TW
ITTER
What is Twitter,a Social Network or a News
Media?
Haewoon Kwak Changhyun Lee Hosung Park Sue Moon
Department of Computer Science, KAIST, Korea
19th International World Wide Web Conference (WWW2010)
TW
ITTER
2
Twitter, a microblog service
write a short message
TW
ITTER
3
Twitter, a microblog service
read neighbors’ tweets
TW
ITTER
4
In most OSN
“We are friends.”
TW
ITTER
5
In Twitter
“I follow you.”
TW
ITTER
6
Following on Twitter
“Unlike most social networks, following on
Twitter is not mutual. Someone who thinks you're interesting can follow you, and you don't have to approve, or follow back."
http://help.twitter.com/entries/14019-what-is-following6
TW
ITTER
7
Following = subscribing tweets
recent tweets of followings
8
TW
ITTER
9
http://blog.marsdencartoons.com/2009/06/18/cartoon-iranian-election-demonstrations-and-twitter/marsden-iran-twitter72/
10
The goal of this work
•We analyze how directed relations of following set Twitter apart from existing OSNs.
•Then, we see if Twitter has any characteristics of news media.
PR
OB
LEM
STA
TEM
EN
T
TW
ITTER
11
me⋅di⋅a [mee-dee-uh]
1.a pl. of medium
2.the means of communication, as radio and television, newspapers, and magazines, that reach or influence people widely
http://dictionary.reference.com/
12
The goal of this work
•We analyze how directed relations of following set Twitter apart from existing OSNs.
•Then, we see if Twitter has any characteristics of news media.
PR
OB
LEM
STA
TEM
EN
T
13
Summary of our findings
1.Following is mostly not reciprocated (not so “social”)
2.Users talk about timely topics
3.A few users reach large audience directly
4.Most users can reach large audience by Word of Mouth (WOM) quickly
OU
TLI
NE
TW
ITTER
14
Data collection (2009/06/01~09/24)
•41.7M user profiles (near-complete at that time)
•1.47B following relations
•4262 trending topics
•106M tweets mentioning trending topics
‣ Spam tweets removed by CleanTweets
*http://an.kaist.ac.kr/traces/WWW2010.html
TW
ITTER
15
How we crawled
•Twitter’s well-defined 3rd party API
•With 20+ ‘whitelisted’ IPs
‣ Send total 20,000 requests per IP / hour
TW
ITTER
16
Recent studies
•Ranking methodologies [WSDM’10]
•Predicting movie profits [HYPERTEXT’10]
•Recommending users [CHI’10 microblogging]
•Detecting real time events [WWW’10]
•The ‘entire’ Twittersphere unexplored
TR
AN
SIT
ION
17
Part I.
1.Following is mostly not reciprocated (not so “social”)
2.Users talk about timely topics
3.A few users reach large audience directly
4.Most users can reach large audience by WOM* quickly
2.
AC
TIV
E S
UB
SC
RIP
TIO
N
18
Why do people follow others?
•Reflection of offline social relationships
•Subscription to others’ messages
otherwise,
2.
AC
TIV
E S
UB
SC
RIP
TIO
N
19
Sociologists’ answer
•“Reciprocal interactions pervade every relation of primitive life and in all social systems”
2.
AC
TIV
E S
UB
SC
RIP
TIO
N
20
Is following reciprocal?
•Only 22.1% of user pairs follow each other
•Much lower than
‣ 68% on Flickr
‣ 84% on Yahoo! 360
‣ 77% on Cyworld guestbook messages
2.
AC
TIV
E S
UB
SC
RIP
TIO
N
21
Low reciprocity of following
•Following is not similarly used as friend in OSNs
‣ Not really reflection of offline social relationships
•Active subscription of tweets!
TR
AN
SIT
ION
22
Part II.
1.Following is mostly not reciprocated (not so “social”)
2.Users talk about timely topics
3.A few users reach large audience directly
4.Most users can reach large audience by WOM* quickly
1.
TIM
ELI
NES
S T
OPIC
S
23
Dynamically changing trends
1.
TIM
ELI
NES
S T
OPIC
S
24
User participation pattern can be a
signature of a topic
1.
TIM
ELI
NES
S T
OPIC
S
25
Majority of topics are headline
54.3%“headline news”
31.5%“ephemeral”
6.9%7.3%
“persistent news”
TR
AN
SIT
ION
26
Part III.
1.Following is mostly not reciprocated (not so “social”)
2.Users talk about timely topics
3.A few users reach large audience directly
4.Most users can reach large audience by WOM* quickly
3.
