Systemic Business Process Simulation using Agent-based Simulation

and BPMN

Jácint Duduka

1,2

and Sérgio Guerreiro

1,3

INESC-ID, Lisbon, Portugal

Universidade Aberta, Lisbon, Portugal

Instituto Superior Técnico, University of Lisbon, Lisbon, Portugal

Keywords: Business Process Simulation, Agent-based Modelling, BPMN, Socio-technical Systems, Distributed

Simulation.

Abstract: The current paradigm of a business process model is that it is a representation of a sequence of tasks that act

upon some data input, to produce an output, aiming the production of a new service or product to be delivered

from a producer to a customer. Although this is a valid way of thinking, it neglects to consider in enough

detail the influence of some phenomenon on inputs, e.g. human behaviour, communication, social interactions,

the organisational culture which can have a significant effect on the output delivered by a business process.

As the dynamics of these phenomena are non-linear, they can be interpreted as a complex system. This holistic

way of thinking about business processes opens the doors to the possibility of combining different simulation

methods to model different aspects that influence a process. A BPMN engine and an agent-based simulation

(ABS) engine are chosen to serve the basis of our framework. In its conception, we not only consider the

technical aspects of the framework but also delve into exploring its management and organizational

dimensions, with the intent of facilitating its adoption in enterprises, as a tool to support decision support

systems. We analyse how accurate the simulation results can be when using these two tools as well as what

considerations need to be considered within organizations.

1 INTRODUCTION

Today rapid technological change is being driven by

the information revolution, as we live in

environments that are increasingly technology-

saturated (Kadar et al., 2015). This saturation makes

the question of the relationship between people and

technology more explicit than ever, to the extent that

this relationship is widely reported and extensively

studied in the literature in the domain of socio-

technical systems (Bider, n.d.; Gregoriades &

Sutcliffe, 2008; Henda et al., 2016; Ibl & Čapek,

2017; Norta et al., 2014; The-Evolution-of-Socio-

Technical-Systems-Trist.Pdf, n.d.; Tropmann-Frick

& Thalheim, 2015; Vespignani, 2012). Socio-

technical systems are an approach to the

understanding and design of complex organisations

and technologies that recognise the interplay between

people and technology (Kloeckner & Birkmeier,

2010).

Despite the realisation of the importance of

humans in business processes, as far as we know,

there has been little focus on how agent-based

simulators(ABS) can be used to enable business

process simulation in enterprises. The majority of the

studies (Haiyan Zhao & Jian Cao, 2007; Halaška &

Šperka, 2018; Liu & Iijima, 2015; Sulis & Di Leva,

2018; Tan et al., 2009) focus on integrating ABS with

discrete event simulation(DES), Petri-nets and other

workflow engines. This choice can pose some

challenges for organizations due to extra investment

required to procure software, hire specialized

workforce with DES knowledge and time to convert

existing business processes to DES models. Although

this approach is suitable in some cases, it is less likely

to be adopted in organizations because of the time and

effort commitments it requires. Our assumption of

what constitutes a successful information system

implementation is based on the information systems

analysis framework depicted by Laudon & Laudon,

2013 that state that there should always be three

dimensions to any successful information system.

The first is management, where there should be

tasks performed at a management level of the

122

Duduka, J. and Guerreiro, S.

Systemic Business Process Simulation using Agent-based Simulation and BPMN.

DOI: 10.5220/0010177001220130

In Proceedings of the 12th International Joint Conference on Knowledge Discovery, Knowledge Engineering and Knowledge Management (IC3K 2020) - Volume 2: KEOD, pages 122-130

ISBN: 978-989-758-474-9

organization for the implementation of the system.

These include but not limited to reflecting about what

knowledge acquisition, retention strategies, training

strategies and budget plans are suitable for the

project. The second dimension is organization, where

they state that there should be a reflection on how

issues such as organizations hierarchy, functional

specialties, business processes, organizational culture

and pollical interest groups impact an information

system. Lastly there is the technology dimension,

where hardware, software, data management and

networking issues should be considered.

We observed that current simulation frameworks

involving ABS and BPS, do not take the management

and organization dimensions into account, focussing

more on technology, and therefore to solve this

problem, we ask the following question: "How to

implement a systemic simulation interoperability

framework, between an agent-based simulator and a

business process engine?"

2 SOLUTION PROPOSAL

To address the issue, we frame a solution that

approaches the problem from a holistic view. Some

authors also share the view of a holistic approach to

this issue (Wu, 2015), in the sense that holistic

modelling plays a vital role in developing socio-

technical systems(STS), because of the interplay

between social and technical elements within these

systems and the resulting emergent behaviour.

