Abstract: It is difficult to price big daga due to its diversity and sparsity, With big data auction set as background, for the problem that big data has a characteristic of replicability and that traditional auction models cannot determine how many big data items sellers should auction to get the maximum profits. A discussion is given on what auction strategies big data sellers should adopt under various circumstance to maximize their profits and on the appropriate number of winning bidders. The traditional Vickrey auction model and sequential auction model are modified, and two new auction models, which are suitable for different conditions of the auction, are proposed. Finally, the two auction models designed and relevant earnings expectations have been obtained.

Key words: big data, auction, the auction number