A Guidance System for Business Process Flexibility

Asma Mejri

, Sonia Ayachi Ghannouchi

1,2

and Ricardo Martinho

3,4

Laboratory RIADI-GDL, ENSI, Mannouba 2010, Tunisia

High Institute on Management of Sousse, Tunisia

School of Technology and Management, Polytechnic Institute of Leiria, Portugal

Center for Health Technology and Services Research (CINTESIS), Portugal

Keywords:

Flexibility, Guidance, BPM, Emergency Care Process, BPMS, Paradigm.

Abstract:

During the last decades, ﬂexibility has gained a strong presence, in a variety of disciplines, mainly in the

BPM ﬁeld. The real challenge for BPM consists in providing modeling paradigms and BPMSs with adequate

information and features to deal with the often conﬂicting requirements of ﬂexibility. In this setting, we focus

on providing a guidance approach for enhancing business process ﬂexibility. Our purpose is therefore to

perceive which modeling paradigm(s) and/or business process management system(s) (BPMS(s)) are the most

adequate to the speciﬁc organization needs in terms of ﬂexibility. This approach was implemented in a plug-in

named BPFlexGuide. To evaluate this approach, we have studied the emergency care (EC) process. Users

interested in the EC process were guided to use the AristaFlow BPM suite BPMS. The results of this study

would help designers to choose the best paradigms and BPMS that best ﬁt their needs on ﬂexibility.

1 INTRODUCTION

In the past decades, ﬂexibility has gained a strong

presence in various ﬁelds, notably in the BPM dis-

cipline. Flexibility is a must, because continuously

changing conditions compel organizations to rapidly

and ﬂexibly adapt their business processes (BPs).

Thus, the real challenge for BPM consists in provid-

ing modeling paradigms and BPMSs with adequate

information and features to deal with the often con-

ﬂicting requirements of ﬂexibility. Indeed, several

paradigms have emerged.

In this paper, we are going to select the most pop-

ular paradigms which are the rule-based, constraint

based, case handling and adaptive process manage-

ment. The adaptive paradigm represents one of the

paradigms that enable users to make structural pro-

cess changes to support the handling of exceptions,

the evolution of BPs and user support in exceptional

situations (Schonenberg et al., 2008). A constraint-

based paradigm makes it possible to execute both al-

lowed and optional scenarios in BPs (Pesic, 2008).

A paradigm is called rule-based if the logic of its

control ﬂow, data ﬂow and resource allocation is ex-

pressed by means of business rules, which are recog-

nized as powerful representation forms that can po-

tentially deﬁne the semantics of BP models and busi-

ness vocabulary (Sun et al., 2006). The case han-

dling paradigm supports ﬂexible and knowledge in-

tensive BPs (Van der Aalst et al., 2005). Each of

these paradigms has its speciﬁc characteristics which

are adjusted to speciﬁc organization’s needs.

On the other hand, users have to be aware of how

their ﬂexibility needs can be addressed by the afore-

mentioned paradigms, and then which BPMSs bet-

ter match these (often combined) paradigms to exe-

cute and manage BP models and corresponding in-

stances. Moreover, process engineers, who often de-

cide which BPMS to adapt to their organizations, do

not know upfront most known BPMSs nor their (ﬂexi-

bility) characteristics. Even if they could collect some

information from the Internet it would be very difﬁ-

cult to come up with a scoring mode that could easily

perform a match between their BPs’ ﬂexibility needs

and the best choice for a BPMS. We propose in this

paper the BPFlexGuide guidance system for choosing

the best-suited BPMSs and BP modeling paradigm to

an organization’s BPs.

With our system, process engineers will be able

to input their organization’s needs in terms of BP

ﬂexibility, and take a ranked and properly scored list

of recommended BPMSs and modeling paradigms.

They can also beneﬁt, with our system, from fully

characterized BPMSs in terms of ﬂexibility, which

will allow them to analyze and improve decision mak-

ing regarding particular characteristics that can only

be observed in some of their BPs. To accomplish

these main objectives of our BPFlexGuide guidance

210

Mejri, A., Ghannouch, S. and Martinho, R.

