Next Generation Enterprise Modelling

The Role of Organizational Theory and Multi-Agent Systems

Balbir Barn

, Tony Clark

and Vinay Kulkarni

Department of Computer Science, Middlesex University, The Burroughs, London, U.K.

TRDDC, Tata Consulting Services, India

Keywords:

Organization Theory, Multi-Agent Systems, Actor Theory, Enterprise Modeling.

Abstract:

This position paper proposes that the current enterprise modeling approaches are overly reliant on the know

how or tacit knowledge of enterprise architects for addressing organizational challenges such as business-

IT alignment. Furthermore, current modeling languages only encourage linear thinking. By drawing upon

existing research on (computational) organization theory and multi-agent systems, we propose implementation

requirements for a next generation enterprise modeling language that supports agent based simulation.

1 INTRODUCTION

The modern enterprise is faced with the tricky chal-

lenge of responding to external drivers such as merger

and acquisitions, or potential new markets, by adapt-

ing and managing internal change particularly with

respect to business-IT alignment. Up to now, such

a response has been dependent upon human expertise

based on tacit knowledge and experience or “know

how”. Such a position is not sustainable with the

rapid pace of change attributed to technology and

globalization. This is conﬁrmed with research that

indicates that Strategic business-IT alignment has re-

mained an ongoing concern for organizations (Luft-

man, 2004) and researchers have addressed the im-

portance of alignment and in particular the need for

congruence between business strategy and IT strategy

(Chan and Reich, 2007).

One speciﬁc approach that has been used to bear

upon the problem of business alignment is the role

of Enterprise Architecture (EA) (Lankhorst, 2005).

However, the predominant theme has focused on de-

veloping enterprise models that are descriptive in na-

ture and hence needing human expertise for their in-

terpretation (see ((Veken, 2013), (Zachman, 1987))

for two obvious examples). As a result, current ap-

proaches to Enterprise Modeling (EM) exhibit a high

degree of latency in meeting key objectives such as

alignment, adaptation etc. Thus EA in its current

state does not readily lend itself to supporting the type

of analysis that key decision makers typically utilize.

Such stakeholders demand: ease of comprehension of

the entire business so that decision-making can lead

to efﬁcient and effective change. In particular they

require the ability play out various what-if (i.e., what

will be the consequences of such and such action) and

if-what (i.e., what would have led to such and such sit-

uation) scenarios to arrive at the right response, estab-

lish feasibility of the response, and estimate a ROI of

the response. Thus ways of simulating an enterprise

are needed and currently EA modeling approaches do

not readily support this requirement.

Away from the EA modeling community, orga-

nizational theory and in particular computational or-

ganizational theory manifested in technologies such

as multi-agent systems provides an opportunity to re-

purpose existing research outcomes to address the EA

simulation and alignment conundrum. In doing so,

this position paper proposes that next generation En-

terprise Modeling languages should draw upon con-

cepts from organizational theory and multi-agent sys-

tems in order that appropriate machinery can be im-

plemented to support the simulation requirement.

The remainder of this paper is structured as fol-

lows: Section 2 presents key concepts from organiza-

tional theory and multi-agent systems. Speciﬁcally

it draws upon the established research relationship

that exists between the two areas. Section 3 presents

the main contribution of the paper and proposes a

novel language based approach to enterprise model-

ing for simulation. The proposal draws upon compo-

nents, goal modeling and agent technologies. Section

4 discusses our plans for addressing the research chal-

lenges raised by the approach.

482

Barn B., Clark T. and Kulkarni V..

Next Generation Enterprise Modelling - The Role of Organizational Theory and Multi-Agent Systems.

DOI: 10.5220/0005097404820487

In Proceedings of the 9th International Conference on Software Engineering and Applications (ICSOFT-EA-2014), pages 482-487

ISBN: 978-989-758-036-9

 2014 SCITEPRESS (Science and Technology Publications, Lda.)

2 ORGANIZATIONAL THEORY

AND MULTI-AGENT SYSTEMS

Our starting premise is the view that there is a press-

ing need for next generation EM languages to ad-

dress the requirements described in section 1 which

can be broadly summarized as languages that: are ma-

chine manipulatable, support simulation through exe-

cutability, and are model based. Such requirements

raise two questions: What needs to be simulated?

