TRUST AND REPUTATION ONTOLOGIES FOR ELECTRONIC

BUSINESS

Stefan Schmidt, Tharam Dillon, Robert Steele

Faculty of IT, University of Technology, Sydney, Australia

Elizabeth Chang

IT School of Information systems, Curtin University of Technology, Perth, Australia

Keywords: Trust, Reputation, Credibility, Ontology, P2P, e-Business.

Abstract: The emergence of social networks in centralized and distributed virtual communities is one of the hottest

topics in today’s research communities. Trust and reputation ontologies which capture the social

relationships and concepts among interacting parties offer a standardized and common understanding of a

problem domain such as electronic business in autonomous environments. To improve interoperability,

ontologies can be shared among interacting agents and form the basis for many of the autonomous activities

of intelligent agents. The ontologies presented in this paper concentrate on the formalisation of business

discovery, business selection, and business interaction QoS review concepts. Special focus is put on trust

and reputation relationships which form among the entities involved.

1 INTRODUCTION

E-commerce platforms have grown enormously over

the past decade, and in many areas e-commerce

business models have overtaken traditional models.

The accessibility of the internet results in better

choice and better prices for consumers through

increased competition and reduced costs. Therefore,

e-commerce businesses enjoy increasing popularity

over traditional businesses. However, the sheer

vastness of the internet and its offerings is

overwhelming for many consumers. Consumers

often find it hard to determine the reputation of e-

businesses and their products or services. Hence,

they have little confidence due to a lack of trust. A

solution to these problems offer autonomous agents

which search e-commerce platforms on behalf of

their owners, the consumer. They can discover

products or services, select appropriate business

partners, negotiate and place contracts, oversee

contract execution, and determine the success and

quality of business transactions. All these activities

are very time and resource consuming and the

concept of truly autonomous computing which

addresses this problem is a compelling idea.

Yet the adoption of autonomous computing

concepts into e-commerce models is still in its

infancy. The integration of social values such as

trust, reputation and credibility represent a challenge

to achieve the vision of autonomous interactions

between intelligent agents. Current frameworks,

which put the underlying concepts of the service

oriented architecture (SOA) into practice, often offer

security models which deal with data integrity and

encryption as well as identity management but lack

integration of sophisticated social protection

mechanisms. These social protection mechanisms

are, however, the decisive factor for human

interaction. Centralized environments such as e-

commerce portals or virtual marketplaces which are

designed to complement or even replace time and

resource consuming direct human interactions

currently lack an interoperable, accessible and

extensible interface to their various social protection

mechanisms. This problem is even more important

in next generation digital ecosystems which will

operate in truly decentralized peer-to-peer (P2P)

environments (Schmidt, Steele & Dillon 2007).

While P2P based e-business environments (Barkai

2001) offer great features such as enhanced privacy,

independence, scalability and accessibility, the

308

Schmidt S., Dillon T., Steele R. and Chang E. (2007).

TRUST AND REPUTATION ONTOLOGIES FOR ELECTRONIC BUSINESS.

In Proceedings of the Ninth International Conference on Enterprise Information Systems - ISAS, pages 308-315

DOI: 10.5220/0002345003080315

 SciTePress

management of social security during the formation

of such virtual communities will require highly

sophisticated and standardized frameworks. These

frameworks need explicit and formal specifications

of the concepts found in e-business domains, and the

relations among them (Gruber 1993) with regard to

social security mechanisms such as trust, reputation

and credibility.

The lack of integration of more sophisticated

social protection mechanisms is due to two factors.

First, the fuzzy nature (Chang, Hussain & Dillon

2005c) of these social values increases the

complexity for their integration into autonomous

environments. Second, there is a strong need for the

formalisation of trust, reputation and reputation

concepts tailored to the specific conditions that

characterize relationships for e-business. While a

number of researchers have recognized and

addressed the need for appropriate methodologies to

calculate social ratings based on trust, reputation and

credibility, there is a need for the formalisation of

these concepts through ontologies with special focus

on social protection mechanisms. This paper extends

the initial work by Chang et al. (Chang, Hussain &

Dillon 2005a) who propose a number of generic

ontologies for the integration of trust and reputation

concepts into the SOA.