A F
EW
HU
BS
27
How many followers a user has?
3.
A F
EW
HU
BS
28
P(x)dx
CCDF
•Complementary Cumulative Density Function
•CCDF(x=k) =
3.
A F
EW
HU
BS
29
Reading the graph
3.
A F
EW
HU
BS
30
Plenty of super-hubs
3.
A F
EW
HU
BS
31
More super-hubs than projected by power-law
•Where do they get all the followers? Possibly from...
‣ Search by ‘name’
‣ Recommendation by Twitter
•They reach millions in one hop
3.
A F
EW
HU
BS
32
Are those who have many followers active?
3.
A F
EW
HU
BS
33
How we plotted
3.
A F
EW
HU
BS
34
How we plotted
Avg. = 8×Med. = 9
3.
A F
EW
HU
BS
35
More followers, more tweets
3.
A F
EW
HU
BS
36
Many followers without activity
3.
A F
EW
HU
BS
37
Twitter user rankings by Followers, PageRank and
RT
3.
A F
EW
HU
BS
38
Twitter user rankings by Followers, PageRank and
RT
3.
A F
EW
HU
BS
39
Great discrepancy among rankings
Kan
dal’s
tau
ra
nk
corr
ela
tion
Top k ranking
# followers (RF), PageRank (RPR) # retweets (RRT )
TR
AN
SIT
ION
40
Part IV.
1.Following is mostly not reciprocated (not so “social”)
2.Users talk about timely topics
3.A few users reach large audience directly
4.Most users can reach large audience by WOM* quickly
5.*WOM: word-of-mouth
4.
WO
RD
-OF-
MO
UTH
41
Which is more efficient for WOM?
4.
WO
RD
-OF-
MO
UTH
42
In Twitter
Following
Information
4.
WO
RD
-OF-
MO
UTH
43
Average path length: 4.1
4.
WO
RD
-OF-
MO
UTH
44
Retweet (RT)
•Relay tweets from a following to followers
4.
WO
RD
-OF-
MO
UTH
45
Retweet (RT)
•Relay tweets from a following to followers
Last day of WWW’10
node 0
4.
WO
RD
-OF-
MO
UTH
46
Retweet (RT)
•Relay tweets from a following to followers
Last day of WWW’10
Last day of WWW’10
Last day of WWW’10
Last day of WWW’10
4.
WO
RD
-OF-
MO
UTH
47
Retweet (RT)
•Relay tweets from a following to followers
RT @node0 Last day of WWW’10node 4
4.
WO
RD
-OF-
MO
UTH
48
Retweet (RT)
•Relay tweets from a following to followersRT @node0 Last day of WWW’10
RT @node0 Last day of WWW’10
RT @node0 Last day of WWW’10
4.
WO
RD
-OF-
MO
UTH
49
Retweet (RT)
•Relay tweets from a following to followers
writer
retweeter
wr
4.
WO
RD
-OF-
MO
UTH
50
Retweet (RT)
•Not only 1 hop neighbors
1 hop neighbors
wr
4.
WO
RD
-OF-
MO
UTH
51
Retweet (RT)
•More goes further
2 hop neighbors
51
wr
4.
WO
RD
-OF-
MO
UTH
52
We construct RT tree
•A tree with writer and retweeter(s)
wr
4.
WO
RD
-OF-
MO
UTH
53
Height of RT trees
r
W
r
W
r r
W
rr
r
1 1 2
4.
WO
RD
-OF-
MO
UTH
54
Empirical RT trees
4.
WO
RD
-OF-
MO
UTH
55
96% of RT trees = Height 1
4.
WO
RD
-OF-
MO
UTH
56
Boosting audience by RT
4.
WO
RD
-OF-
MO
UTH
57
Additional readers
3 followers
2 additional readers by retweeter
wr
4.
WO
RD
-OF-
MO
UTH
58
A retweet brings a few hundred additional
readers
4.
WO
RD
-OF-
MO
UTH
59
Time lag between hops in RT tree
4.
WO
RD
-OF-
MO
UTH
60
Fast relaying tweets by RT:
35% of RT < 10 min.
4.
WO
RD
-OF-
MO
UTH
61
Fast relaying tweets by RT:
55% of RT < 1hr.
SU
MM
ARY
62
Summary
1.We study the entire Twittersphere
2.Low reciprocity distinguishes Twitter from OSNs
3.Twitter has characteristics of news media:
‣ Tweets mentioning timely topics
‣ Plenty of hubs reaching a large public directly
‣ Fast and wide spread of word-of-mouth