AGENT-ORIENTED COMPUTING FOR BUSINESS PROCESS

MANAGEMENT – WHAT, WHY AND HOW

Minhong Wang

, Kuldeep Kumar

and Weijia Ran

Division of Information & Technology Studies, The University of Hong Kong, Hong Kong

College of Business Administartion, Florida International University, USA

Keywords: Agent-Oriented Computing, Business Process Management, Complex Systems.

Abstract: Agent-oriented computing has been addressed for seeking significant advances on complex process

management systems. However, recent research on this problem is experience-driven, ad-hoc, and without a

cohesive theoretical base. This study aims to examine the key problems and solutions for complex process

management as well as the mechanism of why and how agent-oriented computing can be applied in

developing complex process management solutions.

1 INTRODUCTION

Business process management (BPM) is regarded as

a systematic approach to improving an

organization's business processes. Most approaches

to BPM use information technologies to automate

business processes in whole or in part, through

building business information systems that offer the

right tasks at the right time to the right participant

along with resources needed to perform these tasks.

They help coordinate and streamline business

transactions, reduce operational costs, and promote

real-time visibility in business performance.

Traditional approaches to BPM systems use

workflow technology to design and control the

process through workflow engine (van der Aalst, et

al., 2002). They follow a highly structured and

predefined workflow model, and therefore are well

suited to applications with standard inputs, processes,

and outputs. However, business processes may

change over time as a result of complex interactions,

resource competition, abnormal events, and other

sources of uncertainty. Against this background, a

number of studies are seeking significant advances

on process management approaches by applying

various technologies.

Although there has been a proliferation of studies

on agent-oriented computing technologies in BPM

(Jennings et al., 2000, Wang et al., 2002), the key

mechanisms have remained unrevealed, i.e. why we

need them for BPM, and how to apply them in

developing BPM solutions. Most prior research on

this problem is experience-driven, ad-hoc, and

without a cohesive theoretical base. There is only

minimal work that examines the root of complexity

of business processes, the need of effective

approaches to complex process management, and

how this need affects the requirements and solutions

for business process management (Kumar et al.,

2006). Unless we have appropriate understanding on

these important issues, further advances on BPM

solutions will be problematic.

2 HOW TO MANAGE COMPLEX

PROCESSES

A business process is a collection of activities that

consist of a series of steps performed by actors

(machines or humans) to produce a product or

service for the customer. Real-world processes are

messier than the input-transformation-output view

might suggest. They are best viewed as networks, in

which a number of roles collaborate and interact to

achieve a business goal. Based on this

understanding, we examine the complexity of BPM

problem, and address the key solutions to BPM in

this section.

Decomposition of Complex Processes. Business

processes are complex systems, made up of a

number of interacting objects with dynamic behavior.

203

Wang M., Kumar K. and Ran W. (2008).

AGENT-ORIENTED COMPUTING FOR BUSINESS PROCESS MANAGEMENT – WHAT, WHY AND HOW.

In Proceedings of the Tenth International Conference on Enterprise Information Systems - SAIC, pages 203-206

DOI: 10.5220/0001679802030206

 SciTePress

The way we manage complexity is to decompose a

complex system or process into smaller modules that

can be designed independently, i.e. modularity.

Modularity ensures easy maintenance and updates of

complex systems by separating the high-frequency

intra-module linkages from the low-frequency inter-

module linkages, and limiting the scope of

interactions between modules by hiding the intra-

module relations inside a module box (Simon,

1981). In this conception, a business process can be

defined as set of subsystems, which can in turn be

hierarchically decomposed into further levels of

detail corresponding to the organizational context.

How to decompose complex processes? “Task”

has been selected as the basic module for building

process management systems. A process can be

decomposed into tasks, task into sub-tasks, and so

on, through many layers in a hierarchy. With the

increased extension of business processes from

intra-organizational to inter-organizational scope, we

need to deal with interactions within an organization

as well as interactions across organizations in

partnership. To reduce the complexity, we need to

distinguish between inter- and intra-organizational

interactions and deal with them by isolating one

from another. We may propose “service” as a high

level view of the building block of a process, where

a process is composed of a set of services; each

service is provided by an organization and can be

further decomposed into tasks.

Flexible Coordination among Tasks. To manage

complex interactions in business organizations,

multiple actors, activities, resources and goals need

to be coordinated. After decomposing a complex

process into a number of task components, we need

to coordinate various interactions between the

components at different levels in a network

hierarchy.

How to coordinate the interactions between the

task components? The more stable and predictable

the situation, the greater the reliance on coordination

based on structured and specifiable schedules, such

as coordination by plan and coordination by

standardization; the more variable and unpredictable

the situation, the greater the reliance on informal and

flexible communication, such as coordination by

feedback and coordination by mutual adjustment

(Minzberg, 1979). Faced with increased

uncertainties in dynamic environments,

organizations must use more flexible coordination

mechanisms to coordinate their processes. Flexible

coordination is portrayed by less centralization in

top management to facilitate bottom-up initiatives.

