using grid for data sharing to support intelligence in decision making

8/4/2019 Using Grid for Data Sharing to Support Intelligence in Decision Making

1/23

Chapter XI

Using Grid for Data Sharingto Support Intelligence

in Decision Making

Nik BessisUniversity of Bedfordshire, UK

Tim French

University of Reading, UK

Marina Burakova-Lorgnier

University of Montesquieu Bordeaux IV, France

Wei Huang

University of Bedfordshire, UK

Copyright 2007, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

IntroductIon

This section provides grounding in intelligence

informed decision making technologies, their

application and integration within the modern

organisations.

Scott-Morton rst articulated the concepts of

decision support systems (DSS) in the early 1970s

AbstrAct

This chapter is about conceptualizing the applicability of grid related technologies for supporting in-

telligence in decision-making. It aims to discuss how the open grid service architecturedata, access

integration (OGSA-DAI) can facilitate the discovery of and controlled access to vast data-sets, to assist

intelligence in decision making. Trust is also identied as one of the main challenges for intelligence in

decision-making. On this basis, the implications and challenges of using grid technologies to serve this

purpose are also discussed. To further the explanation of the concepts and practices associated with the

process of intelligence in decision-making using grid technologies, a minicase is employed incorporat-

ing a scenario. That is to say, Synergy Financial Solutions Ltd is presented as the minicase, so as to

provide the reader with a central and continuous point of reference.

IGI PUBLISHING

This paper appears in the publication, Managing Strategic Intelligence: Techniques and Technologies

edited by M. Xu 2007, IGI Global

701 E. Chocolate Avenue, Suite 200, Hershey PA 17033-1240, USA

Tel: 717/533-8845; Fax 717/533-8661; URL-http://www.igi-pub.com

ITB14662


2/23

0

Using Grid for Data Sharing to Support Intelligence in Decision Making

under the general term of management support

systems (MSS). Further works on bounded

rationality from Simon (1977) and classica-

tion types of DSS from Keen and Scott-Morton

(1978), Alter (1980), Holsapple and Whinston

(1996) have led us to understand that DSS is a

set of concepts associated with supporting the

decision making process via the use of appropri-

ate resources. These (resources) may include but

are not limited to users, data, models, software,

and hardware.

Computer-based developments over the last

four decades have facilitated decision makers

with numerous tools to support operational,

tactical and/or strategic level of enquiries withinthe environment of an organization. In relation

to intelligent decisions, the use of expert systems

(ES) and knowledge management systems (KMS)

have evolved over the years by developments in

computational science including data mining,

data visualization, intelligent agents, articial

intelligence, and neural networks. One of the

purposes of these technologies is to provide

managers (decision makers) with a holistic view

hence, the ability to analyze data derived from a

collection of multiple dispersed and potentiallyheterogeneous sources (Han, 2000).

One of the challenges for such facilitation is

the method of data integration, which aims to

provide seamless and exible access to informa -

tion from multiple autonomous, distributed and

heterogeneous data sources through a query in-

terface (Calvanese, Giacomo, & Lenzerini, 1998;

Levy, 2000; Ullman, 1997). In the context of DSS,

there are two broad classes of approaches to

data integration: Data Warehousing and Database

Federation (Reinoso Castillo, Silvescu, Caragea,Pathak, & Honavar, 2004). Practices in relation to

the data warehouse approach cover the acquisi-

tion, extraction, transformation, and loading of

the data into a centralized repository, which can

then be queried using a unied query interface.

The approach further allows interactive analysis

of multidimensional data of variable granularity

with multifunctionalities such as summarization,

consolidation, and aggregation (Nguyen, Min

Tjoa, & Mangisengi, 2003), as well as, the ability

to represent data in cube format (Nieto-Santiste-

ban, Gray, Szalay, Annis, Thakar, & OMullane,

2004). The key difference of the data federation

approach is, that decision makers can query di-

rectly the dispersed heterogeneous data sources

and hence, users are required to impose their own

ontologies in relation to the data requested.

The informational needs of a decision maker

are not limited to those prementioned and are very

seldom limited to data, but include other type of

resources, which may be required to be accessed

from multiple dispersed sources. The resourcesmay include but are not limited to databases,

software, hardware, or even instruments such as

satellites, seismographers, detectors and PDAs.

For example think of an emergency situation

caused by an earthquake. The emergency man-

agement team will be required to make real-time

intelligent decisions and act accordingly to save

lives, property, and the environment by assessing

multiple dispersed resources (Asimakopoulou,

Anumba, & Bouchlaghem, 2005). This particular

decision making process will require team work-ing and collaboration from a number of dispersed

decision makers whose decisions may be depended

on each others interactions. Resource integration

at that level will support decision makers since

it will allow them to view satellite images of the

affected area, observe seismic activity, forecast,

simulate and run what if scenarios, collaborate

with experts and the authorities. This will as-

sist decision makers to prioritize and ultimately

make decisions, which will be disseminated to

available rescue teams who will take then careof the operational tasks. This dissemination may

typically involve a server broadcasting decisions

to heterogeneous mobile devices such as personal

digital assistants (PDAs).

The volume of the data-sets is typically mea-

sured in terabytes and will soon reach petabytes

(Antonioletti et al., 2005). These data-sets are


3/23


variably geographically distributed and their

complexity is ever increasing. That is to say, that

the extraction of meaningful knowledge requires

more and more computing resources. The com-

munities of users that need to access and analyze

this data are often large and geographically dis-

tributed. The combination of large data-set size,

geographic distribution of users and resources,

and computationally intensive analysis results

in complex and stringent performance demands

that, until recently, have not been satised by any

existing computational and data management

infrastructure.

In tackling these problems, the latest studies

in relation to networking and resource integra-tion have resulted in the new concept of grid

technologies, a term originally coined by Foster

in 1995. Grid computing has been described as

the infrastructure and set of protocols that enable

the integrated, collaborative use of distributed

heterogeneous resources including high-end

computers, networks, databases, and scientic

instruments owned and managed by multiple

organizations, referred to as Virtual Organisa-

tions (Foster, 2002). Avirtual organization (VO)

is formed when different organizations cometogether to share resources and collaborate in

order to achieve a common goal (Foster, Kes-

selman, Nick, & Tuecke, 2002). Hence, the grid

concept as a paradigm has an increased focus on

the interconnection of resources both within and

across enterprises. In the rst phase, scientists

have almost exclusively used grid technologies

for their own research and development purposes.

Now however, the focus is shifting to more general

application domains that are closer to everyday

life, such as medical, business, and engineeringapplications (Bessis & Wells, 2005; ERCIM,

2001). It is anticipated that grid technologies will

facilitate intelligence informed decision making

in a way that managers and their teams will be

able to carry out tasks of increased complexity

more effectively and efciently in the form of

one or many interconnected, separable, or in-

separable VOs (Bessis & Wells, 2005; Brezany,

Hofer, Whrer, & Min Tjoa, 2003). Therefore,

in the context of this chapter, the primary goal

is to demonstrate how grid technologies and the

VO concept can serve as the vehicle to empower

intelligence in decision making.

To operate within a VO requires a decision

maker to interface a service or to act as an agent

of someone else in some capacity. Decision mak-

ers will necessarily be involved in delegacy. To

delegate is to entrust a representative to act on a

decision makers behalf. A key delegacy challenge

is the ability to interface with secure, reliable and

scalable VOs, which can operate in an open, dy-

namic, and competitive environment. To achievethis, a number of security mechanisms have to

be seamlessly integrated within the grid environ-

ment. Previous studies have proposed the use of

public key infrastructure (PKI) and X.509 digital

certicates (Foster, Kesselman & Tuecke, 2001;

Foster et al., 2002) while others have proposed

the use of IBC: Identity-based Cryptography (Lim

& Paterson, 2005).

