Selecting and/or scoring content-relevant advertisements

14 Pages·2014·1.61 MB·English

USOO8706546B2
(12) Unlted States Patent (10) Patent N0.: US 8,706,546 B2
Anderson et a]. (45) Date of Patent: Apr. 22, 2014
(54) SELECTING AND/OR SCORING OTHER PUBLICATIONS
CONTENT-RELEVANT ADVERTISEMENTS Bangalore, Srl_ nl_ vas and Rambow, Owen, Explor_t l_ ng a Probabil_ is_ ti_ c
(75) Inventors, Darrell Anderson, Mountain View, C A Hierarchical Model for Generation. In: Proceedings of the 18th Con
(US)_ Alexander P aul C at obus ference on Computational Linguistics (COLING ’2000), Jul.
’ . . _ . ’ 31-Aug. 4, 2000, (pp. 42-48).*
Mounnan VleW’ CA @S)’ Glao NgUyen’ PCT/ISN220, “Noti?cation of Transmittal of the International
saqamemo’ CA (US)’ Narayanan Search Report and the Written Opinion of the International Searching
Shlvakumar, san Jose, CA (Us) Authority, or the Declaration” for PCT/US06/27900, mailed Mar. 5,
_ _ _ 2007 (1 pg.).
(73) Asslgnee: Google Inc., Mountain View, CA (U S) PCT/ISAJ210, “International Search Report” for PCT/USO6/27900,
mailed Mar. 5, 2007 (2 pgs.).
( * ) Notice: Subject to any disclaimer, the term of this PCT/ISAJ237, “Written Opinion of the International Searching
patent is extended or adjusted under Authority” for PCT/USO6/27900, mailed Mar. 5, 2007 (3 pgS.)
U_s_c_ 15403) by 2071 days_ US. Appl. No. 10/314,427 titled “Methods and Apparatus for Serv
ing Relevant Advertisements”, ?led Dec. 6, 2002.
(21) APPL NO, 11/184,053 US. Appl. No. 10/375,900 titled “Serving Advertisements based on
Content”, ?led Feb. 26, 2003.
- _ U.S. Appl. No. 10/676,571 titled Methods and Apparatus for Char
(22) Flled' JUL 18’ 2005 acteriZing Documents Based on Cluster Related Words, ?led Sep. 30,
(65) Prior Publication Data 2003'
US 2007/0016473 A1 Jan. 18, 2007 * Cited by eXaII?nel‘
(51) Int_ CL Primary Examiner * Alvin L Brown
G06Q 30/00 (2012.01) (74) Attorney, Agent, or Firm * Lerner, David, Littenberg,
(52) US, Cl, KrumholZ & Mentlik, LLP
USPC .......................................................... .. 705/14
(58) Field of Classi?cation Search (57) ABSTRACT
USPC .......................................................... .. 705/14 Ads eligible to be served With a document (for example,
See application ?le for complete search history. because they are relevant to the document) may each be
scored using a price parameter associated With the ad and an
(56) References Cited indication of relevancy of the ad to the document. The indi
cation of relevancy of the ad to the document may be based on
U.S. PATENT DOCUMENTS
an ordinal ranking of a relevancy criteria of the document
used to select the ad, and/or a value of a relevancy criteria of
6,269,361 Bl 7/2001 Davis et a1. _ _
7,373,599 B2 5/2003 McElfresh et 31, the document used to select the ad. The ellglble ads may be
2003/0046161 A1 3/2003 Kamangar et a1. determined by obtaining relevancy criteria for the document
é * ?x? et ~~~~~~~~~~~~~~~~~~~~~ ~~ 705/14 and selecting ads using at least some of the obtained relevancy
e er e a . ' ' '
>I< ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘ H a
2006/0230053 A1* 10/2006 Eldering ........ .. .. 707/101 dlsmbmed manner
2006/0287920 Al * 12/2006 Perkins et a1. ...... .. 705/14
2012/0095827 A1 * 4/2012 Vora et al. ................ .. 705/14.49 21 Claims, 4 Drawing Sheets
130
AD ENTRY,
ADVERTISERS MAINTENANCE AND AD CONSUMERS
DELIVERY SYSTEM(S)

US. Patent
Apr. 22, 2014 Sheet 2 0f 4 US 8,706,546 B2
CONTENT-RELEVANT
AD SERVER
b.) o
ACCEPT REQUEST INCLUDING (\1/510
DOCUMENT INFORMATION
I
OR SIMPLY
ACCEPT
RANKED DETERMINE AD SEARCH CRITERIA (\320
(ORDERED) USING DOCUMENT INFORMATION
CRITERIA
Y
RANK (ORDER) THE CRITERIA M30
Y
ASSIGN A WEIGHT TO (AT
LEAST SOME OF) THE SEARCH N340
CRITERIA USING RANKS
I
SELECT ELIGIBLE (e.g., RELEVANT)
ADS USING THE CRITERIA
SCORE IDENTIFIED ELIGIBLE ADS USING (\370
ASSOCIATED CRITERIA WEIGHT
RETURN 380
FIGURE 3
US. Patent Apr. 22, 2014 Sheet 3 0f4 US 8,706,546 B2
/\, SHARD1
440a AD QUERY
\(CRITERlA)
SHARD 2 <- 4‘
4406\1 _.,. ~ 4_AD REQUEST
. DOCUMENT
AD MIXER < ( )
. REPLY (AD(S)) —>
- 410 8
?
