METAMODEL-BASED DECISION SUPPORT SYSTEM FOR

DISASTER MANAGEMENT

Siti Hajar Othman and Ghassan Beydoun

School of Information Systems and Technology, Faculty of Informatics

University of Wollongong, Wollongong NSW 2522, Australia

Keywords: Modelling Language, Metamodel, Decision Support System, Disaster Management, Model.

Abstract: Generally software model developers use a general purpose language such as Unified Modelling Language

(UML) in modelling their domain application models. But when they come to the situation in which the

models they create do not perfectly fit the modelling needs as they desire, a more specific domain modelling

language offers a better alternative approach. In this paper, we create a Disaster Management (DM)

metamodel that can be used to create a disaster management language. It will serve as a representational

layer of DM expertise leading to a DM decision support system based on combining and matching different

DM activities according to the disaster on hand. A creation process of the metamodel is presented leading to

the synthesis of initial metamodel, as a main component to create a decision support system to unify,

facilitate and expedite access to DM expertise.

1 INTRODUCTION

Various kinds of modeling languages have been

applied in different disciplines including systems

engineering (Weilkiens, 2008), software

engineering, information management, computer

science and business process modelling. In this

research we study disaster management (DM) as a

specific domain, where its own language will be

modelled. Through this paper, we develop a DM

metamodel as a precise definition of the constructs

and rules needed for creating the semantic models of

this domain. Generally, who the users are of a

metamodel depend on the type of created

metamodel. Traditionally users will be among CASE

tool vendors, modelling tool vendors, method

engineers, repository vendors, system integrators,

researchers and end users. In this context, end users

would include emergency managers, disaster

management coordinators or safety managers for

various public and private organizations seeking to

create a DM model to manage anticipated disasters.

The increasing number of disasters recently, such

as earthquakes, tsunamis, floods, bushfires, air

crashes, epidemic, have posed a huge challenge not

only to population at large, but also to public

services and agencies tasked with activities relating

to preventing and managing disaster responses.

Recent failures can be easily identified in the

management of the Swine-Flu (H1N1) pandemic

hitting Australian shores in large numbers through

cruise ships or in the devastating communication

failures in the recent bushfires in Victoria

(Australia). Many such failures are due to expertise

not being available in a timely manner. This is partly

due to inability to recognize and identify correct

expertise, as it is often perceived as too tied to kinds

of events (floods, bushfires, tsunamis, pandemic or

earthquake). Potential for reusing expertise is often

overlooked leading to catastrophic consequences. In

this paper, we present an approach to unify DM

knowledge to create a DM Decision Support System

(DSS) that combines and matches different DM

activities to suit the disaster on hand by using the

DM metamodel.

The approach proposed in this paper is inspired

by a software engineering knowledge management

practice known as method engineering

(Brinkkemper, 1996) which involves storing various

software methodologies as a collection of reusable

process fragments for later reuse to create hybrid

methodologies as new software development

projects arise. In DM, the first step and the focus of

this paper are to appropriately represent DM

knowledge and to warehouse DM knowledge in an

403

Hajar Othman S. and Beydoun G. (2010).

METAMODEL-BASED DECISION SUPPORT SYSTEM FOR DISASTER MANAGEMENT.

In Proceedings of the 5th International Conference on Software and Data Technologies, pages 403-412

DOI: 10.5220/0003006504030412

 SciTePress

appropriate form to later allow mixing and matching

DM experiences. The appropriate representation of

DM knowledge will enable the creation of a

repository of DM experiences. Interfacing this to a

DSS that takes as input new disaster parameters, will

assist in deciding the best DM approach by

combining various actions from previous DM

experiences.

2 RELATED WORKS

Failures in preventing disasters or failures in their

subsequent management are rarely caused by a

single factor. They are often due to an accumulation

of complex chain of events and often accompanied

by changes in external environment factors (Aini,

Fakhrul-Razi et al., 2005). Hence, it is common

wisdom that no two disasters are exactly the same,

and that every disaster requires its own management

process. However, the way disasters impact human

lives and business processes may well be similar and

responses are often transferrable between disasters.

Evacuation of personnel for example is a DM action

that is applicable in many disaster situations. This

paper aims to use a generic representational layer (a

metamodel) to give a unified view of common

concepts and actions that apply in various disasters.