The current state of the art in the topic focusses

on two main categories. In the first category, we can

find solutions that try to conceive a concept

equivalency framework between ABM to business

process modelling notations or vice-versa.

Although

there is some success in doing this (Aksyonov &

Aksyonova, 2013; Dam et al., 2015; Endert et al.,

n.d.; Ghlala et al., 2017; Laroque et al., n.d.), they

agree that there will be concepts that are merely

difficult or even impossible to convert.

The second category consists of integrating ABS

with process simulation engines that do not take into

account the complexities occurring in enterprises,

such as budget limits, training, project deadlines,

skillset availability in the workforce. Between these,

we find mainly DES, Petri-nets and some other

generic process engines.

Instead, in our solution, we propose designing

agent-based models and business processes

separately and let the software agents drive the

business process engine as if they were real users. To

address the management and organizational

dimensions of our solution, we compared usage

trends of business process modelling languages, for

the past 16 years worldwide. Although the

comparison is not exhaustive of all languages, we

focussed on the main ones and collected data using

Google Trends.

Figure 1: Numbers represent search interest relative to the

highest point on the chart for the given region and time. A

value of 100 is the peak popularity for the term. A value of

50 means that the term is half as popular. A score of 0

means there was not enough data for this term. Source:

Google Trends. (2020). Retrieved September 22, 2020,

from https://trends.google.com/trends/explore?date=all&q

=%2Fm%2F08kq3d,%2Fm%2F01xc3f,%2Fm%2F01gt82

Choosing a highly available modelling language

is how we intended to fulfil the management and

organizational dimensions of our framework, the

assumption being that a highly adopted language

requires less investment to implement, less workforce

training, less time to convert models, and more

skillset reusability encourages collaboration. Given

the overall trend of BPMN, we chose it as our

business process modelling language.

2.1 The Simulation Process

Our engine of choice, Camunda, continues

functioning with the purpose for which it has been

designed, and is not aware of the change of process

actors, from real humans to software agents.

The same can be said for the software agents and

the ABS, which continue working and following the

rules defined in its model. Agents send messages to

the BPMN engine when specific pre-set criteria are

met via REST and are not aware of the purpose of

those messages.

This approach is different from the current

mapping approaches, in that it avoids any sort of

concept equivalency problems altogether because

models are not converted, they interact with each

other during the simulation runtime.

One of the challenges in conceiving such

Systemic Business Process Simulation using Agent-based Simulation and BPMN

123

integration between the two systems is that multiple

agent instances will be created during the simulation

process, as well as multiple process instances in the

BPMN engine. The consequence is that messages

may be routed to the wrong process instance if no

attention is paid to the way messages are transferred

between systems. Therefore, the question is: "How

does a system that has received a message know

which request this is the reply for?"

The solution concept we adopted is demonstrated

by Hohpe & Woolf, 2012, where it is suggested that

"the requestor add a Request ID to the request

message, have the replier copy the Request ID to the

Correlation ID field of the response message so that

the requestor can correlate the reply message to the

request message."

In our case, the "Agent ID" is used

as the identifier of the message.

This separation of concerns also has other

advantages, which for instance, opposed to BPMN

integration approaches proposed by some authors

(Onggo et al., 2017), this one does not suggest any

extension artefacts to the BPMN standard. The

ability to bypass these difficulties would significantly

enhance the process of creating better models because

the BPMN standard itself does not need to be

modified or extended in any way and no significant

investment of time is required to train staff in

organisations about features of new extensions,

instead, already existing tools are reused.

2.2 Scope of Work

The focus of our research is oriented towards

verifying functional aspects of the distributed

simulation method being proposed. It intends to

understand only aspects deemed fundamental for the

operation of such a way of simulating socio-technical

systems and ignores testing the broader context of

applicability in the industry. A detailed analysis of

those aspects is outside of the scope of this work, and

hence, they are only briefly outlined here.

3 METHODOLOGY

Our first objective was to determine whether it is

possible to perform web requests from an ABS engine

to a BPMN engine using the REST protocol. A

review was conducted to understand what support is

provided within the leading agent-based simulators

for performing web requests. Then one engine is

chosen based on the effort required to implement

models and how easily it can be scaled. The rationale

behind this criteria is to find an ABS engine for our

work, that has potential to study social systems of a

large size, with low investment requirements to

develop those models, relative to other engines.