4405, SHARD T
E
SHARD1 550 AD MIXER
DOCUMENT CRITERIA
CRITERIA TO AD INFORMATION (e.g.,
SERVING DOCUMENT - > CRITERIA
OPERATIONS \ INDEX)
AD 8530
INFORMATION DOCUMENT To AD REQUEST
(PART1 OF N) CRITERIA SERVING BOCUMENT)
S “ OPERATIONS
560 """""""""""""""" ._
AD FILTERING/ 520
F). REDUCTION A REPLY
0 OPERATIONS AD (AD($))
57 ----------------- " ARBITRATION —>
3 OPERATIONS
540 - ------ -' 580
500 5108
' FIGURE 5
US. Patent
Apr. 22, 2014 Sheet 4 0f 4 US 8,706,546 B2
a
m3m.
M
5momn5aa>o>5mEzd
v8 7 mmEDQE
11 >V m@ow@<moomwS0w5mmo Eom m
w, am
S6vE53905 5
5$v5%25 0
US 8,706,546 B2
1 2
SELECTING AND/OR SCORING etc. US. Provisional Application Ser. No. 60/416,144 (incor
CONTENT-RELEVANT ADVERTISEMENTS porated herein by reference and referred to as “the ’ 144 appli
cation”), titled “Methods and Apparatus for Probabilistic
§1. BACKGROUND OF THE INVENTION Hierarchical Inferential Learner” ?led on Oct. 3, 2002 and
US. patent application Ser. No. 10/676,571 (referred to as
§1.1 Field of the Invention “the ’ 571 application” and incorporated herein by reference),
The present invention concerns advertising, such as online titled “Methods andApparatus for Characterizing Documents
advertising for example. In particular, the present invention Based on Cluster Related Words,” ?led on Sep. 30, 2003 and
concerns selecting and/ or scoring content-relevant advertise listing Georges Harik and Noam Shazeer as inventors
ments (“ads”). describe exemplary ways, that may be used in a manner
§1.2 Background Information consistent with the principles of the present invention, to
Advertising using traditional media, such as television, determine one or more concepts or topics of information.
radio, newspapers and magazines, is well known. Unfortu The ’900 application further describes that a document
nately, even when armed with demographic studies and may be associated with one or more ads using a document
entirely reasonable assumptions about the typical audience of identi?er (e. g., a URL) to determine one or more ads. For
various media outlets, advertisers recognize that much of example, the document information may have been processed
their advertising budget is simply wasted. Moreover, it is very to generate relevance information, such as a cluster (e.g., a
dif?cult to identify and eliminate such waste. PHIL cluster), a topic, etc. The document clusters may then be
Recently, advertising over more interactive media has used as query terms in a large OR query to an index that maps
become popular. For example, as the number of people using 20 topics (e.g., PHIL cluster identi?ers) to a set of matching ad
the Internet has exploded, advertisers have come to appreciate groups, via the index that maps topics to ad groups.
media and services offered over the Internet as a potentially The results of this query may then be used as ?rst cut set of
powerful way to advertise. candidate criteria. More speci?cally, the candidate ad groups
Interactive advertising provides opportunities for advertis may then be used to determine an actual information retrieval
ers to target their ads to a receptive audience. That is, targeted 25 (IR) score for each ad group summarizing how well the cri
ads are more likely to be useful to end users since the ads may teria information plus the ad text itself matches the document
be relevant to a need inferred from some user activity (e.g., relevance information. Estimated or known performance
relevant to a user’s search query to a search engine, relevant to parameters (e. g., selection rates, conversion rates, etc.) for the
content in a document requested by the user, etc.). Query ad group may be considered in helping determine the best
keyword targeting has been used by search engines to deliver 30 scoring ad group(s). Targeting criteria associated with the
relevant ads. For example, the AdWords advertising system best scoring ad group(s) can be used as “criteria” to determine
by Google Inc. of Mountain View, Calif. (referred to as a ?nal set of ads.