We use existing DM and security models (Asghar,

Alahakoon et al., 2006; Russo, Raposo et al., 2006;

Benaben, Hanachi et al., 2008; Beydoun, Low et al.,

2008; Kruchten, Monu et al., 2008; Beydoun, Low

et al., 2009) and DM literature produced by World

Health Organisation and Emergency Management

Australia, as a starting point towards creating a

repository of past DM experiences to be stored as

reusable components and expressed using concepts

identified in a generic DM metamodel. This will be

the first to create a DSS to enable formulating DM

approaches as new situations arise.

Our work also draws on research from method

engineering (Brinkkemper, 1996) and

metamodelling (Nordstrom, Sztipanovits et al.,

1999). Method engineering is an application of

knowledge based technology underpinned by

software engineering results for completion of

knowledge representation and acquisition.

Metamodelling, a central activity promoted by the

efforts of the Object Management Group (OMG)

(Object Management Group (OMG), 2003), has also

been promoted in method engineering. It aims to

create interoperable, reusable, portable software

activities and components. In this context, a

metamodel is a fundamental building block that

makes statements about the possible structure of

models (Stahl, Voelter et al., 2005). It is usually

defined as a set of constructs of a modelling

language and their relationships, as well as

constraints and modelling rules without necessarily

the concrete syntax of the language (Beydoun, Low

et al., 2009). We use metamodelling in our work to

develop existing tentative attempts to represent DM

knowledge in a reusable form to give a unified point

of access supported by an intelligent DSS. In

particular, we illustrate our unification approach by

presenting an initial metamodel that we believe

could generalize most of the concepts used in

existing DM models.

3 METAMODEL-BASED DM

KNOWLEDGE MANAGEMENT

DM is defined as a management of all aspects of

planning and responding to all phases in disaster as

illustrated in Table 1. These phases include

mitigation, preparedness, response and recovery

activities (W3C Incubator Group, 2008). This

definition includes the management of risks and

consequences of a disaster. Large disasters cut

across many boundaries including organizational,

political, geographical, topical and sociological. This

presents serious challenges in interoperability

between various teams and creates difficulties in

collaboration and cooperation across authorities,

countries and systems. Moreover, data collection

and integration problems arise as various

technologies and tools are typically involved in data

gathering and monitoring e.g. Global Positioning

Systems (GPS), Geographical Information Systems

(GIS), data collection platforms and early warning

systems. A solid, general and global framework for

coordinating people involved and interoperates with

data, during and after disaster through is still

inadequate.

In metamodelling DM knowledge, we uncover and

make explicit key aspects of activities, cooperation

and components in DM. Surveying a number of

existing DM models (shown in Table 1), we observe

that some concepts represent a similar DM activities

which are expressed differently. For example, in a

Circular Model for Disaster (Kelly, 1998), the

terminology ‘Emergency Response’ is being used to

represent the response and rescue activity of disaster

victims. But, the same activity however is

represented by using ‘Emergency State’ in Ibrahim-

Razi Model in (Shaluf, 2008).

ICSOFT 2010 - 5th International Conference on Software and Data Technologies

404

Table 1: DM concepts used to represent various DM activities in existing disaster models.

Author Model

DM Phase

Mitigation Preparedness Response Recovery

In (Kelly, 1998)

Circular Model for

Disaster

Disaster mitigation Disaster prevention,

Disaster preparedness

Warning,

Disaster,

Emergency response

Rehabilitation,

Reconstruction,

Development

In (Tierney, Lindell

et al., 2001)

Disaster Phases and

Time Period Model

Hazard

vulnerability,

Hazard mitigation

Emergency

preparedness

Emergency response Disaster recovery

In (Manitoba Health

Disaster

Management, 2002)

Integrated Disaster

Management Model

Hazard assessment,

Strategic plan,

Mitigation

Risk management,

Preparedness

Response Monitoring and

evaluation

In (Doherty, 2004)

An Emergency

Management Model for

Home Health Care

Organizations

Mitigation Preparedness Response Recovery

In (Ahmed, 2008)

Expand-Contract Model

Prevention and

mitigation strand

Preparedness Strand Relief and Response

Strand

Recovery and

Rehabilitation Strand

In (Asghar,

Alahakoon et al.,

2006)

A Comprehensive

Conceptual Model For

Disaster Management

Hazard assessment,

Strategic planning,

Mitigation

Risk management,

Preparedness

Response

Recovery, Monitoring

and evaluation

In (Shaluf, 2008)

Ibrahim-Razi Model

Inception of errors,

Accumulation of

errors

Warnings, Disaster

impending stage,

triggering event

Emergency state,

Disaster

Normal state

In (Ahmed, 2008)

Traditional DM Cycle

Model

Mitigation Preparedness Disaster Impact Reconstruction,

Rehabilitation

Managing and sharing knowledge of this

complex domain is hard. A unified metamodel can

ensure that key concepts are easily presented to

newcomers and can increase portability of various

models across supportive modelling tools. It can also

create better communication amongst practitioners

and research could then focus on improving and/or

realizing a unified body of knowledge (Beydoun,

Low et al., 2009). Advantages of metamodel

developed in this paper are as follows:

• Facilitating global communication among

different disaster emergency users as the

metamodel has generalized all the concepts that

must exist in this domain.

