When Opinion Request Meets Majority Search: Avoiding Fraud in On-Line Review Systems

Abstract

On-line review systems have been proved to be sensitive to fraud and have shown some shortcomings due to their reliance on using numerical ratings as reviews. For that reason, and supported by recent work on the field, we aim to address the problem of fraud in this type of systems, by designing a mechanism based on pairwise comparisons-encapsulated as opinion polls-coupled with an incentive policy attempting to foster the collection of majority opinions over individual experiences.