The integration and formalisation of notions for

trust, trustworthiness, reputation and credibility into

the core specifications of e-commerce frameworks is

a first step to achieve interoperability across existing

e-commerce portals such as ebay, amazon.com or

product search and comparison portals such as

froogle or cnet.com. Currently, these portals and

platforms offer basic reputation mechanisms based

on consumer and/or expert reviews and ratings.

However, these reputation systems are not

interoperable and are generally not sufficient to

protect consumers and businesses from risks such as

fraud, risks related to inadequate product or service

quality, and contract settlement risks.

To overcome these limitations and risks we have

proposed a reputation aware service brokering

architecture for service oriented environments which

are run in a P2P setting (Schmidt, Steele & Dillon

2007). The main purpose of this architecture is its

independence of centralized service brokers and the

integration of trust and reputation measurement

mechanisms to ensure a secure and balanced

community of business service providers and

consumers. This decentralized architecture offers

flexibility, reliability, independence and better

security and, thus, helps to increase consumers’

confidence. Ultimately, this confidence will lead to

the increased adoption of autonomous agents who

will act on the behalf of, both, service providers and

service consumers.

In the following section, we discuss related work

with focus on social protection mechanisms in

virtual communities. In section 3, we introduce trust

and reputation ontologies for business discovery and

selection before addressing a business execution

review ontology in section 4.

2 BACKGROUND

The formation of virtual communities in

decentralized, autonomous environments is one of

the most researched topics in recent years. Several

researchers have proposed the integration of social

information into the semantic web in order to

support the formation of a more secure and stable

digital ecosystem (Abdul-Rahman & Hailes 2000)

(Eysenbach 2001) (Marsh 1994) (Golbeck, Parsia &

Hendler 2003) (Chang, Hussain & Dillon 2005a).

While most of those publications concentrate on

inference methodologies for these social values, few

have addressed the structural relationships of all

involved entities in a more formal semantic structure

of an ontology. In the following we briefly discuss

existing approaches that formalize the integration of

social data into the semantic web as well as the

calculation of this data.

One of the early approaches was proposed by

Marsh (Marsh 1994) who introduces very broad

concepts in his trust model which is based on

observations from social science or even biology.

However, Marsh’s model is not applicable to

specific contexts such as e-business without wide-

ranging prior adjustments and extensions to meet the

requirements of current and future e-business

settings. Abdul-Rahman and Hailes (Abdul-Rahman

& Hailes 2000) propose a trust calculation model

that takes agent reputation and third party opinions

into account but fails to formalize and describe the

complex relationships between all parties in detail.

Instead their work concentrates mostly on the

calculation of trust values.

Eysenbach (Eysenbach 2001) proposes a

specialized ontology that formalizes the

collaboration of agents in the medical domain. He

provides detailed information about the semantics of

interacting entities such as individuals, organisations

and regulators and their relationships with regards to

trust, reputation and general security. He discusses

the opportunities and problems arising from the

semantic web and the ‘web of trust’, but his

TRUST AND REPUTATION ONTOLOGIES FOR ELECTRONIC BUSINESS

309

proposed ontology for e-health is not intended for

usage outside this domain. Golbeck et al. (Golbeck,

Parsia & Hendler 2003) discuss the formation of the

‘web of trust’ in social networks. They extend the

Friend-Of-A-Friend (FOAF) RDF schema (Brickley

& Miller 2006) by integrating nine levels of trust.

However, important concepts for trust calculations

such as reputation and credibility as well as their

relationships with regards to the trust concept are

missing. Furthermore, this research concentrates on

more general trust relationships in distributed social

networks and does, therefore, not reflect the more

complex relationships required for autonomous e-

business interactions.

