Complex Responsive Processes: The Emergence of Enabling

Constraints in the Living Present of a Cyber-Physical Social System

Guido T. H. J. Willemsen

, Luis Correia

and Marco A. Janssen

ISCTE-IUL/ISTA, University Institute of Lisbon, Lisbon, Portugal

Dept. Informática, LASIGE, Faculdade de Ciências, Universidade de Lisboa, Lisbon, Portugal

School of Sustainability/ School of Complex Adaptive Systems, Arizona State University, Tempe, U.S.A.

Keywords: MAPE-K, Cybernetics, Complex Responsive Processes of Relating, Digital Twin, Simulation, Business

Process Management, BPM, BPMN, Cyber-Physical Systems, Social Complexity.

Abstract: Contemporary business process modeling is based on predefined constraints where flexibility is built in.

Current business challenges result from an increase in data which, are a valuable source for decision taking.

Control models from cybernetics could do the job, especially when learning capabilities are added. However,

in an agent-based architecture there is something to add: the social component. This position paper aims to

advance understanding and practical application of how organizations can effectively utilize the abundance

of data in their operational processes while also exploring novel approaches to organizational dynamics and

coordination. More in detail, the paper outlines a model that combines socialComplex Responsive Processes

(CRP) with a cyber-physical control cycle within a multi-agent simulated business process.

1 INTRODUCTION

In the last few decades, organizations have become

considerably more digital. As a result, exponentially

more valuable data is created both within the

organization, at business partners, and in its

environment. With this data, more appropriate

decisions can be taken, and learning curves can be

accelerated. Mainstream organization theory is based

on Systems Thinking, drawing on Kantian

philosophy, where the elements of duality are leading,

i.e. the rationalist and formative teleology, where

action is constrained by given forms. But how can we

use these exponentially increased data in operational

business processes more dynamically as enabling

constraints, and how is the emergence between

process actors organized? In this position paper, a

model is proposed to harness the possible power of

social Complex Responsive Processes of relating

(CRP) in combination with a cyber-physical control

cycle considering a multi-agent simulated business

process.

This position paper is an elaboration on our

earlier paper in which the model was presented

https://orcid.org/0009-0008-5672-5373

https://orcid.org/0000-0003-2439-1168

https://orcid.org/0000-0002-1240-9052

conceptually. This study has outlined the structure of

a self-organized agnostic control cycle for business

processes where CRP techniques are applied based on

the principles of cybernetics and social science. In

this second position paper, the model is taken to a

level of applicability. As an example of this, the

traditional Beer Game will be modelled as a use case

with process modeling standards in a multi-agent

system, controlled by inter-agent knowledge sharing

and a multi-level control cycle described in this paper.

2 BUSINESS PROCESS

MODELING

Business process modeling has been formalized in the

last few decades. A good example of formalization is

the use of BPMN (Business Process Management

Notation) and integration in Process Management

Systems. This, however, has led to often inflexible

and tightly coupled architectures.

324

Willemsen, G., Correia, L. and Janssen, M.

Complex Responsive Processes: The Emergence of Enabling Constraints in the Living Present of a Cyber-Physical Social System.

DOI: 10.5220/0012789400003758

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 14th International Conference on Simulation and Modeling Methodologies, Technologies and Applications (SIMULTECH 2024), pages 324-331

ISBN: 978-989-758-708-5; ISSN: 2184-2841

2.1 Contemporary Process Models

Organizations are complex systems where learning

and knowledge creation is critical for survival.

According to Stacey, the learning ability of

organizations is constrained; "The mainstream theory

of learning and knowledge creation in organizations

is a systems theory, and like other systems theories, it

implicitly assumes the dual causal structure of

Rationalist and Formative Teleology” (Stacey, 2001).

These Kantian principles assume predefined

constraints and enable the unfolding of already

enfolded knowledge by the system. Systems thinking

is the foundation for contemporary business process

management.

2.2 Business Process Management

Systems

The de facto standard for process modeling is BPMN.

The idea behind BPMN is that business processes can

be modelled by business users where the

corresponding execution layer is generated instantly.

Currently, many low-coding BPM platforms use the

BPMN to feed Business Process Management

Systems (BPMS).

