Algorithms and strategies for web advertising
- Web advertising is a multi-billion dollar business and the main revenue source of the most popular web sites such as Google, Yahoo and Facebook. Web advertisers reach their potential customers mainly through two different channels: sponsored search, the display of textual ads next to the search engine organic results and display advertising, the display of graphical ads as part of web pages. In the simplest form of web advertising three parties are involved: the advertisers, the medium, i.e., the search engine or the publisher, and the users. Their interactions can be grouped in three steps: (a) Bidding: The advertisers provide the medium with their ads, their bids, and some preferences regarding the users of interest; (b) Delivery: Given the ads and a user in a particular context, i.e., a query or a web page, the medium selects an ad that respects advertiser preferences and maximizes its own expected profit; (c) Response: The user sees the selected ad and he either ignores it or proceeds to an action related to it, e.g., he clicks a link or buys a product. Such actions fulfill the advertiser goals and, consequently, they drive the whole procedure. In this thesis we study each of these steps and we propose new algorithms and strategies that tackle inefficiencies of current solutions. In bidding, we introduce output URL bidding, a new bidding mechanism for sponsored search, where advertisers bid on search result URLs, as opposed to keywords in the input query. For example, an advertiser may want his ad to appear whenever the search result includes the sites www.imdb.com and en.wikipedia.org, instead of bidding on keywords that lead to these sites, e.g., particular movie titles or actor names. We have shown that output bidding can provide a more concise language among advertisers and the search engine and it can increase the recall of relevant queries. In ad selection, we introduce sponsored search auctions with conflicts. In such auctions, advertisers can condition their bids on the non-appearance of certain undesired ads on the results page. With such constraints the ad selection becomes computationally challenging. However, we introduce algorithms that can deal with the computational cost, while providing advertisers with higher utility and the search engine with increased revenue. Finally, our work includes also novel ways to quantify the user response. Using a field experiment we have measured the impact of display advertising on user search behavior. Our results show that display ads can increase the volume of queries that are related to an ad by 5% to 25% for a two-week period. Surprisingly, the impact of advertising is not limited to exposed users but also to their social friends.
|Type of resource
|electronic; electronic resource; remote
|1 online resource.
|Stanford University, Department of Electrical Engineering
|Statement of responsibility
|Submitted to the Department of Electrical Engineering.
|Thesis (Ph. D.)--Stanford University, 2011.
- © 2011 by Panagiotis Papadimitriou
- This work is licensed under a Creative Commons Attribution Non Commercial 3.0 Unported license (CC BY-NC).
Also listed in
Loading usage metrics...