The required mechanisms include flatter hierarchies,

decentralized autonomy-based units and decision-

based coordination, which are used for narrowing

direct control and encouraging more mutual

adjustment and coordination.

Awareness of Dynamic Environments. As a result

of complex interactions, resource competition,

abnormal events, and other sources of uncertainty,

business processes are usually semi-structured or

unstructured to the extent that there is an absence of

routine procedures for dealing with them. Problem

solving is then regarded as an interaction between

the behaving organism and the environment under

the guidance of a control system. A basic idea

underlying this viewpoint is control of complex

dynamic systems or situations based on situation

awareness.

In a dynamic business process environment,

there is a need for spontaneous decision and

coordination of processes based on situation

awareness. An exact execution order of activities is

impractical, while the interaction or relationship

between the environment and activities is more

reliable in determining how to manage and

coordinate tasks (Wang et al., 2006). In other words,

we need to be able to coordinate the processes by

sensing and comprehending the situations,

determining responses to the situations, as well as

taking initiatives to achieve business goals. The

question of which task to execute and when to

execute it is dependent on the current environment

and underlying business rules rather than a static

process schema.

Flexible Resources Coordination. The rise of

Internet mediated e-Business brings the era of quick

connected global business relationships into

existence. Business networks that are temporarily

integrated and driven by demands have emerged and

operated for the lifespan of the market opportunity.

Along with this change, a business process can be

dynamically established by connecting or composing

services together from different organizations

through alliances, partnerships, or joint ventures.

Attentions on business processes should be extended

from task and procedure to resources discovery,

selection and coordination (Wang et al., 2008). What

is new in this flexible form is reliance on the idea of

separating requirements from satisfiers

(Mowshowitz, 1997). This separation allows for

crafting process structures that enable management

to switch between different resource options for

implementing a process. The success of the model is

highly dependent on the match between the

requirements and satisfiers that deliver the services.

One way to ensure this balance is to model the

ICEIS 2008 - International Conference on Enterprise Information Systems

204

integration or composition of business processes as a

management problem which involves: 1) the

separation of requirements from the means for

realization, and 2) the dynamic selection and

allocation of resources to requirements.

3 MECHNISMS OF AOC

What is AOC? Agent is some actor (performer)

who on the behalf of a principal provides a service.

The agent provides the service by receiving request

of service from the principal. In performing or

providing a service on behalf of a principal, the

agent can ask other agents to provide it with the

service, i.e. be the principal that asks other agents to

perform the service. In this situation, the agent is a

broker. In computer science, the term “agent” is

used to describe a piece of software that acts for a

user or other program in a relationship of agency. It

denotes a software-based entity that enjoys the

properties of autonomy, social ability, reactivity and

pro-activity (Wooldridge, 2002). The key idea of

Agent-Oriented Computing (AOC) suggests the

delegation of tasks and responsibility of a complex

problem and emphasizes autonomy and co-operation

of agents to perform tasks in open and complex

environments.

Why AOC for BPM? AOC represents an emerging

computing paradigm that helps understand and

model complex real-world problems and systems, by

concentrating on high-level abstractions of

autonomous entities. AOC is used to model and

implement intelligent solutions to semi- or ill-

structured problems, which are too complex to be

completely characterized and precisely described.

By using AOC, a complex system can be viewed as

network of agents acting concurrently, each finding

itself in an environment produced by its interactions

with the other agents in the system.

AOC is applicable to a number of complex

(dynamic, open, and interactive) domains like

electronic commerce, manufacturing resource

planning (ERP), supply chain management, project

management, etc. It is particularly well suited for

complex process situations that are not all known a

priori, cannot all be assumed to be fully controllable

in their behaviors, and must interact on a

sophisticated level of communication and

coordination (Wang et al., 2006).

For example, a typical supply chain faces

uncertainty in terms of supply, demand, and process,

etc. Moreover, business entities in a supply chain are

highly interdependent in order to achieve coherence

among them. A large number of interacting

decisions may take place between different entities,

most of which are impossible to foresee at design

time. With mixed and often conflicting objectives,

processes are sophisticated and difficult to manage

using closed-form analytical solutions. Business

managers have no way to deal with the situation, but

fall back on whatever general capacity they have for

intelligent, adaptive, goal-oriented action. The

agent-based approach is directly applicable to this

type of application domain.

4 HOW AOC WORKS FOR BPM

We investigate how AOC can be used to develop

solutions of complex process management.

AOC for Decomposition of Complex Processes.

Business processes display complexity because of

interactions of their internal components and

interaction of the process with its environment. The

highly dynamic and unpredictable nature of business

processes makes AOC appealing. A process can be

decomposed into a set of tasks, task into sub-tasks,

and so on, through several layers in a hierarchy. The

tasks are then assigned to autonomous agents, each

having specific goals to achieve and interact with

one another to manage their autonomy and

interdependencies. To deal with business processes

across different organizations, “service” can be used

as a high level view of the building block of a

process. A process is composed of a set of services,

each of which can be further decomposed into tasks.