A Guidance System for Business Process Flexibility.

DOI: 10.5220/0006316902100217

In Proceedings of the 12th International Conference on Evaluation of Novel Approaches to Software Engineering (ENASE 2017), pages 210-217

ISBN: 978-989-758-250-9

system, we performed the following main research

tasks:

• Compare existing BPMSs regarding ﬂexibility:

This includes weighting the different ﬂexibility

criteria;

• Measure the ability of existing BP modeling

paradigms to deal with ﬂexibility;

• Implement a system that guides the users to

choose the most appropriate BPMS and modeling

paradigms;

• Evaluate the system using a ﬂexible process (in

healthcare domain).

The remainder of this paper is organized as follows.

Section 2 presents basic concepts, including ﬂexibil-

ity in BPM, a summary on Regev et al.’s BP ﬂexibil-

ity taxonomy and most relevant related work. Our re-

search method is described in Section 3 and in section

4 we reveal our main elements and artifacts derived

from this method. The BPFlexGuide implementation

details with ProM are presented in Section 5, and in

Section 6 we apply our approach to a case study. Sec-

tion 7 concludes the paper.

2 RELATED WORK

Guidance addresses the issue of guiding either users

and/or developers, practitioners and/or academicians,

and/or modelers during the different steps of the BPM

life-cycle. The concept of guidance has hence been

researched in the BPM community in the context of

proposing principles and guidelines, providing rec-

ommendation systems or decision support systems.

In (Mendling et al., 2010), authors propose a set of

seven process modeling guidelines. These guidelines

directly aim to bring support to process modelers. In

(Thomas et al., 2014), authors propose a set of ten

principles that characterize BPM as a research domain

and guide its successful use in organizational practice.

On the other hand, authors propose recommen-

dation systems, in (Barba et al., 2013), (Schonen-

berg et al., 2008), (Setiawan et al., 2011), (Conforti

et al., 2015), (Huang et al., 2012), (Koschmider et al.,

2011), (Mertens et al., 2014). For example, in (Barba

et al., 2013), the recommendation system is based on

a constraint-based approach for planning and schedul-

ing the BP activities and considers both the control-

ﬂow and the resource perspective. Recommendations,

in (Schonenberg et al., 2008), are given in order to as-

sist users in selecting activities, during process exe-

cutions, by giving recommendations on possible next

steps.

Authors, in (Conforti et al., 2015), propose a rec-

ommendation system that supports process partici-

pants in taking risk-informed decisions, with the goal

of reducing risks that may arise during process execu-

tion. A system for supporting users at modeling time

was presented in (Koschmider et al., 2011). It pro-

vides information that facilitates the decision for the

right recommendation which is based on the user pro-

ﬁle. Consequently, users can follow up recommenda-

tions that are ranked, based on different criteria. In

order to provide guidance to users during process ex-

ecution, a recommendation system was proposed in

(Mertens et al., 2014). Their main contribution con-

sists of a robust process engine that can deal with

changes that are held in very dynamic and complex

environment, at run-time.

Other research efforts were engaged in developing

decision support systems. For instance, authors, in

(Cingil et al., 2012), have developed a decision sup-

port system that includes a relational database to keep

the predeﬁned relations, its descriptions, the inputs

for weights, its grades and calculated scores. Besides,

authors have proposed in (Yao and Kumar, 2013) a

clinical decision support system that provides deci-

sion support, in order to get adaptable clinical path-

ways. These pathways are selected during BPs exe-

cution and are based on rules that encapsulate med-

ical knowledge and patient conditions that are con-

stantly changing. In most of these approaches, the

focus on ﬂexibility and user assistance can be ob-

served. But, they do not support, in a straightforward

way, BP ﬂexibility. Moreover, the need for increased

attention to ﬂexibility, the competing paradigms and

the wide variety of BPMSs has been recognized by

both academia and industry. Though, to the best

of our knowledge, none of the research works di-

rectly targets the development of a guidance system

that takes into account ﬂexibility and the multitude of

paradigms and BPMSs. Upon the recognition of such

scarcity, there is an urgent need to propose a guid-

ance approach and develop the corresponding tool in

order to guide users to choose the most convenient

paradigms and BPMSs, according to their particular

needs in terms of ﬂexibility. An overview on our guid-

ance approach will be the main focus of the next sec-

tion.

