Networks of Outrage (LONDON).key

Transcription

Networks of Outrage (LONDON).key
What are the topics of populist
anti-immigrant movements on
Facebook?
Cornelius Puschmann, Julian Ausserhofer,
Markus Hametner, Noura Maan
Humboldt Institute for Internet and Society/Der Standard
Social Media and Society Conference
Session 5B
Goldsmiths, University of London
13 July 2016
!Advertisements!
My employer(s):
•
Alexander von Humboldt Institute for Internet and Society, Berlin
•
Hans-Bredow-Institute for Media Research, Hamburg
Conference:
•
Association of Internet Researchers’ conference (AoIR 2016)
5-8 October 2016, Berlin
Project:
•
"Networks of Outrage: Mapping the emergence of new
extremism in Europe" (Alexander von Humboldt-Institut for
Internet and Society, Berlin & Der Standard, Vienna)
period: January-October 2016
funded by VolkswagenStiftung
2
Theory: NPS and counter publics
Networked public sphere
•
The public sphere is characterized by the interrelations of
politics, civil society and (eventually) mass media
(Habermas, 1962/1991)
•
The networked public sphere (NPS) takes a reconfigured
digital media environment into account (Benkler, 2006)
•
The NPS emerges through shared practices (Boyd, 2010)
and exists at the intersection of technology and sociality
•
It has been argued that the NPS is affective (Papacharissi,
2014) and (increasingly?) subject to forms of "political
turbulence" (Margetts et al, 2016)
4
Counter-publics
•
Counter-publics were originally conceived as a counterweight to the overbearing/dominating state
•
Counter-publics in social media emerge and dissipate
rapidly, in addition to being increasingly radical:
"the mass-media public sphere will become more open to
radical opinion as a result of the coincidence of societal
crises and the growth of virtual counter-public
spheres" (Downey and Fenton, 2003, p. 199)
➡
How can the discourse of populist social media
counter-publics be characterized?
5
Our case: PEGIDA
PEGIDA: a networked populist movement
•
Patriotic Europeans Against the
Islamisation of the West (PEGIDA)
•
anti-muslim, anti-immigration, antiglobalization movement
•
founded in Dresden in late '14,
Facebook group launched in early '15
•
franchises have been formed in cities
across Germany and abroad (NL, UK,
IE, AU, …)
•
strong courtship by the AfD
(new right-wing party, similar to UKIP)
7
PEGIDA on social media
•
PEGIDA is very active on social media
(205k likes to the Facebook page
pegidaevdresden)
•
PEGIDA’s Facebook page primarily
captures discourse in support of the
movement
•
Twitter more mixed — discourse and
counter-discourse, English vs. German
8
First study: PEGIDA's sources on Twitter
(Puschmann et al, 2016)
•
manual content analysis: coded for stance towards
PEGIDA and source type
•
non-traditional news
sources, blogs, and social
media are favored by the
supporters of PEGIDA (77%
of all URLs)
•
traditional news sources are
favored by its opponents
(65% of all URLSs)
9
Second study: PEGIDA on Facebook
•
January 2015 - May 2016
•
3,400 posts
•
415,000 comments
•
1,460,000 likes
•
weekly rally attendance:
21,000 (January '15) - 2,000 (July '16)
•
Facebook posts
Facebook comments