In terms of social exchange theory, an inter-

action always contains an element of risk and

uncertainty due to the fact that an interactionpartner might not reciprocate or do so in an in-

sufcient manner (Stewart, 2003). A mediated

interaction as compared to a face-to-face inter-

action is characterised by a signicantly higher

level of uncertainty and risk (Lee & Turban, 2001;

Ratnasingam, 2005), which inevitably brings up

the question of the interrelation between risk and

control essential to an understanding of trust. The

perceived risk of an interaction is based on the

evaluation of its negative consequences, which are

difcult or impossible to control (Koller, 1988).The more negative are the consequences and the

less an individual can control them, the higher

is the perceived risk. The relationship between

trust and risk has a bilateral causal character that

offers large opportunities for building sustain-

able and auto-manageable systems. The greater

the risk of interaction, the more trust is desired


4/23


and the greater is the motivation to build trust

between oneself and ones interaction partner.

The greater the trust among all members of a

particular group, the greater the risk manage-

ment abilities of that group (McLain & Hack-

man, 1999). Tan and Thoen (2003) afrm, in the

context of uncertainty, trust permits one to feel

condent that current action will have a favour-

able outcome. Seen in this light, trust arises in

spite of high risk and uncertainty conditions as

a compensatory mechanism that permits one

party (e.g., the grid service consumer) to engage

in interaction with another party [either a partner

(individual or collectively) or a system (e.g., the

grid service providers)]. Thus, any analysis oftrust formation between grid entities should ide-

ally take into consideration the specicity of the

grid system, the particular network congurations

and the virtual character of the collaboration. At

the same time, it is important to stress that trust

develops not between organizations as such, but

rather as between the individual human actors

or proxy agents who represent them (Hoecht &

Trott, 1999).

The development of a VO partnership within

a grid community can be viewed generically asa model of dyadic interaction between trustor

(the grid service consumer) and trustee (the grid

service provider). A trustor (the grid service con-

sumer) inevitably takes a risk while depending on

the performance of a trustee (his/her grid partner).

This step is predicated upon the necessity to rely

upon another party in order to achieve ones own

interests, and hence, the interdependence between

grid partners. Not only trusting behavior, but

trusting intentions as well involve a high level

of risk. In the situation of high insecurity, trustbuilding is based on cognitive mechanisms, the

wary suspicious side, to assess the situation and

its consequences, thus potentially reducing the

importance of affective regulation (McKnight,

Kacmar, & Choudhury, 2004). However, the

cognitive nature of these mechanisms does not

of itself equate merely to the control of an inter-

action partner by means of security measures

alone. Trust is more likely to develop under in-

security, when an individual does not know how

the partner will behave (Molm et al., 2000). In

negotiated exchange, an outcome is predictable

thanks to agreement terms that minimize risk of

free-riding behavior, except if the agreement is

not completely binding. Negotiation has the re-

versed effect on trust building: it minimizes risk

and, thus, decreases trust and increases distrust.

Furthermore, there are certain regularities of trust

formation in a computer-mediated interaction that

are different from the situation of the face-to-face

communication.

Within this chapter, our main goal is tohighlight that resource integration within grid

environments in general and for assisting intel-

ligence in decision-making in particular have

been frequently limited to technical merits alone.

We hereby elaborate and articulate our ideas at

greater length and propose ways in which trust

issues as a soft, socially related concept can be

better articulated both with reference to the lit-

erature and to a novel semiotic paradigm. Hence,

the chapters main goals are twofold. Firstly, to

discuss how grid technologies, VOs and opengrid service architecturedata access integra-

tion (OGSA-DAI) can assist intelligence in deci-

sion making. We do this, by discussing Simons

(1977) well-known decision-making phases model

intelligence-design-choice alongside with the

concept of bounded rationality. Secondly, to

stimulate conceptual thinking towards a better

understanding of the novelty of this technology

and the need for a relevant soft trust model to

support its emergence. We do this, by discuss-

ing the role of soft trust issues at two distinctintangible and ambiguous levels of abstraction:

at the VO level of abstraction and the Grid (data)

service level of abstraction through the use of the

semiotic paradigm. To further the explanation of

the concepts and practices associated with using

grid technologies to support intelligence in deci-

sion-making, a minicase is employed incorporat-


5/23


ing scenarios. We conclude by discussing the

implications of using grid technologies to assist

intelligence in decision-making.

thE grId concEpt And Its

coMMErcIAl ExploItAtIon

The concept of grid computing has emerged as

an important research area differentiated from

open systems, clusters, and distributed comput-

ing. That is to say, open systems such as Unix,

Windows, or Linux servers, remove dependencies

on proprietary hardware and operating systems,

but in most instances are used in isolation. Each

deployed application has its own set of servers

purchased for a particular purpose within the

enterprise. Multiple applications rarely share

common servers, resulting in silos of statically

linked applications and servers. This conguration

results in poor server utilization. In contrast, the

grid builds upon open source architectures and

addresses the removal of silos within a connected

enterprise (Xu, Hu, Long, & Liu, 2004). It might

also prot by providing available internal resource

to other internal and/or external customers.Unlike conventional distributed systems,

which are focused on communication between

devices and resources, grid computing takes

advantage of computers connected to a network

making it possible to compute and to share data

resources. Unlike clusters, which have a single

administration and are generally geographically

localized, grids have multiple administrators and

are usually dispersed over a wide area. But most

importantly, clusters have a static architecture,

while grids are uid and dynamic with resourcesentering and leaving.

The added value that grid computing pro-

vides as compared to conventional distributed

systems lies in the inherent ability of the grid to

dynamically orchestrate large scale distributed

computational resources across VOs, so as to le-

verage maximal computational power towards the

solution of a particular problem. More specically,

the grid can allocate and reschedule resources

dynamically in real-time according to the avail-

ability or nonavailability of optimal solution paths

and computational resources. Should a resource

become compromised, untrustworthy or simply

prove to be unreliable, then dynamic rerouting and

rescheduling capabilities can be used to ensure

that the quality of service is not compromised.

Prior agreements, including service delivery

and recovery aspects can be pre-arranged at the

VO level of abstraction before and during run-

time execution at the service level of granularity

across the computational nodes that a particular

VO owns. These advanced features that areintegral to grid computing are rarely to be found

in large scale conventional distributed networks,

particularly those that need to cooperate and co-

ordinate dynamically across organizational and

geographical boundaries. Hence, it is the ability

of grid communities to orchestrate their activities

at the VO level and the service level dynamically

(without the need to consider platform dependant

features) that characterizes grid solutions as dis-

tinct from large-scale conventional distributed

computer networks.The grid is a computational network of tools

and protocols for coordinated resource sharing and

problem solving among pooled assets. These can

be distributed across the globe and are heteroge-

neous in character. Specically, grid computing

is widely seen to represent the next wave of

computing and as such has become the subject of

worldwide focus amongst the research community.

It is specically characterized by ad hoc col-

laborations (sharing of computing resources) as

between geographically distributed institutionsand organizations. The grid is a type of a parallel

and distributed system that enables the sharing,

selection, and aggregation of resources distributed

across multiple administrative domains based on

their availability, capability, performance, cost,

and users quality of service requirements (Goyal,

2005). Grid computing uses many computers con-


6/23


nected via a network simultaneously to solve a

single scientic or business related problem.

Whereas global grid initiatives initially tended

to focus on the needs of the UK scientic com-

munity (Fox & Walker, 2003) in an initiative col-

lectively known as E-Science, in the future, the

business community is expected to increasingly

benet too: grid computing is expected to become

a mainstream business-enterprise topology dur-

ing the rest of the current decade (Castrol-Leon

& Munter, 2005). The type of application most

likely to benet from the blurring of the binding

as between application and host is one that usually

requires substantial amounts of computer power

and/or produces or accesses large amounts ofdata. That is to say, execution of an application

in parallel across multiple host machines distrib-

uted within or between enterprises can increase

performance substantially and also make use of

the spare capacity of existing nodes (PC, servers,

etc.) too. Grid applications are often typically

involved with large volumes of data produced

by data-intensive simulations and experiments

(ERCIM, 2004). In order to guarantee seamless

automation and interoperability of the distributed

data, the need for adequate descriptions such assemantic-based data descriptions, models, ser-

vices, and systems becomes crucial.

