1.
Do
All
Birds
Tweet
the
Same?
Characterizing
Twitter
around
the
World
Barbara
Poblete,
Ruth
García,
Marcelo
Mendoza
and
Alejandro
Jaimes
University
of
Chile,
Yahoo!
Research
Barcelona
Universitat
Pompeu
de
Fabra,
Universidad
Federico
Santa
Maria
October
26,
2011,
Glasgow,
CIKM
2. Objective
Identify
differences
and
similarities
in
the
use
of
social
media
by
analyzing
tweets
and
network
of
friends
in
different
countries.
3. What
we
did
Ê Data:
analysis
of
one
year
of
Tweets
for
10
most
active
countries
ü Content:
languages,
sentiment,
structure
(retweets,
hashtags,..)
ü Structure:
network
(modularity,
average
path
length,
reciprocity,
connectivity)
4. Crawling
Strategy
12,964,735 active users -> 6,263,457 with valid location
4,736, 629 users belonging to 10 most active countries.
5,270,609,213 tweets during 2010
.
0.00%
10.00%
20.00%
30.00%
40.00%
50.00%
United
States
Brazil
United
Kingdom
Japan
Indonesia
Canada
México
Netherlands
South
Korea
Australia
TWEETS(%)
USERS(%)
5. Engagement
0
200
400
600
800
1000
1200
1400
1600
1800
2000
Indonesia
Japan
Brazil
Netherlands
UK
US
Australia
Mexico
South
Korea
Canada
(Tweets/User)
6. Languages
0
200,000,000
400,000,000
600,000,000
800,000,000
1,000,000,000
1,200,000,000
1,400,000,000
1,600,000,000
1,800,000,000
English Portuguese Japanese Spanish Dutch German Indonesian
and Malay
Korean
Netherlands >10%,
Indonesia >10%,
Mexico >10%,
South Korea >10%
7. How
do
Tweets
differ
in
function?
Macro
average
Country (Tweets/User) URL (%) Hashtag (%) Mention(%) Retweet(%)
Indonesia 1813.53 14.95 7.63 58.24 9.71
Japan 1617.35 16.30 6.81 39.14 5.65
Brazil 1370.27 19.23 13.41 45.57 12.80
Netherlands 1026.44 24.40 18.24 42.33 9.12
UK 930.58 27.11 13.03 45.61 11.65
US 900.79 32.64 14.32 40.03 11.78
Australia 897.41 31.37 14.89 43.27 11.73
Mexico 865.70 17.49 12.38 49.79 12.61
South Korea 853.92 19.67 5.83 58.02 9.02
Canada 806.00 31.09 14.68 42.50 12.50
8. How
do
Tweets
differ
in
function?
Macro
average
Country (Tweets/User) URL (%) Hashtag (%) Mention(%) Retweet(%)
Indonesia 1813.53 14.95 7.63 58.24 9.71
Japan 1617.35 16.30 6.81 39.14 5.65
Brazil 1370.27 19.23 13.41 45.57 12.80
Netherlands 1026.44 24.40 18.24 42.33 9.12
UK 930.58 27.11 13.03 45.61 11.65
US 900.79 32.64 14.32 40.03 11.78
Australia 897.41 31.37 14.89 43.27 11.73
Mexico 865.70 17.49 12.38 49.79 12.61
South Korea 853.92 19.67 5.83 58.02 9.02
Canada 806.00 31.09 14.68 42.50 12.50
9. How
do
Tweets
differ
in
function?
Macro
average
Country (Tweets/User) URL (%) Hashtag (%) Mention(%) Retweet(%)
Indonesia 1813.53 14.95 7.63 58.24 9.71
Japan 1617.35 16.30 6.81 39.14 5.65
Brazil 1370.27 19.23 13.41 45.57 12.80
Netherlands 1026.44 24.40 18.24 42.33 9.12
UK 930.58 27.11 13.03 45.61 11.65
US 900.79 32.64 14.32 40.03 11.78
Australia 897.41 31.37 14.89 43.27 11.73
Mexico 865.70 17.49 12.38 49.79 12.61
South Korea 853.92 19.67 5.83 58.02 9.02
Canada 806.00 31.09 14.68 42.50 12.50
10. Happiness
Level
ENGLISH TWEETS SPANISH TWEETS
5
5.2
5.4
5.6
5.8
6
6.2
6.4
6.6
6.8
7
01
02
03
04
05
06
07
08
09
10
11
12
Australia
Brazil
Canada
Indonesia
Japan
Mexico
Netherlands
South
Korea
UK
USA
5
5.2
5.4
5.6
5.8
6
6.2
6.4
6.6
6.8
7
01
02
03
04
05
06
07
08
09
10
11
12
Dodds et al. “Temporal patters of happiness
and information in a global social network:
Hedonometrics and Twitter”, 2011
11. Network
Coverage with regard to the initial complete graph
Active nodes
Edges within the same location
17. Conclusions
Ê Some
use
Twitter
more
for
conversation
and
others
for
formal
information
dissemination
(links
of
news,
pics,
etc).