more activity from the organizers, but
less engagement?
10
Attendance at weekly rally
Method: Correlated topic model (CTM)
Texts as feature spaces
12
Correlated topic models/LDA
(Blei et al, 2003/Blei, 2012)
13
Blei, 2012, p.78
Results: Topics in comments by day
1.0
0.6
0.4
0.2
1.0 0.0
0.2
1.0 0.0
0.4
0.6
0.8
frauen−köln−polizei
0.8
afd−maas−partei−spd
0.2
0.2
0.0
0.0
0.4
0.6
0.8
1.0 0.0
0.2
0.4
0.6
0.8
0.6
0.4
0.2
1.0 0.0
0.8
0.6
0.2
0.0
dresden−grüße−stolz−super
0.2
0.4
0.8
0.6
0.4
0.2
1.0 0.0
0.8
0.6
europa−islam−muslime−religion
flüchtlinge−geld−mann
0.8
1.0
0.8
0.6
0.4
0.2
1.0 0.0
0.8
0.6
0.2
1.0 0.0
0.4
1.0
0.8
0.6
0.4
0.2
1.0 0.0
0.8
0.6
0.2
1.0 0.0
0.4
krieg−putin−regierung−russland
0.0
afd−europa−flüchtlinge−türkei
flüchtlinge−pack−politik−politiker
flüchtlinge−politiker−regierung−ungarn
demo−dresden−lügenpresse−medien
0.4
1.0
0.8
0.6
0.4
0.2
1.0 0.0
0.6
0.4
0.2
1.0 0.0
0.6
0.8
1.0 0.0
0.2
0.4
0.6
0.8
geld−kinder−schule
arbeiten−flüchtlinge−geld
0.2
europa−flüchtlinge
0.4
1.0
0.6
0.4
0.2
1.0 0.0
0.4
0.6
0.8
asylanten−grünen
0.2
1.0 0.0
0.8
0.6
linken−politiker−recht−wählen
afd−frauen−islam−kinder
0.0
0.4
familie−flüchtlinge−kinder−tochter
0.8
politik−politiker−recht
0.8
1.0 0.0
0.2
0.4
0.6
0.8
bürger
islam
1.0
0.6
0.4
0.2
1.0 0.0
0.6
0.8
frauen−köln−polizei
0.4
0.2
0.2
1.0 0.0
1.0 0.0
0.8
afd−maas−partei−spd
0.2
0.2
0.0
0.0
0.4
0.6
0.8
1.0 0.0
0.2
0.4
0.6
0.8
0.6
0.4
0.2
1.0 0.0
0.8
0.6
0.2
dresden−grüße−stolz−super
0.0
0.4
flüchtlinge−geld−mann
0.8
1.0
0.8
0.6
0.4
0.2
1.0 0.0
0.8
0.6
0.4
1.0
0.8
0.6
0.4
0.2
1.0 0.0
0.8
0.6
0.4
0.2
1.0 0.0
0.8
europa−islam−muslime−religion
0.2
afd−europa−flüchtlinge−türkei
1.0 0.0
0.2
0.4
0.6
0.8
flüchtlinge−politiker−regierung−ungarn
demo−dresden−lügenpresse−medien
0.6
0.6
0.4
0.2
1.0 0.0
0.4
0.6
0.8
geld−kinder−schule
0.2
1.0 0.0
0.8
0.6
krieg−putin−regierung−russland
0.0
0.4
flüchtlinge−pack−politik−politiker
0.2
europa−flüchtlinge
arbeiten−flüchtlinge−geld
From steady and recurring topics…
0.4
1.0
0.8
0.6
0.4
0.2
1.0 0.0
1.0
0.6
0.4
0.2
1.0 0.0
0.4
0.6
0.8
asylanten−grünen
0.2
1.0 0.0
0.8
0.6
linken−politiker−recht−wählen
afd−frauen−islam−kinder
0.0
0.4
familie−flüchtlinge−kinder−tochter
0.8
politik−politiker−recht
0.8
1.0 0.0
0.2
0.4
0.6
0.8
bürger
islam
1.0
0.6
0.4
0.2
1.0 0.0
0.2
1.0 0.0
0.4
0.6
0.8
frauen−köln−polizei
0.8
afd−maas−partei−spd
0.2
0.2
0.0
0.0
0.4
0.6
0.8
1.0 0.0
0.2
0.4
0.6
0.8
0.6
0.4
0.2
1.0 0.0
0.8
0.6
0.2
0.0
dresden−grüße−stolz−super
0.2
0.4
0.8
0.6
0.4
0.2
1.0 0.0
0.8
0.6
europa−islam−muslime−religion
flüchtlinge−geld−mann
0.8
1.0
0.8
0.6
0.4
0.2
1.0 0.0
0.8
0.6
0.2
1.0 0.0
0.4
1.0
0.8
0.6
0.4
0.2
1.0 0.0
0.8
0.6
0.2
1.0 0.0
0.4
krieg−putin−regierung−russland
0.0
afd−europa−flüchtlinge−türkei
flüchtlinge−pack−politik−politiker
flüchtlinge−politiker−regierung−ungarn
demo−dresden−lügenpresse−medien
0.4
1.0
0.8
0.6
0.4
0.2
1.0 0.0
0.6
0.4
0.2
1.0 0.0
0.6
0.8
1.0 0.0
0.2
0.4
0.6
0.8
geld−kinder−schule
arbeiten−flüchtlinge−geld
0.2
europa−flüchtlinge
0.4
1.0
0.6
0.4
0.2
1.0 0.0
0.4
0.6
0.8
asylanten−grünen
0.2
1.0 0.0
0.8
0.6
linken−politiker−recht−wählen
afd−frauen−islam−kinder
0.0
0.4
familie−flüchtlinge−kinder−tochter
0.8
politik−politiker−recht
0.8
1.0 0.0
0.2
0.4
0.6
0.8