Perhaps the most important function that has

emerged from the grid concept is the notion of

VOs. Grid computing provides a means by which

an open distributed and large scale network of

computational resources owned by VOs can en-

gage in the cooperative processing of typically

large data-sets, using the spare capacity of existing

computers owned by real organizations. There-

fore, a VO is formed when different organizationscome together to share resources and collaborate

in order to achieve a common goal. A VO denes

the resources available for the participants and

the rules for accessing and using the resources.

Resources here are not just computing, storage, or

network resources, but they may also be software,

scientic instruments or business data. Thus, by

engaging in a grid partnership both large and

small organizations can potentially leverage the

vast pooled assets of other partner organizations

without the need to purchase or physically own

these expensive resources. A VO mandates the

existence of a common middleware platform that

provides secure and transparent access to com-

mon resources. In practical terms, a VO may

be created using mechanisms such as certicate

authorities (CAs) and trust chains for security,

replica management systems for data organization

and retrieval and centralised scheduling mecha-

nisms for resource management (Venugopal,

Buyya & Ramamohanarao, 2005). Typical initial

application areas have included E-Science data-grids in which Universitys share their resources

across a grid so as to process vast quantities of

data involved in areas such as molecular model-

ing, climate change modeling, and nancial and

economic modeling.

In terms of standards, grids share the same

protocols with Web services (XML, WSDL,

SOAP, UDDI). This often serves to confuse as

to exactly what the differences between the two

actually are. The aim of Web services (WS) is to

provide a service-oriented approach to distributedcomputing issues, whereas grid arises from an

object-oriented approach. The idea of service-

orientation is not new. Distributed application

developers have long deployed services as part

of their infrastructure. CORBA is an example

of the efforts to standardise on a number of

services that provide the functionality needed to

support loosely-coupled, distributed object-based

applications. Further developments in the area

led to the emergence of WS (Atkinson et al.,

2005). However, WS typically provide stateless,persistent services whereas grids provide stateful,

transient instances of objects. In fact, the most

important standard that has emerged recently

is the open grid services architecture (OGSA),

which was developed by the Global Grid Forum

(GGF). OGSA is an informational specication

that aims to dene a common, standard, and


7/23


open architecture for grid-based applications.

The goal of OGSA is to standardize almost all

the services that a grid application may use, for

example job and resource management services,

communications, and security. OGSA species a

service-oriented architecture (SOA) for the grid

that realizes a model of a computing system as

a set of distributed computing patterns realized

using WS as the underlying technology. An im-

portant merit of this model is that all components

of the environment can be virtualized. It is the

virtualization of grid services that underpins the

ability to map common service semantic behavior

seamlessly on to native platform facilities. These

particular characteristics extend the functional-ity offered by WS and other conventional open

systems. In turn, the OGSA standard denes

service interfaces and identies the protocols

for invoking these services. The potential range

of OGSA services are vast and currently include

data and information services, resource and

service management, and core services such as

name resolution and discovery, service domains,

security, policy, messaging, queuing, logging,

events, metering, and accounting. OGSA-DAI

(data, access and integration) provides a meansfor users to grid-enable their data resources.

OGSA-DAI is a middleware that allows data

resources to be accessed via Web services. How-

ever, newer developments in the area have led to

more sophisticated data integration capabilities

using distributed query processing (DQP). DQP

works as a layer on top of OGSA-DAI, which al-

lows queries to be applied to various XML and

relational data resources as though they were a

single logical resource. This can be done through

an additional set of grid services that extend thescope of OGSA-DAI: one of these services acts

as the point of contact for a client and orchestrates

other services behind the scenes, including ser-

vices that evaluate queries on each data resource.

Data integration scenarios can be managed at

either the client or service end; DQP illustrates

an extension to OGSA-DAI at the service end,

enabling data integration (Antonioletti et al.,

2005).

Eay Ai f e gi y

be-ci bai Iy

(1999-)

The grid is being utilized internally and externally

by business organizations to aid their nancial

decision making and modeling. A number of

major Banks in the UK in the U.S. and Europe

have been early adopters (1999-2006) of inter-

nal and external grid computing models so as to

better utilize underused computational nodes in

the context of nancial services modeling and

decision making. As the chairman of the inu-

ential Landesbank Baden Wurtenburg (LBBW)

has recently concisely expressed, the grid and

nancial service industry are a marriage made

in heaven: The banking and nance industry is

predestined from Grid computing solutions. Our

business processes can be parallelized and thus

made faster and more efcient than ever before

(Platform, 2005). That is to say, by seeking to

use underused resources as part of a grid (wherethe VOs are typically comprise different internal

departments), these organizations hope to create

and run advanced simulations and otherwise

distribute increasingly data-intensive computa-

tional tasks across their existing computational

nodes without the need to purchase additional or

dedicated resources. Many of these grid projects

are of a highly commercially sensitive charac-

ter and therefore the details are often withheld

from the public domain. The interested reader

is however, refereed to two reports (Davidson,2002; Carbonnier, 2005) in which grid projects

within JP Morgan and Chase Manhatten Banks

respectively are described in some detail and

which may be viewed as being fairly typical in

illustrating the rationale behind early adoption of

grid applications within international banking. In


8/23


essence the commercial rationale behind many of

these projects is to leverage extra value from exist-

ing computational resources by using the spare

capacity of a vast network of computational

nodes to support data-intensive operations. The

nancial imperative for wider commercial use of

the grid is now undeniable and has recently been

articulated as follows:

Grid computing is not just about an asset change

in enterprise environments; it is about support-

ing a new business model, since there is no

killer application for grids. The key question for

Finance Directors and CFOs is how to break

out of the cycle of asset acquisition and into acapacity service provision model in order to

save money against a new budget system. The

benets of grid computing are about helping to

bring CAPEX (capital expenditurei.e., the cost

of the network, infrastructure and terminals) and

OPEX (operating expenditurei.e., the cost of

keeping the network running) down to acceptable

levels. The grid-based pay-per-use/utility model

is attractive because it can transfer cost from a

CAPEX to an OPEX model, but we dont believe

it will ever be an all or nothing situation forusers. (Fellows, 2005)

gi Ai y sME (sma a

Meim Eeie):

te ne Wae?

In the next wave of commercial adoption of

the grid within the nancial services industry

(2005-onwards), small and medium enterprises

(SMEs) are also now seeking to engage in external

grid partnerships, so as to gain access to vastlyincreased computational power at minimal cost.

In the most common case, the type of application

most likely to benet from the blurring of the

binding as between application and host is one that

usually requires substantial amounts of computer

power and/or produce or access large amounts of

data. That is to say, execution of an application in

parallel across multiple host machines distributed

within or between enterprises can increase per-

formance substantially and also make use of the

spare capacity of existing nodes (PC, servers, etc.)

too. In order to guarantee seamless automation

and interoperation of distributed data, the need

for adequate descriptions such as semantic-based

data descriptions, models, services and systems

becomes crucial.

EnAblIng IntEllIgEncE In

dEcIsIon MAkIng usIng grId

tEchnologIEs

The objective of this section is to discuss and

exemplify the potential of how grid technologies,

VOs and open grid service architecturedata

access integration (OGSA-DAI) within a dynami-

cally changing environment can assist intelligence

in decision-making. We do this, by discussing

Simons (1977) well known decision making

phases intelligence-design-choice alongside

with the concept of bounded rationality.

