18. 具有现代意义的最早报纸商业广告之一
《Public Advertiser》
Advertising for coffee,
London 1657
The ad explains “The Vertue of
the COFFEE drink” :
what coffee is,
how it grows,
how it cures numerous
maladies, including
Dropsy, Gout, and Scurvy, …
38. Internet Advertising Revenues Hit Historic
High in Q3 2012 at Nearly $9.3 Billion (U.S)
Interactive Advertising Bureau(在线广告供给方的行业协会),IAB推动了数字化市
场营销行业的发展,制定市场效果衡量标准和在线广告创意的标准比。如它制定素
材的规格,以及广告效果的评估方式。这些标准是与它的会员商议的结果。
39. Half-year Internet ad revenue & Ad Category Breakouts
Year Revenue % Growth
HY 2012 $17,028 14%
HY 2011 $14,941 23%
HY 2010 $12,127 11%
HY 2009 $10,900 -5%
HY 2008 $11,510 15%
HY 2007 $9,993 26%
HY 2006 $7,909 37%
HY 2005 $5,787 26%
HY 2004 $4,599 40%
HY 2003 $3,292 11%
HY 2002 $2,978 -20%
HY 2001 $3,720 -7%
HY 2000 $4,013 147%
HY 1999 $1,627 110%
HY 1998 $774 125%
HY 1997 $344 320%
HY 1996 $82
Total $111,624
HY 2011* HY 2012
% $ % $
Search 46% $6,843 48% $8,128
Display Related 36% $5,349 33% $5,586
-Banner Ads 22% $3,266 21% $3,622
-Digital Video Commercials 6% $891 6% $1,053
-Rich Media 5% $727 3% $495
-Sponsorship 3% $465 2% $416
Mobile* 4% $636 7% $1,242
Classifieds 8% $1,235 7% $1,160
Referrals/Lead Generation 5% $800 5% $834
E-mail 1% $79 0% $78
67. Computational Advertising Central Challenge
Find the "best match" between a given user in a given context and a
suitable advertisement.
− a表示Sponsor(advertiser),c表示媒体(context),u代表受众(user)。
− 公式的含义:给定user,给定context,选择一组ad,使得ROI(投资回报率)最高
Contexts
− Search Engines
− Publisher pages
− Mobile
− SNS
− Intelligent TV
− …
68. Challenge Decomposed
1. Representation
− Represent the user, the context, and the ads in an effective &
efficient way
2. Definition
− Define the mathematical optimization problem to capture the actual
marketplace constraints and goals
3. Solution
− Solve the optimization problem in an effective & efficient way
70. Example: Normalized as an IR Problem
Representation
− Treat the ads as documents in IR
Optimization/solution
− Retrieve the ads by evaluating the query over the ad corpus
Details
− Analyze the “query” and extract query-features
• Query = full context (content, user profile, environment, etc)
− Analyze the documents (= ads) and extract doc-features
− Devise a scoring function = predicates on q-features and d-features +
weights
− Build a search engine that produces quickly the ads that maximize the
scoring function
71. Most Related Conferences
KDD (Knowledge Discovery in Database)
SIGMOD (Management of Data)
VLDB (Very Large Database)
WWW (World Wide Web)
WSDM (Web Search and Data Mining)
SIGIR (Information Retrieval)
CIKM (Conference of Information and Knowledge
Management)
EC (Electronic Commerce)