2.3 Action or Activity Orientation

An effective and solid BPMS requires a clear

architecture that encompasses all functionality in an

orchestrated manner to achieve specific business

objectives. However, this clear-cut architecture is

often missing in organizations. Also, most process

models are activity oriented. Activity orientation

means that complex real-life contexts are harnessed

in predefined models. It defines how things should be

done (while descriptive models describe what has

been done) (Gotel, Finkelstein, 1996), as the process

execution results in process instances, the unique

enactment of the process model. The context

variability is fitted in the model by decision (split) and

join gateways. This presumes that each unique variant

of a process instance is deterministic, and the process

dynamics is constrained by its process model. To

become more effective as an organization, an action-

oriented process model definition is necessary

(Nurcan, 2008).

In an action-oriented architecture a connection is

made between the knowledge extracted from event

data and actions. This enables a context specific

management of actions. An action-oriented approach

can be split into decision-orientation or conversation-

orientation. This should reduce development and

transactions costs as the adaptive capabilities increase

in a loosely couples process, where decisions drive

the adjacent possible action.

2.4 Flexibility in Process Modelling

Most process architecture frameworks presume

knowledge of future situations and are not flexible by

nature. Nurcan elaborates on process flexibility and

identifies the characteristics of a flexible process

architecture: posteriori flexibility by adaptation, or a

priori flexibility by selection which are driven by the

modeling paradigm. In this paradigm, the decision,

conversation, and user experience-oriented approach

enable the process executor to instantly adapt to its

contextual situation.

A critical characteristic that Nurcan identifies is

the flexibility technique, which is only applicable a

priori and can be applied in three ways: late binding,

late modeling, and case handling. Late binding

selects the process patterns that are applicable for the

specific instance and composes a process model on

the fly. This technique requires a loosely coupled

process architecture. Late modeling relates to a

coarse-grained modeling approach, where the degrees

of freedom for process execution are high. For each

instance, the details are defined within the higher-

level constraints. In the case handling technique, the

data and flow of a process is combined in a case. This

case is state driven, where the appropriate case is

selected to achieve the next goal, given the current

state. State events will then drive case selection.

3 CYBER-PHYSICAL SOCIAL

CONTROL LOOP

When business processes are deployed in real-life

environments, an effective control mechanism is

essential. Mechanisms from cyber-physical systems

control can be integrated with social interaction

techniques to support the genuine system dynamics.

3.1 MAPE-K

ext

Model

In cybernetics, useful control mechanisms can be

found. Recently, these mechanisms have been

extended with learning capabilities that will support

adaptability and context sensitivity.

3.1.1 Cybernetic Control Cycle

A successful action-oriented process can adapt itself

toward its environment. To become adaptive by

Complex Responsive Processes: The Emergence of Enabling Constraints in the Living Present of a Cyber-Physical Social System

325

nature, the flexibility techniques for process

management of Nurcan can be applied. In the control

cycle, process parameters will be retrieved from its

policies as business rules using the reflexion and rule-

based techniques and process patterns are selected

from the repository with late binding. These policies

and process patterns will then be improved by a

learning cycle (Senge, 1990).

In an adaptive process, decisions for process

composition and execution are driven by contextual

information. To gain grip on organizational processes

constituted of temporal actor behavior, control cycles

are required (Liu, Barabasi, 2016). These control

cycles use knowledge of the environment and the

internal state of the system to decide on the actions to

be taken. A well-known control cycle process is

MAPE-K. The MAPE-K control cycle consists of five

components; the environment is Monitored (M) and

Analyzed (A), actions are Planned (P) and Executed

(E). All these activities are based on an agent-specific

Knowledge Base (K or KB) (Kephart et al., 2003). KB

includes data such as topology information, historical

logs, metrics, symptoms, and policies, which are used

by the Monitoring component and deployed by the

Execution component.

MAPE-K could be applied to several levels of the

processes, both on a central and decentralized level

(Weyns et al., 2012). When the MAPE-K control is

organized on a decentral level, the execution of the

subsystem is driven by agent-specific goals which

shapes the behavior of higher-level processes. The

decentralized process enables the agent to learn,

based on its domain specific goals. This MAPE-K

loop is modelled as a Markov Decision Process

(MDP) using Bayesian learning. Centralized control,

on the other hand, will take care of synchronization

of these activities (Weyns et al., 2010).

In current research on the MAPE-K, attention to

the influence of social environmental factors is

limited. More specifically, how do environmental

factors like the participation of agents in a group

influence the perception of environmental data and

the evaluation principles of each single agent?