Secondly, what technologies can we deploy to sup-

port execution? In answering the ﬁrst question we

need to revert to Organizational Theory to understand

the meaning of Organization and its constituting ele-

ments. Human organizations in particular have been

the subject of detailed analysis from a range of disci-

plines including engineering, economics, psychology

and sociology (Scott, 1992). The resulting analysis

of the literature for organizational theory leads to a

persuasive argument for the use of agent technology,

particularly multi-agent systems as candidate tech-

nology for supporting the simulation/executability re-

quirements identiﬁed.

Ours is an organizational society such that orga-

nizations are the dominant characteristic of modern

societies. One rationale for the existence of organi-

zations posited by Carley and Gassser is that they ex-

ist to overcome the cognitive, physical, temporal and

institutional limitations of individual agency (Car-

ley and Gasser, 1999). While there are many ways

in which these limitations can be overcome and the

structure, form or architecture of an organization con-

tributes to such efforts, decades of research indicate

that there is no optimal organizational design. In-

stead, the challenge morphs into one of adaptability

and response to change. First we present here a nec-

essarily brief overview of some of the key deﬁnitions

and perspectives on organizations that underpin how

we intend to articulate the concept of an organization

in the context of the model driven enterprise. We ﬁrst

begin with a deﬁnition of the term organization rec-

ognizing that there are multiple deﬁnitions depending

upon the perspective taken. The deﬁnition is reported

from (Parsons and Jones, 1960):

Organizations are social units (or human

groupings) deliberately constructed and re-

constructed to seek speciﬁc goals.

We explore this deﬁnition further by considering how

the study of organizations has generally investigated

the constituent elements of an organization and three

dominant theoretical perspectives informing research.

Leavitt identiﬁes some core features of organizations

(Leavitt, 1964):

Social: Structure regularized aspects of relationships

among participants in an organization that may be

both normative (embodying what ought to be) or

factual order (actual structures).

Participants: individuals who in return for a variety

of inducements make contributions to the organi-

zation. Participants may belong to more than one

organization.

Goals: An organizational goal is a desired state of

affairs which the organization attempts to realize.

Goals are central to how an organization functions

and are often vague or very speciﬁc.

Technology: This is the means by which work is

performed in an organization. Technology can

be interpreted as a manufacturing plant, the soft-

ware systems enabling workers to perform work

or even technical knowledge and skills of partici-

pants.

Environment: Organizations exist in a speciﬁc

physical, socio-technical and cultural environ-

ment to which they must respond and adapt. All

aspects of an organization are inﬂuenced and

contextualized by the environment. For exam-

ple, software systems are purchased from external

providers or developed by technicians trained in

some other organization.

These features are generic to organizations and can

form the basis for extracting key concepts of an orga-

nization. Carley (Carley and Gasser, 1999) presents a

similar set. Note that these features may vary in some

way when viewed through a particular perspective or

metaphor. The last century has seen three dominant

perspectives (and overlaps) dominating research in

organization theory: Organizations as Rational Sys-

tems; Organizations as Natural Systems and Orga-

nizations as Open Systems. A rational system per-

spective denotes a focus on efﬁciency and optimiza-

tion and ultimately presents a reductionist model. The

open systems perspective is of most relevance to us as

it ranges from a simple clockwork view (a dynamic

system with predetermined motions), cybernetic view

(a system capable of self-regulation in terms of exter-

nally prescribed criterion such as a thermostat) to an

open system (a system capable of self-maintenance

based on throughputs of resources such as a living

cell) (Buckley, 1967).

These views are categorized by Gazendam

(Gazendam et al., 1998) and conform to essentially

two categories: Classical Organizational theories and

Systems theories. He suggests that classical the-

ories have a strong correspondence to a machine

metaphor where the organization as a whole con-

sists of agents performing tasks in ﬁxed structures

NextGenerationEnterpriseModelling-TheRoleofOrganizationalTheoryandMulti-AgentSystems

483

consisting of agent tasks, communication paths and

spatio-temporal orderings. Systems theories on the

other hand view the organization comprising of sub-

organizations fulﬁlling a speciﬁc function. Gazendam

furthermore notes that: “System theories of organiza-

tion are relatively poor because they only pay atten-

tion to the system level, and remain rather abstract.”