“Google”), delivers ads targeted to keywords from search A content-relevant an ad server can use the set of one or
queries. Similarly, content targeted ad delivery systems have more “criteria” to request ads. The provided ads may partici
been proposed. For example, US. patent application Ser. No. 35 pate in an arbitration (e.g., an auction) to place the ads in
10/314,427 (incorporated herein by reference and referred to available ad spots, to provide the ads with enhanced features
as “the ’427 application”), titled “METHODS AND APPA or treatments (e.g., enhanced colors, enhanced fonts, images,
RATUS FOR SERVING RELEVANT ADVERTISE animation, etc.), etc.
MENTS”, ?led on Dec. 6, 2002 and listing Jeffrey A. Dean, There are many ways of selecting a set of ads given a set of
Georges R. Harik and Paul Buchheit as inventors; and Ser. 40 one or more “criteria.” For example, a requestor may request
No. 10/ 375,900 (incorporated by reference and referred to as that an ad be sent back if K of the M criteria sent match a
“the ’900 application”), titled “SERVING ADVERTISE single ad group. One version of AdSense from Google deter
MENTS BASED ON CONTENT,” ?led on Feb. 26, 2003 and mined a set of ads give a set of one or more criteria as follows.
listing Darrell Anderson, Paul Buchheit, Alex Carobus, Claire Suppose a list of the 60 best criteria is provided. Such criteria
Cui, Jeffrey A. Dean, Georges R. Harik, Deepak Jindal and 45 could be grouped into a sequence of queries, such as:
Narayanan Shivakumar as inventors, describe methods and
apparatus for serving ads relevant to the content of a docu rawiqueryOI"critPORicritliORicritZ . . .
iORicritw”;
ment, such as a Web page for example. Content targeted ad
delivery systems, such as the AdSense advertising system by
Google for example, have been used to serve ads on Web 50 rawiqueryl:“critlsiORicritl7iORicrit18 . . .
iORicrit3O”;
pages.
As can be appreciated from the foregoing, serving ads
relevant to concepts of text in a text document is useful rawiqueryZI“crit3liORicritDiORicrit33 . . .
because such ads presumably concern a current user interest. iORicrit45”; and
Consequently, such online advertising has become increas 55
ingly popular. However, such content-targeted ad delivery
rawiquery3:“crit46iORicrit47iORicrit48 . . .
systems can be improved. For example, the ’900 application iORicritGO”. [1]
describes how so-called “targeting criteria” (or simply “cri for example. Each of the queries could be processed in
teria”, whether used in the singular or plural form), used to sequence until any ad, without regard to the number of ads, is
look up relevant ads, may be determined in an exemplary 60 returned. For example, raw_queryO could be processed. If any
embodiment. Speci?cally, the ’900 application describes that ads matched any of critO through critls, these ads would be
an off-line (perhaps nightly) dump of a complete ads database returned (for subsequent processing) and the other raw que
is used to generate an index that maps topics (e.g., a PHIL ries would not be processed. If no ads were returned for
cluster identi?ers) to a set of matching ad groups. This may be raw_query0, raw_queryl would be processed, and so on.
done using one or more of (i) a set of serving constraints 65 Although this approach has worked well, it has room for
(targeting criteria) within the ad group, (ii) text of the ads improvement. More speci?cally, opportunities to ?ll ad spots
within the ad group, (iii) content on the advertiser’s Web site, may be lost, criteria-to-document relevancy information can
US 8,706,546 B2
3 4
be ignored (or at least diluted), and multiple requests can lead FIG. 2 is a diagram illustrating an environment in which, or
to complexity and wasted resources. Each of these limitations with which, embodiments consistent with the present inven
is addressed below. tion may operate.
With regard to lost opportunities to ?ll ad spots, by stop FIG. 3 is a ?ow diagram of an exemplary method for
ping the process once one of the raw_queries returns at least determining and scoring ads in a manner consistent with the
one ad, there is a potential for un?lled ad spots. For example, present invention.
suppose that a document has six (6) ad spots (and a requestor FIG. 4 is a block diagram illustrating a distributed ad serv
ing system.
wants six (or more) ads). Suppose further that raw_queryO
(ultimately) produces no ads, but raw_queryl (ultimately) FIG. 5 is a bubble diagram of exemplary operations that
may be performed in a manner consistent with the present
produces one ad. The process is stopped at this point and ?ve
invention, as well as information that may be used and/or
ad spots are left un?lled. Suppose that raw_query2 would
generated by such operations, in a distributed ad serving
have produced 20 ads. Since raw_query2 is never processed,
system.