Especially when the MAPE-K model is applied to

distributed control loops with decentralized decision-

making, it could be valuable to see how the adaptation

rules and results are shared amongst the other agents.

3.1.2 Learning Capabilities

Recent initiatives aimed at fine-tuning the MAPE-K

model and diving into the characteristics of the KB.

Research by Kloes et al. (Kloes et al., 2015) presents

a MAPE-K extension, where the KB is described with

four adaptation mechanisms: the Environment model

Env

, System model K

Sys

, Goal model K

Goal

and

Adaptation model K

Adapt

. Also, they added two

components to enable meta-adaptation: Evaluation

and Learning. Recently, Kloes et al. also added the

Verification component to this (Kloes et al., 2018).

With these extensions, the MAPE-K model logic

becomes adaptive and applies dynamic, context-

specific rules. The first results from this study show

that the adaptability of the process improves but

should be validated to a higher extent to achieve

generic applicability. From now on, the learning

extension is referred to as the MAPE-K

ext

model.

MAPE-K

ELVMAPE-K

MAPE-K

triggered

Monitor

Managed

System

Analyze

symptoms

Plan

Actions

Execute

Process

Control

deployed

Execute

ELV

Join

Figure 1: MAPE-K with learning capability.

Within the Knowledge component, two elements are

subject to external factors: Knowledge of the

Environment and Knowledge of Goal, while two

other elements are internally oriented: Knowledge of

the System and Knowledge on the Adaptation actions.

The MAPE-K

ext

model shows how autonomous

decision-making techniques in a runtime

environment can be used to adapt to continuously

changing environments in a quantitative manner.

Guards monitor the environment and activate or de-

activate specific system- or sub-goals. A guard

specifies when an activity can be executed (Ricci et

al., 2008). So, these guards are trained to make the

system context sensitive. In the study of Kloes (Kloes

et al., 2018), a model for Goal requirements definition

is proposed, where a parent goal can consist of sub-

goals. These sub-goals could mutually reinforce and

measured as weighted contributors to the parent-goal

but can also be exclusive contributors. Together, the

joint success rates of the set of sub-goals will

determine the total success of the parent-goal and

therefore the success of planned actions.

3.2 Social Business Process Interaction

A business process will often rely on a human-

machine interaction, where the machines act in a

cyber-physical environment while humans behave in

a social construct. A complex responsive process

approach could integrate both worlds.

SIMULTECH 2024 - 14th International Conference on Simulation and Modeling Methodologies, Technologies and Applications

326

3.2.1 Complex Responsive Processes of

Relating

Organizations operate in a complex environment,

which is characterized by emergence, nonlinearity,

and self-organization (Oukharijane et al., 2019). In

organization science, the organization, as the locus of

attention, has been studied as a Complex Adaptive

System (CAS), where micro-dynamics of local

interactions between the organizational actors result

in global patterns. A MAPE-K

ext

control cycle is often

situated in this organizational context. Although this

approach distinguishes the several steps of

complexity, the single organizational actor is

constituted as a rule-driven agent (Macintoch,

MacLean, 2001). Before, we referred to Nurcan who

states that an adaptive process should focus on an

action-oriented agent. However, the full range of

human experiences is hardly captured while the

environment is perceived as social and complex

patterns, in which behavior of a human actor is both

physical and cognitive. Complex intelligence, where

knowledge is created out of social interaction,

includes this human factor, but lacks a suitable

integration with the idea of CAS. This has been

identified by Stacey et al as Complex Responsive

Processes of Relating (CRP) (Stacey, 2003), where

activity of actors is influenced by the behavior of

other actors, individuals, or groups. CRP, however, is

taking both perspectives on human interaction and

emergence into consideration (Stacey, Griffin, 2005).

According to Homan ( (Homan, 2016), p. 495),

“the complex responsive process perspective does not

assume the [agents] to be more or less mechanistic

entities (automatons) reacting in a rule-driven fashion

to their neighbors, but endows the [agent] with

thoughts, reflections, emotions, anxieties, ambitions,

socialization, history, political games, spontaneity,

unpredictability, and uncertainty, also understanding

(human) interactions with others as intrinsic power

relations”. In the CRP setting, actors will search for

others to create a critical mass or are complementary

in capabilities or skills to overcome uncertainty.

These groups are formed around common themes,

which are shared, repeated, and endure in its values,

beliefs, traditions, habits, routines, and procedures

(Stacey, 2003).