With this in mind we go on to describe a typi-

cal SME nancial services application in which actitious organization (Synergy Ltd) seeks to

engage in a VO partnership with several universi-

ties so as to seek to leverage the computational

power of the grid for competitive advantage. The

scenario serves as an integrative element within

this chapter, since the remaining sections make

explicit reference to it.

sME seai: syey Fiae

si l

Synergy Finance Solutions Ltd. is a (ctitious)

small and medium enterprise (SME) that develops

and sells advanced computer share trading pack-

ages to both private and corporate investors. These

packages are designed to support individual and

corporate investors wishing to track and predict

future equity (share) price movements across


9/23


global equity markets. Their current package is

designed to meet the needs of individual inves-

tors and is called PrivateInvestor. The license

to use the package is sold to private investors

and the package is typically installed on their

local PC workstation. PrivateInvestor uses

advanced fractal modeling techniques to track

real time Global share price changes on a daily

basis so as to establish patterns. These patterns

are then used together with 12-month historical

price data-sets and advanced fractal modeling

techniques to guide each private investor as to

exactly when best to trade shares, so as to gain

maximum prot at minimal nancial risk. The

package adapts itself to the risk prole of eachindividual investor as it learns more about their

real-time share-trading activities. Synergy makes

their data-set of historical share price patterns

for each share traded available for downloading

into the PrivateInvestor package, on demand,

to each investors workstation.

The managing director of Synergy is Mark

who is a rational manager (Keen & Scott-Morton,

1978) and very familiar with Simons (1977) three-

phase systematic decision-making process. He has

applied it successfully many times in the past.Mark thinks that it is the time to apply it again

for the benet of Synergy. Mark starts with the

rst phase that is the intelligent phase. His goal

is to clearly dene the problem by identifying

symptoms and examining the reality. The rst

phase begins with the identication of his orga-

nizational goal and objective that is to provide an

accurate service to his PrivateInvestor package

customers. Mark thinks that his company does

well in this respect and therefore, he feels that to a

certain extend his organizational goal can be met.However, Mark feels also dissatised. He identi-

es a difference between what he desires/expects,

and what is occurring. This is due to the fact that

a number of PrivateInvestor package customers

have not invested in the best possible way. Mark

made an attempt to determine whether a problem

exists. During his investigation, the sales depart-

ment informs him that Synergy has lost some

customers in the last year. The sales department

conrmed that the scale of loss is not signicant.

For some managers, losing a few customers is not

a major concern but for Mark this is considered

to be a symptom of an underlying problem. Mark

decided to revisit the kind of service that Synergy

offers to PrivateInvestor package customers.

Mark meets with Synergys executive team that

consists of the marketing, nancial advisor, po-

litical analyst and sales managers. He also meets

with Synergys three data analysts who analyse the

12-month data-set. Outcomes from the meeting

have led them to appreciate that the 12-month data-

set limits the accuracy of their advanced fractalnancial models; customers who have left and

gone to competitors who use a 10-year data-set;

competitors use more data analysts; competitors

invest more money in buying additional hardware

resources; and nally, competitors have access to

more modeling tools to choose from; On this basis,

Synergy realizes the need to make an intelligence

informed decision that will keep it abreast of its

competitors. Synergy fully understands that they

need somehow to provide a more accurate service

to its customers. This should be a good enoughsolution to retain existing PrivateInvestor pack-

age customers and maybe even, to increase the

number of its customers.

With this in mind, Synergy moves to the design

phase that is, the second phase of Simons (1977)

systematic decision-making process. This phase

involves nding or developing and analyzing

possible courses of action towards the identica-

tion of possible solutions against the identied

problem space. Synergy operates also under

the process-oriented decision-making thinking(Keen & Scott-Morton, 1978) and fully appreci-

ates Simons (1977) bounded rationality theory.

Synergy appreciates that despite the attractiveness

of optimization as a decision-making strategy, its

practical application is problematic. This is due to

the fact that it is not feasible to attempt to search

for every possible alternative for a given decision.


10/23


Simon exemplied this by dening the term of

problem space. A problem space represents a

boundary of an identied problem and contains

all possible solutions to that problem: optimal,

excellent, very good, acceptable, bad solutions,

and so on. The rational model of decision-making

suggests that the decision maker would seek out

and test each of the solutions found in the domain

of the problem space until all solutions are tested

and compared. At that point, the best solution will

be known and identied. However, what really

happens is that the decision maker actually simpli-

es reality since reality is too large to be handled

by human cognitive limitations. This narrows the

problem space and clearly leads decision-makerto attempt to search within the actual problem

space that is far smaller than the reality.

In the context of this chapter, the attempted

problem space is incomplete and refers to the

actual problem search space. Thus, the decision

maker will most likely not choose the optimal

solution because the narrowed search makes it

improbable that the best solution will ever be

encountered. The approach will lead the decision

maker to settle for a satisfactory solution rather

than searching for the best possible solution.

Similarly, Synergys 12-month data-set make

it impossible for data analysts to identify and

produce the most accurate packages for Priva-

teInvestor customers. On the same basis, data

analysts use a limited number of advanced fractal

nancial models as compared all those that are

theoretically possible available.

At this stage, Synergy has decided to identify

the course of action, which will lead in improv-

ing their existing solution without seeking the

optimum solution. Using this rationale, Synergy

feels that providing access to its own vast 50-year

data collection of historic share-prices that iscurrently unusable can be used to produce more

accurate packages for PrivateInvestor custom-

ers. It is believed that this will increase the actual

problem search space. Thus, the data analysts,

PrivateInvestor package customers and the

decision makers will most likely choose a better

solution because the extended search of the actual

problem search space increases the possibility

Figure 1. VO Grid partners extended search space (Extended version of Simons bounded rationality

theory, 1977)


11/23


that a better solution will be encountered. Figure

1 (reproduced over-page) illustrates Synergys

intelligence and choice decisions.

However, because of the capacity and process-

ing power limitations of the servers at Synergy,

only the last 12 months share-price patterns

have been available for download. Admittedly,

the move to allow access to this 50-year historic

data-sets, adds some complications. For example,

the amount of data required to be analyzed and

charted is potentially vast: patterns relating to

each share are analyzed in real time, daily, weekly,

monthly, yearly, and so forth. Synergys managing

director invites his IT manager to the meeting.

He conrms that buying additional hardwareresources required for these modeling processes

would be a very expensive and risky business.

Synergy decides that it might be a good idea to

extend its search space to look for more alterna-

tive solutions to choose from. Synergy invites

its IT manager to collaborate with two external

academics that are highly regarded in the area of

data management and decision-making modeling.

The outcome of this discussion leads Synergy to

believe that grid technologies may prove viable

as an alternative solution.Synergy moves to the choice that is the third

and last phase of Simons (1977) systematic de-

cision-making process. At this stage, Synergy

needs to make a decision based on the alterna-

tives derived from the previous phase. Synergy

has three options to choose from:

Take the risk and do nothing Buy additional hardware resources, even

consider to invest in more data analysts and

in the deployment of additional cutting-edgefractal nancial models

Enter into a grid partnershipSynergy decides that it is better to enter into

a grid partnership with several universities by

purchasing the computing spare time of their

computational nodes. This is because this will

allow data analysts and PrivateInvestor package

customers to apply their advanced fractal nancial

models to a wider search area (a 50-year data-set

as compared to 12 months). It might then still be

possible not identify the best possible solution

but it is more likely that a better solution will

be identied because the extended search of the

actual problem search space will increase the op-

portunities for a better solution to be encountered.

This in turn, will provide more opportunities to

allow investors to consider where to invest, what

are the possible advantages, disadvantages, risks

and ultimately, decide when to invest.

