Learning optimally from self-interested data sources in on-line ad auctions

In the analysis of current online ad auctions essential parameters such as click-through rates are often assumed to be known. The disregard of the uncertainty that is present in reality leads to sever