New algorithms for mining the reputation of participants of online auctions (Q943870)

Two novel measures are proposed for reputation rating in the online auction environment: credibility and density. The credibility measure is defined for buyers and sellers iteratively: a buyer is more credible if he participates successfully in many auctions involving credible sellers and vice versa. The density measure evaluates the number of neighbors for a given seller, i.e. the number of other sellers who are linked to the given one by a sufficient number of common buyers. The author discusses how these measures can be used to safeguard against such popular schemes of reputation rating deception as ballot stuffing and clique creating. Application to data on www.allegro.pl, the leading Polish online auctions provider in presented.