Presentation_Slides_Zhaoyuan_Zhangx

Download Report

Transcript Presentation_Slides_Zhaoyuan_Zhangx

Trust and Reputation Management
in an Web-Enhanced World
Zhaoyuan Zhang
Email: [email protected]
Preview
1
Introduction of trust and reputation management
2
Models for trust and reputation management
3
Utilization of the models
Why trust and reputation
management?
Emails
E-Commerce
Cloud Computing
Gain
REASONS
knowledge about unknown web services
More efficient investments
Helpful in making decisions
Less risks
Agents
Trusting agent:
“An entity who has faith or belief in
another entity in a given context and
a given time slot”
(Chang, Elizabeth. Trust and Reputation for ServiceOriented Environments Technologies for Building
Business Intelligence and Consumer Confidence.
Page: 27. Johns Wiley & Sons; 2006.)
Trusted agent:
“An entity in whom faith
or belief has been placed by
another entity in a given context
and at a given time slot”
(Chang, Elizabeth. Trust and Reputation for
Service-Oriented Environments Technologies
for Building Business Intelligence and
Consumer Confidence. Page: 28. Johns Wiley
& Sons; 2006.)
A time slot is “a period of time … defined as the
breadth or duration of time over which the trust value
is collected”
Chang, Elizabeth. Trust and Reputation for Service-Oriented Environments Technologies for
Building Business Intelligence and Consumer Confidence. Page: 40. Johns Wiley & Sons;
2006
Trust is “defined as the belief the trusting agent has in the
service provider’s willingness and capability to deliver a
mutually agreed service in a given context and in a given time
slot”.
(Chang, Elizabeth. Trust and Reputation for Service-Oriented Environments
Technologies for Building Business Intelligence and Consumer Confidence. Page: 27.
Johns Wiley & Sons; 2006.)
Trust Recommendation
Trust recommendation is “a method that helps to
form a trust relationship by deriving the initial trust
value, also known as reputation based on references
or recommendations collected from other parties”.
(Chang, Elizabeth. Trust and Reputation for Service-Oriented Environments
Technologies for Building Business Intelligence and Consumer Confidence. Page: 40.
Johns Wiley & Sons; 2006.)
Models for Trust and Reputation Management
Trust Models
Deterministic
Approach
Bayesian
Approach
Fuzzy
Approach
Deterministic Approach
Trust and Reputation for Service-Oriented Environments Technologies for Building Business Intelligence
and Consumer Confidence. Page: 251. Johns Wiley & Sons; 2006.
Deterministic Approach
 Two agents:
 Trusting agent: web service user
 Trusted agent: one or several third party agents
 Trusted agent gives the trust recommendation
 Trusting agent analyzes the trust recommendation from
trusted agents
 Analysis is based on a period of time
Bayesian Approach
Bayesian Theorem
Use a web service’s historical performance
to predict the future performance
A positive historical experience
Use historical trust value
has a value of 1
(history) of a web service as the
condition:
A negative historical experience
has a value of 0
Fuzzy Approach
Define fuzzy sets of expected results of a
web service from different criteria:
•
•
•
Reliability
Availability
Security
Use a web service’s historical performance
statistics
For each record of the history, compare it from
each aspect with the defined result sets
Utilization of the models
Amazon’s Mechanism
“Our very best sellers have a close to 0% Negative Feedback
Rate…
A 0-2% negative feedback rate indicates a good
.
performance of a seller. While a negative rate of greater than
5% means there is some problem with the business
practice…The feedbacks are 12-month based…”
http://www.amazon.com/gp/help/customer/display.html?nodeId=1161284&#good
Amazon’s Mechanism
Historical trust statistics
Fuzzy sets indicating different levels of
potential performance results
Trust value is based on a time slot
Combination of different models
Conclusion
 Introduction of trust and reputation management
• Why trust management?
• Definition of trusting agents, trusted agents, time slot, trust, trust
recommendations.
 Models for trust and reputation management
• Deterministic Approach: trust recommendation
• Bayesian Approach: historical trust value statistics
• Fuzzy Approach: fuzzy sets, historical trust value statistics
 Utilization of the models
 Combination of different models
References
 Chang, Elizabeth. Trust and Reputation for Service-Oriented Environments
Technologies for Building Business Intelligence and Consumer Confidence.
Johns Wiley & Sons; 2006.
 M.wojcik. H.S. Venter. J.H.P. Eloff. Trust Model Evaluation Criteria: A Detailed
Analysis of Trust Evaluation. 2008
 Agudo, Isaac. Fernandez-Gago, Carmen. Lopez, Javier. A Model for Trust
Metrics Analysis.
 Zhu, Manling. Jin, Zhi. Trust Analysis of Web Services Based on a Trust
Ontology. 2007.
 Machala, Daniel. E-Commerce Trust Metrics and Models. Xeroxr Research and
Technology
 Receiving Amazon.com Feedbacks from Buyers.
http://www.amazon.com/gp/help/customer/display.html?nodeId=1161284&#good
 Nefti, Samia. Meziane, Farid. Kasiran, Khairudin. A Fuzzy Trust Model for ECommerce.
 Li, Wenjuan. Ping, Lingdi. Trust Model to Enhance Security and Interoperability
of Cloud Environment. 2009
 Thrope, Sean. Modeling a Trust Cloud Context. 2010.
LOGO