The second objective was to understand if time

intervals between events occurring in the ABS are

conserved in BPMN. This was mainly an essential

step of our research as it has been highlighted several

times in the literature (Baker, n.d.; Brodsky, n.d.; Lin

& Guo, 2010; Tolk, 2013), that time management is a

usual problem to address in distributed simulation

engines.

The experimental method was used for this

purpose because our primary goal is to determine

whether our proposed solution works at a functional

level. It means that to determine whether time

intervals between events vary between the two

engines, we needed to understand how those time

intervals change over time between the two engines,

and an experiment would give us the control needed

to set up those conditions and test our theory. Its been

noted in the literature(Dennis & Valacich, 2001) that

the objective of experimental research is to enable

testing and extending a theory. Also, Williamson &

Johanson, 2017 proceed in stating that it is a method

that seeks to establish a cause-and-effect relationship

between variables, which is the case in our study.

By no means, experimental research is the best or

worst method, yet it is the most adequate for some

reasons:

 A cause-effect relationship needed to be

understood. Specifically, we wanted to

understand whether time intervals between

events occurring in an ABS are kept constant

upon triggering equivalent events in a BPMN

engine.

 A specific set of conditions are being studied.

We only want to verify that REST API

requests can be transmitted between the two

systems and that the intervals between two

events are respected between the two systems.

3.1 Choosing a Real-life Inspired

Business Process

It was important to inspire our experiment in a real

business process, as some authors point out (Guala,

2002) there is higher confidence in an experiment if a

real component is used. Thus metrics and business

process model from a real case study(RCS) had been

chosen(Bhat et al., 2014) which is a Lean Six-

Sigma(LSS) process improvement study conducted

in the Health Information Department (HID) of a

Medical College Hospital in India which consisted in

using LSS to improve the patient registration process

KEOD 2020 - 12th International Conference on Knowledge Engineering and Ontology Development

124

of the hospital.

The RCS concluded that the mother tongue

patients and receptionists spoke, had an impact on the

process cycle time. This variable was adequate for

this experiment because it satisfies the criteria for the

case study selection which was that it had to describe

the impact of user behaviour in a business process

output, in this case, it was communication between

patients and receptionists, given that they spoke

different languages and how this impacts the number

of patients registered per unit of time.

3.2 Modelling Communication between

Agents

Communication between actors is a relevant aspect to

be modelled in business processes (Gregoriades &

Sutcliffe, 2008) and some authors have already

studied it (Barange et al., 2014; Elleström, 2019;

Frieder et al., n.d.; Kennedy, 2012; March, 2004)

within the context of agent-based modelling.

In our RCS all the staff were proficient in the local

languages, namely Kannada and Tulu, in addition to

English. The study also observed that out of 16 staff

working in the department, only two of them knew

Malayalam, five knew Konkani, six knew Hindi, one

knew Malayalam and Konkani, and the only one

knew all three languages in addition to the local

language. Thus, six staff with a different combination

of language expertise were selected for the study. The

cycle time in handling patients, who were proficient

in only local languages, only Malayalam, only

Konkani and only Hindi was observed for ten patients

in each group(Bhat et al., 2014).

The study concluded that cycle time for

registering patients, who only spoke Malayalam,

Konkani and Hindi was significantly larger than those

who knew local languages and therefore, that is the

behaviour we model in our ABS, more specifically,

which is difficulty in communication between agents

as a function of their fluency in certain idioms.

3.3 Modelling the Business Process

Once the behaviour above is configured in the ABS,

the business process below is modelled in BPMN.

Due to its versatility and ease of use, we chose

Camunda Modeler V4.0.0 to design our model and

Camunda BPM 7.12 server as the business process

engine for our experiment.

When converting the activity diagram to BPMN,

some tasks were omitted as those played no active

part in the experiment because they did not send or

receive messages from or to agents.

Figure 2: Activity diagram of a chosen business process

Source: (Bhat et al., 2014).

Figure 3: Simplified BPMN model of the RCS.

Systemic Business Process Simulation using Agent-based Simulation and BPMN

125

The ABS engine begins execution, by having agents

communicating with each other and invoking the BPMN

engine when the PR pair finishes communicating, by

sending messages using REST. Cycle time results are

collected and stored in a database for posterior analysis.

4 RESULTS

Netlogo had been highlighted by several authors

(Abar et al., 2017; Lytinen & Railsback, n.d.;

Railsback et al., 2006) as being versatile enough for

small and large experiments, as well as presenting a

low learning curve. These characteristics were

relevant for our choice as we needed to find an engine

that is not only robust but also readily available in the

industry to facilitate adoption within organisations,

and also if further studies are conducted in the future.