• Simplifying teaching new created model of this

domain among the model users through a set of

syntax and semantic rules provided.

• Providing guidelines for creating a

comprehensive disaster management model

which can cover the whole phases of DM (e.g:

Earthquake Emergency Response Model -

Response phase and Bushfire Risk Reduction

Model - Mitigation phase).

• Enabling users to create new customised DM

model based on choosing and combining sets of

concept component based on their own model

requirement.

• Highlighting scope for improvement in a DM

practice through validation against other DM

metamodels.

3.1 A Metamodel-based DM Decision

Support System

Developing a DM metamodel is our first step

towards creating a DSS to unify, facilitate and

expedite access to DM expertise. This metamodel

will describe the various DM activities and desired

outcomes and serve as a representational layer of

DM expertise, enabling an appropriate DM DSS

based to guide combining and matching different

DM activities according to the disaster scenario on

hand. The DM metamodel will be complemented

with a Disaster Retrieval Model that will be used to

choose appropriate procedures and suit with

different kinds of disaster (natural or man-made) on

hand. The computational architecture of our system

and the integration of a DSS platform will be context

independent. For instance, different countries have

their own organization in coordinating and act as an

advisory board for handling disaster activities.

Most countries have a government agency to

manage major disasters. For example, in Australia

there is EMA (Emergency Management Australia),

in the USA there is FEMA (Federal Emergency

Management Agency) in Canada there is the PSC

(Public Safety Canada). Hence for the purpose of

developing our DM metamodel, models of different

DM activities as applied by different countries are to

be combined and stored into one database namely

DM Activities Repository. This will be a collection

of organizational, operational, planning, logistics

and administration procedures and policies executed

METAMODEL-BASED DECISION SUPPORT SYSTEM FOR DISASTER MANAGEMENT

405

by these countries through their DM processes.

These will be identified and organized according to

the DM metamodel consisting of common concepts

used in all four disaster phases.

The generic DM metamodel based on identified

common concepts is the destination point of

scattered concepts used in many DM activities

worldwide. A process towards concept

generalization is applied to make our DM

metamodel more applicable (described in the next

section). Activities from different sources (and

countries) will be stored as Procedure Fragments in

the DM Activities Repository. The DSS will assist in

deriving the best disaster procedure fragment

solution according to the disaster on hand. It will use

a set of rules that will specifically determine what is

the best solution based on disaster description input

entered by a user of the system (e.g: local disaster

manager, emergency coordinator and researcher)

and the repository.

We adapt a Case-Based Reasoning and a Model-

Based Reasoning technique in the way we determine

the best decision solution for our DSS system. In

this system architecture, we integrate an exception

tolerant technique (Gao and Xu, 2010) to handle any

exception problems which may occurred during the

enactment of this DM workflow. Some of the

situations in disaster context which require the

exception handling include:

• Hospital emergency department fails to get clear

information concerning the location in which the

actual disaster had happened;

• Commands receive from high authorities for

coordination of emergency operations is vague;

• Emergency equipment and resource failures;

• Emergency Call Centre receives unclear disaster

call information;

• GIS fails to acquire a real data from the disaster

real location;

• Missing of DM policies and procedures in

handling certain emergency operation or;

• Aid distributions fail to deliver.

To produce the DSS system and a populated DM

knowledge repository, the first step is we construct

the DM metamodel. Besides the metamodel, we

classify and formulate all DM activities procedures

into a unified repository. A knowledge based

interfaced to the repository is developed in order to

support a retrieval and integration of the procedure

fragments. In the next section, we overview existing

relevant metamodels and describe the process of

formulating our DM metamodel, before the

metamodel is presented in Section 4.

