Randomized revenue monotone mechanisms for online advertising

Abstract

Online advertising is the main source of revenue for many Internet firms; thus designing effective mechanisms for selecting and pricing ads become an important research question. In this paper, we seek to design truthful mechanisms for online advertising that satisfy Revenue Monotonicity (RM) - a natural property which states that the revenue of a mechanism should not decrease if the number of participants increase or if a participant increases her bid. In a recent work Goel and Khani [5], it was argued that RM is a desired goal for proper functioning of an online advertising business. Since popular mechanisms like VCG are not revenue-monotone, they introduced the notion of Price of Revenue Monotonicity (PoRM) to capture the loss in social welfare of a revenue-monotone mechanism. Goel and Khani [5] then studied the price of revenue-monotonicity of Combinatorial Auction with Identical Items(CAII). In CAII, there are k identical items to be sold to a group of bidders, where bidder i wants either exactly di ∈ {1,…, k} number of items or nothing. CAII generalizes important online advertising scenarios such as image-text and video-pod auctions. In an image-text auction we want to fill an advertising slot with either k text-ads or a single image-ad. In video-pod auction we want to fill a video advertising break of k seconds with video-ads of possibly different durations. Goel and Khani [5] showed that no deterministic RM mechanism can attain PoRM of less than ln(k) for CAII, i.e., no deterministic mechanism can attain more than 1 ln(k) fraction of the maximum social welfare. Goel and Khani [5] also design a mechanism with PoRM of O(ln2(k)) for CAII.