Chang et al. (Chang, Hussain & Dillon 2005a)

have published the most extensive and formalized

trust and reputation related definitions, ontologies

and calculations in this area so far. Their work

concentrates on general trust and reputation concepts

and relationships for in service oriented

architectures. In this paper we extend their general

trust and reputation ontologies with detailed

concepts that identify and realize specific e-

business-related requirements. In our previous work

we have proposed the Deco Arch framework

(Schmidt, Steele & Dillon 2007), a new approach for

the formation of virtual communities based on

contextual interdependencies between the reputation

of businesses and the contexts they belong to.

3 REPUTATION AND TRUST

ONTOLOGY FOR E-BUSINESS

In the following we describe the static concepts and

relationships for a trust and reputation ontology for

e-business. We base our concepts on the generic

trust and reputation ontologies introduced by Chang

et al. (Chang, Hussain & Dillon 2005a) as depicted

in Figure 1. An agent to be assessed for

trustworthiness, credibility or reputation is called

reputation queried agent or trusted agent and acts as

service provider or seller. The agent assessing the

reputation queried agent’s trustworthiness is called

trusting agent and acts as a consumer. Peer agents

which share information about their past experiences

with the requesting agent are called recommending

agents (Schmidt et al. 2006) (Chang, Hussain &

Dillon 2005a).

In the following subsections we introduce a

number of extensions to this generic ontology with

the necessary details which are required for e-

business specific scenarios such as service, product

or business provider discovery, business selection,

and review of the contract execution on completion.

Figure 1: Generic Trust Ontology for SOA (Chang,

Hussain & Dillon 2005a).

3.1 Business Discovery

The discovery of services, products or business

partners is generally initialized by the consumer.

The consumer defines his preferences and

constraints in a business need profile. This business

need profile is described in detail through a set of

criteria where each criterion has an importance

assigned to it. A criterion may be of simple nature if

it is described through semantic attributes providing

details about the service or product. Such semantic

attributes are for example a book title, a company

name, a quantity or a maximum price. In addition a

criterion may also be expressed as a policy which

links itself to other complex concepts such as quality

of service, privacy, currency, delivery, payment,

time restrictions, etc. Depending on the context other

policies can be added by agents to satisfy individual

requirements. On the other hand businesses are

specified through a business profile which is

composed of a set of criteria and thus follow a

similar semantic structure as the business need

profile defined by the consumer (Figure

2).

Figure 2: Business Discovery Ontology for E-business.

ICEIS 2007 - International Conference on Enterprise Information Systems

310

The information provided in the business need

profile is used to query e-commerce marketplaces

for potential businesses. Results can then be

matched by comparing the criteria expressed in the

business need profile and the business profile which

describe the services, products or business partners.

These matching calculations serve as an initial filter

to reduce the number of candidate business partners.

In a next step the consumer needs to select the

final product or service as well as the business

partner. Depending on the existing information

about a potential business partner, service or product

the consumer agent chooses between two different

approaches for the service selection. The first

approach is a service selection without referral. We

demote the term ‘referral’ as a recommendation or

an opinion that the recommending agent offers to the

trusting agent about the quality or the

trustworthiness of a product or service offered by the

recommendation queried agent. An opinion could

also contain trustworthiness information about the

recommendation queried agent itself. The trusting

agent will only use the business selection without

referral approach if it already possesses sufficient

data about the service, product or service provider

from previous transactions in the same context and

the same timeslot. Furthermore it is imperative that

this data is reasonable current since we assume that

social ratings decay over time (Schmidt et al. 2006).

3.2 Business Selection without Referral

Figure 3 depicts an extended e-business ontology

which defines the relationships between the

consumer and the business concept for the service

selection without referral scenario. In order to

increase the confidence during service selection and

contract negotiation a trust relationship needs to be

established between both parties. This trust

relationship is strongly dependant on reputation

values for the various entities involved. These

entities which are rated through reputation values are

classified as follows:

Consumer concept

• A consumer is represented through an agent

which is rated by its reputation.

Business concept

• A business is rated by its reputation, this includes:

 individual service or product ratings

 service provider or manufacturer ratings.