The idea is to utilize the spare-capacity of uni-

versity computers in real-time, on an on-demandbasis. Their grid partners will then orchestrate the

optimal workow (scalability) needed between

themselves, making best use of any spare capacity

available, so as to process and analyse this 50-year

historic data-set for each individual share. There

are a number of middleware solutions supporting

the coordination and allocation of jobs to be done

in a dispersed environment including Condor-G,

Globus Toolkit, and Unicore. These historic pat-

terns are then to be fully integrated with real-time

minute-by-minute share trading patterns so asto generate a prediction (typically buy, sell, hold,

etc.) back to Synergy. Thus, it is more likely for

their data analysts to select and produce a more

accurate prediction that is clearly caused by intelli-

gence data sharing. Synergy intends to make these

(more accurate) predictions available to existing

private investors who have previously purchased

the PrivateInvestor package at additional cost

that is as an optional premium Gold service

option on an on-demand basis. Historical data is

initially held centrally at Synergy but it can bedistributed via the grid partnership agreement

across any virtual organisation (VO) partners

as necessary that is made available to any grid

partner or partners on demand. Each University

partner may choose to delegate the data-analysis

and processing of this data to another partner in

real-time, depending on the availability of their


12/23

0


real-time processing capacity. If required, a uni-

versity partner may decide to move data to the

distributed environment as required to meet time

related constraints. The concept is rather similar

in principle to the UK National Grid, whereby

electricity is generated and distributed across

many providers and consumers according to real-

time demand. In this case, Synergy is deemed to

be the consumer and their university partners are

deemed to be their suppliers. Not electricity of

course, but of share pattern analytics derived from

both historical and real timeshare data.

By entering into a grid partnership, Synergy

will be provided with even more opportunities to

make intelligence informed decisions and producemore accurate predictions. Using Simons (1977)

three-phase systematic decision-making theory

(intelligence-design-choice), Synergys data

analysts will have access to a wider selection of

available nancial strategies including more data

mining tools and models available through the grid

partnership. For example, university academic,

research, and technical members of staff will

provide such support and share their expertise

with Synergy. On the other hand, Synergy could

make available a number of incomplete and ob-

solete data-sets that can be used by the university

partners for educational and research purposes.

That is to say, tutors could demonstrate to students

how to apply advanced fractal nancial modeling

using real world data-sets. Similarly, researchers

could undertake experimental research to further

advance nancial models for the benet of Synergy

and the wider community.

Overall, the VO approach will extend the op-

portunities to see things from a multiperspective

point of view that will ultimately advance the in-

volved partners. It is anticipated that the intended

approach will expand available opportunities by

extending the actual search space and by facilitat-ing methods required to deliver a better quality

of service. The ability to share and compute a

vast data-set alongside with the incorporation

of advanced modelling tools and utilisation of

expertise across the grid application environ-

ment will support Synergys managers and data

analysts. PrivateInvestor package customers

and grid partners will make intelligence informed

decisions. For Synergy, this will result in a no

cost solution that will provide a higher quality

Figure 2. The climate between the VO partners


13/23


of service as compared to their competitors. The

move to make historical data available to distrib-

uted partners can be a risky and challenging one.

In particular, data access, integration, analysis,

and charting using a variety of dispersed sources

including legacy systems can cause resource allo-

cation and recovery complications. However, there

are a number of paradigms whereby data access

and integration (DAI) can be implemented within

a grid environment. The concept is rather similar

in principle to E-Science, whereby dispersed data

owners make their heterogeneous data sources

available to other researchers in a VO via the use

of OGSA-DAI (a method for data replication and

virtualization). In addition, sophisticated dataintegration capabilities using DQP, as a layer

on top of OGSA-DAI will allow grid partners

to query, data and/or text mine to the dispersed

resources as though they were a single logical

resource. Figure 2 illustrates the potential of the

grid within a dynamically changing environment

via the use of a rich picture.

Finally, another issue of concern is quality of

service (QoS) including the aspects of multilevel

access, user-friendly interface, security, and reli-

ability. Synergy clearly needs to select and formeffective and evolving partnerships with trusted

university grid service providers. Within these

providers, Synergy seeks to orchestrate the pro-

vision of services in an optimally trustworthy

manner. To achieve this Synergy may need to

check not only that their VO partner local security

and access controls are adequate but also seek to

check and examine wider QoS, and reliability is-

sues too. Indeed, Synergy needs to check on the

organizational reputation of their VO providers

before entering into a grid partnership with anyparticular University potential partner.

The following section seeks to exemplify how

the OGSA-DAI can facilitate the discovery of and

controlled access to distributed sources in general

and Synergys 50-year vast data-set in particular,

to assist intelligence in decision making amongst

the VO partners.

ui ogsA-dAI Faiiae Ae

syey va daa-e

Analysis of the 50-year data-sets requires a

complex series of processing steps in which each

generates intermediate data products of a size com-

parable to the input data-sets. These intermediate

data products need to be stored, either temporarily

or permanently, and made available for discovery

and use by other analysis processes. OGSA-DAI

is the standard infrastructure to support effective

manipulation, processing and use of this vast,

distributed data resource. This will allow shared

data, networking, advanced fractal nancial

models, and compute resources to be delivered to

Synergys data analysts in an integrated, exible

manner. The method will enable Synergys data

analysts to make intelligence informed decisions

and to produce more accurate predictions for the

benet of Synergys customers.

The aim of the OGSA-DAI middleware is to

assist with the access and integration of dispersed

data sources available on the grid. OGSA-DAI

is compliant with Web services inter-operability

(WS-I) and the Web services resource framework

(WSRF). OGSA-DAI is a middleware, whichsupports the integration and virtualization of data

resources, such as relational, XML databases, le

systems or indexed les. Various interfaces are

provided and many popular database management

systems are supported including MySQL, Oracle,

DB2, XML. Data within each of these resource

types can be queried, updated, transformed,

compressed, and/or decompressed. Data can be

also delivered to clients or other OGSA-DAI Web

services, URLs, FTP servers, GridFTP servers, or

les. On the OGSA-DAIs Web site there are fullinstructions of how to download and install the

middleware. Set-up prerequisite software includes

JDK 1.4, Tomcat, Apache Ant and some additional

libraries such as JDBC drivers, etc.

According to the latest specications, OGSA-

DAI provides the following types of services:


14/23


Data access and integration service group

registry (DAISGR): The service allows

data resources that are represented by ser-

vices to be registered and discovered.

Grid data service factory (GDSF): The

service acts as a persistent access point

to a data resource and contains additional

related metadata that may not be available

in the DAISGR.

Grid data service (GDS): The service

acts as a transient access point to a data

resource.

On this basis, we can now proceed to describe

a scenario to introduce various issues of relevanceto Synergys data-sets access and integration

services. These include data collection, advanced

fractal nancial modelling, data generation, and

data analysis. Figure 3 demonstrates these OGSA-

DAI related service interactions between the

different VO grid partners. Figure 3 also notates

these services so as to provide the reader with a

central and continuous point of reference.

Let us assume that the environment comprises

ve simple hosting environments: one that runs

the Synergys data analyst user application (A1);three that encapsulates computing and storage

resources (B, C, D); all three also encapsulate

data-set services; in which one of them (B)

encapsulate Synergys fractal nancial models

and other partners data mining tools; and nally,

a different one (E) that remains idle but could

take over compute related tasks and/or host data

moved from another partner environment. To

complicate the scenario, we assume also that the

latter hosting environment (E) runs a partners

user application (A2) to assist in applying advancednancial modelling tools on a demand basis. It

also encapsulates advanced nancial models.

Firstly, we expect that each data-set is stored in

a different VO grid partner (service provider) and

it is registered with the Grid Data Services Fac-

tory (GDSF) so that they can be found. Similarly,

it is anticipated that Synergys advanced fractal

nancial models and any other data mining tools

have been registered as a service so they can be

found too. Let us assume that Synergys data

analyst as a service requestor needs to obtain

X information on share prices of a particular

stock over the period of ten years. At this stage,

it is important to note that Synergys data analyst

does not need to know which data-set(s) are able

to provide this information and where these are

located. It might be the case that information is

stored in more that one data-set (DS).