From the Social Feedback Theory (Banisch et al.,

2020) we learned that the behavior of the agent is

influenced by the group the agent belongs to, from

now identified as trust groups. Agents perceive their

environment through the lens of the group and act,

accordingly, based on its dominant logic (Bose et al.,

2017). Gergen describes this behavior as social

constructionism (Gergen, 1999). According to

Gergen, relationships in the group and the reality of

group members are socially constructed and are

limited by culture, history, and human embeddedness

in the physical world. Not the individual mind but the

relationship becomes the main driver for dynamics.

The gesture and response dynamics in group activities

are triggered by environmental artifacts and lead to

the application and creation of patterns and the

disclosure of new artifacts to the environment, which

is, as Stacey states, the true source of knowledge

creation (Stacey, 2001). So, in the CRP theory, to

understand the dynamics of a system, one should

focus on the interaction of actors in groups instead of

individual behavior (Stacey, 2003).

In the MAPE-K

ext

model the focus is set on a

mechanistic control loop, as it originated from

cybernetics. However, the MAPE-K

ext

model is

applicable in closed systems with clear constraints

and is based on simplified models of reality, which do

not represent the living present. When applied to a

social system, the subjective pole is missing as agent

specific considerations are only partly taken into

account. By adding human behavior to the model, the

inter-agent dynamics will change, as the closed

system is opened, and the process becomes subjective

to the adjacent-possible with enabling constraints

(Kaufmann, 2016).

By adding the inter-agent dynamics from the CRP

theory to MAPE-K

ext

we could increase the learning

capabilities of each agent and spread knowledge

between trust groups more quickly.

Figure 2: MAPE-K

ext

with CRP exchange integration.

In this research the feasibility of CRP in the MAPE-

ext

cycle is developed and assessed in simulated

business processes. This results in a cyber-physical

social model and will be identified as MAPE-K

ext

CRP.

3.2.2 Integration in a BPMS Engine

A business process with a MAPE-K

ext

CRP control

cycle can be modelled in BPMN. Also, it is possible

to use these BPMN flows in a simulation environment

Complex Responsive Processes: The Emergence of Enabling Constraints in the Living Present of a Cyber-Physical Social System

327

or improves it flexibility by using late binding or late

modeling (Patiniotakis et al., 2012).

To model the MAPE-K

ext

CRP in BPMN, the

following approach is used:

1. Develop a model to simulate a managed process.

This managed process is the operational

environment which requires controls for action,

more specifically decision taking. Logic will be

moved out of the managed process to the MAPE-

K control loop. As a result, the managed process

will become a pure constructor for information

processing and/or physical creation.

2. Add a MAPE-K control cycle to the managed

process to allocate decision making.

The MAPE-K control cycle will include the logic

stored in the KB (K) where context data is stored

and monitored (Monitor), analyzed and

interpreted with policies (Analyze), translated

into behavioral change (Plan), and applied for

execution (Execute).

3. Implement an internalized learning process to

update the agents KB.

The KB contains both contextual information

(externalized) and policies (Internalized). This

KB will be updated during runtime. As a learning

cycle is added to update the agents KB, the

internalized policies and rules become dynamic.

This learning capability exists of three elements:

Evaluating, Learning, and Validating (ELV) to

update the KB. This learning on top of MAPE-K

results in the MAPE-K

ext

model.

4. Add interfacing of agent specific KB data within

the agent population.

Updating the KB in the MAPE-K

ext

is internalized.

However, the learning capabilities of other agents

could be valuable to speed up the improvement of

a KB. Successful strategies can be shared among

agents by updating KB entity records like specific

policies of process patterns. Several Agent Based

Modelling (ABM) techniques are available to

facilitate this data exchange (Pires et al., 2023).

5. Add social bonding between agent cliques.

Create groups of agents by adding trust levels.

Grouping of agents can be defined in diverse ways

i.e., imposed by the modeler, self-organized by

agents creating formalized groups or even

informal group definition (like influencers).

When an agent is part of a group, the trust level

between agents improves. This could result in free

sharing of KB data between group agents, called

Trust groups (Hoogendoorn et al., 2008).

This model is based on the ABM principles and

detailed with the ODD (Overview, Definition and

Details) technique (Grimm et al., 2020).

Figure 3: MAPE-K

ext

CRP model in BPMN.