• A business is part of a group alliance (Schmidt,

Steele & Dillon 2007) which is rated by a

collective reputation.

• A business is represented through a supplier agent

which is rated by its reputation.

Figure 3: Business Selection without Referral.

For example, in a scenario where a company

clerk wishes to purchase a new stack of printer

paper, he assigns the task of discovering suitable

office material suppliers, selecting the appropriate

paper quality and type and its supplier, negotiating a

contract and the monitoring of the contract

execution, to his autonomous agent. The agent

already has historical data available for all potential

suppliers and their products offered from past

interactions. Based on this existing data from past

experiences it evaluates its business risk as low and,

therefore, it chooses to select the supplier without

the need to obtain additional opinions from

recommending agents.

The business concept can have several public

reputation values. First, the business (e.g.

manufacturer or producer) itself has a reputation of

3.79 on a scale of 0-5. Furthermore, a specific

product offered by this business has a reputation of

2.45. Second, the business has a group alliance

reputation of 2.89 which is a weighted average of

the ‘printer paper supplies’ context which is

calculated across reputation values for businesses in

the same group alliance. And finally, the specific

supplier agent has a reputation value of 2.46. All

four reputation values are of interest to the service

consumer which uses these values as part of its

trustworthiness value calculations. The

trustworthiness value is used primarily for service

selection but also provides decision support during

contract negotiations (Schmidt et al. 2005b). On the

other hand, the supplier agent has information about

the service consumer from past interactions and is,

thus, able to calculate a public reputation value of

3.76 which provides the supplier with important

information about its reliability and standing within

the community. This information is especially useful

TRUST AND REPUTATION ONTOLOGIES FOR ELECTRONIC BUSINESS

311

during contract negotiations where the reseller

specifies the payment conditions.

Figure 4: Trust Relationship Concept Details.

Similar to the generic trust ontology (Figure 1),

the extended ontology for service selection without

referral (Figure 3) defines a trust relationship

between the consumer and the business concept.

This trust relationship is defined by a context, a

timeslot and a trustworthiness value (Figure 4). As

mentioned earlier, trustworthiness, reputation and

credibility values loose significance over time as

ratings and opinions about services or products are

updated, or businesses are ranked differently based

on their recent performance. This dynamic

behaviour of the trustworthiness and the reputation

value is represented by the trend concept depicted in

the detailed view of the trust relationship concept in

Figure 4. The trend concept provides a valuable

indication about the recent changes in the

trustworthiness or reputation value and can have the

following states [decreasing, neutral, increasing]

(Schmidt, Steele & Dillon 2007). A second concept

called confidence expresses the strength of a

reputation or trustworthiness value. This strength

value is depending on the number of past

experiences or opinions from which the value was

previously calculated. The more past experiences

with a potential business, supplier, service or

product exist, or the more opinions these

calculations are based on, the higher the confidence

in the resulting reputation or trustworthiness value.

3.3 Business Selection with Referral

In the more common case where the consumer agent

does not possess sufficient data about the service,

product, provider, manufacturer, or supplier from

previous transactions in the same context and the

same timeslot, it will ask neighbouring agents

(recommending agents) to provide opinions on these

entities. In this case there is a need to take the

credibility and trustworthiness of the recommending

agent into account and, therefore, we need to extend

the previously discussed ontology as depicted in

Figure 5.

Figure 5: Business Selection with Referral.

The consumer and the business need to build a

trust relationship which is based on third party

opinions supplied by recommending agents as well

as its own past experiences with the business if the

consumer and the business had previous contact. If

the consumer and the business had a previous

relationship but this information alone was

considered as not sufficient for a comprehensive

trustworthiness evaluation it can still use this

information in the same structure as introduced in

the previous section (see Figure 4). In order to

complete the information about a potential business

partner, the trusting agent (consumer representative)

needs to extend this relationship by allowing third

party recommending agents to contribute their

opinions about their previous interactions with the

recommendation queried agent (business

representative). These opinions are then integrated

into the previously introduced trust relationship

concept as depicted in Figure 6.