4 METAMODELLING PROCESS

In this section, we present details of our model

creation process and all components which support

its development. To create our DM metamodel, we

adapt a metamodelling approach from the work

used to develop a Framework for Agent Modelling

Language (FAML) in (Beydoun, Low et al., 2008)

and (Beydoun, Low et al., 2009). The approach of

our metamodelling creation process consists of 7

main steps:

Step 1: Identifying models by using Model Importance

Factor (MIF) to find the best collection of DM

models;

Step 2: Extraction of general concepts relevant to all

identified DM model which have been derived in

Step 1 (Asghar, Alahakoon et al., 2006; Russo,

Raposo et al., 2006; Benaben, Hanachi et al.,

2008; Beydoun, Low et al., 2008; Kruchten, Monu

et al., 2008; Beydoun, Low et al., 2009) will be

reviewed in this section;

Step 3: Candidate concepts are short-listed;

Step 4: Differences between concepts are reconciled;

Step 5: Chosen concepts are designated into relevant sets

Step 6: Relationships among concepts are identified

leading to the initial DM metamodel (Figure 1);

Step 7: Validating the metamodel.

4.1 Step 1: Identifying Models by using

Models Selection Criteria (Model

Importance Factor)

Existing models used to describe disasters provide a

starting point to identify commonly used concepts in

DM. Many disaster models have been developed by

many researchers and organizations globally. To

select a subset of 10 models for our metamodelling

process, we formulate a criterion named as Model

Importance Factor which its derived formula is as

shown in (1). It calculates the weight to categorize

which model is the most relevant to be chosen as

required for Step 1 (model for relevant set) and Step

7 (model for validation) models. In developing the

MIF, we adapt the idea of Journal Impact Factor

measuring the frequency of which the average article

in a journal has been cited in a particular year. Our

MIF compare the impact of the models in the same

domain. It is defined as follows:

ICSOFT 2010 - 5th International Conference on Software and Data Technologies

406

cited

The total number of Times the model or

metamodel is cited (Paper & Journal);

For a model appear in a publication

without citation, default weight is set to:

• Thesis - 10; Report - 15;

current

The current Year calculation is made;

published

The Year model is published;

Elevel :

Weight of Effort is calculated based on

level of model developer by using weight:

• 0.1 - Individual; 0.2 - National

Organization, 0.3 – International

Organization;

Rcoverage :

The weight of

elevancy represents how

pertinent and applicable the model to the

DM metamodel development

requirement;

Participant (P)

The number of Participants involved in

developing a model.

MIF = (T

cited

* (E

level

* P) * Rcoverage)

((Y

current

+ 1) - Y

published

))

(1)

In this formula, we set 6 criteria in categorizing

which models contain the highest priority to be

chosen as a set of best DM models. The first

criterion is T

cited

which indicates the total number of

times the model is cited, specifically model comes

from journal or conference paper. Except for a

publication which cannot be determined its citation

number, we set a default weight according to the

types of the publication (e.g: Thesis is set 10 and

Report produced by a government and other

organization is set 15). Then this followed by

determining the current year when the calculation is

made through Y

current

and the year when the model is

published by Y

published

. Another criterion we set is

level

that will evaluates the weight of effort in which

the level of model developer is calculated by using

index of 0.1 for individual (e.g: university

researcher), 0.2 for national organization (e.g: EMA)

and 0.3 for international organization (e.g: WHO).

A weight of a total number of people involved in

developing the model is also considered. To support

this condition, P is constructed to represent the

number of participants involved in developing the

model. The last criterion we set is R

coverage

calculate the weight of how relevant, pertinent and

applicable is the model to the DM metamodel

development requirement. For example, a ‘Manitoba

Health Disaster Management Model’ (Manitoba

Health Disaster Management, 2002) was set 0.3

weight contribution as it could cover most of the

whole DM aspect in his model, whereas an

‘Evacuation Model’ gets less weight compared to

Manitoba model as it cover only a small portion

from the whole DM domain concepts. Below is two

sample of MIF calculation for Manitoba model

(1.00), Russo model (MIF: 0.24) and Kructhen

model (MIF: 0.12).

cited

current

published

standard

P R

coverage

MIF

15 2010 2002 0.2 10 0.3

1.00

10 2010 2006 0.1 4 0.3

0.24

3 2010 2008 0.1 4 0.3

0.12

Even though we have set up the MIF criteria, we

still consider a contribution from variety of sources

in selecting a subset of DM models to be used in

Step 1 process. For instance from thesis, government

(state, country) or private organization reports,

papers and journals. This to ensure that the

metamodel we develop will not only consider an

academic impact point of view (e.g: number of how

many paper of model is cited) but it also includes

organization and government impact such as from a

government, private or any emergency organization

reports.

4.2 Step 2: Selecting Concepts from

Models

We implement manual extraction in deriving the

concepts from each model we identify in step 1.