3 PROPOSED TWO-STEP

RESEARCH APPROACH

In this section we describe our research approach,

which is based on two steps: (1) a general classiﬁca-

tion of BPMSs and process modeling paradigms; (2)

A Guidance System for Business Process Flexibility

211

the proposal of a guidance procedure that foresees in-

puts from users regarding their needs in terms of BP

ﬂexibility, and later guiding them to choose the best-

suited BPMS for their organizations.

Figures 1 and 2 present the main activities devel-

oped within these major two steps. These activities

can be summarized as follows:

• Classiﬁcation of BPMSs

– Identiﬁcation of ﬂexibility criteria: we began

by identifying a set of criteria that we derived

from Regev et al.’s taxonomy, in order to eval-

uate ﬂexibility within BPMSs. From this activ-

ity we have speciﬁed eleven Flexibility Criteria

(FC) (detailed in the next section);

– Development of the questionnaire ”‘Evalua-

tion of Business Process Management Systems

(BPMS)”’: here, we developed a questionnaire

for BPMS developers and researchers to an-

swer, based on their advanced knowledge of

the BPMSs they develop and research upon.

We also selected these BPMSs taking into ac-

count their prominence in literature, as well

as their ability to support BP ﬂexibility. We

could then analyse the answers to the question-

naires and classify BPMSs accordingly, regard-

ing their support on ﬂexibility and modeling

paradigms;

• Proposal of a guidance procedure

– User’s BP ﬂexibility needs: here we have to

identify ﬂexibility needs from the users of our

BPFlexGuide system. This step implied efforts

of mapping the well-known ﬂexibility taxon-

omy of Regev et al. into a set of questions to

be understood by general users without a pro-

found knowledge of BP ﬂexibility;

– Similarity study: here we compare the obtained

answers from the users’ needs questionnaire

and the previous classiﬁcation of BPMSs, in or-

der to derive a score in terms of ﬂexibility;

– Speciﬁc classiﬁcation: we then classify and

sort the BPMSs and modeling paradigms with

an algorithm (detailed in the next section),

and present the associated recommendations to

users.

In the next section, we provide the results and cal-

culations derived from each of these research activi-

ties.

Figure 1: Overview of step 1 of our research method.

Figure 2: Overview of step 2 of our research method.

4 BASIC ELEMENTS OF OUR

TWO-STEP RESEARCH

APPROACH

We present in this section the main elements on which

our research is based. These include the identiﬁcation

of ﬂexibility criteria for evaluating BPMSs, the de-

sign of the questionnaire and obtained answers, and

the calculations and developed algorithms to assess

the BPMSs to users’ needs regarding BPs ﬂexibility.

4.1 BP Flexibility Criteria

We began by identifying a set of criteria that we de-

rived from Regev et al.’s taxonomy, in order to eval-

uate ﬂexibility within the selected BPMSs. We have

speciﬁed eleven Flexibility Criteria (FC), which con-

cern the following questions:

• FC1: To which extent do the BPMS modelers de-

scribe the process control ﬂow?

• FC2: To which extent does the BPMS support de-

scriptive modeling and execution of process activ-

ENASE 2017 - 12th International Conference on Evaluation of Novel Approaches to Software Engineering

212

ities?

• FC3: To which extent does the BPMS support de-

scriptive modeling and execution of the precondi-

tions of the activities?

• FC4: To which extent does the BPMS sup-

port descriptive modeling and execution of

data/information exchanged between process ac-

tivities?

• FC5: To which extent does the BPMS support de-

scriptive modeling and execution of roles associ-

ated to process activities?

• FC6: Does the BPMS support changes to pro-

cess models which will affect all new process in-

stances?

• FC7: Does the BPMS support changes at the in-

stance level, and that will only affect certain se-

lected instances, in order to accommodate excep-

tional situations?

• FC8: Can the BPMS support incremental change

and/or revolutionary change?