Looking at our first objective, we were able to

gather data about existing ABM systems concerning

the programming languages they use to create their

model. Understanding which programming language,

they use was fundamental as we assumed that it

would be the primary vector by which the ABM could

send web requests, the assumption being that if the

underlying modelling language supports web

requests, then the engine supports them too.

Table 1: Agent-Based Modelling engines vs Programming

language they use (‘Comparison of Agent-Based Modeling

Software', 2020).

Name Programming Language

AnyLogic Java

Cougaar Java

Framsticks Framscript

JADE Java

MASON Java

Netlogo Netlogo,

Python(PyNetlogo)

Repast Java, .Net, Python

From the short review above, key findings

emerge: 100% of the ABM engines support

programming languages that can submit web requests

or support extensions that allow for external scripting

engines to be embedded in the agent-based model,

which in turn supports sending web requests.

Besides, we specifically studied the

documentation of the ABM of choice, Netlogo 6.1.

We found that it does not support any capability to

perform web requests natively, although there were

some attempts (NetLogo/Web-Extension,

2012/2020) to introduce similar functionality using

extensions, however not to send generic web requests.

On the other hand, it was also found that one of the

extensions supported is the Python scripting

engine(Jaxa-Rozen & Kwakkel, 2018) through the

PyNetLogo extension. As Python is a generic

scripting language, it not only allowed to make web

requests via REST protocol but also to establish full

integration between the two applications, control

message correlation, transformation and logging.

Although the results above confirm that majority

of ABM engines support web requests, our method

also relies on the BPMN engine supporting a REST

API that allows a consumer to start a process.

Table 2: List of major BPMN engines vs support for process

invocation through REST.

Engine Support REST process

invocation

ActiveVOS Y

Activiti Y

Bizagi BPM Suite Y

Bonita BPM Y

Camunda Y

Flowable Y

Imixs-Workflow Y

jBPM Y

Orchestra N

Sydle SEED Undetermined

From this, we can understand that the majority of

the BPMN engines do provide support for a REST

API that allows invocation of processes, and in our

experiment, we used Camunda Server 7.12.

These results together demonstrate the adequacy

for implementing our method using the majority of

ABM and BPMN engines.

With regards to verifying the accuracy of our

simulation results, we had two objectives:

 O1: Determine the correlation between engine

type and task execution interval;

 O2: Determine whether the engine type has a

significant effect on the task execution interval

of different groups of agents.

Regarding O1, the intension was to analyse how

event intervals varied between the two engines,

considering agent behaviour individually. For this,

we collected and compared time deltas between

events in each engine. A Pearson Correlation

Coefficient was then calculated to understand if event

intervals vary significantly between engines. The

Pearson correlation analysis was conducted between

DeltaABS and DeltaBPMN variables. Cohen's

standard was used to evaluate the strength of the

relationship, where coefficients between .10 and .29

KEOD 2020 - 12th International Conference on Knowledge Engineering and Ontology Development

126

represent a small effect size, coefficients between .30

and .49 represent a moderate effect size, and

coefficients above .50 indicate a large effect

size(Cohen, 1988). One of the assumptions made in

this work when estimating the Pearson correlation is

that a Pearson correlation requires that the

relationship between each pair of variables is linear

(Conover & Iman, 1981). This assumption is violated

if there is curvature among the points on the

scatterplot between any pair of variables.

Figure 3: Presents the scatterplot of the correlation. A

regression line has been added to assist the interpretation.

The result of the correlation was examined based

on an alpha value of 0.05. A significant positive

correlation was observed between DeltaABS and

DeltaBPMN (r

= 1.00, p < .001, 95% CI [1.00,

1.00]). The correlation coefficient between DeltaABS

and DeltaBPMN was 1.00, indicating a large effect

size. This correlation indicates that as DeltaABS

increases, DeltaBPMN tends to increase. Table 3

presents the results of the correlation. Note. n = 7137.

Table 3: Pearson Correlation Results Between DeltaABS

and DeltaBPMN.

Combination r

95% CI p

DeltaABS-

DeltaBPMN

1.00 [1.00, 1.00]

.001

The results of our experiment suggest the

correlation coefficient between DeltaABS and

DeltaBPMN was 1.00, indicating a large effect size.