The following lists the steps required for a

service requestor to interact with appropriate

data services:

Action 1: Synergys data analyst as a ser-

vice requestor will need to request the data

access and integration service grid register

(DAISGR) for source of data about X.

Action 2: Register will return a handle to

the service requestor.

Action 3: Register will send a request to

the factory (GDSF) to access the relevant

data-sets that are registered with it.

Action 4: Factory will create a grid data

service (GDS) to manage access to relevantdata-sets.

Action 5: Factory will return a handle of

the GDS to Synergys data analyst.

Action 6a: Synergys data analyst as a ser-

vice requestor will perform the query to the

respective GDS using a database language

such as SQL.

Action 7: The GDS will interact with the

data-set(s).

Action 8a: The GDS will return querys

results in a XML format to the servicerequestor.

In the event that GDSF has identied more than

one of the data-sets (DS1, DS2, DS3) that contain

the relevant information, Synergys data analyst

will either select a particular GDS (for example,

GDS1) based on the analysts preference(s) or


15/23


request for data to be integrated into a sink GDS

(6b). That is to say, a sink GDS will handle the

communications (6c) between data analyst and the

multiple GDSs (GDS1, GDS2, GDS3), which will

further interact (7) with their respective data-sets

(DS1, DS2, DS3) so as to return querys results in

a XML format (8b) to Synergys data analyst.

Similarly, a service requestor can submit a

request for a particular nance model that is either

a service of Synergy or registered with another

VO grid partner. A service requestor can be either

a Synergy data analyst (A1) or a partners advisor

(A2) who is available to offer advice or to assist

Synergys data analyst in applying a special type

of nancial modeling tool on an on-demand basis.Once data and models have been collected via the

GDS, Synergys data analyst or a partners advisor

could then for example run their simulation tests.

In the event that a service will or communication

fails another registered resource (service provider)

will take over of the outstanding task(s). For ex-

ample, if during compute perform, one resource

(D) from the grid partners becomes unavailable,

another idle registered resource (E) from the same

or different partner will carry on the computation.

This is due to the fault tolerance grid service that

allows a task to carry over to a different registered

and available resource.

The approach as a whole allows the discovery

of resources and allocation of tasks on a reliable

and exible manner. Using available computing

power, grid partners will minimize time related

constraints when Synergys data analysts run

their prediction tests, which ultimately will en-

able them to make more informed decisions.

The availability of equity enhances computing

power alongside accessing a larger selection of

data-sets, that can be data-mined using additional

data mining tools and advice from experts on an

on-demand basis will likely assist Synergy to

produce more accurate predictions. It is also amethod for the other participated VO grid partners.

Thus, Synergy could make available a number

of incomplete and obsolete data-sets that can be

used by the university partners for educational

and research purposes. That is to say, researchers

(A3) could undertake experimental research to

further advance nancial models for the benet

of Synergy and the wider community. Similarly,

tutors could demonstrate to students how to ap-

ply advanced fractal nancial modeling in real

world data-sets (A4).

Figure 3. OGSA-DAI interactions between VO grid partners


16/23


However, despite the fact that the primary aim

of OGSA-DAI is to make data more accessible,

it must also provide controls over data access

to ensure that the condentiality of the data is

maintained, and to prevent users who do not

have the necessary privileges to change or even

to view data content. Some related trust issues

are discussed next.

te re f sf t Ie

i Ieiee Ifme

deii Mai

Trust is a very complex and nonhomogenous phe-

nomenon that covers many elds of social knowl-

edge and enquiry. The concept has previously

been variously been identied with: a general

disposition; a rational decision about cooperative

behaviour; an affect-based evaluation about an-

other person; a characteristic of social systems

(Rousseau, Sitkin, Burt, & Camerer, 1998), and as

a clan organising principle (McEvily, Perrone,

& Zaheer, 2003). Trust relates to a willingness to

rely on others, and to the condent and positive

expectations about the intentions or behaviour of

another, also, to the willingness to be vulnerableand to acquire risk (Mayer, Davis, & Schoorman,

1995; Rousseau et al., 1998). In spite of the fact

that trust can be analyzed in relation to risk-taking

intentions and/or behaviours, the theoretical link

between trust and risk often remains somewhat ill

dened. The interdependence between trust and

risk is interpreted in many different ways. First,

risk is considered to be an essential condition of

trust emergence (Coleman, 1990), when none or

almost none of the assurance mechanisms are

available to build an interaction between partners.Secondly, trust entails a willingness to take risks

based on the sense of condence that others will

respond as expected and will act in mutually sup-

portive ways, or at least, that others do not actually

intend to do harm (McKnight et al., 2004). The

assumption that trust and risk are closely related

phenomena is not solely a theoretical model, but

has been supported through empirical evidence.

Thus, Koller (1988) found that the degree of risk

affects the degree of trust toward an interaction

partner and stressed that both phenomena relate

to the domain of social perception. An individual

concludes that the individual trusts the interaction

partner, if the individual nds that interaction

with the partner in a risky situation. Indeed, trust

appears to be situated somewhere between com-

plete control and uncertainty. Indeed, trust may

well begin only when mere condence ends. In

many ways trust is seen as being intimately de-

pendant on an information gap as between trustor

and trustee. An individual aware of all relevant

facts does not need to trust, while an individualnot knowing anything about the issue in question

is unable to trust, but only to hope or believe

(Clapses, Bachman, & Wehner, 2003). It has also

been demonstrated (McLain & Hackman, 1999)

that in the context of a lack of information about

the interaction partner, trust emerges in a high-

risk insecure environment, and at the same time,

plays the role of a risk-reducing mechanism.

On this basis, an important element of this

chapter is to highlight that a VO within a grid

environment in general and decision making inparticular is frequently not limited by technical

consideration only. We prementioned that to oper-

ate within a VO, a decision maker is involved in

delegacy. To delegate is to entrust a representative

to act on decision makers behalf. The interac-

tion between individual delegates (as members

of the grid community) to build mutual trust is

central to the analysis itself. We share the view

that, individual elements may offer solutions to

problems but are at best limited as a whole. In

other words, a VO includes, but does not equateto the level of interactions (people-to-people)

and the level of grid services alone. It is also

enriched by a number of phenomena related to

organizational behavior. It might inherit concerns

related to risk and (in)security and might require

further the exploration of trust into the domain

of human cognition and behavior. Hence, a VO


17/23


can be viewed analyzed as a special kind of a

social network and in this respect, with particu-

lar references to its structure, cognitive aspects,

and relations. Thus, it seems important to revisit

trust related issues within the application of grid

environments.

There is little previous research that compre-

hensively accounts for and models the holistic

nature of trust-building processes and regularities

within a grid application environment. Hence, to

safeguard interests and alleviate inconsistencies

caused within a VO as a distributed environment,

we hereby propose a two-level model of abstrac-

tion, a kind of multidisciplinary deconstruction,

that seeks to identify the grid community andsingles out the technological and social mecha-

nisms of trust formation with grid services.

soFt trust At tWo lEvEls oF

AbstrActIon

The purpose of this section is to stimulate con-

ceptual thinking towards a better understanding

of the novelty of this technology and the need for

a relevant soft trust model to support its emer-gence. We do this, by elaborating and articulating

our ideas in relation to the role of soft trust issues

at two distinct intangible and ambiguous levels

of abstraction: at the VO level of abstraction and

the grid (data) service level of abstraction through

the use of the semiotic paradigm.

Emeee f via oaizai

(vo) lee f Aai

A grid service provider needs to ensure that un-authorized access to services and data does not

take place. Additionally, a providers reputation

is clearly at stake and there is a need to maintain

quality, timeliness, reliability, and integrity of the

service according to whatever kind of agreement

has been entered into with consumers and other

providers in an orchestrated manner. There is an

obligation for a service provider to ensure quality

and continuity of service under a wide variety of

conditions. Legal and economic factors may be

relevant too. Intrusion detection is an important

area of responsibility, particularly so in grid

contexts where an unauthorized user may be

potentially able to gain access not only to services

but also to the underlying data-sets themselves.

