Frequency capping in online advertising (extended abstract)

Niv Buchbinder, Moran Feldman, Arpita Ghosh, Joseph Naor

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

We study the following online problem. Each advertiser ai has a value vi, demand di, and frequency cap fi. Supply units arrive online, each one associated with a user. Each advertiser can be assigned at most di units in all, and at most fi units from the same user. The goal is to design an online allocation algorithm maximizing total value. We first show a deterministic upper bound of 3/4-competitiveness, even when all frequency caps are 1, and all advertisers share identical values and demands. A competitive ratio approaching 1 - 1/e can be achieved via a reduction to a model with arbitrary decreasing valuations [GM07]. Our main contribution is analyzing two 3/4-competitive greedy algorithms for the cases of equal values, and arbitrary valuations with equal demands. Finally, we give a primal-dual algorithm which may serve as a good starting point for improving upon the 1 - 1/e ratio.

Original languageEnglish
Title of host publicationAlgorithms and Data Structures - 12th International Symposium, WADS 2011, Proceedings
Pages147-158
Number of pages12
DOIs
StatePublished - 2011
Event12th International Symposium on Algorithms and Data Structures, WADS 2011 - New York, NY, United States
Duration: 15 Aug 201117 Aug 2011

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume6844 LNCS

Conference

Conference12th International Symposium on Algorithms and Data Structures, WADS 2011
Country/TerritoryUnited States
CityNew York, NY
Period15/08/1117/08/11

All Science Journal Classification (ASJC) codes

  • Theoretical Computer Science
  • General Computer Science

Fingerprint

Dive into the research topics of 'Frequency capping in online advertising (extended abstract)'. Together they form a unique fingerprint.

Cite this