This correlation indicates that event intervals are kept

constant between the ABS and BPMN engines. It

confirms our suspicion that the messages flow

between systems without significant changes in task

execution intervals.

For O2, the intension was to analyse how event

intervals varied between the two engines, considering

the collective behaviour of agents. This analysis was

deemed relevant because it is likely that in real-world

agent-based models, the behaviour is modelled for a

collection of agents. Many agent-based simulation

tools provide support for the concept of "breed"

which allows a modeller to create different varieties

of agents that behave differently. We segregated our

agents by communication difficulty, i.e. each group

of agents took a different amount of time to fill in the

registration form, and it varied according to time

ranges the table below:

Table 4: Agent Group VS Delay Range.

Patient Language Time Range(ticks)

Hindi 450-550

Konkani 150-250

Malayalam 350-450

Other 0-50

To evaluate point number two, a multivariate

analysis of variance (MANOVA) was conducted to

assess if mean differences exist on task execution

interval for Hindi, Konkani, Malayalam and Others

between the different source engines. The MANOVA

test is an appropriate statistical analysis when the

purpose of the research is to assess if mean

differences exist on more than one continuous

dependent variable by one or more discrete

independent variables (DeCarlo, 1997).

The main effect for source engine was not

significant, F(4, 899) = 0.00, p = 1.000, η

p = 0.00,

suggesting the linear combination of Malayalam,

Konkani, Other, and Hindi was similar for each level

of source engine. The MANOVA results are

presented in Table 5.

Table 5: MANOVA Results for Malayalam, Konkani,

Other, and Hindi by source engine.

Variable Pillai F df

Residual

p η

Source 0.00 0 4 899 1 0

The results indicate that the linear combination of

Malayalam, Konkani, Other, and Hindi was similar

for each level of source engine, which leads us to

conclude that even if the agents were operating in

groups, those differences would not be affected

during message transmission between systems.

5 CONCLUSIONS

We intended to investigate how could we integrate an

ABS engine with a BPMN engine, to perform

business process simulations and obtain statistically

Systemic Business Process Simulation using Agent-based Simulation and BPMN

127

significant results. The main aim of such integration

is to create a mechanism to simulate and study

complex phenomena within business processes.

Based on the quantitative analysis of event

intervals between the two systems and also based on

the event intervals of groups of agents between the

two engines, it can be concluded that integrating an

ABS engine with a BPMN engine, produces

statistically coherent simulation results with respect

to time management between the systems.

Despite the success demonstrated, some

significant limitations should be highlighted. We

could not evaluate how well our ﬁndings apply in a

real implementation project within an organisation as

our experiment has firmly focussed on addressing

functional and simulation results significance aspects.

It is possible that the practical implementation

constrains of our technique outweigh the benefits of

using it, so, therefore, it is suggested that further

research is undertaken to look into those aspects.

Due to the novelty of the simulation framework

proposed, we also encountered difficulties in

determining how it compares to other studies in the

same field. On the one hand, this can be a significant

step forward for a holistic business process simulation

paradigm, but on the other, for the time being, it

leaves some gaps in knowledge that can only be filled

in by further research.

The main achievements, including contributions,

may be summarised as follows. First, we created a

new way of simulating business processes. The

innovation in our method is that it allows for a holistic

simulation to happen, where complex phenomena can

be made part of the business process simulation. We

did consider not only the technical aspects of the

solution but also its organizational and management

contexts. This, in turn, opens doors to study more

complicated problems within enterprises, that are

difficult to study analytically, such as the effect of

emergence, feedback loops and self-organization on

process performance and at a broader sense, it

enriches our scientific knowledge base in process

optimisation methodologies.

It has been shown that it is possible to use and

reuse existing simulation tools to enable this holistic

type of simulation. It remains unclear to which degree

our framework is related to low implementation costs,

for instance, we speculate that our approach requires

low investment in purchasing new tools and training

staff as tools we propose are readily available in the

market. It is suspected that this is an attractive

proposition not only for large organisations but also

for small to medium businesses that cannot afford

expensive software solutions. Finally, we

contemplate whether our approach is simple and easy

to be adopted in academia or for individual

researchers, as a tool to study conservation laws in

business processes. All data collected and analysis

results were made available online: Duduka, Jacint.

(2020) 00xE8/BPABSIF: Business process & Agent-

based Interoperability Framework. Retrieved

September 22, 2020, from https://github.com/

00xE8/BPABSIF.

6 FUTURE WORK

The author identified two categories of work to be

proposed based on the experiences collected during

the research. The first category is related to problems

identified during the work undergone, and the second

relates to further areas of research that would expand

the scope of the work and enrich the features of our

method.