Here we present some of the samples. The first

metamodel we observe is Benaben’s metamodel

(Benaben, Hanachi et al., 2008) expressed using

Web Ontology Language (OWL) and focuses on

crises management. This metamodel elaborates a

common and sharable reference model built to

characterize crisis situations in three interrelated

views namely System, Treatment System and Crisis

Description. Benaben’s metamodel covers the whole

crisis characterization and collaborative processes

that deal with it, aiming to integrate partners through

information system interoperability. We derive 21

concepts from Benaben model. Some of them are

Collaborative Process, Procedure, Task of Actor,

Actor on site, Event, Trigger, Crisis, Danger, Risk,

Effect, Indicator, Gravity Factor, Civilian Society,

People, Service of Actor and Resource.

The second metamodel we use is Kruchten’s

(Kruchten, Monu et al., 2008) which conceptualises

disasters as encompassing multiple stakeholder

domains depicted in four main views: Disaster

Visualization, Physical View, Communication and

Coordination Simulator and Disaster Scenario. The

metamodel aims to create a common language to

communicate, analyze and simulate

interdependencies about disaster scenario without

having to disclose all critical and confidential data

between parties involved. This metamodel attempts

METAMODEL-BASED DECISION SUPPORT SYSTEM FOR DISASTER MANAGEMENT

407

to unify the terminology sharpening the definition of

terms and their semantic relationships.

The third metamodel we consider is Asghar’s

(Asghar, Alahakoon et al., 2006) which focuses on

the arrangement of disaster activities in a logical

sequence. This metamodel is built by linking

disaster management actions with hazard and risk

assessment activities. The model also incorporates

environmental conditions, making it possible to

analyse and separate the environmental issues from a

disaster. And from this metamodel we derive 12

concepts including Strategic planning, Risk

Management, Mitigation, Preparedness, Response,

Recovery, Early warning, Resource management,

Environmental affects, Damage assessment,

Coordination and Hazard assessment. Another

metamodel we use is Russo’s (Russo, Raposo et al.,

2006) which focuses on configuring collaborative

virtual workspaces specifically in DM of oil and gas

offshore structures. It investigates how a distributed

workspace environment can support DM involving

distributed collaborative technical teams to work as

a collaborative virtual team. This metamodel is

focused on one-specific-disaster approach.

Targeting a generic metamodel in our work is

inspired by (Beydoun, Low et al., 2008) and

(Beydoun, Low et al., 2009), where a generic

metamodel was developed for representing and

securing Multi Agent System (MAS). In fact, several

generic security concepts identified in (Beydoun,

Low et al., 2008) have their equivalent in DM. For

example, recovering from an intrusion attack in a

MAS requires restoring data logs. Analogies to this

exist in restoring many lost community services in

disaster scenarios, requiring maintaining back up

organizational structures. Our work takes DM

modelling a step further aiming to generalize various

types of DM activities concepts into one generic

encompassing metamodel.

4.3 Step 3: Short-listed Candidate

Concepts

The collection of existing DM models that we have

revised, assist us towards deriving the common

concepts used in all these models. It gives a total of

55 common concepts from 5 models we identified

and will be reconciled in the step 4.

4.4 Step 4: Reconcile Difference

between Concepts

In step 4, we reconcile differences between

definitions where possible. In choosing the common

concept definition to be used, study to all definition

of concepts that we have derived is crucial. If there

is a contradictory use of concept definition between

two or more sources occurs, then a process to

harmonize and fit the definition in the metamodel is

required. As for an example, a concept of People is

defined differently in a few models. Benaben

defined it as ‘All the group of persons which can be

threatened by the crisis situation’, but in Kructhen

model, the concept is defined as ‘Cell that contains

people’. Thus we will choose the definition used by

Benaben model as it fits our metamodel. Below are a

few examples of concept and its definition:

i) Event - An incident or situation, which occurs in

a particular place during a particular interval of

time.

ii) Effect -

An event that can produce other effects

or a noticeable consequence of a disaster.

iii) Risk - The potential disaster losses, in lives,

health status, livelihoods, assets and services,

which could occur to a particular community or a

society over some specified future time period.

iv) Disaster - A situation where serious disruption of

the functioning of a community or a society

involving widespread human, material, economic

or environmental losses and impacts, which

exceeds the ability of the affected community or

society to cope using its own resources.

4.5 Step 5: Designate Chosen Concept

into Relevant Sets

In Step 5, a process of designating all derived and

reconciled concepts is performed according to theirs

respective set. In DM metamodel, we choose to

present them in one layer which encompassing all

DM concepts that we have processed. Figure 1

illustrates each of these concepts.