Corporate governance policies and orientation,

trusted accountancy practices, all serve to dene

a providers relationships to its suppliers, custom-

ers, and business partners (Will, 2003). Trust or

mistrust of a VO at an organizational, depart-

mental and workgroup level may well inuence

whether or not a VO is suitable as a grid partner.Furthermore, a VO is clearly embedded within a

society and culture. A provider needs to consider

how their virtual identity may be veried, and

trusted by potential consumers of grid services.

In particular managing user expectations and

soft requirements poorly can lead to consumer

frustration and indeed even result in frustration

and a degree of mistrust (Tiong, 2005).

In order for Synergy to select and form an

effective and evolving partnership with trusted

providers and orchestrate the provision of servicesin an optimally trustworthy manner it is necessary

to look beyond mere agent-to-agent level of trust

formation and technological mediators to wider

concerns. The value of this approach is intended

to help Synergy to select and verify a suitable

university partner or set of partners to orches-

trate their activities (grid workows) in such a

manner to maximize trust while minimizing risk

of various that is to optimally match candidate

partners against sets of relevant trust, reputation

and reliability criteria.For example Synergy might wish to check the

status (VO reputation) of their potential univer-

sity grid service partners in terms of any of the

above mentioned dimensions: nancial viability,

research reputation, ranking in university league

tables, implementation of local security policies,

and so forth. Equally, a university might wish to


18/23


check whether Synergy meets their own internal

ethical and corporate governance standards by

referring to suitable public domain sources. By

using a semiotic trust ladder, it should be possible

for both Synergy and candidate or actual university

partners to more systematically check and verify

trust dimensions at the social, pragmatic, and

syntactic levels of abstraction. For example it will

be possible for both Synergy and their partners

to look beyond the ne-grained issued of which

XML based standard to select and to address more

fundamental issues that encompass risk, quality

of service and trust issues. Essentially the idea

is for the grid partners to quantify and manage

hidden or implicit trust expectations, to assessthe potential commercial and reputational risks

of their engagement as well as of course selecting

the most appropriate technological trust mediators

to support grid workow activities.

It should also be possible for both Synergy (the

grid service consumer) and University partners

(grid service providers) to dynamically assess and

re-assess their relationships in the light of new and

changing evidence or wider trust domains so as

to generate for example a crude trust/risk rating

for a grid service before, during invocation andafter service invocation. However, to really add

value to existing trust management in the context

of agent to agent (autonomous) trust brokerage

and negotiation a much more ne-grained means

of enabling an agent with these wider contexts

is needed. Organizational reputation and orga-

nizational cultures change and evolve over time.

Local contexts, methods and ways of working

also evolve continually. Ideally therefore, as a

grid service is invoked an agent should be able

to reverify at least some elements of an e-serviceproviders wider trust domain (or just in time)

during run time execution.

gi (daa) seie lee f

Aai viewe t

e semii le

Human trust is a far more elusive and subtle

concept than is articulated in frameworks such

as Web services-trust, as it generally involves

the reference not merely to local contexts but

also wider organizational and social settings

within which e-service transactions of all kinds

typically take place. Existing approaches to the

trusted grid services, which emphasise the value

of establishing secure communications between

autonomic entities do not appear to attempt to

explicitly seek to verify local events, credentials

against wider social, cultural, and organizational

dimensions. Indeed, Liu (2003, 2006) has called

for a wider examination of so-called soft issues

of grid computing and more specically identi-

es the semiotic paradigm as being a potentially

useful conceptual probe within which to address

these wider concerns. Without seeking to enable

agents with wider organizational trust contexts

(what we herein choose to call a trust domains)

we cannot say that these agent based approaches

truly simulate real human trust, but rather, only alimited subset of the characteristics of human trust

that are necessary but not sufcient to claim that

a particular grid service is in fact trustworthy.

Based on Lius (2003, 2006) more general

approach to soft issues of the grid, this work

maps these concerns to the well known classic

semiotic ladder (Stamper, 1973) so as to instan-

tiate a new variant, namely the semiotic trust

ladder shown within Table 1 below, to illustrate

the value of the semiotic paradigm in helping

stakeholders to better conceptualise trust issueswithin virtual organizational settings. Essentially

the novel semiotic trust ladder offers a way of

conceptualizing and modelling trust meaning


19/23


making at a variety of levels of abstraction by

identifying actors, signs, and articulating ways

in which norms and metanorms mediate all acts

of communication.

In Table 1, for each layer of the trust ladder,

some exemplar trust issues are identied and

aligned to the grid service lifecycle. By extend-

ing this approach it is possible to develop a fully

comprehensive account of trust issues during the

entire grid service lifecycle. Indeed, by attempting

to identify and map trust issues to the trust lad-

der, it is hoped that previously implicit or poorly

understood or articulated trust issues may be more

clearly revealed to VO partners at an earlier stage

in the grid service lifecycle than hitherto.

IMplIcAtIons And chAllEngEs

oF usIng grId tEchnologIEs

to support IntEllIgEncE In

dEcIsIon MAkIng

One of the major implications in using grid

technologies as a vehicle to assist intelligence

in decision-making is the ability to enlarge the

actual search space boundaries within the term

of problem space as described by Simon (1977).

Problem space represents a boundary of an identi-

ed problem and contains all possible solutions

to that problem: optimal, excellent, very good,

acceptable, bad solutions, and so on. By searching

in a narrow space, the decision maker will most

likely not choose an optimal solution because the

narrowed search of the actual problem search

space makes it improbable that the best solution

will ever be encountered.

Clearly the grid potentially vastly increases the

size and complexity of the problem spaces that can

realistically be addressed not only by SMEs, but

by all types of organization. Problems that havehitherto been regarded as being intractable either

because of the size of the data-sets needed, their

distributed nature or the sheer complexity of the

multidimensional analysis required can now be

re-examined. Within E-Science these problem

spaces encompass traditional scientic domains

such as nuclear physics but now also typically

include areas such as climate change, where

vast quantities of data and simulations requiring

multidimensional analysis are needed.

Table 1. Macro-dimensions of VOs via a semiotic trust ladder

Exemplar Grid Service

Trust Issues

Semiotic

Trust Ladder

Applicability

(VO Grid Lifecycle)Signs

To what extent does the Service

conform to the desired VO

cultural/cross-cultural norms?

Are there any legal safeguards?

Social world trust: Beliefs and

expectations

Planning stage Cultural/Social trust

Policy signs

Reputation of Grid service

provider/consumer?Any ethical conicts?

Pragmatics: Goals, intentions,

trusted negotiations, trustedcommunications

Planning, build, run time Reputation signs

How reliable, valid are the

services and will they meet

quality norms?

Semantics: Meanings, truth/

falsehood, validity

Build and run time Authentication/validity signs

Secure agents: How trusted are

they?

Syntactics: Formalisms, trusted

access to data, les, software

Build and run time Trusted access signs

Intrusion detection/prevention

adequate?

Empirics: Entropy, channel

capacity

Run time Messaging/trafc management

signs


20/23


Within the business community large Banks

have been amongst the rst to exploit the enhanced

power of the grid to leverage extra value from

vast legacy systems. Now as has been shown

through our illustrative case study, SMEs are

able to address previously intractable problems

and to leverage competitive advantage from grid

computing. This is only the beginningdecision

makers will soon be able to address or re-address

complex multidimensional problems within their

businesses using grid solutions as their standard

or normative preferred tool. Thus, the grid should

not be seen as being merely a tool of scientists or

academicians but rather as a new and powerful

business decision support tool, having real cut-ting edge potential to solve business problems

and enhance competitive advantage. However,

for the power of the grid to be fully realized by

business decision makers, a risk assessment is

needed. For as has been shown in this chapter,

trust issues remain one of many risk factors that

need to be considered before grid computing is

adopted. Since the grid by denition involves the

creation of virtual partnerships between VOs, like

any partnership there are risks as well as rewards.