• FC9: How would the duration of change that

the BPMS support be characterized: temporary

and/or permanent?

• FC10: Is the BPMS able to deal with immediate

and/or deferred change?

• FC11: Can the BPMS support planned / ad-hoc

changes?

It is important to mention that all FCs have the same

weight. We have speciﬁed for each FCs a scale in or-

der to get consistent results. FC1, FC2, FC3, FC4 and

FC5 range from 0 (not descriptive) to 5 (very descrip-

tive). FC6 and FC7 can be 1 (yes) or 0 (No). FC8,

FC9, FC10 and FC11 can take 1 (if it satisﬁes one

of the characteristics describing this criteria) , 0 (if it

doesn’t satisfy none of the characteristics) or 2 (if it

doesn’t satisfy both characteristics).

4.2 Questionnaire for BPMS Providers

The questionnaire has been designed in order to

evaluate ﬂexibility in BPMSs from a provider point

of view. The questionnaire was speciﬁcally de-

signed to seek responses from the most senior person-

nel responsible for the development of the selected

BPMSs. We had it ﬁlled out by eight senior devel-

opers/researchers that provides us with answers for

eight BPMSs (or modeling tools): DECLARE, ES-

ProNa, Jrules, AristaFlow BPM suite, Philhamon-

icFlows, jBPM, ProdProc and FLOWer. Most of the

BPMSs from which we got answers were developed

in the context of research projects. Others were de-

veloped by professional developers from the industry.

Table 1 contains the responses for each BPMSs. Ac-

cording to the questionnaire’s results, Table 2 summa-

rizes the answers to the following question: “<What

is (are) your BPMS(s) underlying modeling/execution

paradigm(s)?”> From Table 2, we can conclude that,

from the selected BPMSs, most of them support

constraint-based modeling (5 out of 8) and also 5 out

of 8 are mono-paradigm oriented. We can also ob-

serve that 2 out of 3 multi-paradigm systems combine

the constraint and rule based paradigms.

4.3 Match between Users’ BP

Flexibility Needs and BPMSs’

Announced Flexibility

We aim at comparing the elements of the users’ needs

in terms of ﬂexibility in BPs and the elements of the

responses of the questionnaire presented in Table 1.

Our study is based on a similarity metric in order to

measure distance, more speciﬁcally the overlap met-

ric. According to (Raj Kumar, 2012), the overlap met-

ric simply tests for equality between two values, so

that different values get distance 1 whereas equal val-

ues get distance 0:

Deﬁnition the overlap distance

overlap

(x, y) = 1 when x 6= y

overlap

(x, y) = 0 when x = y

4.4 Algorithm of Speciﬁc Classiﬁcation

of BPMSs

In order to calculate this similarity score for each

BPMS, we follow the methodology that includes the

following steps: (1) Compare the users’ ﬂexibility cri-

teria with the different BPMS ﬂexibility criteria. (2)

Attribute a calculated similarity score for each crite-

rion. The score is 0 or 1, where 0 indicates no similar-

ity and 1 indicates identical elements. (3) Sum all the

criteria to obtain a similarity score SIM for the con-

sidered BPMS. The next step consists in comparing

the calculated scores for each BPMS in order to sort

them.

Algorithm 1. We consider two sets S and U. S is

the set of the different BPMSs. Each S

consists of

the values taken by the ﬂexibility criteria for a given

BPMS. U is the set of the criteria entered by the user.

Let n be the number of studied systems, NCR be the

number of studied ﬂexibility criteria.

Furthermore let S = (S

, S

, ..., S

)

U = (FCU

, FCU

, ..., FCU

NCR

)

A Guidance System for Business Process Flexibility

213

Table 1: Responses of the questionnaire for each BPMS.

DECLARE ESProNa JRules AristaFlowBPM suite PHiharmonicFlows ProdProc jBPM FLOWer

FC1 2 3 4 4 2 2 5 1

FC2 5 3 4 5 3 4 5 2

FC3 2 4 3 3 5 4 3 2

FC4 2 3 1 5 5 3 5 2

FC5 2 3 2 5 4 3 3 2

FC6 0 0 1 1 1 1 1 1

FC7 1 1 1 1 1 0 1 1

FC8 1 1 1 2 1 0 2 2

FC9 1 1 1 2 1 1 1 1

FC10 1 1 1 2 1 1 1 1

FC11 1 2 1 2 1 0 2 1

Table 2: BPMS classiﬁcation according to their modeling paradigms.