Regarding problems encountered during the

experiment, we found that although our results point

in the direction that our proposed method can be used

to simulate complexity in business processes, the

author feels that further investigation should be

conducted into some aspects that came to light during

the current study:

1. Netlogo and many other agent-based

simulation engines are synchronous systems. This

means that agents perform actions in sequence

without true parallelism, and therefore if the business

process being simulated require messages to be sent

in parallel, this may create challenges. We are

proposing further studies to understand the extent to

which this can create issues;

2. Impact of errors in simulation results. It is

understood that there is a margin of error in every

experiment; however, the author suggests a broader

study that looks are factors that can cause Netlogo to

behave abruptly and understand how these can

influence simulation results. These factors could be

hardware, software, resource availability;

3. The implementation of functionality within

BPMN to handle incoming and outgoing messages. It

has been found that custom scripts embedded in to

"receive the message" tasks in BPMN are not invoked

when messages arrive but straight after the token

arrives in the task. This can influence cycle time

results and other problems, and a better way to handle

messages in BPMN should be studied;

In terms of improvements to be made to our

method, it is proposed that future work consists in

exploring other simulation methods that are best

suited to stimulate different types of factors that

KEOD 2020 - 12th International Conference on Knowledge Engineering and Ontology Development

128

influence a business process. More specifically,

system dynamics is a method well suited to study how

quantitative variables are impacted by the overall

dynamics of the process and thus, variables such as

costs and budgets, can be included in the simulation

to create an even richer understanding of the overall

dynamics of the business process.

In order to better comprehend the suitability of

this simulation approach in real-world situations,

there is a need to employ it in a project from the

design phase, so that aspects as the influence of

process designer skills and time to create models can

be factored into the effectiveness. These are aspects

not covered in this study, as we only focus on

understanding the feasibility of building a solution

that supports such a simulation approach and whether

simulation results are reliable enough compared to

real ones. Therefore, a case study employing our

approach is another suggestion for future work.

ACKNOWLEDGEMENT

This work was supported by the European

Commission program H2020 under the grant

agreement 822404 (project QualiChain) and by

national funds through Fundação para a Ciência e a

Tecnologia (FCT) with reference UIDB/50021/2020

(INESC-ID).

A special thanks also goes to Liaison Group

supported our research.

REFERENCES

00xE8. (2020). 00xE8/BPABSIF. https://github.com/

00xE8/BPABSIF (Original work published 2020)

Abar, S., Theodoropoulos, G. K., Lemarinier, P., & O’Hare,

G. M. P. (2017). Agent Based Modelling and

Simulation tools: A review of the state-of-art software.

Computer Science Review, 24, 13–33.

https://doi.org/10.1016/j.cosrev.2017.03.001

Aksyonov, K. A., & Aksyonova, O. P. (2013). Application

of BPMN and EPC graphical notations for multi agent

simulation of business processes. 2013 23rd

International …. https://ieeexplore.ieee.org/abstract/

document/6652872/

Baker, T. M. (n.d.). Time Management in Distributed

Simulation Models. 5.

Barange, M., Kabil, A., Keukelaere, C. D., & Chevaillier,

P. (2014). Collaborative Behaviour Modelling of

Virtual Agents using Communication in a Mixed

Human-Agent Teamwork. 17.

Bhat, S., Gijo, E. V., & Jnanesh, N. A. (2014). Application

of Lean Six Sigma methodology in the registration

process of a hospital. International Journal of

Productivity and Performance Management, 63(5),

613–643. https://doi.org/10.1108/IJPPM-11-2013-

0191

Bider, I. (n.d.). Functional Decomposition of a Socio-

Technical System: What is Missing? 8.

Brodsky, Y. I. (n.d.). Fundamentals of Simulation for

Complex Systems. Complex Systems, 6.

Cohen, J. (1988). Statistical power analysis for the

behavioral sciences (2nd ed). L. Erlbaum Associates.

Conover, W. J., & Iman, R. L. (1981). Rank

Transformations as a Bridge between Parametric and

Nonparametric Statistics. The American Statistician,

35(3), 124–129.

https://doi.org/10.1080/00031305.1981.10479327

Dam, H. K., Ghose, A., & Qasim, M. (2015). An Agent-

Mediated Platform for Business Processes. In

INTERNATIONAL JOURNAL OF INFORMATION

TECHNOLOGY AND WEB ENGINEERING (Vol. 10,

Issue 2, pp. 43–61).

https://doi.org/10.4018/IJITWE.2015040103

DeCarlo, L. T. (1997). On the meaning and use of kurtosis.