Theories based on the machine metaphor have

formed the basis of research on (multi-) agent-based

system in the late 1980s and 1990s to study alternative

viewpoints for describing organizations (Wooldridge,

2009). Here an agent is an autonomous and intelli-

gent being such as a human or a simulator of a human

realized by software (a computer agent) (Gazendam

et al., 1998). Systems that are comprised entirely of

computer agents have been used as simulations of or-

ganizations and correspondingly offer interesting per-

spectives on the study of organizations. Multi-agent

systems (MAS) and the associated Computational Or-

ganizational Theory (Carley and Gasser, 1999) pro-

vide the collective apparatus for investigation.

Computational Organizational Theory (COT)

aims to understand and model both human organi-

zations and artiﬁcial organizations (multi-agent sys-

tems) that exhibit collective organizational properties

such as the need to act collaboratively. Typical out-

puts of such research are the generation of new con-

cepts and theories about organizing and organizations.

Historically many applications and models have been

constructed but our review of the current enterprise

modeling literature indicates that COT has not yet

been applied to some of the tricky problems of en-

terprise modeling such as Business-IT alignment dis-

cussed in the introduction.

There are immediate information processing re-

quirements that are deducible from the deﬁnition of

organization such as: information ubiquity, tasks,

uncertainty distribution of organizational intelligence

and necessity of communication through a model-

based perspective. COT also suggests that: organi-

zations are modelable, and so are manipulatable; are

able to be designed to ﬁt speciﬁc needs and there is an

assumption that the costs of modeling and researching

organizations in simulation mode rather than in vivo

are lower (Carley and Gasser, 1999).

Key characteristics of organizations such as that

described by Leavitt and Carley emphasize structure

and behavior. (Hoogendoorn et al., 2007) propose

these two aspects as necessary pre-requisites for mod-

eling change when using MAS. In their proposal, or-

ganizations are described solely by the way groups

and roles are arranged to form a whole. Related to

this, Giorgini et al. use the i* framework (Yu, 1997)

to deﬁne a series of architectural organizational styles

which act as metaclasses and offer a set of design pa-

rameters for coordinating goals, actions and behavior

and therefore govern how an organization functions

(Kolp et al., 2006; Argente et al., 2006). Our posi-

tion contributes to enterprise modeling technologies

by drawing upon research outputs from COT to meet

the needs of an adaptive organization located in a sys-

tematic understanding of socio-technical nature of an

organization (Bean, 2010).

If MAS and COT are an appropriate way forward,

then there are additional requirements for methods

that can support COT based approaches. Those tasked

with modeling enterprises need guidance that: “al-

low the description of social structures, permit the use

of tools to perform project management, and include

IDE or CASE tools that facilitate the analysis and

design of MASs (Luck et al., 2005)”. Furthermore,

all methodologies need to contain enough abstrac-

tions to model and support MASs, which are usually

structured as societies of agents that play roles and

exchange information following predeﬁned protocols

(Isern et al., 2011). Isern et al. then go onto review a

range of agent-oriented methodologies by evaluating

their underlying meta models. Analysis of these meta

models guides us toward the essential features of the

language proposed in section 3.

We have posited that current approaches to EM

presents a linear form of enquiry requiring tacit

knowledge based on an Architect’s know how that

prevents scaling up to rapidly address “what if” type

of questions. Adopting technologies based on MAS

requires robust models for representing the complex-

ity and dynamic nature of organizations as they re-

spond to external business drivers. In particular then

MAS can be used to provide simulation models for

exploration of complex environments. Simulation

models can be explanatory models that can help iden-

tify kinds of behaviour expected under speciﬁc con-

ditions or they can predictive models that determine

more precisely the the kind of behaviour a system will

display in the future (Siebers and Aickelin, 2008).

Luck et al. propose a grouping of the agent-

technologies, tools and techniques that can address

these types of simulation for EM for theory building

about an enterprise at three levels: Organizational-

level (focusing on larger aggregations of structures;

Interaction-level (collaboration, communication and

decision making between agents) and Agent-level

(learning and reasoning (Luck et al., 2005). Cross cut-

ting concerns such as agent programming languages

and methodologies (noted earlier) provide practical

steps towards realization of agent systems.

In the next section, we discuss this partitioning has

been used to inﬂuence our proposal.