Ê Higher
conversational
level
seems
to
be
related
to
more
happy
tweets
(less
formality?).
Ê Twitter
networks
seems
to
be
less
reciprocal
and
more
hierarchical
Ê Smaller
networks
tend
to
have
more
reciprocity
Ê High
reciprocity
seems
to
lead
to
more
activity
per
user
Ê Reciprocity
does
not
mean
more
connectivity:
Indonesia
has
reciprocity
in
small
communities
with
low
connectivity
among
them
As we all know Social Media is massively used for several purposes, in protests, by business , in work and for personas purposes. It has been reported that 145 million users login to Twitter every day , producing more than 90 million tweets per day. Increasng power of social Media : 1 billion users The web is changing. It’s no longer just a place for information seeking and shopping but a platformwhere connections are made, friendships formed and information and opinion exchanged.Social networks have become more embedded in our everyday lives, whether it’s Facebook, Orkut orLinkedIn, we now contact more people in our personal life through our social networks (our researchshows that on average we stay in contact socially with 52 people via these networks) than we dothrough any other means including face to face contact, email and phone.The new social web makes different demands on both consumers and advertisers. Consumers arenot merely finding, they are contributing; writing, uploading pictures, videos, creating regular statusupdates and livestreaming their every day happenings.It’s essential for brands to understand why and where different groups of consumers participate inthis new world. It’s not merely a question of identifying the best places to target – the classic mediaplanning/buying approach – but truly knowing what motivates them to be part of it.
Prior work has shown that analysis of partial crawls of social networks can underestimate measures like degree distribution, but continue to preserve accuracy for other metric , such as density, reciprocity and connectivity. We believe that by preserving the active component of the graph, we are analyzing the most relevant part of the social structure. For the crawling we used the Twiter API (4J), collecting the list of followers-followees for each user.
Order porcolumna
Order porcolumnaSiting external sources ->
Order porcolumna
For some countries, reciprocity seems to be significant specially for Asian countries such as South Korea and Japan and Canada. Generally, the symetric nature of social ties affects network structure, sometimes increasing connectivty, and reducing the diameter. We can observe that communities with high clustering coefficient and less reciprocity may indicate more hierarchical type Relationships between users.
Modularity, coefficient defined by Girvan and Newman, which evaluated how well a graph can be partitioned. A value of 0.4 or greater is generally considered meaningful. In our analysis we can appreciate that Indonesia and Canada display high modularity, which indicated that the communities found in these countriesAre more compact and closed than in other countries. On the contray, Mexico, South Korea and UK indicare less separation between their communities
Indonesia presents the higher diameter, indicating that this network is very partitioned, which agrees with the modularity coefficient.We can also see that average path lengths are proportional to diameter values.( In general, the number of shortest paths is proportional to the number of edges in the graph. So for example, the countries with more edge coverage (USA; Brazil,, UK) have the higuest shortest paths values), We also seea direct relationship with reciprocity, in some countries large diameters also present high reciprocity such as indonesia. This suggests that graph structure strongly influences the relationship of diameter and reciprocity. So given our observationThat Indonesia had a high modularity, the large diameter supports the idea that this country has compact and isolated communities. On the other hand, Canada also shows a very significant modularity value but its diameter andaverage path length value are very Similar to countries that do not have a community structure. The main difference btw indonesia and canada is that the firstHas a lower clustering coeficient and density than the second.
In this illustration it is interesting to observe that all countries concentrate their most isgnificantammount of links towards themselves, with the exception of Canada,Australia and UK, which connect to the US almost as much as to themselves
MoreSelf Connected
More connected
Twitter seems to be used more as a conversation channel in certain countries in contrast to others such as USA where more formal communication seems to be used(links)Reciprocity does not always means more connectivity and small diameters, in fact, Indonesia presents reciprocity between small compacted communities