only one ad is shown. This is a wasted opportunity to show
FIG. 6 is a block diagram of apparatus that may be used to
relevant ads, which is a lost opportunity to generate revenue
perform at least some operations, and store at least some
for the ad serving system.
information, in a manner consistent with the present inven
With regard to ignoring or diluting criteria-to-document
tion.
relevancy information, although the groups of queries are
ordered such that an earlier processed group (e.g., raw §4. DETAILED DESCRIPTION
_query0) has more relevant criteria than a potentially later 20
processed group (e.g., raw_queryl, raw_query2, etc.), within The present invention may involve novel methods, appa
a group, all criteria are treated equally. Consider, for example, ratus, message formats, and/or data structures for selecting
raw_query0. Since the criteria have been ranked by relevancy, and/or scoring content-relevant ads. The following descrip
critl may be much more relevant to the content of the docu tion is presented to enable one skilled in the art to make and
ment than crit9. However, suppose that raw_queryO returns 25 use the invention, and is provided in the context of particular
applications and their requirements. Thus, the following
two adsiad A and ad B. Suppose further that ad A was
returned because it had a targeting criteria that matched critl, description of embodiments consistent with the present
invention provides illustration and description, but is not
while ad B was returned because it had a targeting criteria that
intended to be exhaustive or to limit the present invention to
matched crit9. Thus, ad A is more relevant to the document
content than ad B. Unfortunately, however, this fact is ignored 30 the precise form disclosed. Various modi?cations to the dis
closed embodiments will be apparent to those skilled in the
in an arbitration in which ad A and ad B compete.
art, and the general principles set forth below may be applied
Finally, with regard to complexity, processing multiple
to other embodiments and applications. For example,
requests (e.g., a second and perhaps even a third request)
although a series of acts may be described with reference to a
leads to complexity and extra load on processing and com
35 ?ow diagram, the order of acts may differ in other implemen
munications resources, particularly in a distributed environ
tations when the performance of one act is not dependent on
ment. the completion of another act. Further, non-dependent acts
Accordingly, given a set of ordered criteria, it would be may be performed in parallel. No element, act or instruction
useful to improve an ad server to generate more ads, to gen used in the description should be construed as critical or
erate more relevant ads, and/ or to reduce the load on process 40 essential to the present invention unless explicitly described
ing, communication and/or storage resources. as such. Also, as used herein, the article “a” is intended to
include one or more items. Where only one item is intended,
§2. SUMMARY OF THE INVENTION the term “one” or similar language is used. Thus, the present
invention is not intended to be limited to the embodiments
Embodiments consistent with the present invention may be 45 shown and the inventors regard their invention to include any
used to score ads eligible to be served with a document (for patentable subject matter described.
example, because they are relevant to the document). In at In the following de?nitions of terms that may be used in the
least some embodiments consistent with the present inven speci?cation are provided in §4.l. Then, environments in
tion, each ad is scored using a price parameter associated with which, or with which, the present invention may operate are
the ad and an indication of relevancy of the ad to the docu 50 described in §4.2. Exemplary embodiments of the present
ment. In at least some embodiments consistent with the invention are described in §4.3. Thereafter, a speci?c example
present invention, the indication of relevancy of the ad to the illustrating the utility of one exemplary embodiment of the
document is based on an ordinal ranking of a relevancy cri present invention is provided in §4.4. Finally, some conclu
teria of the document used to select the ad, and/ or a value of sions regarding the present invention are set forth in §4.5.
a relevancy criteria of the document used to select the ad. 55 §4.l De?nitions
In at least some embodiments consistent with the present Online ads, such as those used in the exemplary systems
invention, the eligible ads are determined by obtaining rel described below with reference to FIGS. 1 and 2, or any other
evancy criteria for the document and selecting ads using at system, may have various intrinsic features. Such features
least some of the obtained relevancy criteria. may be speci?ed by an application and/ or an advertiser. These
In at least some embodiments consistent with the present 60 features are referred to as “ad features” below. For example,
invention, the ads are selected, and perhaps ?ltered, in a in the case of a text ad, ad features may include a title line, ad
distributed manner. text, and an embedded link. In the case of an image ad, ad
features may include images, executable code, and an embed
§3. BRIEF DESCRIPTION OF THE DRAWINGS ded link. Depending on the type of online ad, ad features may
65 include one or more of the following: text, a link, an audio ?le,
FIG. 1 is a diagram showing parties or entities that can a video ?le, an image ?le, executable code, embedded infor
interact with an advertising system. mation, etc.
