More Related Content Similar to Who is fake discover astroturfing or attempts of fake influence presentation (20) Who is fake discover astroturfing or attempts of fake influence presentation1. Who is FAKE?
Discover Astroturfing or Attempts of Fake
Influence!
Lutz Finger Soumitra DuttaMiningData.biz
16. Bots 1.0 – Spammer going Social
@you malware.com
@you-as-well malware.com
D fresh-contact malware.com
17. Bots 1.0 – Spammer going Social
10.000 messages
in 4 month
@PeaceKaren_25
Jacob
Ratkiewicz
et.
al
-‐
2011
18. 10.000 messages
in 4 month
Bots 1.0 – Spammer going Social
@PeaceKaren_25
Jacob
Ratkiewicz
et.
al
-‐
2011
20. Look for Un-Normal
Differences:
• Time: Regular or Bursty
• Heavy Hashtag Usage
• Blacklisted URL
• Spam Words
• Few Friends
Grier
et
al.
2010
(Berkeley)
Training:
• @spam
• Manual classification
• Honey Pots
24. Bot 2.0 – Social Bots
Analytics ToolsConversation Bots
30. Can Bots do Astroturfing?
• Reach
• Intention
• Ease of Action
31. Can Bots do Astroturfing?
INTENTION / INFLUENCE
• Opinion leaders (Katz 1955)
• Influentials (Merton 1968)
• Mavens & connectors (Gladwell 2000)
• Reach
• Intention
• Ease of Action
32. Can Bots do Astroturfing?
To create
Intention
Is not easy
• Reach
• Intention
• Ease of Action
33. Can Bots do Astroturfing?
Readiness
Multiple Sources
Topic Dependence
• Reach
• Intention
• Ease of Action
To create
Intention
Is not easy
35. But it is NOT impossible
Arjun
Mukherjee
et.al.
38. Influence the News
0
20.000
40.000
60.000
80.000
100.000
120.000
N
Korean
leader
Kim
Jong-‐il
dies
AnMmaNer
atom
trapped
for
first
Mme,
say
scienMsts
Neutrinos
beat
light
speed
again
Earth-‐like
planet
found
in
the
"habitable
zone"
No
rhinos
remain
in
West
Africa
Eurozone
debt
web:
Who
owes
what
to
whom?
Writer
Christopher
Hitchens
dies
BBC
apology
for
Clarkson
comments
'Witch's
coNage'
found
in
Pendle
In
pictures:
ApocalypMc
Manchester
Social
Media
Index
Clicking
Sharing
CommenMng
50%
on
the
courtesy
of
55%
40. Thus how to spot them?
INDIVIDUAL
• Not ‘loud’
• Might be human
• Missing trainings data
41. Thus how to spot them?
GROUP
• Similarity of group
• Description, Focus,
Tweets…
INDIVIDUAL
• Not ‘loud’
• Might be human
• Missing trainings data