract

Finite state in analyzinmodeling aexisting chchange evawhen a chamanagemenexacting thethe represenmanagemen © 2013 PuSelection Keywords: W

1. Introdu

Changeto a businensure thaand prompconfiguratarises the Here Chanis more prmust be domust not dynamic vaids servic

013 Internat

ventionaS

Thirum a

cd

machine and Png diverse web a web service hange managemaluation and moange whereas nt only includee overall functinting the overalnt and also the e

ublished by Eln and peer rev

Web Services; Ch

uction

e managementness process

at: Business ript handling otion managemneed for an e

nge Managemronounced to one very quicbe affected t

variations to tce provider an

tional Conf

l Usage Service C

maran.Ma , aDepartment of C

bDepartment ofcDepartment of CdDepartment of C

Petri nets are onservice researcchange manag

ment approacheonitoring. Alsosimple finite s

es emergency aion as in case ofll change manaevaluation of c

lsevier B.V. view under r

ange Mangement

t is a set of prduring its m

isk is managedof all changesment system; effective fram

ment frameworthe profitabil

ckly within ththere by manthe business lnd serves him

ference on A

of FiniteChange M

DhavachelComputer science

f Computer sciencComputer scienceComputer science

ne among the vach areas. In thgement framews to satisfy the

o complex strustate machine cand minor chaf Petri nets. Thugement. In this

changes implem

responsibility

t;Change factors;

rocesses that amaintenance pd and minimis; all changesand all autho

mework managrk is proposedlity of the orge given time

naging the chlogic so that i

m to satisfy cli

Agricultural

e State MManage

lvan.Pb, Aie and Engineering

ce and Engineeriand Engineering

e and Engineering

arious conceptuhis paper a com

work is analyzee normal chang

ucture called Pecan be used fo

anges. Using Fus for the evalupaper, we are g

mented using cer

y of Informat

;Finite State Mach

are employedphase. The pzed; Standards to service aorized changeging those chd for making mganization. Soconstraint as

hanges at theit is feasible tients’ newfang

l and Natura

Machineement Fr

ishwarya.Dg, Pondicherry En

ng, Pondicherry g, Pondicherry Eng, Pondicherry En

ual and computmparison of usied with experimges, they fail toetri nets is usedor the same reFSM, only a suation of changgoing to see howrtain change fac

tion Enginee

hine;Petri nets.

to ensure thapurpose of thdized methodsassets and cones support buhanges withouminor alteratio the changesthe services c runtime. Weto make frequgled quests. S

al Resource

over Peramewor

Dc, K.Rajakngg College, Indi

University, Indiangg College, Indingg College, Indi

ational models ing finite state mental analysio there is no p

d to model the opresentation si

specific logic ces implementew FSM is efficctors when com

ering Researc

at significant che change mas and procedunfiguration ite

usiness needs ut affecting thons to business in business consumed by e focus on cuent modificaSuch environm

es Engineeri

etri net Inrk

kumarid ia.

. a. ia

that have beenmachine and Ps. Though therproper support overall system ince the procescan be extracteed, FSM is veryient in modelin

mpared to Petri n

ch Institute

changes are imanagement pr

ures are used fems are recoand goals. H

he business fuss logic but wlogic of the wthe providers reating enviro

ations in a serment is useful

ing

nWeb

n widely used Petri nets for re are many for efficient functionality ss of change ed instead of y efficient for ng the change nets.

mplemented rocess is to for efficient rded in the

Hence there unctionality. whose effect web service and clients

onment for rvice which in the run-

Available online at www.sciencedirect.com

© 2013 The Authors. Published by Elsevier B.V.Selection and peer review under responsibility of Information Engineering Research Institute

Open access under CC BY-NC-ND license.

Open access under CC BY-NC-ND license.