US 8,706,546 B2
5 6
When an online ad is served, one or more parameters may pages, spending at least a predetermined amount of time on a
be used to describe how, when, and/or where the ad was Website or Web page, registering on a Website, etc. Often, if
served. These parameters are referred to as “serving param user actions don’t indicate a consummated purchase, they
eters” below. Serving parameters may include, for example, may indicate a sales lead, although user actions constituting a
one or more of the following: features of (including informa conversion are not limited to this. Indeed, many other de?ni
tion on) a document on which, or with which, the ad was tions of what constitutes a conversion are possible.
served, a search query or search results associated with the The ratio of the number of conversions to the number of
serving of the ad, a user characteristic (e.g., their geographic impressions of the ad (i.e., the number of times an ad is
location, the language used by the user, the type of browser rendered) and the ratio of the number of conversions to the
used, previous page views, previous behavior, user account, number of selections (or the number of some other earlier
any Web cookies used by the system, user device character event) are both referred to as the “conversion rate.” The type
istics, etc.), a host or af?liate site (e.g., America Online, of conversion rate will be apparent from the context in which
Goo gle, Yahoo) that initiated the request, an ab solute position it is used. If a conversion is de?ned to be able to occur within
of the ad on the page on which it was served, a position a predetermined time since the serving of an ad, one possible
(spatial or temporal) of the ad relative to other ads served, an de?nition of the conversion rate might only consider ads that
absolute size of the ad, a size of the ad relative to other ads, a have been served more than the predetermined time in the
color of the ad, a number of other ads served, types of other past.
ads served, time of day served, time of week served, time of A “property” is something on which ads can be presented.
year served, etc. Naturally, there are other serving parameters A property may include online content (e.g., a Website, an
that may be used in the context of the invention. 20 MP3 audio program, online games, etc.), o?line content (e. g.,
Although serving parameters may be extrinsic to ad fea a newspaper, a magaZine, a theatrical production, a concert, a
tures, they may be associated with an ad as serving conditions sports event, etc.), and/or o?line objects (e.g., a billboard, a
or constraints. When used as serving conditions or con stadium score board, and out?eld wall, the side of truck
straints, such serving parameters are referred to simply as trailer, etc.). Properties with content (e.g., magaZines, news
“serving constraints” (or “targeting criteria”). For example, in 25 papers, Websites, email messages, etc.) may be referred to as
some systems, an advertiser may be able to target the serving “media properties.” Although properties may themselves be
of its ad by specifying that it is only to be served on weekdays, of?ine, pertinent information about a property (e. g., attribute
no lower than a certain position, only to users in a certain (s), topic(s), concept(s), category(ies), keyword(s), relevancy
location, etc. As another example, in some systems, an adver information, type(s) of ads supported, etc.) may be available
tiser may specify that its ad is to be served only if a page or 30 online. For example, an outdoor jaZZ music festival may have
search query includes certain keywords or phrases. As yet entered the topics “music” and “jaZZ”, the location of the
another example, in some systems, an advertiser may specify concerts, the time of the concerts, artists scheduled to appear
that its ad is to be served only if a document, on which, or with at the festival, and types of available ad spots (e.g., spots in a
which, the ad is to be served, includes certain topics or con printed program, spots on a stage, spots on seat backs, audio
cepts, or falls under a particular cluster or clusters, or some 35 announcements of sponsors, etc.).
other classi?cation or classi?cations (e.g., verticals). In some A “document” is to be broadly interpreted to include any
systems, an advertiser may specify that its ad is to be served machine-readable and machine-storable work product. A
only to (or is not to be served to) user devices having certain document may be a ?le, a combination of ?les, one or more
characteristics. Finally, in some systems an ad might be tar ?les with embedded links to other ?les, etc. The ?les may be
geted so that it is served in response to a request sourced from 40 of any type, such as text, audio, image, video, etc. Parts of a
a particular location, or in response to a request concerning a document to be rendered to an end user can be thought of as
particular location. “content” of the document. A document may include “struc
“Ad information” may include any combination of ad fea tured data” containing both content (words, pictures, etc.) and
tures, ad serving constraints, information derivable from ad some indication of the meaning of that content (for example,
features or ad serving constraints (referred to as “ad derived e-mail ?elds and associated data, HTML tags and associated
information”), and/or information related to the ad (referred data, etc.) Ad spots in the document may be de?ned by
to as “ad related information”), as well as an extension of such embedded information or instructions. In the context of the
information (e.g., information derived from ad related infor Internet, a common document is a Web page. Web pages often
mation). include content and may include embedded information
The ratio of the number of selections (e. g., clickthroughs) (such as meta information, hyperlinks, etc.) and/ or embedded
of an ad to the number of impressions of the ad (i.e., the instructions (such as JavaScript, etc.). In many cases, a docu
number of times an ad is rendered) is de?ned as the “selection ment has an addressable storage location and can therefore be
rate” (or “clickthrough rate”) of the ad. uniquely identi?ed by this addressable location. A universal
A “conversion” is said to occur when a user consummates resource locator (URL) is an address used to access informa
a transaction related to a previously served ad. What consti tion on the Internet.
