Logo image
Maximizing the residual surplus in position auctions
Thesis   Open access

Maximizing the residual surplus in position auctions

Safa T. Aman
Master of Science (M.S.), Drexel University
Jun 2020
DOI:
https://doi.org/10.17918/00000248
pdf
Aman_Safa_2020630.40 kBDownloadView

Abstract

Mechanism design focuses on designing algorithms to allocate a finite set of resources to a set of strategic agents while maximizing some objective function. This thesis studies mechanisms which maximize residual surplus, which is given as the total value of the allocation chosen minus the payments charged to agents. Specifically, this work describes the design of an optimal mechanism for position auctions, where each agent has a valuation for receiving some service, and agents are allocated some probability of receiving that service. Position auctions most commonly model sponsored search auctions, where the agents are allocated advertisement slots for a web search result and the service provided is a click on the agent's ad. This work defines the truthful Bayesian optimal mechanism for position auctions as a multi-price lottery. It then explores how this could be used to design a prior-free mechanism for position auctions that performs well without the need for the Bayesian assumption by relaxing the more complex lottery to a single-price lottery, and suggests limitations of this approach.

Metrics

29 File views/ downloads
40 Record Views

Details

Logo image