100 M. Thirumaran et al. / IERI Procedia 4 ( 2013 ) 99 – 109

time management of web services and to exactly spot the solution to the service provider’s maintenance element. There are many models that subsist for Business Process Management in which the process is recycled for the overall process changes. However, the problem with these solutions is that they only support the process level flexibility and not the application/service level flexibility. On the other hand this Business Logic Management framework tries to append that service level flexibility. These changes are done at the Business analyst level instead of being done at the Developer level, which reduces the hierarchy level in change management and thus implies a reduction in time and cost requirements. Here comes the need for an efficient model representation which provides clear visualization of the process of change management and the evaluation of changes implemented for the analyst to understand the system easily. Petri nets provide a graphical notation for the formal description of the dynamic behavior of systems and are particularly well suited to systems which exhibit concurrency, synchronization, mutual exclusion and conflict whereas finite state machines is a mathematical model of computation used to design both computer programs and sequential logic circuits. Though Petri nets have been used in the application areas such as Software design, Workflow management, Process Modeling, Data analysis, Concurrent programming, Reliability engineering, Diagnosis, Discrete process control and Simulation, finite state machine plays an efficient role in representing the evaluation of change implemented. Also Petri net is not only a significant burden placed on the analyst in order to specify complex models, but in addition the graphical representation may become too complex to be useful and understandable. In order to provide simple understandable and useful representation for the inexperienced non specialist analyst, finite state machine can be used for implementing the framework. This paper fully concentrates on the advantages of using FSM over Petri nets for change management and evaluation of changes implemented.

2. Literature survey

Xumin Liu et al. [1] proposed an Ev-LCS, an end-to-end framework that specifies, reacts to, and verifies top-down changes in a LCS. This framework first propose a formal model which provides the grounding semantics to support the automation of change management and a set of change operators that allow specifying a change in a precise and formal manner by proposing a set of algorithms to automatically implement them. It then proposes a change enactment strategy that actually implements the changes. Dimitris Apostolou et al. [2] proposed an ontology-based approach for developing and maintaining e-Government services that can effectively deal with changes which enables the systematic response of e-Government systems to changes by applying formal methods for achieving consistency when a change is discovered and also enables the knowledgeable response of service designers and implementers to changes by utilizing design rationale knowledge. Florian Rosenberg et al. [3] proposed a domain-specific service selection mechanism and system implementation to address the issue of runtime adaptation of composite services that implement mission-critical business processes by making use of a domain-specific language called VieDASSL which can be used by domain experts to define the runtime adaptable selectors based on the QoS attributes in the adaptive QoS model. This approach ensures that changes in the QoS model and selectors can be handled at runtime without the disruption of the business processes by assuming the units of measurement for each QoS attribute are fixed and cannot be dynamically adapted. Bassam Atieh Rajabi and Sai Peck Lee [4] proposed two formalism of BPEL namely Graph based and Rule based formalism. The rigidity in graph based models incurs problems lack of runtime criteria (flexibility, dynamic and adaptability, which compromises the ability of the graph based processes to react to dynamic changes in BP and exceptional circumstances whereas the common objective of BRMS is to integrate complex process logic into a process model to support dynamic changes. To increase flexibility in process execution, Constraint based BP management approach is used which supports the evolution of BP. In [5], An adaptive and flexible framework is proposed to integrate OO diagramming technique and PN modeling language ( i.e. the graph based which has the visual appeal of being intuitive and explicit, even for those who have little or no technical background and rule based modeling language which requires good understanding of propositional logic and the syntax of logical expressions) in order to increase the representation capability for graph based modeling to support the dynamic changes in the runtime instances. In order to generate the BPEL code automatically, Yanhuna Du et al [6] proposed the Petri Net approach that automatically can verify the composition of partially incompatible services and also serves the time in such a way which will generate the BPEL code. This approach consist of 3 phases: Modeling composition of services is modeled as open WorkFlow Nets (oWFNs) which are composed by using the mediation transaction (MTs). Automatic verification of composition uses graph called Modular reachability graph (MRG) of composition is constructed and analyses the compatibility that can significantly improve the problem of state space explosion. In order to save the time, techniques called Event-condition Action (ECA) rule based is