ICSOFT-EA2014-9thInternationalConferenceonSoftwareEngineeringandApplications

484

Figure 1: Component Abstraction (Core Concepts).

3 NEXT GENERATION EM

We posit that any approach that is derived from ideas

from the previous sections relies on being able to rep-

resent and process an organization that is expressed

in terms of a component-based abstraction. We envis-

age a product-line approach (Reinhartz-Berger et al.,

2013) whereby a suite of tools based on this abstrac-

tion is used to facilitate a collection of different orga-

nization analysis and simulation activities. Each ac-

tivity will constitute a domain, e.g., cost analysis, re-

source analysis, mergers and acquisition, regulatory

compliance. In principle, each new domain will re-

quire a new domain speciﬁc language to represent the

concepts. How should such a proliferation of domains

be accommodated by a single component abstraction?

Our proposal is to construct an extensible kernel

language that is used as the target of translations from

a range of domain speciﬁc languages (DSLs). Each

DSL supports an organization analysis and simulation

use-case. We then aim to construct a virtual machine

for the kernel language so that it is executable. Model

execution supports organization simulation and some

analysis use-cases. Links to external packages such as

model-checkers will complete the analysis use-cases.

The use of a single kernel language provides a sin-

gle focus of development effort and can help min-

imize the problem of point-to-point integration of

analysis methods. Our proposal is that a small core

collection of concepts, including component, inter-

face, goal, event, function as shown in ﬁgure 1, are

a suitable basis for most types of analysis and simu-

lation use-case and therefore the kernel language will

be deﬁned in terms of these concepts.

Given its ability to accommodate multiple simula-

tion and analysis use-cases, we envisage the language

being the basis of a suite of organizational modeling,

simulation and analysis tools, presented in the form of

a single integrated extensible meta-tool EA Simula-

tion Environment (EASE-Y) shown in ﬁgure 2. Since

organizational information is likely to be very large

(at least many tens of thousands of model elements)

Figure 2: The EASE-Y Architecture.

it is important the tool is implemented efﬁciently, is

scalable, supports distributed concurrent development

and is ﬂexible in terms of its architecture. To this end

we aim that the kernel language should be compiled

to a machine language running on a dedicated kernel

engine, the language integrates with standard repos-

itory technology, and can run equally well on single

machines, networked machines and via the cloud.

Organizations consist of many autonomous com-

ponents. Components are organized into dynamically

changing hierarchical groups, operate concurrently,

and manage goals that affect their behavior. We aim

for the kernel language to reﬂect these features by

having an operational semantics based on the Actor

Model of Computation (AMC) (Hewitt, 2010) and its

relation to organizations, or iOrgs (Hewitt, 2009). Ac-

tors have an address and manage an internal state that

is private and cannot be shared with other actors in the

system.

Execution proceeds by sending asynchronous

messages from a source actor to the address of a tar-

get actor. Synchronous messages can be achieved

by sending an actor in an asynchronous message to

which the result should be sent. Each message is han-

dled in a separate execution thread associated with the

target of the message and the message itself (collec-

tively referred to as a task). During task-execution an

actor may choose to change its state and behavior (be-

coming a new actor) that is immediately available to

process the next message sent to the target address.

Our claim is that the AMC provides a suitable

basis for execution and analysis of the concepts dis-

cussed in section 2. Actors, sometimes individually

and sometimes collectively, can be used to represent

the features of a component. The rest of this section

lists the key features that must be supported by the

kernel language and how the actor approach can sup-

port them:

[adaptability] This is required because organiza-

tional components may change dynamically during a

simulation. Resources, individuals, and even depart-

NextGenerationEnterpriseModelling-TheRoleofOrganizationalTheoryandMulti-AgentSystems

485

ments may move location, and have an affect on re-

sults. Furthermore, the behavior of a component may

change over time as information changes within the

system. An actor can, in principle, change behavior

as a result of handling each message.

[modularity] Each part of an organization is in-

tended to perform a business function that can be ex-

pressed in terms of a collection of operations. The

internal organization in terms of people, IT systems

and the implementation of various business processes

is usually hidden. The AMC provides an interface

of message handlers for each actor. Both the state

and the implementation of the message interface are

hidden from the outside. The speciﬁcation of an ac-

tor in terms of its external interface can be expressed

in terms of LTL formulas that constitute the external

goal for a component.