BPMS / Paradigm Constraint based Rule based Case handling Adaptive Process management Total

DECLARE yes yes 2

EsProNa yes 1

Jrules yes yes 2

AritaFlow BPM suite yes 1

Philharmonic Flows yes 1

ProdProc yes 1

jBPM yes 1

Flower yes yes 2

Total 5 3 2 1

= (FC

, FC

, ..., FC

NCRi

We deﬁne the similarity score for each system as

follows: SIM

∑

NCR

j=1

sim

i j

where sim

i j

= d

overlap

(FCU

, FC

) Next, we do

the sorting of the S

according to the calculated simi-

larities. SORT

= sort

{

SIM

}

where i ∈ [1, n]

4.5 Algorithm of Classiﬁcation of

Paradigms

We consider the following steps in order to calculate

the paradigms ﬂexibility measurement score for each

paradigm:

• Calculate for each BPMS the similarity score

(SIM) (as done in the algorithm for classiﬁcation

of BPMSs).

• Sum the similarity scores if the BPMS belongs to

the paradigm.

• Compare the ﬂexibility measurement scores of the

different paradigms.

Algorithm

Algorithm 2. Here we consider an additional set P of

considered paradigms. Let n be the number of studied

systems, m the number of studied paradigms, NCR be

the number of studied ﬂexibility criteria.

Furthermore let S = (S

, S

, ..., S

)

U = (FCU

, FCU

, ..., FCU

NCR

)

= (FC

, FC

, ..., FC

NCRi

P = (P

, P

, .., P

Each S

belongs to one or more paradigms. We

deﬁne the paradigms’ ﬂexibility measurement score

(PFMS) for each paradigm as follows:

PFMS

∑

k=1

whereSS



SIM

i f S

∈ P

0 otherwise



(1)

Next, we do the sorting of the S

accord-

ing to the calculated similarities. SORT

sort



PFMS



where j ∈ [1, m]

5 IMPLEMENTATION

The approach conducted in this research work is sup-

ported and tested by a corresponding tool implemen-

tation. Therefore, we did not develop a tool from

scratch. Whereas, we expanded it upon an existing

open-source solution which is the ProM tool, in its

latest version 6.4 (Verbeek et al., 2010a).

The starting point is a spreadsheet-based (MSEx-

cel) document that contains the results of the ques-

ENASE 2017 - 12th International Conference on Evaluation of Novel Approaches to Software Engineering

214

Figure 3: BPFlexGuide architecture.

tionnaire (which were presented in Tables 1 and 2). In

order to allow for further implementation in the ProM

tool, it has to be transformed to a data structure that

conforms to the eXtensible Event Stream (XES) for-

mat (Verbeek et al., 2010b). XES is an XML-based

format for storing structured data such as event logs

and the standard input format for ProM (as of Version

6). The conversion is performed in the ﬁrst stage us-

ing Disco1: another import and data-exchange ProM

plugin (G

unther and Rozinat, 2012). As shown in

Figure 3, this XES document serves as an input to

our BPFlexGuide. When starting BPFlexGuide, an

internal repository of all the BPMSs’ ﬂexibility crite-

ria is constructed. This is done when extracting the

BPMSs’ ﬂexibility criteria for each BPMS from the

XES document. The BPMSs’ ﬂexibility criteria have

been implemented as classes. The conﬁguration of all

the BPMSs and the paradigms is then conducted.

Next, the user can pick, from a list, the ﬂexibility

criterion which best ﬁts her/his needs on ﬂexibility.

The provided user interface allows conﬁguring user

needs on ﬂexibility (based on the taxonomy of Regev

et al. (Regev et al., 2006)). Her/his choices are then

stored.

