Hector GarciaMolina Stanford University Ali Dasdan Santanu Kolay Ebay Inc Search Engine Results Page SERP Organic Results Sponsored Ads Query Sponsored Search Ads 2 ID: 737220
Download Presentation The PPT/PDF document "Output URL Bidding Panagiotis Papadimitr..." is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
Slide1
Output URL Bidding
Panagiotis Papadimitriou, Hector Garcia-Molina, (Stanford University)Ali Dasdan, Santanu Kolay(Ebay Inc)Slide2
Search Engine Results Page (SERP)
Organic Results
Sponsored Ads
Query
Sponsored Search Ads
2Slide3
Keyword bidding
Advertiser
Search Engines
the social network
lord of the rings
the matrix
l
otr
III
...
...
# keywords = ~ 10K
KEYWORDS
3Slide4
Example SERPs
en.wikipedia.org
/wiki/
The_Social_Network
www.imdb.com
/title/tt1285016/
www.imdb.com
/title/tt133093/
en.wikipedia.org
/wiki/
The_Matrix
en.wikipedia.org
/wiki/
The_Lord_of_the_rings
en.wikipedia.org
/wiki/
The_Lord_of_the_rings
www.imdb.com
/title/tt167260/
www.imdb.com
/title/tt120737/
the social network
the matrix
the lord of the rings
lotr
iii
4Slide5
Output URL bidding
Advertiser
Search Engines
imdb.com AND wikipedia.org
# URLs = 2
URLs
5Slide6
Topics
ImplementationEvaluation6Slide7
Implementation challenge
7
SERPSlide8
Alternative implementation solutions
1. Serialization8O: Organic Search ComponentS: Sponsored Search Component
Latency
Simplicity
SERP
2
. Parallelization
SERP
O:
Organic Search Comp. (
Or
+
Op
)
Or’:
Small replica of
Or
S:
Sponsored Search Component
V:
Ad validation
More resources
No latencySlide9
Topics
ImplementationEvaluation9Slide10
Bid language model
Output Expressione.g., a := (u1 u2) u3 (h1 h2)u: URLe.g., en.wikipedia.org/wiki/
The_Social_Networkh: hoste.g., en.wikipedia.org
10Slide11
How to evaluate/study output bidding?
Use existing keyword campaigns to generate realistic output expressions to study11
The social network
lord of the rings
the matrix
l
otr
III
…
…
Output Expression Generator
imdb.com AND wikipedia.org
Slide12
INPUT:
set of keywords R (from a keyword campaign)OUTPUT: expression a that “covers” R, i.e., q R, matches(a, results of
q)
Generator input & output
Candidate expressions
Output
Expression
a
1
:= u
1
u
2
u
3
a
2
:= u
1
u4
a
3
:= u
5
12Slide13
Compactness
Contain few URLsSpill minimizationDo not match “irrelevant” queries
Which expression to select?
Candidate expressions
Output
Expression
Size
|
a
|
spill(
a,
R)a
1 := u1 u2
u
3
3
{}
a
2
:= u
1
u4
2{q
5}
a
3
:= u
5
1
{q
4
,q
5
, q
6
}
13Slide14
Output expression generation
problem statementQuery Set Output Covermin. γ|a| + (1-γ)|spill(a,
R)|subj. to matches(
a
,
q
),
q
Rγ : regularization parameterNP-hard to solve
Reduction from Set Cover, Red-Blue Set Cover (see paper)Developed Greedy Algorithm (see paper)
14Slide15
Is spill always “bad”?
Example:q1: lord of the ringsq2: the matrixq3: the social networkq4: …
q5: lotr
q
6
: …
15
Output
Expression
Size
|
a
|
spill(a,R)
a
1
:= u
1
u
2
u
33{}
a
2 := u1
u
4
2
{q
5
}
a
3
:= u
5
1
{q
4
,q
5
, q6}
→(
l
ord of the
rings)Slide16
Spill may be good!
Cluster queries Q using the bipartite graph (see paper)Divide spill(a, R) into: positive: relevant negative: irrelevant
16
Output
Expression
Size
|
a
|
spill(
a,R)
+
-
a
1
:= u
1
u
2
u
3
3
{}
{}
a
2
:= u
1
u
4
2
{q
5
}
{}
a
3
:= u
51
{q5}
{q4, q6}Slide17
Experimental evaluation goals
Recall output expression looks likea := (u1 u2) u3
(h1
h2
)
Evaluation questions
URLs, hosts or mixed?
Specific or generic? (# conjuncts)
Long or compact? (# disjuncts)Comparison criteria
Compactness vs spill tradeoffPositive vs negative spill
17Slide18
Experimental setup
Dataset (from Yahoo query logs)2,251 ads13M queries, 63M URLs (7M hosts)18
The social network
lord of the rings
the matrix
l
otr
III
…
…
Output Expression Generator
imdb.com AND wikipedia.org
γ
- URLs
- hosts
- mixed
# conjunctsSlide19
Compactness vs
spill tradeoffURLs, hosts or mixed?Mixed expression curves dominate othersSpecific
vs generic?2 conjuncts suffice
Long or compact?
Next slide…
19Slide20
Positive
vs negative spill20For |a|>60, more than 50% of spill is positiveURLs, hosts or mixed?Mixed expression curves dominate others
Long or compact?|
a
|=70-80 suffice
(1/3 the size of equivalent keyword set)Slide21
More experiments in paper
Combining keyword and output biddingE.g., a movie advertiser uses:1. imdb.com wikipedia.org cover 80% of queries2. keyword1, keyword2, … cover 20% of queriesCombined expressions
are as compact as output expressionsyield less (negative) spill than output expressions
21Slide22
Conclusions
Output URL bidding can be implemented efficientlyAdvantages over keyword biddingBid compactnessMore relevant queries (positive spill)Combining keyword and output bidding seems to be the most promising directionSlide23
Thank you!
Contact: papadimitriou@stanford.edu http://panagiotis.info