4.6 Step 6: Define Relationships among

Concepts

After combining all DM concepts to its designated

diagram, the relationships exist among the concepts

are then created. For instance, between Disaster and

ElementsAtRisk concept, we set the Association

relationship by using (

) symbol with

‘AffectWellness’ to indicate that a disaster could

affect all elements which are at risk by a disaster. A

Specialization relationship (

) is used to signify

that Civilian societies, Infrastructures, Natural sites

and People are components which is categorized as

‘is a kind’ of the elements.

ICSOFT 2010 - 5th International Conference on Software and Data Technologies

408

Figure 1: DM Metamodel proposed.

4.7 Step 7: Validation

Validity threats to the resultant metamodel are

internal and external. The creation process ensures

that internal threats are resolved and that the

metamodel is consistent and coherent. However, to

deal with external threats, it is important to ensure

that the metamodel is capable of generating a wide

range of DM models. This threat is currently being

dealt with by ensuring that the metmaodel can

generate 10 models which were not used in its

construction. This validation is not part of this paper.

5 RESULTANT DM

METAMODEL

In this section we represent our initial DM

metamodel as the output in which Steps 1 to Step 6

processes as (described in Section 4) are iteratively

applied into it and also the refinement of this

metamodel. Our resultant metamodel contains the

relationships among concepts which can represent

the semantic of the domain as shown in Figure 1.

The core class in this DM Metamodel is the

Organisation which represents the loose

‘organisation’ where DM concepts are

operationalised. All key concepts in DM are grouped

in the Organisation concept. Other key DM

concepts are aggregated within this class and they

include: DMProcedure, DMRequirement,

DMPolicy, Actor, DMTeam, DomainKnowledge,

Resource, ActorRole and MessageCommunication.

DMProcedure can represent the collections of

implemented procedures of DM activities including

for example Mitigation, Preparedness, Rescue,

Response and Evacuation. DMTeam defines a

collection of ActorRole class which typically

describes human roles that work towards a DMGoal.

ActorTask class in our metamodel is derived from a

DMGoal class. Here we also model a

DisasterPreventionGoal as a class that can be

achieved by DisasterPreventionTask.

Some of the classes from the crisis metamodel

developed by Benaben in (Benaben, Hanachi et al.,

2008) is taken into consideration while we develop

the model of the actual disaster event (left hand side

of Figure 1). To model this, we grouped all

components consisting of People, Infrastructure,

METAMODEL-BASED DECISION SUPPORT SYSTEM FOR DISASTER MANAGEMENT

409

NaturalSite and CivilianSociety into ElementsAtRisk.

We introduced the ‘is a kind of’ (specialization)

relationships which tied up these four components

with ElementsAtRisk class. Thus,

DisasterActionService, a class of collaboration

among several actors will provide support and help

to this affected group of elements through the

ElementsAtRisk class. Disaster, a tragedy that affects

this ElementsAtRisk typically occurs due to

accumulation factors represented by a Trigger and

have consequences that are described by Effect and

vary in intensity represented by ComplexityFactor

and GravityFactor.

The metamodel is generic and generalizes

various kinds of disaster concepts that can be refined

according to the context on hand. It explicitly covers

the management of disaster in all four different

phases including mitigation, preparedness, response

and recovery. We anticipate that various concepts in

DM, their relationships and attributes, different

types of data models can be generated using

refinement of concepts in this metamodel.

5.1 Refinement of DM Metamodel:

Bushfire Disaster Case Study

To illustrate and validate the semantics of our

metamodel concepts, we refine concepts described

into the scenario of the recent bushfire disaster in

Marysville (Victoria, Australia) (shown in Table 2).

It illustrates the refinement of concepts we proposed

in our DM metamodel in a specific disaster domain.

As a result, Figure 2 shows the corresponding

refinement in a diagram, illustrating the

independence of our metamodel from any specific

disaster metamodel. It can be used to derive one of

many possible disaster models. Figure 2 is one of

many models that could be derived by using the DM

metamodel. For this bushfire model, it shows us

various factors caused this tragedy in Marysville

come from a combination of a hot weather

temperature (47 Celsius), low humidity level

calculated by using Fire Danger Index (less than

6%), strong north-westerly wind (with average 100

km/h) and an extremely dry of fuel moisture in

Marysville bushland area. All these combination are

identified as an example for the GravityFactor

concepts we introduced in our DM metamodel.