bürger
islam
1.0
0.6
0.4
0.2
1.0 0.0
0.2
1.0 0.0
0.4
0.6
0.8
frauen−köln−polizei
0.8
afd−maas−partei−spd
0.2
0.2
0.0
0.0
0.4
0.6
0.8
1.0 0.0
0.2
0.4
0.6
0.8
0.6
0.4
0.2
1.0 0.0
0.8
0.6
0.2
0.0
dresden−grüße−stolz−super
0.2
0.4
0.8
0.6
0.4
0.2
1.0 0.0
0.8
0.6
europa−islam−muslime−religion
flüchtlinge−geld−mann
0.8
1.0
0.8
0.6
0.4
0.2
1.0 0.0
0.8
0.6
0.2
1.0 0.0
0.4
1.0
0.8
0.6
0.4
0.2
1.0 0.0
0.8
0.6
0.2
1.0 0.0
0.4
krieg−putin−regierung−russland
0.0
afd−europa−flüchtlinge−türkei
flüchtlinge−pack−politik−politiker
flüchtlinge−politiker−regierung−ungarn
demo−dresden−lügenpresse−medien
0.4
1.0
0.8
0.6
0.4
0.2
1.0 0.0
0.6
0.4
0.2
1.0 0.0
0.6
0.8
1.0 0.0
0.2
0.4
0.6
0.8
geld−kinder−schule
arbeiten−flüchtlinge−geld
0.2
europa−flüchtlinge
0.4
1.0
0.6
0.4
0.2
1.0 0.0
0.4
0.6
0.8
asylanten−grünen
0.2
1.0 0.0
0.8
0.6
linken−politiker−recht−wählen
afd−frauen−islam−kinder
0.0
0.4
familie−flüchtlinge−kinder−tochter
0.8
politik−politiker−recht
0.8
1.0 0.0
0.2
0.4
0.6
0.8
bürger
islam
1.0
0.6
0.4
0.2
1.0 0.0
0.6
0.8
frauen−köln−polizei
0.4
0.2
0.2
1.0 0.0
1.0 0.0
0.8
afd−maas−partei−spd
0.2
0.2
0.0
0.0
0.4
0.6
0.8
1.0 0.0
0.2
0.4
0.6
0.8
0.6
0.4
0.2
1.0 0.0
0.8
0.6
0.2
dresden−grüße−stolz−super
0.0
0.4
flüchtlinge−geld−mann
0.8
1.0
0.8
0.6
0.4
0.2
1.0 0.0
0.8
0.6
0.4
1.0
0.8
0.6
0.4
0.2
1.0 0.0
0.8
0.6
0.4
0.2
1.0 0.0
0.8
europa−islam−muslime−religion
0.2
afd−europa−flüchtlinge−türkei
1.0 0.0
0.2
0.4
0.6
0.8
flüchtlinge−politiker−regierung−ungarn
demo−dresden−lügenpresse−medien
0.6
0.6
0.4
0.2
1.0 0.0
0.4
0.6
0.8
geld−kinder−schule
0.2
1.0 0.0
0.8
0.6
krieg−putin−regierung−russland
0.0
0.4
flüchtlinge−pack−politik−politiker
0.2
europa−flüchtlinge
arbeiten−flüchtlinge−geld
…to busty and resurgent topics
0.4
1.0
0.8
0.6
0.4
0.2
1.0 0.0
1.0
0.6
0.4
0.2
1.0 0.0
0.4
0.6
0.8
asylanten−grünen
0.2
1.0 0.0
0.8
0.6
linken−politiker−recht−wählen
afd−frauen−islam−kinder
0.0
0.4
familie−flüchtlinge−kinder−tochter
0.8
politik−politiker−recht
0.8
1.0 0.0
0.2
0.4
0.6
0.8
bürger
islam
Clustering daily topics by similarity
flüchtlinge−politiker−regierung−ungarn
asylanten−grünen
geld−kinder−schule
flüchtlinge−pack−politik−politiker
linken−politiker−recht−wählen
afd−maas−partei−spd
politik−politiker−recht
dresden−grüße−stolz−super
europa−flüchtlinge
krieg−putin−regierung−russland
bürger
afd−europa−flüchtlinge−türkei
arbeiten−flüchtlinge−geld
flüchtlinge−geld−mann
frauen−köln−polizei
afd−frauen−islam−kinder
demo−dresden−lügenpresse−medien
europa−islam−muslime−religion
islam
familie−flüchtlinge−kinder−tochter
200
150
100
20
50
0
Results
•
Aggregating comments by day/week seems promising
•
CTM topics on the PEGIDA page combine social concerns
(citizenship, work, money, children), concrete events
(cologne), domestic (spd, maas, linken, grüne) and foreign
politics (turkey, hungary, russia)
•
There is further analytic potential in:
•
the expression of topics across documents
•
the degree of similarity of topics and documents
•
the relation of metadata (user, pre/post-event) to topics
21
Thank you for listening!
flickr/barnyzk
References
Benkler, Y. (2006). The wealth of networks: How social production transforms markets and freedom. New Haven, CT:
Yale University Press.
