Reputation Bootstrapping For Trust Establishment Among Web Services

Authored By 

  • Zaki Malik
  • Athman Bouguettaya
  • Saksham (13114049)
  • Priyanka Jain (13114039)
  • Rahul Agarwal (13114041)
  • Rishabh Bansal (13114045)
  • Veena Rani Meena (13114070)

Introduction

Reputation System

Reputation Systems computes and publishes reputations(predictor of future behavior) for a set of objects (example- service providers,goods or entities) within a community based on collection of opinions.

  • Reputation Bootstrapping is assessing the reputation of newly deployed web Services.
  • It is a major issue in service oriented environments because historical information often isn't present regarding newcomer behaviors.
  • The author examine different techniques for fairly and accurately bootstrapping newcomer's reputation in a service oriented environment.

Introduction

Procedure

Newcomer with No History Register

For each consumer define 

Ri=Di/Ti

Di=No of defected Transactions that consumer i had undertaken

Ti=Total no of Transactions that consumer i had undertaken

If a newcomer has completed sufficient transactions, we compute weighted average of all Ri's, weights can be assigned according to the contributing rater's creditability.

Else We can ask community provider to bootstrap newcomer's reputation

Assigning Initial Reputation

Case 1 : Default Initial Reputation

  • Upon Registration , the newcomer can present some credentials that let it buy initial reputation from the community provider. The newcomer might belong to the same service provider group as an existing reputable service.
  • Newcomers can present credentials of any service willing to endorse them.
  • Alternatively, the community provider can assign an average of all providers' reputations

Case 2 : Initial Reputation Evaluation 

  • The provider determine the evaluation period, and the newcomer has no knowledge of how long it is or how many transactions occur during this time.
  • The community providers uses services with high credibility(elders)  to evaluate the newcomer. The provider then weighs the feedback according to elder's credibility.
  • In some situations elders might have to pay for newcomer's services. After a transaction completes, the credit agency tells the newcomer about the fake charged account and evaluators return any tangible items delivered.

Results

None : Default Initial Reputation

Punishing : Low Initial Reputation

Adapt :  Our Approach of assigning initial reputation

Applications of the propsed technique

  • To Predict the future behavior of web service providers.
  • Allowing web services to determine to which extent they can trust other services before interacting with them.
  • Allowing newcomers to compete with existing services for market share even when there is no historical information available about their behavior

Limitations

  • A one of a kind service will have no community to provide initial reputation.
  • The elders might not provide proper initial reputation so as to lower the competitions in the market.
  • The newcomers might provide better services initially so as to get higher initial reputation.
  • The adaptive approach doesn't give good BSR(Bootstrp success rate) when Ri is in the middle range where most of the services fall.
  • Newcomers can pay companies to endorse them wrongly.

Possible Improvements

  • Taking average of reputations in various transaction intervals for calculating initial reputation may provide better result.
  • For one of a kind service we can provide a mediocre value as initial reputation.
  • We can check if reputation provided by elders matches actual reputation in long terms and if so we can increase reputation of elders.
  • If a newcomer belong to some service provider group which is initially in the community we can take weighted average of different services provided in that community as the Initial reputation.

Thank You

Computer Networks- Group 9

By Saksham Arya

Computer Networks- Group 9

  • 508