We have implemented the algorithms presented in

the previous section. The result of the execution of

these algorithms is the ranking of the different BPMSs

and of the modeling paradigms. Finally, these results

are shown to the user. In case the same ranking is

obtained for two BPMSs, the user picks one of them.

6 CASE STUDY RESULTS

Adopting BPM technology in the healthcare sector

is starting to address some of the characteristics of

healthcare processes, including their high degree of

ﬂexibility (Lenz and Reichert, 2007). Hence, health-

care domain requires high ﬂexibility (Kirchner et al.,

2013). Therefore, this section shows the result of the

BPFlexGuide plugin applied to the EC process. The

goal of the case study was to elicit ﬂexibility needs of

an EC process and to help users to select most ade-

quate paradigms and BPMSs that best suit the needs

in terms of ﬂexibility.

6.1 Eliciting Flexibility Needs in the

Emergency Care Process

In this work, we are going to consider the EC process

which initiates with the patient registration and ends

with her/his discharge of the emergency department

of Farhat Hached hospital. The EC process was ana-

lyzed and modeled at ﬁrst in (Kirchner et al., 2013),

using the BPMN language.

The main technique used to elicit ﬂexibility needs

on the EC process was interviews. The target group of

the respondents of these interviews was professionals

who are working in the EC process. Table 3 resumes

the answers of the respondents towards the taxonomy

of Regev et al.. Regarding the abstraction level of

modeling, when asking physicians this question ”¡Do

changes affect all the patients or only one patient?”¿.

They held that the changes concern neither only one

patient nor all the patients. Regarding the dimensions,

our study focused on the ﬁve known dimensions de-

ﬁned in the Regev et al.’s taxonomy which are the

functional, operational, behavioral, informational and

organizational dimensions. Interviewees believe that

it would be useful to know about the functional per-

spective. For instance, it would be important to intro-

duce or remove some activities of the EC process.

Concerning the operational dimension, intervie-

wees think that a change in the process level does not

require the review of the implementation of applica-

tions of information technology that are used in the

emergency department. Respondents answered that

changing the order of activities can affect changes in

the EC process. Thus, they afﬁrm that it is very useful

to focus on the behavioral perspective when modeling

the EC process. For the informational perspective, re-

spondents believe that it would be very important that

changes concern the emergence of new types of in-

formation or data that can be exchanged between the

different actors involved in the EC process. Concern-

ing the organizational perspective, interviewees see

that it would be useful that the changes also concern

the roles associated to process activities. Concern-

ing the properties of change, according to intervie-

wees, changes in the model concern the entire model

in some cases. For instance, the model could be to-

tally changed through the effective implementation of

A Guidance System for Business Process Flexibility

215

Table 3: Flexibility needs of the EC process, according to

Regev et al.’s taxonomy.

Criteria Responses

FC1 2

FC2 0

FC3 3

FC4 4

FC5 3

FC6 0

FC7 0

FC8 2

FC9 1

FC10 1

FC11 2

the disaster plan.

Changes can also concern a part of the model.

This can occur, for example, in the cases of conta-

gious or mental diseases. Regarding the duration of

change, changes in the EC process should be tempo-

rary. They should occur in a limited interval of time

(even for months). Because of patient’s needs of im-

mediate care, the changes should be immediate. They

should not be delayed in time. In addition, intervie-

wees believe that the plans must be developed and

must be planned well in advance. However, actions

can occur at unexpected times. To sum, regarding

the anticipation of change, both planned and ad-hoc

change has to be provided.

6.2 Applying our Guidance Approach

to the EC Process

After eliciting the roles and causes for changes in

the EC process, and deriving the associated ﬂexibility

needs, we used this information to advise on a possi-

ble BPMS/tool to be used. Our BPFlexGuide offers a

classiﬁcation of the selected paradigms: constraint-

based, rule-based, case handling and adaptive pro-

cess management. This classiﬁcation could be gen-

eral (i.e. taking into account all the dimensions) or

speciﬁc (taking into account one of the three dimen-

sions of the taxonomy of Regev et al.). Regarding the

speciﬁc classiﬁcation, the constraint-based paradigm

was retained as the ﬁrst one that is the most suitable

for the EC process users’ needs which are presented

is the previous section. The second paradigm is the

rule-based paradigm. The case handling paradigm

was held as the third paradigm. Finally, the adaptive

process management paradigm was classiﬁed as the

last one.