tutes a conversion may vary from case to case and can be A “Web document” includes any document published on
determined in a variety of ways. For example, it may be the the Web. Examples of Web documents include, for example,
case that a conversion occurs when a user clicks on an ad, is a Website or a Web page.
referred to the advertiser’s Web page, and consummates a “Document information” may include any information
purchase there before leaving that Web page. Alternatively, a included in the document, information derivable from infor
conversion may be de?ned as a user being shown an ad, and mation included in the document (referred to as “document
making a purchase on the advertiser’s Web page within a derived information”), and/or information related to the
predetermined time (e.g., seven days). In yet another alterna document (referred to as “document related information”), as
tive, a conversion may be de?ned by an advertiser to be any well as an extensions of such information (e.g., information
measurable/ observable user action such as, for example, 65 derived from related information). An example of document
downloading a white paper, navigating to at least a given derived information is a classi?cation based on textual con
depth of a Website, viewing at least a certain number of Web tent of a document. Examples of document related informa
US 8,706,546 B2
7 8
tion include document information from other documents automotive line, and a separate advertising campaign for its
with links to the instant document, as well as document infor motorcycle line. The campaign for its automotive line may
mation from other documents to which the instant document have one or more ad groups, each containing one or more ads.
links. Each ad group may include targeting information (e.g., a set
Content from a document may be rendered on a “content of keywords, a set of one or more topics, etc.), and price
rendering application or device”. Examples of content ren information (e.g., cost, average cost, or maximum cost (per
dering applications include an Internet browser (e.g., impression, per selection, per conversion, etc.)). Therefore, a
Explorer, Netscape, Opera, Firefox, etc.), a media player single cost, a single maximum cost, and/or a single average
(e.g., an MP3 player, a Realnetworks streaming audio ?le cost may be associated with one or more keywords, and/or
player, etc .), a viewer (e. g., anAbobe Acrobat pdf reader), etc. topics. As stated, each ad group may have one or more ads or
A “content owner” is a person or entity that has some “creatives” (That is, ad content that is ultimately rendered to
property right in the content of a media property (e. g., docu an end user.). Each ad may also include a link to a URL (e.g.,
ment). A content owner may be an author of the content. In a landing Web page, such as the home page of an advertiser,
addition, or alternatively, a content owner may have rights to or a Web page associated with a particular product or server).
reproduce the content, rights to prepare derivative works of Naturally, the ad information may include more or less infor
the content, rights to display or perform the content publicly, mation, and may be organized in a number of different ways.
and/or other proscribed rights in the content. Although a FIG. 2 illustrates an environment 200 in which the present
content server might be a content owner in the content of the invention may be used. A user device (also referred to as a
documents it serves, this is not necessary. A “Web publisher” “client” or “client device”) 250 may include a browser facility
is an example of a content owner. 20 (such as the Explorer browser from Microsoft, the Opera Web
“User information” may include user behavior information Browser from Opera Software of Norway, the Navigator
and/ or user pro?le information. browser from AOL/ Time Warner, the Firefox browser from
“E-mail information” may include any information Mozilla, etc.), an e-mail facility (e.g., Outlook from
included in an e-mail (also referred to as “internal e-mail Microsoft), etc. A search engine 220 may permit user devices
information”), information derivable from information 25 250 to search collections of documents (e.g., Web pages). A
included in the e-mail and/ or information related to the content server 230 may permit user devices 250 to access
e-mail, as well as extensions of such information (e.g., infor documents. An e-mail server (such as GMail from Google,
mation derived from related information). An example of Hotmail from Microsoft Network, Yahoo Mail, etc.) 240 may
information derived from e-mail information is information be used to provide e-mail functionality to user devices 250.
extracted or otherwise derived from search results returned in 30 An ad server 210 may be used to serve ads to user devices 250.
response to a search query composed of terms extracted from The ads may be served in association with search results
an e-mail subject line. Examples of information related to provided by the search engine 220. However, content-rel
e-mail information include e-mail information about one or evant ads may be served in association with content provided
more other e-mails sent by the same sender of a given e-mail, by the content server 230, and/or e-mail supported by the
or user information about an e-mail recipient. Information 35 e-mail server 240 and/or user device e-mail facilities.