101 M. Thirumaran et al. / IERI Procedia 4 ( 2013 ) 99 – 109

developed. Haochen Li, Zhiqiang Zhan et al. [7] proposed a solution that automates the change management process with machine learning method and assesses the solution from the business-driven perspective. This system analyzes the RFC and obtains the related CIs automatically and also automatically retrieves the most similar change solution. By automating the Change Management process with Business Driven Perspective, this system reduces the service interruption time and cost while doing the changes. To achieve the accuracy and efficiency, system returns valuable change solution that supports the collection of RFC (Request for Change) by comparing it with other system that will provide the optimal solution. Thus, the solution with the neural network of retrieving the similar RFC gives practical decision and also validates the change solution when selected for automatic change management process. Sabri MTIBAA and Moncef TAGINA [8] present a change management framework for a citizen-centric healthcare service platform. A combination between Petri nets model to handle changes and reconfigurable Petri nets model to react to these changes are introduced to fulfill healthcare goals. In [9], Using Petri nets based process modeling allows application of different analysis techniques. These techniques can be used to examine the behavior of the process and to calculate its performance measures. Transforming Petri net model into graph model gives possibility of using classical graph processing algorithms. This paper discusses the reasons, which impose Petri nets as a conceptual standard for modeling and analysis of workflow. Petri nets notation is used for representation of the main routing constructs as well as for the workflow process description. The correspondence of Petri-net-based model and graph model is demonstrated.S. Mtibaa and M. Tagina [10] present a distributed telemedicine environment reaping from both the benefits of Service Oriented Approach (SOA) and the strong telecoms capabilities. We propose an automated approach to handle changes in a distributed telemedicine environment. A combined Petri nets model to handle changes and Reconfigurable Petri nets model to react to these changes are used to fulfill telemedicine functional and non functional requirements.

3. Features of finite state machine vs petri nets

3.1 Schema driven

Since all the modifications are done at the schema level for the analyst’s ease of understanding the functionality of the program, finite state machine plays a major role implementing the schema in case of change management and change evaluation. With the help of finite state machine, it is possible to transform the schema into finite state machine or vice versa. Thus finite state machine is fully schema driven whereas Petri nets are not schema driven since it is process driven (i.e. the process can be transformed to Petri nets or vice versa) which is not easily understood by the analyst.

3.2 Extracting Particular Logic

In the process of change management, before doing changes it is necessary to extract a particular logic where we are going to do changes since change management framework mainly concentrates on doing emergency changes or simple changes. This is possible when we make use of finite state machine since it represents the logic through rules, functions or parameters whereas it is not possible to extract a particular logic by Petri nets. Petri nets can extract the origin of the process or the whole process even though our aim is to do minor changes.

3.3 Trace point

By making use of finite state machine, we can trace each and every statement in the code easily. Especially in the change management of web services, we can trace a specific statement in the code where we are going to do changes whereas if we use Petri nets, it is not possible to reach particular statement in the code since it can exhibit only the part of the code in which the change has to be made. Also in the process of evaluation of the changes made, if any error occurs, it can traced at a specific location where the error has occurred by making use of finite state machine whereas in case of Petri nets, it will denote a specific part of the code where the error has occurred.

3.4 Rollback

If an error occurs in the process of evaluation of the changes made, current state (state containing error) of the

102 M. Thirumaran et al. / IERI Procedia 4 ( 2013 ) 99 – 109

program can be rolled back to the previous state without affecting the original process of the program (i.e. without modifying the original code of the program)if we use finite state machine. But this may not be possible in case of Petri net since it rolls back to the origin of the whole process. thus whenever an error occurs while using Petri nets, the control reaches the origin of the process thereby increasing the time and space complexity in order to reach the state (from which it has been rolled back) for doing further different changes to make the system work according to change request.

3.5 Reachability

In finite state machine, all the states are reachable since finite state machine consists of only finite number of states whereas sometimes it is not possible in case of Petri nets since they are state transition systems. State transition systems differ however from finite state automata in several ways: In a state transition system the set of states is not necessarily finite, or even countable. In a state transition system the set of transitions is not necessarily finite, or even countable.

3.6 Possibility of deadlock

There is a possibility of deadlock in case of Petri net (this may not be always possible) since it has been used to model a variety of concurrent and discrete event distributed systems. But there is no possibility of deadlock in finite state machine since it is a mathematical model of computation used to design both computer programs and sequential logic circuits.

3.7 Time and Space complexity

Finite state machine can exist in only state at a given time whereas Petri nets can exist in more than one state at a given time. Thus finite state machine is less complex than Petri nets. Also finite state machine consumes less time and space because of its simple structure and simple processing where Petri net consumes more time and space because of its complex structure and complex processing.

3.8 Structure

Finite state machine consists of two main components namely state and transition whereas Petri netsconsists of four main components namely places, transition, arcs and tokens. In other words, since the finite state machine consists of only states, an input symbol and transitions, it is very simple and easily understandable whereas Petri net is a directed bipartite graph with two node types called places and transitions. The nodes are connected via directed arcs. Connections between two nodes of the same type are not allowed. Places are represented by circles and transitions by rectangles. The directed arcs describe which places are pre- and/or post conditions for which transitions (signified by arrows) occurs. These things make Petri nets structurally more complex than finite state machine.

