An Iterative Reputation Ranking Method via the Beta Probability Distribution

Ranking user reputation and object quality has drawn increasing attention for online rating systems. By introducing an iterative reputation-allocation process, in this paper, we present an iterative reputation ranking algorithm in terms of the beta probability distribution (IBeta), where the user reputation is calculated as the probability that the user will give fair ratings to objects and the high reputation users’ ratings have larger weights in dominating the corresponding quantity of fair/unfair ratings. User reputation is reallocated based on their ratings and the previous reputations. The user reputation and users’ quantities of fair/unfair ratings are iteratively updated until they become stable. The experimental results for the synthetic networks show that both the AUC values and Kendall’s tau
$\tau $
of the IBeta algorithm are larger than those generated by the RBPD method with different fractions of random ratings. Moreover, the results for the empirical networks indicate that the presented algorithm is more accurate and robust than the RBPD method when the rating systems are under spamming attacks. This paper provides a further understanding on the role of the probability for the online user reputation identification.