Whereas, a GlobalClimateChange is an instance of

ComplexityFactor in this case. Hence, a combination

of GravityFactor and ComplexityFactor concepts

will contribute to a result of bushfire factors. In this

model, the tragedy is made from a combination of a

factors comes from BushfireFactor concept.

Table 2: A refinement of DM metamodel concepts to a

specific-domain disaster (Bushfire in Marysville,

Victoria).

DM Metamodel

Concept

Bushfire Disaster Concept

DMProcedure

BushfireProcedure (e.g:

EvacuationProcedure and

FirstAttackFireProcedure)

DMPolicy

BushfirePolicy (ex: Bushfire-

RecoverPolicy, RescuePolicy,

PreparednessPolicy)

ActorDefinition

EmergencyTeam

DMTeam

MarysvilleBushfireTeam

Domain

Knowledge

BushfireDomainKnowledge

Resource

MarysvilleWaterResource,

FireFightingResource

ActorRole

RescueTeamRole

Message

Communication

StateWeatherofBureau

People

MarysvillePeople

DMGoal

BushfireManagementGoal

ActorTask

RescueTask (eg: FireServiceTask,

HealthDepartmentTask)

Disaster

PreventionTask

BushfirePreventionTask

The BushfireDisaster concept then affects the

MarysvilleAffectedElement including its

infrastructure (e.g: schools, shops, roads), natural

site (e.g: river, tree), civilian society (e.g: local

communities) and also people in Marysville. The

specific model (shown in Figure 2) suggests that to

create such a comprehensive DM organization a few

elements are required to this main concept. Those

elements are such DM policy (through

BushfirePolicy), various DM resources (e.g:

BushfireRescueResource and MarysvilleWater

Resource), emergency rescue team (as

MarysvilleBushfireTeam), role of emergency actor

(as BushfireRescuerRole), DM procedure (as

BushfireProcedure – with the example of fire attack

procedure and evacuation procedure in bushfire

case), and DM requirement (BushfireManagement

Requirement).

This organization ideally can make use of an

ontology of bushfire disaster (BushfireDomain

Knowledge) to support the DM team.

BushfireManagementGoal is a concept represents

our DMGoal. It is a specification of the state where

DM tries to achieve. This concept can be derived

from an action task conducts by emergency rescue

team. Thus, this situation is represented through

RescueTask concept which contains

‘IsDerivedFrom’ relationship to Bushfire

ManagementGoal concept. The model (of Figure 2)

also shows that bushfire risk can be reduced by

performing a rescue task action. Then,

BushfireActionService is a concept which

representing the service provided by a rescue actor.

ICSOFT 2010 - 5th International Conference on Software and Data Technologies

410

Figure 2: Metamodel refinement, a model of bushfire disaster in Marysville, Victoria, Australia.

With the aims to drive the disaster situation to a

more stable and handled state, they provide (through

‘Serves’ relationship) their rescue service

(BushfireActionService) to the components which

exposed to a disaster (MarysvilleAffectedElement).

During refinement of the metamodel, if there is

any difference occur to the original components of

the metamodel such as any process of adding new,

changing or even deleting any current class in new

model created by the user, all these cases will be

stored in system repositories (Heicke, 2009). As

discussed by Heicke, who studied the difference,

changes in metamodel, suggesting that to represent

difference in metamodel, every modelling element

of the original metamodel need to have options of

insertion, deletion or modification. It could be done

by constructing new constrains and rules in

managing the situation. Therefore, by implementing

this process the evolution of the metamodel could be

recorded accurately and precisely. We also adapt

this process to manage all dynamic issues occur in

our metamodel.

6 CONCLUSIONS

In this paper, we have illustrated a novel approach in

modeling a disaster management language through

DM Metamodel. We observed that many existing

disaster models are not based on any standard

metamodel but rather constitute proprietary solutions

mainly focused on frameworks and other model

example aspects. An important task of these works is

the construction of navigation metamodel from the

conceptual data of DM model. In order to simplify

this activity, we have proposed DM metamodel that

can describes all contained DM model concepts and

the way they are arranged, related and constrained.

We presented the Metamodel-based DM DSS

architecture where the DM metamodel will be used

to represent, store and later retrieve DM knowledge.