3.9 Theoretical foundations for logic

Since finite state machine have full support for the theoretical foundation for logic, it is more is applicable for modeling sequential processing associated with logical set theory whereas since Petri net have no support for theoretical foundation of logic, it is more applicable for modeling concurrent processing associated with event and condition.

103 M. Thirumaran et al. / IERI Procedia 4 ( 2013 ) 99 – 109

4 .Case stu

The figurecase of Pecase of finchange req

PETRI N

P

E

WHOLECHANGFigure 1(a)

udy

e 1(a) and 1(b)etri nets, whennite state machquest with the

NET FOR WAP

E

P2

E2

P4 P5

E4 E5

P9 P

E8

E PART IS EES

) shows the renever a changehine, only a spe description

WHOLE PROPPROVAL

P1

E1

P3

E3

P6

E6

P

10 P11

P14

E9

End

EXTRACTED

epresentation e request compecific logic sthat certain a

CESS OF LO

P7 P8

E7

P12

P13

WHILE DO

of loan approvmes, the wholeset is extractedamount of tax

OAN FSM

OING BUT EXTRDOIN

val using both process is ex

d. For exampl must be add

FOR THE WA

1

F6

F

10 1

ONLY A SPERACTED WHNG CHANGES

h Petri net andxtracted from te, In a Bankined to all kind

WHOLE LOGAPPROVAL

q0

0

F3

10

q1

1

1 1

F7 F8

1

1

12

q2

1

1

10

F11

21

H

ECIFIC PARTHILE S. Figure 1(b)

d finite state mthe origin wheng system, conds of loans.

GIC OF LOA

F9

13

F10

T IS

machine. In ereas in nsider a

AN

104 M. Thirumaran et al. / IERI Procedia 4 ( 2013 ) 99 – 109

educatifrom the oP10, P11, transitionsit is very though the

4.1 Evalua

BU

While evaluationdoesn’t nemaking usexisting sydoing chancode consi

ional loans” (origin. In the P12, P13, P1

s (refer table much larger

e main aim is t

ation of chang

q

F

q1

1

1

F6

10

F7

11

1

F

E

USINESS POLFig

evaluating thn of changes meed to have mse of the changystem beyondnges or depenistency and si

(refer table 1 above menti

14} which rep3 and 4). Thin case of Peto do minor ch

ges

0

0

3

01

1

1

1F9

F8

1312

1q2

20

F11

21

H

LICY ENFORgure 2(a) he changes mmade withoutmuch knowlege factors and

d a particular cndendability. Tmilarity meas

and 2 ). Whiioned examplepresents the shus with theetri nets sincehanges.

F10

RCEMENT

made using ct any complex

edge in prograd FSM, it is eacontext with tThe change fasure.

ile using Petre, the extractset of placeshelp of FSM

e it involves e

hange factorsxity and alsoamming for uasy to detect tthe help of nuactors used ar

i nets, it is neion includesand {E1, E2,

M, time and sextracting the

1

F6

10

ORD

s, FSM playsprovides a cl

understandingthe errors or v

umber of transre Business po

ecessary to ex{P1, P2, P3, E3, E4, E5, space complee whole proce

02

q0

0

F30

01

q1

1

11

F7 F8

11 12

q2

1

F11

20

21

H

DER OF EXEFigure 2(b)

s a major rolear visualiza

g the functionviolation of positions in stateolicy enforcem

xtract the whP4, P5, P6, PE6, E7, E8 ,

exity is reducess from the o

F4

12

F9

13

1

F10

ECUTION

ole in implemation for the anality of the olicies or devie transition tabment, order of

hole process P7, P8, P9, E9} set of

ced whereas origin even

menting the analyst who system. By iation of the ble after the f execution,

105 M. Thirumaran et al. / IERI Procedia 4 ( 2013 ) 99 – 109

Figure 3 Me

CODE

1

F6

10

asure of change f

CONSISTEN

q0

0

F3

01

q1

1

1

1

F7 F8

11 12

1q2

1 20

F11

21

H

Figure 2(c)

factors using finit

NCY

F9

13

1

F10

te state machine

SI

F6

IMILARITY M

q0

0

F3

10

q1

1

1

1

F7 F8

1011 12

q2

1

F11

21

H

Figure 2(d

MEASURE

1

F9

13

1

20

F10

E

d)