Using the BPFlexGuide plugin, we have entered

the ﬂexibility needs of the EC process. The goal of

the case study was to help process participants and de-

cision makers to choose the most appropriate BPMS

that best ﬁt their needs on ﬂexibility.

The plug-in provided, as output, a ranking of the

most suitable BPMSs that can be adopted. Speciﬁ-

cally, the BPMSs had also been ranked according to

the three dimensions of the taxonomy of Regev et al..

We notice that the different BPMSs got various rank-

ings. As a result, EC process modelers were guided to

use the ADEPT2/ AristaFlow BPM suite BPMS ﬁrst,

jBPM and ESProNa second. In this work, we model

the EC process using the AristaFlow BPM suite. The

AristaFlow BPM suite allows thus for a high degree

of ﬂexibility during process execution. It provides a

complete set of high-level change patterns for deﬁn-

ing ad-hoc deviations. Users may dynamically add

new activities or jump forward in the ﬂow of control.

7 CONCLUSIONS AND FUTURE

WORK

We provided in this paper a framework for guid-

ing users to choose the most appropriate BPMSs and

paradigms that ﬁt best their needs on ﬂexibility. In

this sense, it can be used to help practitioners (i.e. pro-

cess analysts, designers, engineers and users) and aca-

demics to select the BPMSs and paradigms according

to their BP ﬂexibility needs.

The proposed approach included the use of a ques-

tionnaire for BPMS providers, the classiﬁcation of

these BPMSs according to a ﬂexibility taxonomy,

as well as the BPFlexGuide tool that could suggest

matches between ﬂexibility needs of a process and

the most suited BPMSs. We validated our approach

by choosing the EC process. The reason for choosing

this kind of BP is in fact due to its important need of

ﬂexibility.

The obtained results from our BPFlexGuide tool

for this case study indicated that the constraint based

paradigm is the most adequate modeling paradigm,

and the recommended BPMS was the AristaFlow

BPM suite. Consequently, we modeled the EC pro-

cess with AristaFlow BPM suite BPMS. We consider

that the obtained model offers in fact a high degree of

ﬂexibility.

We can conclude that these evaluation results are

then promising. The advantages of such an approach

consist in providing better quality models and more

precisely models enabling ﬂexibility. Although vali-

dation of our approach was achieved by using a ﬂex-

ible emergency care process, we intend to apply it to

other BPs and organizations with different ﬂexibility

needs. Examples include, for instance, organizations

from the engineering and industry ﬁelds or higher ed-

ENASE 2017 - 12th International Conference on Evaluation of Novel Approaches to Software Engineering

216

ucation domains. These may imply including other

ﬂexibility needs that are not covered by the BP ﬂexi-

bility taxonomy that we adopted in this proposal.

Many issues are still open and can be subject

of future work. First, we are planning to general-

ize the use of the BPFlexGuide tool also to BPMS

providers, in order to allow them to fully character-

ize their BPMSs through the use of the tool. Inte-

grating other dimensions into the provided process

guidance is technically possible. Besides, it would

be more adequate to accept intermediate values be-

tween 0 and 1. This could be ameliorated using fuzzy

logic. The challenges here include not to overload

BPMS providers and users with ﬂexibility criteria,

and also to score BPMSs according to different BP

ﬂexibility taxonomies. Moreover, we plan to pro-

pose a post-validation approach for the BPMS rec-

ommended by our BPFlexGuide tool. This will im-

ply measuring the ﬂexibility of BPs modeled and ex-

ecuted in that recommended BPMS, by counting the

number of changes that a user needs to perform in

those processes (using that BPMS) to achieve the de-

sired ﬂexibility.

REFERENCES

Barba, I., Weber, B., Del Valle, C., and Jim

enez-Ram

ırez,

A. (2013). User recommendations for the optimized

execution of business processes. Data & Knowledge

Engineering, 86:61–84.