Blei, D. M. (2012). Probabilistic topic models. Communications of the ACM, 55(4), 77. http://doi.org/
10.1145/2133806.2133826
Blei, D. M., & Lafferty, J. D. (2007). A correlated topic model of Science. The Annals of Applied Statistics, 1(1), 17–
35. http://doi.org/10.1214/07-AOAS114
Blei, D. M., Ng, A. Y., & Jordan, M. I. (2003). Latent dirichlet allocation. Journal of Machine Learning Research, 3,
993–1022. http://doi.org/10.1162/jmlr.2003.3.4-5.993
Boyd, D. M. (2010). Social network sites as networked publics: Affordances, dynamics, and implications. In Z.
Papacharissi (Ed.), A networked self: Identity, community, and culture on social network sites (pp. 39–58). London:
Routledge. http://doi.org/10.1162/dmal.9780262524834.119
Downey, J., & Fenton, N. (2003). New media, counter publicity and the public sphere. New Media & Society, 5(2),
185–202. http://doi.org/10.1177/1461444803005002003
Habermas, J. (1991). The structural transformation of the public sphere: An inquiry into a category of bourgeois
society. Cambridge, MA: MIT Press.
Margetts, H., John, P., Hale, S., & Yasseri, T. (2016). Political turbulence: How social media shape collective action.
Princeton, NJ: Princeton University Press.
Papacharissi, Z. (2014). Affective publics: Sentiment, technology, and politics. Oxford: Oxford University Press.
Puschmann, C., Ausserhofer, J., Maan, N., & Hametner, M. (2016). Information laundering and counter-publics: The
news sources of islamophobic groups on Twitter. In International AAAI Conference on Web and Social Media.
Menlo Park, CA: AAAI Press. Retrieved from http://www.aaai.org/ocs/index.php/ICWSM/ICWSM16/paper/view/
13224/12858
23
Results: Topics by user
Second study: PEGIDA on Facebook
Facebook comments per user (log scale).
Facebook likes per post over time.
Facebook comment length in characters (log scale).
25
User-topic heatmap
1
2
arsch
europa
bürger
gutmenschen
dresden
leute
friedlich leute
menschen
liebe medien
merkel
menschen
politik
polizei stadt
ungarn
wahrheit
Heatmap and cluster dendrogram of 10 topics across 100 users (random
sample, top 5% by number of posts >=150 characters)
26
6
7
anfang
europa
flüchtlinge
gutmenschen
islam
kultur
länder
merkel
politik
regierung
spd
asylanten
flüchtlinge
grenzen
gute
heimat
länder
menschen
merkel
politiker
raus
recht
regierung
richtig
sachsen
3
4
5
berlin bürger
europa
medien
merkel tag
europa
flüchtlinge
flüchtlingen
islam
kultur
mädchen
menschen
muslime
religion
syrien
terroristen
frauen geld
kinder leute
mann
männer
menschen
merkel pack
regierung tag
8
9
10
afd
deutscher
ändern
euro
gute
frauen gott
afd
kinder
hand
kopf
asylanten cdu
kosten
krieg
gewalt
linke
lügen nazi
grünen
parteien
politik
islam
presse
politiker
inke linken
recht
polizei
merkel
regierung
problem
muslime
sicherheit spd
regierung
recht spd
wählen
schuld staat
wahrheit
verloren
zukunft
verstehen
Weekly topics as hierarchical clusters
Hierarchical cluster dendrogram using Ward's method showing the degree of
similarity among topics. Basis is Euclidean distance, calculated from the log
likelihood scores of terms (the beta statistic) within topics.
27
1
2
3
4
5
flüchtlinge
merkel
pack
politiker
afd
europa
islam
merkel
dresden
islam
medien
muslime
politik
politiker
dresden
flüchtlinge
merkel
dresden
merkel
6
7
8
9
10
europa
flüchtlinge
islam
merkel paris
regierung
terroristen
dresden
islam
asylanten
flüchtlinge
geld
politiker
afd
europa
flüchtlinge
merkel
europa
islam
merkel
11
12
13
14
15
islam
muslime
politiker
dresden
flüchtlinge
merkel
dresden
leute
menschen
frauen
köln
merkel
polizei
flüchtlinge
merkel
16
17
18
19
20
kinder
merkel
politiker
dresden
flüchtlinge
menschen
merkel
ungarn
flüchtlinge
islam
kinder
leute politik
politiker raus
recht
regierung
dresden
europa
flüchtlinge
menschen
merkel
regierung
islam
menschen
politik
politiker
Terms associated with twenty topics in the corpus (log
likelihood >= 0.005).
Text mining workflow
read
tokenize
remove punctuation/numbers/whitespace
remove stopwords
(stem)
Build DTM/TDM
28
Latent Dirichlet Allocation
(Blei et al, 2003/Blei, 2012)
•
topics as feature bundles with distributions across texts
•
variable number of k topics
•
identifies latent topical differences
•
correlated topic model algorithm (CTM) allows for
systematically correlated topics
29