As a proof of concept, we presented our first

version of this metamodel and showed refinement of

its concepts in the domain of bushfires in particular

the recent tragic bushfires in Victoria. Also, we have

discussed several application scenarios in which our

METAMODEL-BASED DECISION SUPPORT SYSTEM FOR DISASTER MANAGEMENT

411

metamodel provides a valid support. We are

currently working on a more comprehensive

validation which will involve taking 20 existing DM

models and ensure that our metamodel can be

refined to generate all of them. There are some

issues that need to be investigated to fully realize the

potentiality of this approach. These include: (i) a

complete set of rules, processes and methodologies

for instantiation of user domain models; (ii) the

limitations of the metamodel; (iii) tools exist to

facilitate the development and use of the domain

model and (iv) methodology exist in validating the

user domain models.

REFERENCES

Ahmed, I. (2008): Disaster Risk Management Framework.

International Training Workshop on Disaster Risk &

Environmental Management, Melaka, Malaysia.

Aini, M. S., A. Fakhrul-Razi, M. Daud, N. M. Adam and

R. A. Kadir (2005): Analysis of Royal Inquiry Report

On The Collapse Of A Building In Kuala Lumpur:

Implications For Developing Countries. Disaster

Prevention and Management, 14 (1): 55-79.

Asghar, S., D. Alahakoon and L. Churilov (2006): A

Comprehensive Conceptual Model for Disaster

Management, Journal of Humanitarian Assistance.

Benaben, F., C. Hanachi, M. Lauras, P. Couget and V.

Chapurlat (2008): A Metamodel and its Ontology to

Guide Crisis Characterization and its Collaborative

Management, Proceedings of the 5th International

ISCRAM Conference Washington, USA.

Beydoun, G., G. Low, B. Henderson-Sellers, H.

Mouraditis, J. J. G. Sanz, J. Pavon and C. Gonzales-

Perez (2009): FAML: A Generic Metamodel for MAS

Development. IEEE Transactions on Software

Engineering, 35(6): 841-863.

Beydoun, G., G. Low, H. Mouraditis and B. Henderson-

Sellers (2008): A Security-Aware Metamodel For

Multi-Agent Systems. Information and Software

Technology, 51 (5): 832-845.

Brinkkemper, S. (1996): Method Engineering:

Engineering of Information Systems Development

Methods and Tools. Information and Software

Technology, 38 (4): 275-280.

Doherty, M. (2004): An Emergency Management Model

for Home Health Care Organizations. Home Health

Care Management Practice, 16 (5): 374-382.

Gao, S. and D. Xu (2010): Real-Time Exception

Management Decision Model (RTEMDM):

Applications in Intelligent Agent-Assisted Decision

Support in Logistics and Anti-Money Laundering

Domains. The 44th Hawaii International Conference

on System Sciences (HICSS), Hawaii.

Heicke, M. (2009). Automated Sequential Composition of

Deltas and Related Optimization Operations,

Universitat Paderborn, Germany.

Kelly, C. (1998): Simplifying Disasters: Developing a

Model For Complex Non-Linear Events, Australasia

and the Pacific Region Conference, Cairns,

Queensland.

Kruchten, P., C. W. K. Monu and M. Sotoodeh (2008): A

Conceptual Model Of Disasters Encompassing

Multiple Stakeholder Domains. International Journal

of Emergency Management, 5(1-2): 25-56.

Manitoba Health Disaster Management Disaster

Management Model for the Health Sector: Guideline

for Program Development.

Nordstrom, G., J. Sztipanovits, G. Karsai and A. Ledeczi

(1999). Metamodeling: Rapid Design and Evolution of

Domain-Specific Modeling Environments, Vanderbilt

University: 156.

Object Management Group (OMG) Model Driven

Architecture (MDA) Guide.

Russo, E. E. R., A. B. Raposo, M. Gattass and T.

Fernando A Metamodel for Configuring Collaborative

Virtual Workspaces: Application in Disaster

Management of Oil & Gas Offshore Structures.

Shaluf, I. M. (2008): Technological Disaster Stages and

Management.

Journal of Disaster Prevention and

Management, 17(1): 114-26.

Stahl, T., M. Voelter and K. Czarnecki (2005). Model-

Driven Software Engineering, Technology,

Engineering, Management, John-Wiley & Sons Ltd.

Tierney, K. J., M. K. Lindell and R. W. Perry (2001).

Facing Unexpected Disaster Preparedness and

Response in the United States. Washington DC,

Joseph Henry Press.

W3C Incubator Group Emergency Information

Interoperability Frameworks. http://www.w3.org/2005/

Incubator/eiif/XGR-Framework-20090806/#ack.

Weilkiens, T., Ed. (2008). Systems Engineering with

SysML/UML, Morgan Kaufmann Publishers Inc.

ICSOFT 2010 - 5th International Conference on Software and Data Technologies

412