The managed process will be the locus of control and

could be any operational BPMN process, as the

MAPE-K

ext

CRP model is agnostic. In BPMN the

decomposition of sub models enables an efficient

reuse of standardized processes. In the managed

process, the MAPE-K cycle will be called by

embedding the subprocess, while the ELV learning

extension process is a subprocess of MAPE-K. In this

research, the CRP extension will be called from the

ELV process. With this process architecture, the

MAPE-K

ext

CRP should enable a learning capability

that includes inter-agent exchange of knowledge.

4 ENABLING CONSTRAINTS IN

THE LIVING PRESENT

4.1 Digital Twin Control Model

In a managed process, the control cycle will take care

of the decision making. When the agnostic MAPE-

ext

CRP model is added to an operational

environment, it will be able to generate instance

specific process models and becomes adaptive. By

integrating this process in a simulation environment,

it will function as a Digital Twin.

4.1.1 MAPE-K

ext

CRP Model Architecture

The generic MAPE-K

ext

CRP model is based on four

layers. The managed process consists of events, tasks,

SIMULTECH 2024 - 14th International Conference on Simulation and Modeling Methodologies, Technologies and Applications

328

gateway (decisions or routing elements defined as

splits and joins) and possible sub-processes. This is

the operational process that is running in its

environment and will effectively change the state or

phase space. This managed process needs to be a fine-

grained model and should take care of achieving a

specific objective of the agent.

Within this managed process, one MAPE-K sub-

process is included. This sub-process monitors the

current state of the environment and will define the

preferred action to take. This could be a decision to

apply a specific policy, which is effectuated in

parameter setting or the selection of a process pattern.

However, to be able to achieve this, the managed

process should be loosely coupled, where process

patterns can be selected, defined as a sub-process.

This enables agility and late modeling, which results

in a high level of adaptability. Knowledge of the

events in its environment, policies, and the process

patterns available will be stored in the KB.

Figure 4: CRP BPMN process flow.

The learning mechanism will be triggered from

the MAPE-K flow. This learning extension was added

to MAPE-K by Kloes et al. (Kloes et al., 2018) and is

applicable for each MAPE-K cycle, just after the

monitoring of the managed system and is executed

after a change of the process typology or

stochastically determined, where the level of

randomness can be varied with a system parameter.

This learning process will evaluate the

effectiveness of earlier MAPE-K decisions and

change policies or process patterns when the results

are not satisfying. The learning step will search for

alternative policies or new process pattern

combinations, which will be stored in the KB after

verification of the result, based on a meta-simulation

of the process during the verification step.

In each cycle of the ELV there is a link to the

knowledge of other agents. For each instance, this

process will select policies (decisions or parameters)

and process patterns that are shared with other agents.

This is represented by an outbound process and

stimulates the social characteristics of agent behavior.

Sequentially, new knowledge is retrieved from

others, were new policies and process patterns are

stored in the agent’s KB while keeping the original

source for each acquired knowledge object. This is

called the inbound process of the knowledge

exchange.

However, knowledge is only shared amongst

other agents that belong to the same trust group.

Knowledge in the trust groups is ranked and scores of

applied knowledge will stimulate or discourage the

use of policies and process patterns. During the

Verification phase in the ELV cycle, the scores of

knowledge of each trust group are taken into

consideration.

In addition, this behavior could result in closed

trust groups, where access to new knowledge from

non-trustees is secluded. To disclose this knowledge,

own knowledge is also shared randomly or via social

links to non-trustees (for example with a blackboard

agent (Szymański et al., 2018) or the attitude

formation technique (Pires et al., 2023)) and their

knowledge is verified. Based on the outcome of this

verification, an improved simulated result will

promote the source agent to the trust group.

4.1.2 Real Time Process Simulation

The MAPE-K

ext

CRP model can be modeled as a

digital twin of the managed process with the

possibility to simulate. This model contains many

possible process patterns and stores its relationship

with other main process patterns. For this research,

the model is agent-based and could contain sub-

processes. These sub-processes represent process

agents which must achieve a specific task, in this

case: Monitoring, Analysis, Planning and Execution

in MAPE-K, Evaluation, Learning and Verification in

the ELV and Outbound and Inbound Exchange in

CRP.

The model is linked to the BPMS bi-directionally:

process states (events) are retrieved from the

managed process to the simulation environment and

execution plans are deployed from the MAPE-K

control cycle to the process orchestration engine.