[autonomy] A key feature of an organization

is that the behavior of each sub-component is au-

tonomous. A particular department is responsible for

its own behavior and can generate output without the

need for a stimulus. The AMC is highly concurrent

with each actor being able to spawn multiple threads

and over which other actors have no control (unless

granted by the thread originator).

[distribution] An organization may be distributed

and this may be an important feature of its simulation.

Furthermore, we have a requirement that the tooling

for organizational analysis and simulation should sup-

port distributed concurrent development. The AMC

associates actors with addresses to which messages

are sent. Execution does not rely on the particular

location of the actor (i.e. the mapping between the

address and the actor behavior) that can be in the

same address space, via a network connection or in

the cloud.

[intent] In addition to autonomous behavior, an

organization component exhibits intent. This might

take the form of an internal goal that guides the be-

havior of the component to ensure that it contributes

to the overall mission of the organization. Although

actors do not directly provide support for such goals,

we intend to use results from the ﬁeld of Multi-Agent

Systems (Wooldridge, 2009) where support for goal-

based reasoning is provided within each agent when

determining how to handle messages.

[composition] An organization is an assembly of

components. As noted above, the topology of an or-

ganization may be static or dynamic. Actors can be

nested in more than one way. Actor behaviors are de-

clared and new actors are dynamically created with an

initial behavior (much like Java classes). The scope

of actor behaviors can be nested to provide modular-

ity. Adding a dynamically created actor to the state

of a parent actor provides composition. Such actors

can be sent as part of messages. If the source actor

retains the address, then the communicated actor be-

comes shared between the source and the target of the

message.

[extensibility] Our aim is to support a number of

simulation and analysis use-cases. As such the kernel

language will need to support a collection of indepen-

dent domains. Whilst we expect the DSLs to target

the kernel language it is likely that each domain will

have its own fundamental concepts and actions (so-

called Therbligs, (Stanton, 2006)). We envisage such

domain-speciﬁc features being deﬁned in the kernel

language and then pre-loaded to form an augmented

target language for DSL translations.

[event-driven] Organizational components can-

not rely on when communications occur and where

they originate. In addition, a component may sim-

ply cause an event to occur without knowing who

will consume the event. This is to be contrasted with

message-based communication where the target is al-

ways known to the source and where sometimes the

message carries information about the source that be-

comes available to the target. The AMC is based on

message passing where the source knows the address

of the target. Given that the kernel language is the tar-

get of DSL transformations, support for event-based

communication becomes an architectural issue where

events are simply messages that are sent to an ac-

tor container that is responsible for delivering event-

messages to dynamically changing collections of ac-

tors. Providing that the transformation establishes the

correct assembly of actors and conforms to an ap-

propriate message passing protocol then component

events are supported without needing to make them

an intrinsic part of the kernel.

4 CONCLUDING REMARKS

This position paper has proposed that current genera-

tion enterprise modeling languages and technologies

support a linear form of enquiry that requires tacit

knowledge based on an Architect’s know how. Such

an approach prevents scaling up to rapidly address

“what if” type of questions that face organizations

as they seek to adapt to respond to ongoing change.

At an abstract level, these types of requirements

have been studied in other disciplines, including eco-

nomics, political science, philosophy and linguistics

leading to computation based organizational theo-

ries and technologies for describing agent interaction,

communication and decision-making. We have pre-

sented an argument that traces a route through (com-

ICSOFT-EA2014-9thInternationalConferenceonSoftwareEngineeringandApplications

486

putational) organization theory to propose that next

generation enterprise modeling languages should ad-

dress COT and multi-agent system approaches to pro-

vide a rich simulation platform that supports both ex-

planatory models and predictive models for the “what

if” question. In doing so, we recognize that there are

open-ended research questions around methodology

and proposed the simulation platform. We plan to

validate our proposition in a number of ways. We

are currently developing a collection of representa-

tive case studies based on real-world data in a lab-

oratory setting. One case study illustrates how the

proposed ideas and techniques can help data-driven

decision making in an IT services providing organi-

zation. Another case study will address merger and

acquisition problem in wealth management domain.

