A Novel Reputation Computing Model - (Work-in-Progress)

With the growth of e-commerce, online reputation system has become the important character in many e-commerce sites. The trust between seller and customer established relying on the reputation system. This make the fairness and accuracy of reputation computing model are very important. Current, most reputation computing models cannot reach the objectives, because they didn't consider the number of objects' ratings. In this paper, we propose a novel computing model which use one-way random effects model. This model introduces the random effects, and considers the number of objects' ratings. the random effects is predicted by BLUP(Best linear unbiased Prediction).We have evaluated the difference of top k lists by this model from that by average model in real data sets, and proof the fairness and accuracy of this model using cases.