Deployment takes place by pushing the process

pattern script to the BPMS. With this, the process

flow visualization can be generated in the BPMS and

execution in real-time is possible. Based on this bi-

directional iteration, a genuine, real-life digital twin

is created in which simulation takes place in the living

present.

In this research the Anylogic simulation

environment is used. Anylogic software supports

different paradigms to model large and complex

systems (Borshchev, Fillipov, 2004) and could be

used to run ABM simulations of complex business

processes. With the outcome of these simulations,

business process decisions can be underpinned with

Complex Responsive Processes: The Emergence of Enabling Constraints in the Living Present of a Cyber-Physical Social System

329

independent or contingent data. The Anylogic model

will be defined as one main process (the managed

process) that includes a MAPE-K sub process, which

is decomposed as a separate agent, embedded in the

agency of the main process. Also, this MAPE-K sub

process consists of the Evaluation, Learning and

Verification process (ELV) as a sub-sub process. The

CRP exchange process will then be another sub

process, a part of the ELV cycle and is integrated in

the other agent process execution environment.

4.2 Case Studies

To show the practical use of the theoretic MAPE-K

ext

CRP model it will be applied to a real-life business

case, to show its usefulness in supporting operational

business process decisions.

Traditional business cases are built on predefined,

rigid processes. A well-known business case in

logistics is the Beer-Game, where the supply chain of

beer is modeled with multiple actors, feedback loops,

nonlinearities, and time delays. In a beer game

simulation, the optimum must be found in the order

quantity in a trade off with stock levels and service

levels across all stages in a supply chain (Sterman,

1984). Based on this model, extensive research has

been performed, including agent-based versions,

BPMN, MAPE-K and deep reinforcement learning.

When the MAPE-K

ext

CRP model is applied to the

beer game, it is not intended to prove its added value

compared with other beer game improvement

techniques. The only objective is to show the

applicability of the model in a business case.

In this research, the beer game will be modelled

with the MAPE-K

ext

CRP model in several steps; first

the late binding process architecture is applied in the

traditional beer game; next the MAPE-K control cycle

and the ELV extension are included; finally, the CRP

inter-agent knowledge exchange process is added,

where knowledge is shared within trust groups. The

analysis is based on the mathematical model

described by Edali and Yasarcan (Edali, Yasarcan,

2014).

Comparison of the outcomes should indicate the

possible added value of the CRP exchange by its

ability to increase knowledge sharing and

acceleration of the learning process. In addition, the

modification of the agent’s Knowledge Base is

measured by the number of new, changed, and

deleted policies and process patterns. Also, the source

of these knowledge base records is reported, as it

could origin from the agent itself or a trusted agent.

The added value of the MAPE-K

ext

CRP model is the

ability to share knowledge between agents in a

controlled manner.

5 CONCLUSIONS

Agent-based models could process operational data in

a complex, self-organized way. In cybernetics, many

applicable models can be found like the MAPE-K

ext

control model. However, the exchange of data

between agents is limited in these models. And just

this inter-agent dynamics seems to have a great

potential to accelerate learning and improvement

initiatives. In this paper we propose the integration of

the MAPE-K model with social complexity CRP

techniques. This research investigates the exchange

of knowledge between agents to apply in each own

MAPE-K

ext

control cycle. The processes and

techniques of knowledge exchange are applied in

both managed process and its simulated model. In the

final stage of this research, the theoretical model will

be applied to an operational simulation model, based

on a real-live business case.

6 FURTHER RESEARCH

In this paper, two elements of knowledge exchange

are selected as knowledge entities that will be used in

the MAPE-K

ext

control cycle: policies and process

patterns. More research must be done on other

knowledge entities like topologies, sensors, or

effectors.

Also, the requirements for a loosely coupled

process architecture are incomplete and should be

extended in a more fine-grained manner. This would

increase the level of flexibility in process pattern

selection and deployment.

The third area of elaboration is the use of trust

groups, as this topic has much more depth than used

in this research. Using several techniques to create or

join trust groups could stimulate the speed and quality

of the exchange of knowledge entities. Also, more

CRP techniques could be applied to increase the level

of inter-agent dynamics.

REFERENCES

Banisch et al. (2020). How social feedback processing in

the brain shapes collective opinion processes. Leipzig:

ODYCCEUS - Opinion Dynamics and Cultural

Conflict in European Spaces.