In the future, grid computing will only be seen toserve and support decision makers if these risks

are properly assessed and accommodated. Like

all enabling technologies, investment needs to

be made in properly harnessing the power of the

grid without exposing the business to undue risk.

This is one of the challenges that still remain to

be solved if grid computing is indeed to become

a normative tool of the business community, not

just a play-thing of academia and scientic stake -

holders. Indeed, there is a greater need now for

the business community to assume a more activerole in the development and commercialization of

the grid. While scientists have hitherto dominated

the grid community, this dominance may soon

increasingly be challenged.

conclusIon

This chapter has endorsed the logic that the con-

cepts and practices associated with grid related

technologies can assist managers in making

intelligence informed decisions within a virtual

organisation (VO). This approach will extend the

opportunities to see things from a multi-perspec-

tive point of view that will ultimately challenge,

mature and advance the involved partners. It is an-

ticipated that the decision to use grid technologies

will unfold new opportunities as it will enlarge the

actual search space boundaries within the term of

problem space as described by Simon (1977). By

default, a problem space represents the boundary

of an identied problem and contains all possible

solutions to that problem. It might then still be

possible not identify the optimal solution but it

is more likely to increase the opportunities for a

better solution to be encountered. Overall, it will

facilitate methods towards normative thinking as

required for a better quality of service.

In the context of this chapter, we have referred

to a VO as the ability to share and exploit com-

modities within a dynamic distributed environ-

ment via networks. Commodities as services areshared and exploited via the use of policies and

may include but are not limited to computational

nodes, stored data, expertise, and other resources.

We have referred to them as transient, uid serv-

ices since they enter and leave based on their

availability and a number of policies.

A core element of this chapter has been to

highlight those VOs within grid environments

that are frequently not limited by technical con-

sideration alone. We took the holistic view that

VOs are also a kind of a social network. Therefore,trust was examined as a soft issue with respect to

its structure, cognitive aspects, and relations. In

particular, we discussed the role of soft trust issues

at two distinct intangible and ambiguous levels of

abstraction: at the VO level of abstraction and the


21/23


grid (data) service level of abstraction through the

use of the semiotic paradigm. We concluded that

trust remains a subtle and elusive concept, yet it

is vital that decision makers attempt to concep-

tualize trust issues explicitly, particularly when

considering implementing complex distributed

systems, such as the grid. Furthermore, semiotics

may well provide a useful paradigmatic vantage

point within which to conceptualize about these

vital trust issues at the empiric, pragmatic, and

organizational levels.

rEFErEncEs

Alter, S. L. (1980). Decision support systems:

Current practices and continuing challenges.

Reading, MA: Addison-Wesley.

Antonioletti, M., Atkinson, M. P., Baxter, R.,

Borley, A., Chue Hong, N. P., Collins, B., et al.

(2005). The design and implementation of grid

database services in OGSA-DAI. Concurrency

and Computation: Practice and Experience,

7(2-4), 357-376.

Asimakopoulou, E., Anumba, C. J. & Bouchla-ghem, D. (2005). Studies of emergency manage-

ment procedures in Greece, Italy and the UK. In

Proceedings of the3rdInternational Conference

on Construction in the 21st century, Advancing

Engineering, Management & Technology , Athens,

Greece (pp.15-17).

Atkinson, M., DcRoure, D., Dunlop, A., Fox, G.,

Henderson, P., Hey, T., et al. (2005). Web service

grids: An evolutionary approach. Retrieved Janu-

ary 5, 2007, from http://www.nesc.ac.uk/techni-

cal_papers/UkeS-2004-05.pdf

Bessis, N. & Wells, J. (2005, March 22-24). Grid

technologies revive the basic IT infrastructure.

Information Systems Unplugged: Developing

Relevant Research. InProceedings of the UKAIS

2005:10th International Conference in Informa-

tion Systems, Newcastle, UK.

Brezany, P., Hofer, J., Whrer, A. & Min Tjoa, A.

(2003). Towards an open service architecture for

data mining on the grid. In Proceedings of the

DEXA Workshops 2003 (pp. 524-528).

Calvanese, D., Giacomo, G. & Lenzerini, M.

(1998). Information integration: Conceptual mod-

elling and reasoning support. In Proceedingsof

the CoopIS98.

Carbonnier, J. P. (2005).BNP paribas bolsters grid

project. Retrieved January 5, 2007, from http://

www.datasynapse.com/pdf/DWT_BNP_Pari-

bas_Bolsters_Grid_ Project _05_3005.pdf

Castro-Leon, E. & Munter, J. (2005) Grid computing: Looking forward part 1: Technology

overview. Intel White Paper. Retrieved January

5, 2007, from http://www.intel.com/cd/ids/devel-

oper/asmona /eng/202676.htm

Clapses, C., Bachman, R. & Wehner, T. (2003).

Studying trust in virtual organisations.Interna-

tional Studies of Management and Organisation,

33(3), 7-27.

Coleman, J. S. (1990). Foundations of social

theory. Cambridge: Harvard University Press.

Davidson, C. (2002, October).JP Morgan unveils

compute backbone project. Waters Financial

Intelligence. Retrieved January 5, 2007, from

http://www.platform.com/Company/Case.Studies

/CS.JPMorgan.06.08.2005.htm

ERCIM News. (2004, October). Special theme:

Grids: The next generation (No. 59), pp. 43-44.

ERCIM News. (2001, April). GRIDS: e-science

to e-business.

Fellows, W. (2005, March 21 ). Enterprise grid

computing to shape IT business models. Finance

Director Europe. Retrieved January 5, 2007,

from http://www.financedirectoreurope.com

/articles/451group.htm

Foster, I., Kesselman, C. & Tuecke, S. (2001).

The anatomy of the Grid: Enabling scalable


22/23

00


virtual organisations. International Journal of

Supercomputer Applications, 15(3).

Foster, I. (2002, July 22). What is the grid? Athree point checklist. Grid Today, 1(6). Re-

trieved January, from http://www.gridtoday.com/

02/0722/100136.html

Foster, I., Kesselman, C., Nick, J. M. & Tuecke,

S. (2002, June 22). The physiology of the grid:

An open grid services architecture for distributed

systems integration. Retrieved January 5, 2007,

from http://www.globus.org/alliance/publications

/papers/ogsa.pdf

Fox, G. & Walker, D. (2003).E-science gap analy-sis.Cardiff University. Retrieved January 5, 2007,

from http://grids.ucs.indiana.edu/ptliupages/pub-

lications/Gap Analysis30June03v2.pdf

Goyal, V. (2005, April 4). The promise of grid

computing. Message posted to gridcomputing@

yahoogroups.com

Han, J. (2000).Data mining: Concepts and tech-

niques: Morgan Kaufmann.

Hoecht, A. & Trott, P. (1999). Trust, risk and con-

trol in the management of collaborative technologydevelopment.International Journal of Innovation

Management, 3(3), 257-270.

Holsapple, C. W. & Whinston, A. B. (1996).

Decision support systems: A knowledge-based

approach. St. Paul, MN: West Publishing.

Keen, P. G. W. & Scott-Morton, M. S. (1978).

Decision support systems: An organisational

perspective. MA: Addison-Wesley.

Koller, M. (1988). Risk as a determinant of trust.

Basic and Applied Social Psychology, 9(4), 265-

276.

Lee, M. K. O. & Turban, E. (2001). A trust model

for consumer internet shopping. International

Journal of Electronic Commerce, 6(1), 76-91.

Levy, A. (2000). Logic-based techniques in data

integration. In Jack Minker (Ed.), Logic based

articial intelligence. Kluwer Publishers.

Lim, H. W. & Paterson, K. G. (2005). I

using grid for data sharing to support intelligence in decision making

Documents