106 M. Thirumaran et al. / IERI Procedia 4 ( 2013 ) 99 – 109

Table 1

TABL

Rules

Functions Parameters

Policy Dependenc y

Account Creation

R1

Get_details()

F1

name P1

NULL

NULL

address P2 occupation P3 occupation_address P4 Annual_income P5 Accnt_type P6 Guardian_name P7

create()

F2

name P1

PO1

F2->F1,F4

address P2 occupation P3 occupation_address P4 Annual_income P5 Accnt_type P6 Guardian_name P7 Accnt_no P8

Account Validity

R2

Valid_account() F3 Accnt_no P8 PO2 F3->F2,F5

Valid_address() F4 address P2 NULL F4->F1

Loan input details

R3

Get_loandetails

F5

property P9

PO4

F5->F2

Property_owner P10 Property_address P11 Property_worth P12 Type_of_loan P13 Actual_loan_amount P14 Accnt_no P8

Eligibility checking

R4

Transaction_checking()

F6 Accnt_no P8

NULL

F6->F3,F5 No_of_transactions P15 No_of_days P16

Account_balance()

F7

totalBalance P17

PO3

F7->F3,F5 Accnt_no P8 Type_of_loan P13 minBalance P18 Actual_loan_amount P14

Property_checking()

F8

Accnt_no P8

PO4

F8->F3,F5

property P9 Property_worth P12 Property_address P11 minWorth P19 Type_of_loan P13

Income_checking()

F9

Accnt_no P8

NULL

F9->F3,F5 Occupation P3 Annual_income P5 Type_of_loan P13 Occupation_address P4

Loan Approval

R5

Funds_available()

F10

Accntno P8

NULL

F10->F3,F5 type_of_loan P13 Interest_rate P20 Actual_loan_amount P14

Total_loan_amount_paid()

F11

Accnt_no P8

PO5

F11- >F5,F6,

F7,F8, F9, F10

type_of_loan P13 interest_rate P20

Actual_loan_amount P14 Total_loan_amount P21

107 M. Thirumaran et al. / IERI Procedia 4 ( 2013 ) 99 – 109

TABLE 4 TRANSITIONS SYMBOLSGetting loan input details E1 Validating address E2 Validating account E3 Income checking E4 Property worth checking E5 Balance checking E6 Transactions checking E7 Calculating interest E8 Calculating total loan amount to be paid

E9

Table 2

TABLE 2

POLICIES DESCRIPTION

PO1 Account should be created only if the address is a valid address.

PO2 Single person should hold the account

PO3 Total balance should be greater than the specified minimum balance except for the person with military as occupation

PO4 Property owner should be the loan borrower except in case of student for educational loan

PO5 No tax or interest rate for educational loans

Table 3 Table 4

TABLE 3

PLACES SYMBOLS Loan request P1 Valid Address list P2 Valid Account list P3 Valid Account for eligibility checking P4 Minimum income list P5 Minimum property worth P6

Minimum balance P7 Minimum number of transactions P8 Account with valid income P9 Account with valid property worth P10 Account with valid balance P11 Account with valid number of transactions

P12

Interest rates for various loan types P13 Eligible account for loan approval P14

Figure 3 shows the measures of the change factors code consistency (cc1) , similarity measure (sm1) , order of execution (ooe1) and business policy enforcement ( bpe1) through finite state machine. For every change request, business logic related with that request is retrieved and finite state machine is constructed whereas while using petri net, states and transitions are constructed for whole business processes. The evaluation of time taken for petri net construction , finite state machine construction , rule detection time , function detection time and overall change execution at the model level for every change request is given in the experimental analysis.

5 .Experimental analyses

The overall time taken for execution of a change in petri nets and finite state machines vary based on the following considerations. Overall execution time of the change using petri net is given using the following formula

Tot Tdt Rtt

Where Tot Time taken for token transition Tdt Time taken for transition decision Rtt Time taken for real transition

Whereas overall execution time of change using finite state machine is given using the following formula -

Rtt

Where Rtt Time taken for real transition As there are no tokens in finite state machines and no time is taken for decision making in a transition. In this case, directly state transition table is considered and the transition from one node to another node is made

108 M. Thirumaran et al. / IERI Procedia 4 ( 2013 ) 99 – 109

easily. Similarly the space taken by petri net and finite state machine during a change is given using the following formulae –

Space complexity For petri nets = Pi CR TOST SST TDST TLT

Where P Process i Process id

CR Change Request TO ST Storage taken for processes’ tokens

SST Storage taken for state and their details TDST Storage taken for transition and their details

TLT Storage taken for transition logging

Space complexity For finite state machine = BLi CR SST TLT

Where BL Business logic retrieved for change request i Process id

CR Change Request SST Storage taken for state and their details TLT Storage taken for transition logging

Apart from time complexity and space complexity, the time taken for detecting a rule within the model, the time

taken detecting a function, time taken for the model construction are also evaluated and shown in the following table. The overall experiment is done using Netbeans IDE6.9.0.