Figure 6: Extended Trust Relationship Concept Details.

ICEIS 2007 - International Conference on Enterprise Information Systems

312

In the extended trust relationship concept the

third party opinion concept is the second input for

the trustworthiness value along with the reputation

value calculated from past experiences. If no past

experiences with the recommendation queried agent

exist, the trustworthiness value may be calculated

solely from third party opinions. A third party

opinion is evaluated by several factors. Firstly, there

is a need to assess the credibility of the

recommending agent in its capability and

willingness to provide correct information and,

hence, the trustworthiness of the opinion (Chang,

Hussain & Dillon 2005a). Secondly, the trusting

agent needs a notion for the confidence or strength

of the opinion provided. This confidence will be

high if the opinion is provides datasets containing

information about multiple interactions with the

recommendation queried agent instead of just one.

Similar to the reputation value the recommender

credibility value is also refined by a confidence and

trend value to deal with its aforementioned dynamic

behaviour.

The trust and credibility relationship between the

trusting agent and the recommender agent is based

on the pre-existing trust relationship between both

parties (if present) and the credibility of the

recommending agent to share truthful information.

Opinions contain data about past experiences

between the recommending agent and the

recommendation queried agent which the

recommending agent is prepared to share. The

recommending agent has a high interest in sharing

truthful information since his credibility is at stake

(Schmidt et al. 2005a). The credibility value

ultimately influences its reputation value and, thus,

its standing within the community. Agents with a

low reputation value face several problems such as

exclusion from information sharing, lower authority

during contract negotiations and lower chance of

being chosen by consumers if they are businesses or

even rejection by businesses if they are consumers.

Therefore agents have an apparent interest to

increase their credibility and reputation values by

sharing opinions about their past experiences. If a

recommending agent is totally unknown to the

consumer agent and, thus, has no credibility, then

the opinion has no influence in the actual

trustworthiness value calculations of the consumer.

However, the opinions can be evaluated after the

actual business interaction and, hence, the credibility

value of the recommending agent can be adjusted

accordingly. Growing trustworthiness, reputation

and credibility ratings will influence the standing

and success of the business in future. The

interdependencies of the trust and credibility

relationship are depicted in Figure 7.

Figure 7: Trust & Credibility Relationship.

The trust and credibility relationship is

influenced by the accuracy of past opinions which

the recommending agent shared with the consumer

agent. The accuracy is generally assessed after the

business interaction took place and the actual quality

of service or performance can be compared with the

original opinions provided by the recommending

agents (Schmidt et al. 2006). Furthermore, the

dynamic nature of the opinion accuracy value is

recognized by a trend and a confidence value similar

to the trend and confidence values used to refine the

reputation of the recommending agent.

Figure 8: QoS Review Ontology.

Another factor in the relationship between the

consumer agent and the recommending agent is their

previous trust relationship which provides details

about the general reliability and trustworthiness of

both interacting parties. However, if no trust

TRUST AND REPUTATION ONTOLOGIES FOR ELECTRONIC BUSINESS

313

relationship between both parties exists or the trust

relationship refers to a different context or different

time slot, both parties have to rely solely on

reputation calculations based on third party opinions.

In this case both parties need to evaluate their

business risks and limit their interactions

accordingly.

The relationship between the recommending

agent and the recommendation queried agent is the

same trust relationship we introduced earlier; hence,

we omit a detailed description here and refer to the

previous discussion.

4 QUALITY OF INTERACTION

REVIEW ONTOLOGY

A second building block to ensure successful and

autonomous interactions between agents is the

monitoring and review of the quality of service

(QoS) and contract adherence according to the

mutually agreed contract or service level agreement.