AOC for Flexible Task Coordination. To

coordinate the interactions in dynamic situations,

flatter hierarchies, decentralized autonomy-based

units and decision-based coordination are required,

where AOC is directly applicable. AOC supports

decentralized control and asynchronous operations

by a group of autonomous software entities, which

are able to perform decision-based coordination of

their activities. In complex process management, it

is impossible to predefine all activities and

interactions at design time. Instead, we define the

goal or role of each agent, based on which a set of

rules can be specified for governing the behavior of

the agent in dynamic situations.

AOC for Awareness of Dynamic Environment.

The complexity of business processes comes not

only from interactions of their internal components,

but also from interactions of the process with its

environment. To manage business processes in a

dynamic environment, we need to be able to

AGENT-ORIENTED COMPUTING FOR BUSINESS PROCESS MANAGEMENT – WHAT, WHY AND HOW

205

continuously perceive the environment, and make

real-time decisions on the process. Agent-based

software entity is able to sense and interpret the

situations and determine appropriate actions upon

the situations based on predefined rules or

knowledge. Moreover, AOC supports prediction of

future state of the environment for purpose of

proactive actions. Different from passive response to

current events, proactive behavior has an orientation

to the future, anticipating problems and taking

affirmative steps to deal positively with them rather

than reacting after a situation has already occurred.

It refers to the exhibition of goal-oriented behaviors

by taking initiatives (Wang et al., 2006, 2002).

AOC for Resource Coordination. A business

process can be dynamically established by

connecting or composing Web services provided by

different organizations over the Internet. However, it

is a complex problem to search appropriated services

from a large number of resources as well as schedule

and coordinate them under various constraints. The

complexity arises from the unpredictability of

solutions from service providers (e.g. availability,

capacity, and price), and constraints on services in a

process (e.g. temporal relationship, time and cost

constraint). To deal with the problem, AOC can be

used for coordination based on distributed decision

making. In process integration, decision and

coordination among services can be modelled as a

distributed constraint satisfaction problem, in which

solutions and constraints are distributed into a set of

services and to be solved by a group of agents

(brokers) on behalf of service requesters and

providers. Finding a global solution to the integrated

process requires that all agents find the solutions that

satisfy not only their own constraints but also inter-

agent constraints (Wang et al., 2008). In this

situation, the special type of agent works as an

intermediary between service requester and service

provider, coordinating on behalf of two parties

regarding service requirements, qualities, costs, and

other constraints.

5 CONCLUSIONS

This work has investigated the rationale for

developing solutions to complex process

management by using agent-oriented technologies.

The main problems and key solutions for complex

process management are examined. Meanwhile, the

mechanisms of Agent-Oriented Computing, together

with its relationship with BPM systems development

are clarified. Based on this, we present a clear

picture on why and how Agent-Oriented Computing

technology can be applied to developing process

management systems, especially in complex

situations. This work will benefit professionals,

researchers, and practitioners by guidelines and

methods for designing and developing technical

solutions to complex business process management.

ACKNOWLEDGEMENTS

This research is supported by a UGC CERG

research grant (No. RGC/HKU7169/07E) from the

Hong Kong SAR Government, and a Seed Funding

for Basic Research (200611159216) from The

University of Hong Kong.

REFERENCES

Jennings, N.R., Faratin, P., Norman, T.J., O'Brien, P., &

Odgers, B., 2000. Autonomous Agents for Business

Process Management. International Journal of

Applied Artificial Intelligence, 14 (2), 145-189.

Kumar, K., & Narasipuram, M.M., 2006. Defining

Requirements for Business Process Flexibility.

Seventh Workshop on Business Process Modeling,

Development, and Support, CAiSE 2006

Mintzberg, H., The Structuring of Organizations, New

Jersey, Prentice Hall, 1979

Mowshowitz, A., 1997. Virtual organization.

Communications of the ACM, 40 (9), 30-37.

Simon, H.A., 1981. The sciences of the Artificial, MIT

Press, Cambridge, Massachusetts, London, England.

van der Aalst, W.M.P., & van Hee, K.M., 2002. Workflow

Management: Models, Methods, and Systems, MIT

press, Cambridge, MA.

Wang, M., & Wang, H., 2002. Intelligent Agents

Supported Flexible Workflow Monitoring System.

Proceedings of the14th International Conference on

Advanced Information Systems Engineering (CAiSE),

Toronto, Lecture Notes in Computer Science,

Springer-Verlag, 2348, 787-791.

Wang, M., Liu, J., Wang, H., Cheung, W., & Xie, X.,

2008. On-demand E-Supply Chain Integration: A

Multi-Agent Constraint-Based Approach. Expert

Systems with Applications, 34(4), 2683-2692.

Wang, M., & Wang, H., From Process Logic to Business

Logic - A Cognitive Approach to Business Process

Management. Information & Management, 43(2), 179-

193.

Wooldridge, M., 2002. An introduction to multiagent

systems, England, J. Wiley.

ICEIS 2008 - International Conference on Enterprise Information Systems

206