derived from or related to e-mail information may be referred As discussed in the ’ 900 application, ads may be targeted to
to as “external e-mail information.” documents served by content servers. Thus, one example of
§4.2 Exemplary Advertising Environments in which, or an ad consumer 130 is a general content server 230 that
with which, the Present Invention May Operate receives requests for documents (e.g., articles, discussion
FIG. 1 is a diagram of an advertising environment. The 40 threads, music, video, graphics, search results, Web page
environment may include an ad entry, maintenance and deliv listings, etc.), and retrieves the requested document in
ery system (simply referred to as an ad server) 120. Adver response to, or otherwise services, the request. The content
tisers 110 may directly, or indirectly, enter, maintain, and server may submit a request for ads to the ad server 120/210.
track ad information in the system 120. The ads may be in the Such an ad request may include a number of ads desired. The
form of graphical ads such as so-called banner ads, text only 45 ad request may also include document request information.
ads, image ads, audio ads, video ads, ads combining one of This information may include the document itself (e. g.,
more of any of such components, etc. The ads may also page), a category or topic corresponding to the content of the
include embedded information, such as a link, and/or document or the document request (e.g., arts, business, com
machine executable instructions.Ad consumers 130 may sub puters, arts-movies, arts-music, etc.), part or all of the docu
mit requests for ads to, accept ads responsive to their request 50 ment request, content age, content type (e.g., text, graphics,
from, and provide usage information to, the system 120. An video, audio, mixed media, etc.), geo-location information,
entity other than an ad consumer 130 may initiate a request for document information, etc.
ads. Although not shown, other entities may provide usage The content server 230 may combine the requested docu
information (e.g., whether or not a conversion or selection ment with one or more of the advertisements provided by the
related to the ad occurred) to the system 120. This usage 55 ad server 120/ 210. This combined information including the
information may include measured or ob served user behavior document content and advertisement(s) is then forwarded
related to ads that have been served. towards the end user device 250 that requested the document,
The ad server 120 may be similar to the one described in the for presentation to the user. Finally, the content server 230
’ 900 application. An advertising program may include infor may transmit information about the ads and how, when, and/
mation concerning accounts, campaigns, creatives, targeting, 60 or where the ads are to be rendered (e. g., position, selection or
etc. The term “account” relates to information for a given not, impression time, impression date, size, conversion or not,
advertiser (e.g., a unique e-mail address, a password, billing etc.) back to the ad server 120/210. Alternatively, or in addi
information, etc.). A “campaign” or “ad campaign” refers to tion, such information may be provided back to the ad server
one or more groups of one or more advertisements, and may 120/210 by some other means.
include a start date, an end date, budget information, geo 65 The of?ine content provider 232 may provide information
targeting information, syndication information, etc. For about ad spots in an upcoming publication, and perhaps infor
example, Honda may have one advertising campaign for its mation about the publication (e.g., the content or topics or
US 8,706,546 B2
10
concepts of the content), to the ad server 210. In response, the look for example) may be used to send and/or receive e-mail.
ad server 210 may provide a set of ads relevant the content of Therefore, an e-mail server 240 or application may be thought
the publication for at least some of the ad spots. Examples of of as an ad consumer 130. Thus, e-mails may be thought of as
o?iine content providers 232 include, for example, magazine documents, and targeted ads may be served in association
publishers, newspaper publishers, book publishers, of?ine with such documents. For example, one or more ads may be
music publishers, of?ine video game publishers, a theatrical served in, under, over, or otherwise in association with an
production, a concert, a sports event, etc. e-mail.
Owners of the of?ine ad spot properties 234 may provide Although the foregoing examples described servers as (i)
information about ad spots in their o?iine property (e.g., a requesting ads, and (ii) combining them with content, one or
stadium scoreboard banner ad for an NBA game in SanAnto both of these operations may be performed by a client device
nio, Tex.). In response, the ad sever may provide a set of ads (such as an end user computer for example).
relevant to the property for at least some of the ad spots. §4.3 Exemplary Embodiments
Examples of o?iine properties 234 include, for example, a §4.3.l Exemplary Methods
billboard, a stadium score board, and out?eld wall, the side of FIG. 3 is a ?ow diagram of an exemplary method 300 for
truck trailer, etc. selecting and scoring ads in a manner consistent with the
Another example of an ad consumer 130 is the search present invention. A request including document information
engine 220. A search engine 220 may receive queries for is accepted. (Block 310) Then, ad search criteria may be
search results. In response, the search engine may retrieve determined using the accepted document information. (Block
relevant search results (e.g., from an index of Web pages). An 320) The determined criteria may be ranked. (Block 330) For
exemplary search engine is described in the article S. Brin and 20 at least some of the criteria (e.g., at least two of the criteria),
L. Page, “The Anatomy of a Large-Scale Hypertextual Search a weight may be assigned using the rankings. (Block 340)
Engine,” Seventh International World Wide Web Conference, Eligible ads are selected (e.g., identi?ed) using the criteria.
