Monetizing User Activity on Social Networks - Challenges and Experiences, 2009 IEEE/WIC/ACM International Conference on Web Intelligence, Sep 15-18 2009

1. Meena
Nagarajan,
Amit
P.
Sheth
KNO.E.SIS
Center
Wright
State
University
M.
Nagarajan,
K.
Baid,
A.
P.
Sheth,
and
S.
Wang,
"Monetizing
User
Activity
on
Social
Networks
-­‐
Challenges
and
Experiences“,
2009
IEEE/WIC/ACM
International
Conference
on
Web
Intelligence,
Milan,
Italy

2.  On
social
networks
 Use
case
for
this
talk
 Targeted
content
=
content-­‐based
advertisements
 Target
=
user
profiles
 Content-­‐based
advertisements
CBAs
 Well-­‐known
monetization
model
for
online
content

3. May
30,June
02
2009

4. June
01,
2009

5.  Interests
do
not
translate
to
purchase
intents
 Interests
are
often
outdated..
 Intents
are
rarely
stated
on
a
profile..
 Cases
that
work
 New
store
openings,
sales
 Highly
demographic-­‐targeted
ads

6. June
01,
2009

7. CONTENT-­‐BASED
ADS
ON
THEIR
PROFILES
June
01,
2009

8.  Non-­‐trivial
 Non-­‐policed
content
▪ Brand
image,
Unfavorable
sentiments1
 People
are
there
to
network
▪ User
attention
to
ads
is
not
guaranteed
 Informal,
casual
nature
of
content
▪ People
are
sharing
experiences
and
events
▪ Main
message
overloaded
with
off
topic
content
I
NEED
HELP
WITH
SONY
VEGAS
PRO
8!!
Ugh
and
i
have
a
video
project
due
tomorrow
for
merrill
lynch
:((
all
i
need
to
do
is
simple:
Extract
several
scenes
from
a
clip,
insert
captions,
transitions
and
thats
it.
really.
omgg
i
cant
figure
out
anything!!
help!!
and
i
got
food
poisoning
from
eggs.
its
not
fun.
Pleasssse,
help?
:(
1Learning
from
Multi-­‐topic
Web
Documents
for
Contextual
Advertisement,
Zhang,
Y.,
Surendran,
A.
C.,
Platt,
J.
C.,
and
Narasimhan,
M.
,
KDD
2008

9.  Cultural
Entities
HOW
 Word
Usages
in
self-­‐
presentation
WHY
 Slang
sentiments
WHAT
 Intentions

10.  Identifying
intents
behind
user
posts
on
social
networks
 Content
with
monetization
potential
 Identifying
keywords
for
advertizing
in
user-­‐
generated
content
 Interpersonal
communication
&
off-­‐topic
chatter

11.  User
studies
 Hard
to
compare
activity
based
ads
to
s.o.t.a
 Impressions
to
Clickthroughs
 How
well
are
we
able
to
identify
monetizable
posts
 How
targeted
are
ads
generated
using
our
keywords
vs.
entire
user
generated
content

12. Identification,
Evaluation

13.  Scribe
Intent
not
same
as
Web
Search
Intent1
 People
write
sentences,
not
keywords
or
phrases
 Presence
of
a
keyword
does
not
imply
navigational
/
transactional
intents
 ‘am
thinking
of
getting
X’
(transactional)
 ‘i
like
my
new
X’
(information
sharing)
 ‘what
do
you
think
about
X’
(information
seeking)
1B.
J.
Jansen,
D.
L.
Booth,
and
A.
Spink,
“Determining
the
informational,
navigational,
and
transactional
intent
of
web
queries,”
Inf.
Process.
Manage.,
vol.
44,
no.
3,
2008.

