上海科技大学知识管理系统

The growing importance of big data in various industries necessitates the development of dedicated pricing models, as current mechanisms often fail to capture its true value. This paper addresses this challenge by introducing an auction model tailored to the unique characteristics of big data, specifically its replicability and externality -- the phenomenon where the value of data depreciates as the number of copies increases.

Although the VCG mechanism is truthful for big data auctions with externalities, its practice of charging different payments to different buyers results in price discrimination. Price discrimination can boost short-term revenue, but it also brings about issues such as perceived unfairness, market segmentation, decreased consumer trust, potential legal risks, and so on. These concerns motivate us to design a uniform pricing mechanism for big data auctions. Therefore, our study explores two scenarios: one where the seller has no prior knowledge of buyers' valuations, and another where the seller is aware of the buyers' value distributions. We introduce four truthful auction mechanisms tailored for big data with externalities, with two mechanisms specifically designed for each scenario. Additionally, we analyze the advantages of these mechanisms through competitive analysis and expectations based on prior information.