We intend to run co-design workshops with Business

Management domain experts in order to evaluate their

response to our proposals. We are currently extending

µLEAP (Clark and Barn, 2014) to be the target kernel

language. We plan to design and implement the ker-

nel language meta-model as a virtual machine, pos-

sibly using multiple Java VMs as targets. Scalability

of the platform in terms of the number of runtime vir-

tual machines is a critical factor for acceptability of

the platform in practice. Recent research by Bolz and

Tratt (Bolz and Tratt, 2013) seems a promising direc-

tion in this regard.

REFERENCES

Argente, E., Julian, V., and Botti, V. (2006). Multi-

agent system development based on organizations.

Electronic Notes in Theoretical Computer Science,

150:55–71.

Bean, S. (2010). Re-thinking enterprise architecture using

systems and complexity approaches. Journal of En-

terprise Architecture, 6(4):7–13.

Bolz, C. F. and Tratt, L. (2013). The impact of meta-tracing

on vm design and implementation. Science of Com-

puter Programming.

Buckley, W. (1967). Sociology and modern systems theory.

Prentice-Hall.

Carley, K. M. and Gasser, L. (1999). Computational or-

ganization theory. Multiagent systems: A modern

approach to distributed artiﬁcial intelligence, pages

299–330.

Chan, Y. E. and Reich, B. H. (2007). It alignment: what

have we learned? Journal of Information technology,

22(4):297–315.

Clark, T. and Barn, B. S. (2014). Outsourcing service pro-

vision through step-wise transformation. In Proceed-

ings of the 7th India Software Engineering Confer-

ence. ACM.

Gazendam, H. W., Jorna, R. J., et al. (1998). Theories about

architecture and performance of multi-agent systems.

University of Groningen.

Hewitt, C. (2009). Norms and commitment for iorgs

(tm) information systems: Direct logic (tm) and

participatory grounding checking. arXiv preprint

arXiv:0906.2756.

Hewitt, C. (2010). Actor model of computation: scal-

able robust information systems. arXiv preprint

arXiv:1008.1459.

Hoogendoorn, M., Jonker, C. M., Schut, M. C., and Treur,

J. (2007). Modeling centralized organization of orga-

nizational change. Computational and Mathematical

Organization Theory, 13(2):147–184.

Isern, D., S

anchez, D., and Moreno, A. (2011). Organiza-

tional structures supported by agent-oriented method-

ologies. Journal of Systems and Software, 84(2):169–

184.

Kolp, M., Giorgini, P., and Mylopoulos, J. (2006). Multi-

agent architectures as organizational structures. Au-

tonomous Agents and Multi-Agent Systems, 13(1):3–

25.

Leavitt, H. J. (1964). Applied organization change in indus-

try: structural, technical and human approaches. New

Perspectives in Organizational Research, 55:71.

Luck, M., McBurney, P., Shehory, O., and Willmott, S.

(2005). Agent technology: computing as interaction

(a roadmap for agent based computing).

Luftman, J. (2004). Assessing business-it alignment ma-

turity. Strategies for information technology gover-

nance, 4:99.

Parsons, T. and Jones, I. (1960). Structure and process in

modern societies, volume 3. Free Press New York.

Reinhartz-Berger, I., Cohen, S., Bettin, J., Clark, T., and

Sturm, A. (2013). Domain Engineering. Springer.

Scott, W. R. (1992). Organizations. Prentice-Hall Engle-

wood Cliffs, NJ.

Siebers, P.-O. and Aickelin, U. (2008). Introduc-

tion to multi-agent simulation. arXiv preprint

arXiv:0803.3905.

Stanton, N. A. (2006). Hierarchical task analysis: Devel-

opments, applications, and extensions. Applied er-

gonomics, 37(1):55–79.

Veken, K. V. d. (2013). Enterprise architecture modelling

to support collaboration-the archimate language as a

tool for communication.

Wooldridge, M. (2009). An introduction to multiagent sys-

tems. John Wiley & Sons.

Yu, E. (1997). Towards modelling and reasoning support

for early-phase requirements engineering. In Require-

ments Engineering, 1997., Proceedings of the Third

IEEE International Symposium on, pages 226–235.

IEEE.

Zachman, J. A. (1987). A framework for information sys-

tems architecture. IBM systems journal, 26(3):276–

292.

NextGenerationEnterpriseModelling-TheRoleofOrganizationalTheoryandMulti-AgentSystems

487