14.  Action
patterns
surrounding
an
entity
 How
questions
are
asked
and
not
topic
words
that
indicate
what
the
question
is
about
 “where
can
I
find
a
chotto
psp
cam”
 User
post
also
has
an
entity

15. Set
of
user
posts
from
SNSs
Not
annotated
for
presence
or
absence
of
any
intent

16. Generate
a
universal
set
of
n-­‐gram
patterns;
freq
>
f
S
=
set
of
all
4-­‐grams;
freq
>
3

17. Generate
set
of
candidate
patterns
from
seed
words
(why,when,where,how,what)
Sc
=
all
4-­‐grams
in
S
that
extract
seed
words

18. User
picks
10
seed
patterns
from
Sc
Sis
=
‘does
anyone
know
how’,
‘where
do
i
find’,
‘someone
tell
me
where’….

19. Sc
=
all
4-­‐grams
in
S
that
Sis
=
‘does
anyone
know
how’,
‘where
extract
seed
words
do
i
find’,
‘someone
tell
me
where’….
Gradually
expand
Sis
by
adding
Information
Seeking
patterns
from
Sc

20. Sis
=
‘does
anyone
know
how’,
‘where
do
i
find’,
‘someone
tell
me
where’….
For
every
pis
in
Sis
generate
set
of
filler
patterns

21. ‘.*
anyone
know
how’
‘does
anyone
.*
how’
‘does
.*
know
how’
‘does
anyone
know
.*’
‘does
anyone
know
how’
Look
for
patterns
in
Sc
-­‐ Functional
compatibility
of
filler
-­‐ words
used
in
similar
semantic
contexts
-­‐
Empirical
support
for
filler

22.  Functional
properties
/
communicative
functions
of
words
 From
a
subset
of
LIWC1
 cognitive
mechanical
(e.g.,
if,
whether,
wondering,
find)
▪ ‘I
am
thinking
about
getting
X’
 adverbs
(e.g.,
how,
somehow,
where)
 impersonal
pronouns
(e.g.,
someone,
anybody,
whichever)
▪ ‘Someone
tell
me
where
can
I
find
X’
1Linguistic
Inquiry
Word
Count,LIWC,
http://liwc.net

23.  Sc
=
{‘does
anyone
know
how’,
‘where
do
I
find’,
‘someone
tell
me
where’}
 pis
=
`does
anyone
know
how’
 ‘does
*
know
how’
 ‘does
someone
know
how’
▪ Functional
Compatibility
-­‐
Impersonal
pronouns
▪ Empirical
Support
–
1/3
 ‘does
somebody
know
how’
▪ Functional
Compatibility
-­‐
Impersonal
pronouns
▪ Empirical
Support
–
0
▪ Pattern
Retained
 ‘does
john
know
how’
▪ Pattern
discarded

24.  Over
iterations,
single-­‐word
substitutions,
functional
usage
and
empirical
support
conservatively
expands
Sis
 Infusing
new
patterns
and
seed
words
 Stopping
conditions

25.  does anyone know how  no idea how to
 anyone know how to  someone tell me how
 i dont know what  have no clue what
 know where i can  does anyone know if
 tell me how to  i dont know if
 i dont know how  know if i can
 anyone know where i  anyone know if i
 does anyone know where  im not sure if
 does anyone know what  i was wondering if
 anybody know how to  idea what you are
 anyone know how i  let me know how
 im not sure what  and i dont know
 does anybody know how  now i dont know
 does anyone know why  but i dont really
 i was wondering how  was wondering if someone
 does anyone know when  would like to see
 tell me what to  see what i can
 im not sure how  anyone have any idea
 i was wondering what  wondering if someone could
 was wondering how i
 i do not want

26.  Information
Seeking
patterns
generated
offline
 Information
seeking
intent
score
of
a
post
 Extract
and
compare
patterns
in
posts
with
extracted
patterns
 Transactional
intent
score
of
a
post
▪ LIWC
‘Money’
dictionary
▪ 173
words
and
word
forms
indicative
of
transactions,
e.g.,
trade,
deal,
buy,
sell,
worth,
price
etc.