Psychological Methods, 2(3), 292–307.

https://doi.org/10.1037/1082-989X.2.3.292

Dennis, A. R., & Valacich, J. S. (2001). Conducting

Experimental Research in Information Systems.

Communications of the Association for Information

Systems, 7. https://doi.org/10.17705/1CAIS.00705

Elleström, L. (2019). Modelling Human Communication:

Mediality and Semiotics. In A. Olteanu, A. Stables, &

D. Borţun (Eds.), Meanings & Co. (Vol. 6, pp. 7–32).

Springer International Publishing.

https://doi.org/10.1007/978-3-319-91986-7_2

Endert, H., Kuster, T., Hirsch, B., & Albayrak, S. (n.d.).

Mapping BPMN to Agents: An Analysis. 17.

Frieder, A., Lin, R., & Kraus, S. (n.d.). Agent-human

Coordination with Communication Costs under

Uncertainty. 8.

Ghlala, R., Aouina, Z. K., & Said, L. B. (2017). Multi-

Agent BPMN Decision Footprint.

… Symposium on

Agent and Multi-Agent ….

https://link.springer.com/chapter/10.1007/978-3-319-

59394-4_23

Gregoriades, A., & Sutcliffe, A. (2008). A socio-technical

approach to business process simulation. Decision

Support Systems, 45(4), 1017–1030.

https://doi.org/10.1016/j.dss.2008.04.003

Guala, F. (2002). Models, Simulations, and Experiments. In

L. Magnani & N. J. Nersessian (Eds.), Model-Based

Reasoning (pp. 59–74). Springer US.

https://doi.org/10.1007/978-1-4615-0605-8_4

Haiyan Zhao, & Jian Cao. (2007). A business process

simulation environment based on workflow and multi-

agent. 2007 IEEE International Conference on

Industrial Engineering and Engineering Management,

1777–1781.

https://doi.org/10.1109/IEEM.2007.4419498

Halaška, M., & Šperka, R. (2018). Is there a Need for

Agent-based Modelling and Simulation in Business

Systemic Business Process Simulation using Agent-based Simulation and BPMN

129

Process Management? Organizacija, 51(4), 255–269.

https://doi.org/10.2478/orga-2018-0019

Henda, H. B. G., Sayed, Y., & Ibtissem, F. (2016). Using a

Multi-Perpectives Approach for Building a Socio-

Technical Information System: Proceedings of the

Sixth International Symposium on Business Modeling

and Software Design, 217–220.

https://doi.org/10.5220/0006224002170220

Hohpe, G., & Woolf, B. (2012). Enterprise Integration

Patterns: Designing, Building, and Deploying

Messaging Solutions. Addison-Wesley.

Ibl, M., & Čapek, J. (2017). A Behavioural Analysis of

Complexity in Socio-Technical Systems under Tension

Modelled by Petri Nets. Entropy, 19(11), 572.

https://doi.org/10.3390/e19110572

Jaxa-Rozen, M., & Kwakkel, J. H. (2018). PyNetLogo:

Linking NetLogo with Python. Journal of Artificial

Societies and Social Simulation, 21(2), 4.

Kadar, M., Muntean, M., Cretan, A., & Jardim-Gonçalves,

R. (2015). Automated Negotiation with Multi-agent

Systems in Business Processes. In P. Angelov, K. T.

Atanassov, L. Doukovska, M. Hadjiski, V. Jotsov, J.

Kacprzyk, N. Kasabov, S. Sotirov, E. Szmidt, & S.

Zadrożny (Eds.), Intelligent Systems’2014 (pp. 289–

301). Springer International Publishing.

https://doi.org/10.1007/978-3-319-11313-5_27

Kennedy, W. G. (2012). Modelling Human Behaviour in

Agent-Based Models. In A. J. Heppenstall, A. T.

Crooks, L. M. See, & M. Batty (Eds.), Agent-Based

Models of Geographical Systems (pp. 167–179).

Springer Netherlands. https://doi.org/10.1007/978-90-

481-8927-4_9

Kloeckner, S., & Birkmeier, D. (2010). Something Is

Missing: Enterprise Architecture from a Systems

Theory Perspective. In B. J. Krämer, K.-J. Lin, & P.