SIMULTECH 2024 - 14th International Conference on Simulation and Modeling Methodologies, Technologies and Applications

330

Borshchev, Fillipov. (2004). From System Dynamics and

Discrete Event to Practical Agent Based Modeling:

Reasons, Techniques, Tools. 22nd International

Conference of the System Dynamics Society. Oxford,

UK.

Bose et al. (2017). Collective Decision-Making. Current

Opinion in Behavioral Science(16), 30-34.

Edali, Yasarcan. (2014). A Mathematical Model of the Beer

Game. Journal of Artificial Societies and Social

Simulation, 17(4).

Gergen. (1999). An Invitation to Social Construction.

Thousand Oaks, CA: Sage.

Gotel, Finkelstein. (1996). An Analysis of the

Requirements Traceability Problem. First IEEE

International Conference ICRE'94. Colorado Springs.

Grimm et al. (2020). The ODD Protocol for Describing

Agent-Based and Other Simulation Models: A Second

Update to Improve Clarity, Replication, and Structural

Realism. Journal of Artificial Societies and Social

Simulation, 2(7).

Homan. (2016). Locating complex responsive process

research in the approaches of theorising about

organizations. Int. Journal of Business and

Globalisation, 17(4), 491-513.

Hoogendoorn et al. (2008). Modeling Dynamics of Relative

Trust of Competitive Information Agents. In

Cooperative Information Agents XII. CIA 2008. Lecture

Notes in Computer Science (p. Vol. 5180). Berlin:

Springer.

Kaufmann. (2016). Humanity in a Creative Universe. New

York: Oxford University Press.

Kephart et al. (2003). The vision of autonomic computing.

IEEE Computer Society, 41-50.

Kloes et al. (2015). Adaptive Knowledge Bases in Self-

Adaptive System Design. 41st Euromicro Conference

on Software Engineering and Advanced Applications.

Kloes et al. (2018). Comprehensible and dependable self-

learning self-adaptive systems. Journal of Systems

Architecture, 85-86, 28-42.

Liu, Barabasi. (2016). Control principles of complex

systems. Review of Modern Physics, 88(July-

September), 1-58.

Macintoch, MacLean. (2001). Conditioned emergence:

researching change and changing research.

International Journal of Operation & Production

Management, 21(10), 1343-1357.

Nurcan. (2008). A Survey on the Flexibility Requirements

Related to Business Processes and Modeling Artifacts.

Proceedings of the

41st Hawaii International

Conference on System Sciences. Honololulu.

Oukharijane et al. (2019). Towards a New Adaptation

Engine for Self-Adaptation of BPMN Processes

Instances. Proc. of the 14th Int. Conference On

Evaluation of Novel Approaches to Software

Engineering.

Patiniotakis et al. (2012). An Aspect Oriented Approach for

Implementing Situational Driven Adaptation of

BPMN2.0 Workflows. 6th International Workshop on

Event-Driven Business Process Management

collocated with BPM 2012. Tallinn, Estonia.

Pires et al. (2023). Knowledge sharing in a dynamic,

multi-level organization: an agent-based modeling

approach. Springer Science+Business Media.

Ricci et al. (2008). The A&A Programming Model and

Technology for Developing Agent Environments in

MAS. ProMAS 2007. Honolulu, HI, USA.

Senge. (1990). The Fifth Discipline. The art and practice of

the learning organization. London: Random House.

Stacey. (2001). Complex Responsive Processes in

Organizations: Learning and Knowledge Creation.

New York: Routledge.

Stacey. (2003). Complexity and group processes: A

radically social understanding of individuals. New

York: Brunner-Routledge.

Stacey, Griffin. (2005). A Complexity Perspective on

Researching Organizations. London: Routledge.

Sterman. (1984). Instructions for running the beer

distribution game. Cambridge: Sloan School of

Management (MIT).

Szymański et al. (2018). Multi-Agent Blackboard

Architecture for Supporting Legal Decision Making.

Computer Science, 19(4), 457-477.

Weyns et al. (2010). Dagstuhl Seminar on Software

Engineering. On Patterns for Decentralized Control in

Self-Adaptive Systems.

Weyns et al. (2012). FORMS: Unifying reference model for

formal specification of distributed self-adaptive

systems. ACM Transactions on Autonomous and

Adaptive Systems, 7(1), 1 - 61.

Complex Responsive Processes: The Emergence of Enabling Constraints in the Living Present of a Cyber-Physical Social System

331