This monitoring process takes place during the

contract execution or service delivery and is of

specific importance for the adjustment of QoS

information. This QoS data is continuously updated

and used to adjust reputation and credibility values

for the business partner during long term business

relationships or long running contracts. For example,

a consumer agent may monitor the QoS of an

internet connection and compare its results with the

service level agreement promised by

telecommunications service provider. Another

example is where a consumer agent constantly

monitors the performance of a financial advisor who

is responsible for investing superannuation funds in

a profitable manner. In both cases the service

providers may also monitor the adherence to the

contract by the consumer who agreed to pay

monthly fees.

In other cases constant monitoring of the

contract adherence may not be required since its

execution is expected to be completed within a very

short period of time. One example for this is the

previously discussed example where a consumer

agent has purchased a stack of printer paper. The

delivery and payment conditions to which both

parties agreed, in this case, are, that the paper must

be delivered within one week by the supplier and the

price must be paid by bank transfer within two

weeks by the consumer. Another example may be

the order of a custom built computer. The consumer

agent will not only review the timely delivery but

also check whether the computer is built according

to the order.

The results of the QoS monitoring or review

process are used to update trustworthiness data for

future reference if the business interaction is

completed. If the business interaction is still ongoing

the constant update of the trustworthiness value with

contract monitoring information may prove

important to detect and solve problems or even

terminate the contract prematurely. Furthermore, the

QoS data can be used to provide opinions about the

trusted agent to other agents. In case of an extended

trust relationship, the review or monitoring

information is furthermore used to assess the quality

of opinions delivered by recommending agents

about the recommendation queried agent. If an

opinion received from a recommending agent differs

significantly from the actual performance of the

trusted agent, the credibility value for these

recommending agents will be adjusted accordingly

(Schmidt et al. 2006).

Public reputation values are also adjusted as a

result of the QoS monitoring and review process.

For example, the reputation of a supplier agent is

increased if it performs better than expected, that is,

it conforms to all contract conditions despite having

a mediocre previous reputation value. Furthermore,

the reputation of a service or product may be

adjusted according to their quality, which may affect

the overall reputation of all products or services that

are categorized in the same context (alliance).

Moreover, significant changes to trustworthiness,

reputation or credibility values will affect the trend

values that indicate the most recent developments of

these social ratings. For example, if a reputation

value changes from 4.6 (very good reputation) to 3.6

(good reputation) than the trend value is adjusted to

‘decreasing’ (Schmidt, Steele & Dillon 2007) which

indicates the negative development of the reputation

value. On the other hand, a reputation value may

increase from 2.9 (some reputation) to 3.6 along

with a new trend value of ‘increasing’. Despite

matching reputation values of, both, agents or

entities, their trend values differ significantly and,

thus, give the evaluating agent an indication about

the future development of both reputation values.

In order to achieve a flexible, consistent, and

efficient QoS review we employ the CCCI

(Correlation, Commitment, Clarity, and Influence)

metrics introduced by Chang et al. (Chang, Hussain

& Dillon 2005b). The central objective of the CCCI

metrics is the measurement of the correlation

between the service contract both agents agreed to

before their business interaction (expected

behaviour) and the actually delivered services or

products during or after the completion of the

business interaction (actual behaviour). The overall

correlation measurement is performed through the

assessment of three variables which play an

ICEIS 2007 - International Conference on Enterprise Information Systems

314

important role in the review process of the business

interaction; commitment, clarity and influence.

The commitment measures the fulfilment of

individual criteria to which both parties mutually

agreed upon in the contract. For example, if a one of

the criteria defined in the contract is a policy which

specifies delivery conditions then it is easy for the

service consumer to rate the commitment to this

criterion by comparing the expected delivery with

the actual delivery. Another important value is the

clarity of individual contract criteria, which need to

be clearly specified, commonly understood and

mutually agreed upon between both business

partners. This is not always as straight forward as

one would expect, for example, if a criterion

specifies the delivery time as ‘autumn’ there are two

problems; first, the delivery date is not quite clear;

and second even the year of delivery is unclear, it

might be this year or in five years. The third and last

central value which is measured as part of the CCCI

metrics is influence. The influence value allows both

parties to denote specific contract criteria as more

important than others. The more important contract

criteria are crucial for the QoS measurement during

or after the completion of the contract.