Table 5 Performance evaluation of FSM and Petri net

The above table gives performance evaluation analyzed for 30 change requests where RdtinFSM gives

rule detection time in FSM ( Finite state machine ) and RdtinPnet gives rule detection time in Petri net model. Similarly Fdt time is function detection time, Et is execution time of the change, Pct is Petri net construction time , Fct is finite state machine construction time and Sc is space complexity which is measured in terms of turing machine tapes. Since Petri net uses , direct processes, the number of tapes taken is very larger when compared the number of tapes used by finite state machine .

109 M. Thirumaran et al. / IERI Procedia 4 ( 2013 ) 99 – 109

6 . Conclusion

The work compares the advantages of using finite state machine over petri nets in web service change management frameworks. The change management model designed supports the business analysts directly in performing the changes instead of depending on the IT analysts. Hence a schema driven process is induced for managing changes with respect to the logics of the services instead of considering the business process workflows of the services. When the process is schema driven, it is analyzed that finite state machine is more efficient than Petri nets because of their sequential processing nature. A change management scenario is built using netbeans IDE and changes are incorporated into the logics via both Petri net and finite state machine. A performance analysis of these two conceptual models are made based on time taken for execution of changes , time taken for construction of the models , time taken for rule detection , time taken for function detection and space used by the models during change execution . From the evaluation of the objective factors, it is inferred that finite state machine suits well than petri nets for the change management model, as emergency changes need to handled in an efficient way in less time.

References

[1] Xumin Liu, Athman Bouguettaya, Jemma Wu, and Li Zhou, “Ev-Lcs: A System For The Evolution Of Long-Term Composed Services,” IEEE Transactions on Services Computing,Vol. PP, Issue 99, 2011. [2] Dimitris Apostolou , Gregoris Mentzas,Ljiljana Stojanovic, Barbara Thoenssen, Tomás Pariente Lobo,”A collaborative decision framework for managing changes in e-Government services,” Government Information Quarterly 28 (2011) 101–116, PUBLISHED @ ELSEVIER, 2011. [3] Bruno Wassermann, Heiko Ludwig, Jim Laredo, Kamal Bhattacharya, Liliana Pasquale, “Distributed Cross - Domain Change Management,” 2009 IEEE International Conference on Web, Services, 978-0-7695-3709-2/09, 2009 IEEE. [4] Bassam Atieh Rajabi, Sai Peck Lee, “Change Management in Business Process Modeling Survey”, 2009 International Conference on Information Management and Engineering, 978-0-7695-1/09, 2009 IEEE. [5] Bassam Atieh Rajabi, Sai Peck Lee, “Runtime Change Management Based on Object Oriented Petri Net,” 2009 International Conference on Information Management and Engineering, 978-0-7695-1/09,2009 IEEE. [6]Haochen Li, Zhiqiang Zhan, “Bussiness-Driven Automatic IT Change Management Based on Machine Learning,” 978-1-4673-0269-2/12, 2012 IEEE. [7] Yanhua Du, Xitong Li, and PengCheng Xiong, “Petri Net Approach to Mediation-Aided Composition of Web Services,” IEEE Tansactions on Automation Science and Engineering, Vol. 9, No. 2, Apr. 2012. [8] Sabri MTIBAA and Moncef TAGINA, “Managing Changes in Citizen-Centric Healthcare Service Platform using High Level Petri Net”, (IJACSA) International Journal of Advanced Computer Science and Applications, Vol. XXX, No. XXX, 2011. [9] Katalina Grigorova, “Process Modelling using Petri Nets”, International Conference on Computer Systems and Technologies – CompSysTech 2003. [10] S. Mtibaa and M. Tagina, “An Automated Petri-Net Based Approach for Change Management in Distributed Telemedicine Environment”, Journal of Telecommunications, Volume 15, Issue 1, July 2012.