Cingil, I., Ozturan, M., and Erdem, A. S. (2012). A deci-

sion support system for evaluation of business process

management systems. International Information In-

stitute (Tokyo). Information, 15(2):537.

Conforti, R., de Leoni, M., La Rosa, M., van der Aalst,

W. M., and ter Hofstede, A. H. (2015). A recom-

mendation system for predicting risks across multiple

business process instances. Decision Support Systems,

69:1–19.

unther, C. W. and Rozinat, A. (2012). Disco: Discover

your processes. BPM (Demos), 940:40–44.

Huang, Z., Lu, X., and Duan, H. (2012). Using recommen-

dation to support adaptive clinical pathways. Journal

of medical systems, 36(3):1849–1860.

Kirchner, K., Herzberg, N., Rogge-Solti, A., and Weske,

M. (2013). Embedding conformance checking in a

process intelligence system in hospital environments.

In Process Support and Knowledge Representation in

Health Care, pages 126–139. Springer.

Koschmider, A., Hornung, T., and Oberweis, A. (2011).

Recommendation-based editor for business pro-

cess modeling. Data & Knowledge Engineering,

70(6):483–503.

Lenz, R. and Reichert, M. (2007). It support for healthcare

processes–premises, challenges, perspectives. Data &

Knowledge Engineering, 61(1):39–58.

Mendling, J., Reijers, H. A., and van der Aalst, W. M.

(2010). Seven process modeling guidelines (7pmg).

Information and Software Technology, 52(2):127–

136.

Mertens, S., Gailly, F., and Poels, G. (2014). Generating

business process recommendations with a population-

based meta-heuristic. In Business Process Manage-

ment Workshops, pages 516–528. Springer.

Pesic, M. (2008). Constraint-based workﬂow management

systems: shifting control to users. PhD thesis, Tech-

nische Universiteit Eindhoven.

Raj Kumar, R. V. (2012). Classiﬁcation algorithms for data

mining: A survey. International Journal of Innova-

tions in Engineering and Technology (IJIET).

Regev, G., Soffer, P., and Schmidt, R. (2006). Taxonomy of

ﬂexibility in business processes. In BPMDS.

Schonenberg, H., Weber, B., Van Dongen, B., and Van der

Aalst, W. (2008). Supporting ﬂexible processes

through recommendations based on history. In Busi-

ness process management, pages 51–66. Springer.

Setiawan, M. A., Sadiq, S., and Kirkman, R. (2011). Fa-

cilitating business process improvement through per-

sonalized recommendation. In Business Information

Systems, pages 136–147. Springer.

Sun, X., Liu, X.-Z., Jiao, W.-P., Huang, G., and Mei, H.

(2006). A rule-based approach to supporting adapt-

able web service composition. CHINESE JOURNAL

OF COMPUTERS-CHINESE EDITION-, 29(7):1084.

Thomas, K., Oliver, M., Jens, P., Maximilian, R., vom

Brocke, J., Schmiedel, T., Recker, J., Trkman, P.,

Mertens, W., and Viaene, S. (2014). Ten principles

of good business process management. Business Pro-

cess Management Journal, 20(4):530–548.

Van der Aalst, W. M., Weske, M., and Gr

unbauer, D.

(2005). Case handling: a new paradigm for busi-

ness process support. Data & Knowledge Engineer-

ing, 53(2):129–162.

Verbeek, H., Buijs, J. C., Van Dongen, B. F., and Van

Der Aalst, W. M. (2010a). Xes, xesame, and prom

6. In Information Systems Evolution, pages 60–75.

Springer.

Verbeek, H. M. W., Buijs, J. C. A. M., van Dongen, B. F.,

and van der Aalst, W. M. P. (2010b). Prom 6: The

process mining toolkit. In Proc. of BPM Demonstra-

tion Track 2010, volume 615, pages 34–39. CEUR-

WS.org.

Yao, W. and Kumar, A. (2013). Conﬂexﬂow: Integrating

ﬂexible clinical pathways into clinical decision sup-

port systems using context and rules. Decision Sup-

port Systems, 55(2):499–515.

A Guidance System for Business Process Flexibility

217