Narasimhan (Eds.), Service-Oriented Computing –

ICSOC 2007 (Vol. 4749, pp. 22–34). Springer Berlin

Heidelberg. https://doi.org/10.1007/978-3-642-16132-

2_3

Laroque, C., Himmelspach, J., Pasupathy, R., Rose, O., &

... (n.d.). BPMN PATTERN FOR AGENT-BASED

SIMULATION MODEL REPRESENTATION. In

Academia.edu. http://www.academia.edu/download/

41502429/BPMN_PATTERN_FOR_AGENT-

BASED_SIMULATION_20160124-27721-

13py4ti.pdf

Laudon, K., & Laudon, J. P. (2013). Management

Information Systems, Global Edition. Pearson.

Lin, Q., & Guo, J. (2010). Accuracy analysis of distributed

simulation systems. 75446I.

https://doi.org/10.1117/12.885297

Liu, Y., & Iijima, J. (2015). Business process simulation in

the context of enterprise engineering. Journal of

Simulation, 9(3), 206–222.

https://doi.org/10.1057/jos.2014.35

Lytinen, S. L., & Railsback, S. F. (n.d.). Agent-based

Simulation Platforms: An Updated Review

. 10.

March, O. (2004). Natural language as an agent

communication language. http://minerva-

access.unimelb.edu.au/handle/11343/38886

NetLogo/Web-Extension. (2020). [Scala]. Center for

Connected Learning.

https://github.com/NetLogo/Web-Extension (Original

work published 2012)

Norta, A., Mahunnah, M., Tenso, T., Taveter, K., &

Narendra, N. C. (2014). An Agent-Oriented Method for

Designing Large Socio-technical Service-Ecosystems.

2014 IEEE World Congress on Services, 242–249.

https://doi.org/10.1109/SERVICES.2014.50

Onggo, B. S. S., Proudlove, N. C., D’Ambrogio, S. A.,

Calabrese, A., Bisogno, S., & Levialdi Ghiron, N.

(2017). A BPMN extension to support discrete-event

simulation for healthcare applications: An explicit

representation of queues, attributes and data-driven

decision points. Journal of the Operational Research

Society. https://doi.org/10.1057/s41274-017-0267-7

Railsback, S. F., Lytinen, S. L., & Jackson, S. K. (2006).

Agent-based Simulation Platforms: Review and

Development Recommendations. SIMULATION,

82(9), 609–623.

https://doi.org/10.1177/0037549706073695

Sulis, E., & Di Leva, A. (2018). An Agent-Based Model of

a Business Process: The Use Case of a Hospital

Emergency Department. In E. Teniente & M. Weidlich

(Eds.), Business Process Management Workshops (Vol.

308, pp. 124–132). Springer International Publishing.

https://doi.org/10.1007/978-3-319-74030-0_8

Tan, W., Xu, W., Yang, F., Li, S., & Du, Y. (2009). A

Framework for Business Process Simulation Based on

Multi-Agent Cooperation. In S. Ahmed & M. Noh

(Eds.), Multiagent Systems. I-Tech Education and

Publishing. https://doi.org/10.5772/6606

The-Evolution-of-Socio-Technical-Systems-Trist.pdf.

(n.d.). Retrieved 18 August 2020, from

https://www.lmmiller.com/blog/wp-

content/uploads/2013/06/The-Evolution-of-Socio-

Technical-Systems-Trist.pdf

Tolk, A. (2013). Interoperability, Composability, and Their

Implications for Distributed Simulation: Towards

Mathematical Foundations of Simulation

Interoperability. 2013 IEEE/ACM 17th International

Symposium on Distributed Simulation and Real Time

Applications, 3–9. https://doi.org/10.1109/DS-

RT.2013.8

Tropmann-Frick, M., & Thalheim, B. (2015). Socio-

technical System Design for Generic Workflows. 2015

26th International Workshop on Database and Expert

Systems Applications (DEXA), 134–138.

https://doi.org/10.1109/DEXA.2015.43

Vespignani, A. (2012). Modelling dynamical processes in

complex socio-technical systems. Nature Physics, 8(1),

32–39. https://doi.org/10.1038/nphys2160

Williamson, K., & Johanson, G. (2017). Research Methods:

Information, Systems, and Contexts. Chandos

Publishing.

Wu, P. (2015). A framework for model integration and

holistic modelling of socio-technical systems. In

Decision Support Systems (Vol. 71, pp. 14–27).

https://doi.org/10.1016/j.dss.2015.01.006

KEOD 2020 - 12th International Conference on Knowledge Engineering and Ontology Development

130