5 CONCLUSION

In this paper we have proposed a number of

ontologies to formalize and facilitate autonomous

interactions between intelligent agents in centralized

and decentralized e-business environments. These

ontologies focus on the integration of social factors

such as trustworthiness, reputation and credibility

concepts during the formation and stabilization of

unsupervised virtual communities. We provided

detailed descriptions of concepts and their

relationships with regards to essential problems such

as business discovery, business selection (with and

without recommendations from third party peers)

and the review of the quality of service during

and/or after the business interaction. These

ontologies offer a common set of concepts and their

relationships and reflect the complex nature of social

network with specific focus on e-business. The

adherence to such ontological concepts will improve

interoperability between the various platforms and

frameworks and, therefore, improve transparency,

accessibility and increased confidence for all

involved parties. Due to space limitations of

conference proceedings we present an example

application of the proposed ontologies on the DEco

Arch website (Schmidt 2006).

REFERENCES

Abdul-Rahman, A. & Hailes, S. 2000, 'Supporting Trust in

Virtual Communities', 33rd Hawaii International

Conference on System Sciences.

Barkai, D. 2001, Peer-to-Peer Computing: Technologies

for Sharing and Collaborating on the Net, Intel Press.

Brickley, D. & Miller, L. 2006, 'FOAF Vocabulary

Specification', RDFWeb Namespace Document.

Chang, E., Hussain, F. & Dillon, T.S. 2005a, Trust and

Reputation for Service-Oriented Environments:

Technologies For Building Business Intelligence And

Consumer Confidence, John Wiley & Sons.

Chang, E., Hussain, F.K. & Dillon, T. 2005b, 'CCCI

metrics for the measurement of quality of e-service',

Intelligent Agent Technology, IEEE/WIC/ACM

International Conference on, pp. 603-610.

Chang, E.J., Hussain, F.K. & Dillon, T.S. 2005c, 'Fuzzy

nature of trust and dynamic trust modeling in service

oriented environments', Proceedings of the 2005

workshop on Secure web services, pp. 75-83.

Eysenbach, G. 2001, 'An ontology of quality initiatives

and a model for decentralized, collaborative quality

management on the (semantic) world-wide-web', J

Med Internet Res, vol. 3, no. 4, p. E34.

Golbeck, J., Parsia, B. & Hendler, J. 2003, 'Trust

Networks on the Semantic Web', Proceedings of

Cooperative Intelligent Agents, vol. 2003.

Gruber, T.R. 1993, 'A translation approach to portable

ontology specifications', Knowledge Acquisition, vol.

5, no. 2, pp. 199-220.

Marsh, S. 1994, 'Formalizing Trust as a Computational

Concept', University of Stirling.

Schmidt, S. 2006, The DEco Arch framework.,

<http://www-staff.it.uts.edu.au/~sschmidt/decoarch/>.

Schmidt, S., Steele, R. & Dillon, T. 2007, 'DEco Arch:

Trust and Reputation Aware Service Brokering in

Digital Ecosystems ', paper presented to the Inaugural

IEEE International Digital Ecosystems and

Technologies Conference, Cairns, Australia.

Schmidt, S., Steele, R., Dillon, T. & Chang, E. 2005a,

'Applying a fuzzy trust model to E-commerce

systems', paper presented to the Joint Conference on

Artificial Intelligence, Sydney, Australia.

Schmidt, S., Steele, R., Dillon, T. & Chang, E. 2005b,

'Building a fuzzy trust network in unsupervised multi-

agent environments', paper presented to the

International Workshop on Web Semantics, Agia

Napa, Cyprus.

Schmidt, S., Steele, R., Dillon, T. & Chang, E. 2006,

'Fuzzy Trust Evaluation and Credibility Development

in Multi-Agent Systems', Applied Soft Computing.

TRUST AND REPUTATION ONTOLOGIES FOR ELECTRONIC BUSINESS

315