"Subclassing and Composition – A Pythonic Tour of Trade-Offs", Hynek Schlawack
La novela pastoril copy studium
1. Beyond the stars:
Interpreting discourse
cohesion in Twitter as an
indicator of community
Cornelius Puschmann, University of Düsseldorf /
Institute for Internet and Society
DIATA 11 Workshop
15. September 2011, Düsseldorf
2. This talk: an idea, a suggestion, bits of data
#1
How are conversations conducted on Twitter and
what issues emerge with establishing coherence?
#2
How and why is @-messaging used among
members of a specific community?
#3
How can conversations (rather than friends/
followers) be used to measure social cohesion?
5. Time and the „all friends“ view
Lovely weather this morning!
6. Time and the „all friends“ view
Lovely weather this morning!
@cornelius Same over here!
7. Time and the „all friends“ view
Lovely weather this morning! @adeline nice! how have you been lately?
@cornelius Same over here!
8. Time and the „all friends“ view
Lovely weather this morning! @adeline nice! how have you been lately?
@cornelius Same over here!
Okay, this is interesting: http://something.com
9. Time and the „all friends“ view
Lovely weather this morning! @adeline nice! how have you been lately?
@cornelius Same over here!
Okay, this is interesting: http://something.com
RT @parker Okay, this is interesting: http://something.com
10. Time and the „all friends“ view
Lovely weather this morning! @adeline nice! how have you been lately?
@cornelius Same over here!
Okay, this is interesting: http://something.com
RT @parker Okay, this is interesting: http://something.com
Breaking news about something: http://something.com (via @zeigor, @parker)
11. Time and the „all friends“ view
Point in time #1 Point in time #2
Lovely weather this morning! @adeline nice! how have you been lately?
@cornelius Same over here!
Okay, this is interesting: http://something.com
RT @parker Okay, this is interesting: http://something.com
Breaking news about something: http://something.com (via @zeigor, @parker)
12. Time and the „all friends“ view
Point in time #1 Point in time #2
Lovely weather this morning! @adeline nice! how have you been lately?
conversation
@cornelius Same over here!
Okay, this is interesting: http://something.com
RT @parker Okay, this is interesting: http://something.com
Breaking news about something: http://something.com (via @zeigor, @parker)
13. Time and the „all friends“ view
Point in time #1 Point in time #2
Lovely weather this morning! @adeline nice! how have you been lately?
conversation
@cornelius Same over here!
Okay, this is interesting: http://something.com
RT @parker Okay, this is interesting: http://something.com
Breaking news about something: http://something.com (via @zeigor, @parker)
Twitter doesn‘t lend itself well to conversation.
15. Premise
Twitter conversation
individual tweet
Thesis:
Examining longer stretches of discourse rather than individual
tweets is methodologically superior in some research contexts.
16. Celebrities vs. casual Twitter users
In [Twitter‘s] follower-following topology analysis we have found a
non-power-law follower distribution, a short effective diameter, and
low reciprocity, which all mark a deviation from known
characteristics of human social networks.
- Kwak, Lee, Park & Moon, 2010
17. Celebrities vs. casual Twitter users
[C]elebrity practice reinforces unequal power differentials. While
Twitter users who do not use the site instrumentally may think of their
followers as friends or family, celebrity practice necessitates viewing
followers as fans.
- Marwick and boyd, 2011
18. Celebrities vs. casual Twitter users
celebrities (RL and SoMe) casual users
many followers fewer followers
# friends < # followers # friends = # followers
small % of followers are larger % of followers are
conversation partners conversation partners
people who communicate people who don‘t
professionally (e.g. artists) communicate professionally
broadcasting, performance conversation
19. The data
1 I conducted a pilot study of 10 “casual“ German
and Austrian Twitter users in Spring 2011.
2 Central issues of the study were
privacy and self-disclosure.
My approach combined email
3 interviews with diachronic content
analysis.
4 I followed this up with a study of 25
video game enthusiasts on Twitter.
20. Types of tweets in the pilot study
33%
n = 4052
54%
12%
2%
@-messages Retweets Tweets containing URLs other Tweets
21.
22.
23. Approach of the gamer study
• studied a community of German-speaking gamers
• tracked 24 users for a period of 6 weeks
• collected ~50k tweets
• 38k of these were @-messages
• recruited subjects through an informer
24. Time S Text
14:15:40 Gary Diese User-Review zu From Dust auf Metacritic hat mich restlos überzeugt! http://...
14:16:51 Bob @Gary Habe Metacritic all die Jahre unterschätzt :P
@Gary @Steve Wollt ihr auf dem Pressetag der GC jetzt eigentlich irgendwas
14:18:30 Chris
kreatives machen?
@Steve @Gary Könnte mir vielleicht ne kleine HD-Cam ausleihen, aber keine Ahnung
14:19:56 Chris
wieviel Speicher die hat und ob gute Qualität
@Bob Ich bin einfach vom Stil des Schreibers überrascht. Wer so "sophisticated"
14:24:57 Gary
schreiben kann, muss Ahnung haben!
@Chris Ähhh... der @Steve klärt das schon ab. Ich fänd's cool, aber ich bin Mittwoch
14:25:50 Gary
schon ziemlich ausgelastet.
14:26:47 Chris @Gary wieso was musst du denn machen?
@Chris Termine, mein Freund! Alles voller Termine. Ich muss auch vor den Anderen
14:27:20 Gary
mein Mittagessen zu mir nehmen ;)
14:27:26 Bob @Gary True Art is Incomprehensible! True Art is Offensive! ^^
27. Observations
Usage of Twitter differs so significantly from one user group
1 to another that the claim of it being a channel for
broadcasting can be contested.
Users conduct surprisingly long and structured
2 conversations with recurring communicative
partners.
A surprisingly large number of these
3 partners are neither friends or followers of
the users.