Information Based Reputation Clause Samples

Information Based Reputation. ‌ [261] illustrates how reputation may be computed by aggregating peer opinions. The proposed method makes use of social network analysis to decide how opinions may be aggregated when there is some sort of dependence between opinions. The basic idea is that a peer may express its opinion about a given aspect of the object being analysed in a given context. The opinion is described as a probability distribution over an eval- uation space . After forming an opinion, an agent may then share this opinion with the rest of the group. A group discussion then takes place, after which the agent may or may not revise their opinions, based on how much can others convince them about changing their opinion. Finally, the group as a whole tries to reach a unified group opinion. Reputation is then defined as the “social evaluation by the group”. Of course, sharing opinions is the central point of this paper. The idea is that shared opinions affect the future opinions of other peers. Hence, [261] proposes a simple and basic communication model for sharing opinions. Opinions then influence each other based on the “semantic similarity” between the two concepts being evaluated. Forming opinions is affected by several measures, such as the time decay factor, the reliability of the information source, etc. The accuracy of an agent’s opinion may sometime be verifiable within a reasonable amount of time. For example, if one gives their opinion about the weather tomorrow, then this opinion may be verified the next day by actually observing the weather and comparing it to what the agent predicted. However, not all opinions may be verified, such as the opinion about the quality of a given scientific paper. In such cases, the opinion may then be evaluated by comparing it to the group opinion. Three different methods are then proposed for the aggregation of opinions, in the hope of reaching one final group opinion. The dependent method aggregates the opinions of agents that have been discussing/sharing their opinions together. If this method fails to return results, then the data is deemed inconsistent and the agents should have further discussions or agree to disagree. Otherwise, the method proposed the group opinion with maximum certainty. If this result is rejected by the agents, then the final proposal is to say that the group opinion lies some where between the result computed by the dependent method and that of the independent method, which assumes that the priors are completely...