Brisbane, Australia and in US. Pat. No. 6,285,999 (both (Block 350) The number of eligible ads may be decreased
incorporated herein by reference). Such search results may (e.g., using ?ltering). (Block 360). The identi?ed eligible ads
include, for example, lists of Web page titles, snippets of text 25 may then be scored using their associated criteria weights
extracted from those Web pages, and hypertext links to those (Block 370) before the method 300 is left (Node 380).
Web pages, and may be grouped into a predetermined number Referring back to block 310, the request may specify the
of (e.g., ten) search results. number of ads desired. The request may specify a type of ad
The search engine 220 may submit a request for ads to the (e.g., text only, audio, video, image, etc.) desired. The docu
ad server 120/210. The request may include a number of ads 30 ment information may be a document identi?er. For example,
desired. This number may depend on the search results, the in the context of the Web, the document information may be
amount of screen or page space occupied by the search a URL. Alternatively, or in addition, the document informa
results, the size and shape of the ads, etc. In one embodiment, tion may include content (extracted) from the document.
the number of desired ads will be from one to ten, and pref Alternatively, or in addition, the document information may
erably from three to ?ve. The request for ads may also include 35 include topics or concepts (e.g., PHIL clusters) of the docu
the query (as entered or parsed), information based on the ment content.
query (such as geolocation information, whether the query Referring back to blocks 310-330, these acts may be per
came from an a?iliate and an identi?er of such an a?iliate), formed as described in §l.2 above with reference to the ’900
and/or information associated with, or based on, the search application. Other means of determining ranked criteria may
results. Such information may include, for example, identi? 40 be used. Indeed, ranked criteria may simply be accepted.
ers related to the search results (e. g., document identi?ers or Referring back to block 340, the criteria weights for each
“docIDs”), scores related to the search results (e.g., informa criteria may be based on how relevant that criteria is to the
tion retrieval (“IR”) scores such as dot products of feature document. For example, a criteria weight for criteria i (cwi)
vectors corresponding to a query and a document, Page Rank may be determined using the formula:
scores, and/ or combinations of IR scores and Page Rank 45 cwiIMACi [2]
scores), snippets of text extracted from identi?ed documents
(e. g., Web pages), full text of identi?ed documents, topics of where MAC (Modi?ed Arbitration Constant) is between 1.0
identi?ed documents, feature vectors of identi?ed docu and 0.0. The MAC can be selected as desired. For example,
tuning the MAC may be used to tune the relative importance
ments, etc.
The search engine 220 may combine the search results with 50 of click through rate (“CTR”) and cost per click (“CPC”)
one or more of the advertisements provided by the ad server when scoring the ads (e.g., for use in an arbitration or auc
120/ 210. This combined information including the search tion). When the MAC is increased, the arbitration becomes
results and advertisement(s) is then forwarded towards the more competitiveithe importance of a CPC offer and CTR
user that submitted the search, for presentation to the user. increases, but better criteria lose some of their advantage.
Conversely, when the MAC is decreased, the arbitration
Preferably, the search results are maintained as distinct from 55
becomes less competitiveithe importance of a CPC offer
the ads, so as not to confuse the user between paid advertise
ments and presumably neutral search results. and CTR decreases, and more relevant criteria gain an advan
Finally, the search engine 220 may transmit information tage.
about the ad and when, where, and/or how the ad was to be Referring back to block 350, a request used for identifying
rendered (e.g., position, selection or not, impression time, 60 ads may include N (e.g., N:29) criteria. The request may take
impression date, size, conversion or not, etc.) back to the ad the form of a query such as:
server 120/210. Alternatively, or in addition, such informa
rawiqueryI‘britOiORicritliORicritZ . . .
tion may be provided back to the ad server 120/210 by some iORicritN” [3]
other means.
Finally, the e-mail server 240 may be thought of, generally, 65 criteriaiweights:{cwo,cw1,cw2, . . [4]
as a content server in which a document served is simply an Using this request, ads for each of the critl- may be determined.
e-mail. Further, e-mail applications (such as Microsoft Out In at least some embodiments consistent with the present