27.  Training
corpus
 8000
user
posts
▪ MySpace
Computers,
Electronics,
Gadgets
forum
 309
unique
new
patterns,
263
unambiguous
 Testing
patterns
for
recall
 ‘To
buy’
Marketplace
–
average
81
%

28. Off-­‐topic
noise
elimination

29.  Identifying
keywords
in
monetizable
posts
 Plethora
of
work
in
this
space
 Off-­‐topic
noise
removal
is
our
focus
I
NEED
HELP
WITH
SONY
VEGAS
PRO
8!!
Ugh
and
i
have
a
video
project
due
tomorrow
for
merrill
lynch
:((
all
i
need
to
do
is
simple:
Extract
several
scenes
from
a
clip,
insert
captions,
transitions
and
thats
it.
really.
omgg
i
cant
figure
out
anything!!
help!!
and
i
got
food
poisoning
from
eggs.
its
not
fun.
Pleasssse,
help?
:(

30.  Topical
hints
 C1
-­‐
['camcorder']
 Keywords
in
post
 C2
-­‐
['electronics
forum',
'hd',
'camcorder',
'somethin',
'ive',
'canon',
'little
camera',
'canon
hv20',
'cameras',
'offtopic']
 Move
strongly
related
keywords
from
C2
to
C1
one-­‐by-­‐one
 Relatedness
determined
using
information
gain
 Using
the
Web
as
a
corpus,
domain
independent

31.  C1
-­‐
['camcorder']
 C2
-­‐
['electronics
forum',
'hd',
'camcorder',
'somethin',
'ive',
'canon',
'little
camera',
'canon
hv20',
'cameras',
'offtopic']
 Informative
words
 ['camcorder',
'canon
hv20',
'little
camera',
'hd',
'cameras',
'canon']

32. Preliminary
work

33.  Keywords
from
60
monetizable
user
posts
 Monetizable
intent,
at
least
3
keywords
in
content
 45
MySpace
Forums,
15
Facebook
Marketplace,
30
graduate
students
 10
sets
of
6
posts
each
 Each
set
evaluated
by
3
randomly
selected
users
 Monetizable
intents?
 All
60
posts
voted
as
unambiguously
information
seeking
in
intent

34.  Google
AdSense
ads
for
user
post
vs.
extracted
topical
keywords

35.  Choose
relevant
Ad
Impressions
 VW
6
disc
CD
changer
 I
need
one
thats
compatible
with
a
2000
golf
most
are
sold
from
years
1998-­‐2004if
anyone
has
one
[or
can
get
one]
PLEASE
let
me
know!

36.  Users
picked
ads
relevant
to
the
post
 At
least
50%
inter-­‐evaluator
agreement
 For
the
60
posts
 Total
of
144
ad
impressions
 17%
of
ads
picked
as
relevant
 For
the
topical
keywords
 Total
of
162
ad
impressions
 40%
of
ads
picked
as
relevant

37.  User’s
profile
information
 Interests,
hobbies,
tv
shows..
 Non-­‐demographic
information
 Submit
a
post
 Looking
to
buy
and
why
(induced
noise)
 Ads
that
generate
interest,
captured
attention

38.  Using
profile
ads
 Total
of
56
ad
impressions
 7%
of
ads
generated
interest
 Using
authored
posts
 Total
of
56
ad
impressions
 43%
of
ads
generated
interest
 Using
topical
keywords
from
authored
posts
 Total
of
59
ad
impressions
 59%
of
ads
generated
interest

39.  User
studies
small
and
preliminary,
clearly
suggest
 Monetization
potential
in
user
activity
 Improvement
for
Ad
programs
in
terms
of
relevant
impressions
 Evaluations
based
on
forum,
marketplace
 Verbose
content
 Status
updates,
notes,
community
and
event
memberships…
 One
size
may
not
fit
all

40.  A
world
between
relevant
impressions
and
clickthroughs
 Objectionable
content,
vocabulary
impedance,
Ad
placement,
network
behavior
 In
a
pipeline
of
other
community
efforts
 No
profile
information
taken
into
account
 Cannot
custom
send
information
to
Google
AdSense

41.  Keywords
to
Ad
Impressions
 Google
Adsense
like
web
service
 Monetization
potential
of
a
keyword
on
the
Web
not
the
same
on
a
social
n/w?
 Ranking
keywords
in
user
post
 We
are
building
an
F8
app
 Collaboration
for
clickthrough
data

42.  Google/Bing:
Meena
Nagarajan
 meena@knoesis.org
 http://knoesis.wright.edu/students/meena/
 Google/Bing:
Amit
Sheth
 amit@knoesis.org
 http://knoesis.wright.edu/amit