A survey of trust and reputation systems for online service provision

[jghms]

Another work which is important to establish an overview of trust and reputation systems. paper

[jghms]

Can there actually be global consensus on trust? Trust is clearly a subjective thing, based on personal experience more than on reputation of the counterparty, at least in the real world. I can trust someone eventhough he has a bad reputation, just because he has always been trustworthy to me. I value my personal interactions more than those of others. But then again in the digital world we are all the same, there wouldn't be a reason why one node should be treated differently than another.

[jghms]

Reputation is the score that the whole community can assign, while trust is a subjective view of an entity's trustworthyness.

Thus if we create consensus on the trusts of all nodes, we have a global reputation system.

[jghms]

Similarities and differences with recommenation systems Collaborative filtering systems are recommenders, which look for similarities between users and for instance recommend items other similar users have bought. A similar concept is that of a reputation system, although reputation should not be subjective but rather objective, a score given based on the transaction history.

[jghms]

Types of trust Subjective and specific "How do you see the election candidate X's ability to handle the economy? - (Bad, Average, Good)

Subjective and general Package sent from ebay arrives broken, receiver gives bad rating but problem was with post office, wrong blaming.

Objective and specific Technical tests of physical objects in test rigs.

Objective and general Average of different objective scores.

[jghms]

Reputation computation engines

• Simple summation
• Bayesian systems
• Discrete trust models
• Belief models
• Fuzzy models
• Flow models