charla en el congreso alice 2012

Cómo afectan las redes sociales a la difusión del mensaje

Comunidades de conversación política

en redes sociales

Esteban Moro @estebanmoro Universidad Carlos III de Madrid

Instituto de Ingeniería del Conocimiento

ALICE Madrid 2012

@estebanmoro ALICE Madrid 2012

¿Cómo se propaga la opinión �política en redes sociales?


Crecimiento de Twitter


Twitter y opinión pública

• El Social media es un proxy de "la opinión pública • Flutrend • Marketing en Fb • Blogs

• Twitter :: ejemplos • Twitter y películas"

Asur y Huberman (2010) • Twitter y Bolsas"

Bollen, Mao y Zeng (2010) • Twitter y opinión política"

A. Tumasjan et al. (2010)"


Como se difunde la opinión política en twitter

Demà a la nit emetrem, en directe i des de Lleida, l'acte d'inici de campanya. I tu, per què votaràs

RT Demà a la nit emetrem, en directe i des de Lleida, l'acte d'inici de campanya. I tu, per què votaràs



RT


Como se difunde la opinión política en twitter

• Elecciones al parlamento calatán 2010 •  102k tweets, 10-28 Nov 2010 •  25k cuentas de twitter únicas

Tweets

RT

MT


Estructura del flujo de opinión política

•  Estructurada en comunidades"La difusión de información es local"formándose zonas densas "de comunicación

• Comunidades • Conversaciones

• Cada comunidad corresponde"a un partido político"(difusión partidista) [partisans]

• Medios de comunicación son"periféricos/fuentes a/de este flujo "de información""ML Congosto, M Fernández, E. Moro"Cuadernos EVOCA Nº 4 "

cori

pirata

psc

ciu

ppc

c’s icv

erc

pacma

si


Elecciones Catalanas 2010

-1.0 -0.5 0.0 0.5 1.0

-1.0

-0.5

0.0

0.5

1.0

0

0

PSCCiUPSCCiUPSCCiUPSCCiUERCPPCICV

PSCCiUERCPPCICVC'sSOL

PSCCiUERCPPCICVC'sSOLPPTPACMACORI


Estructura del flujo de opinión política

Figure 1: The political retweet (left) and mention (right) networks, laid out using a force-directed algorithm. Node colors reflectcluster assignments (see § 3.1). Community structure is evident in the retweet network, but less so in the mention network. Weshow in § 3.3 that in the retweet network, the red cluster A is made of 93% right-leaning users, while the blue cluster B is madeof 80% left-leaning users.

tive Twitter users. This structural difference is of particularimportance with respect to political communication, as wenow have statistical evidence to suggest that mentions andreplies may serve as a conduit through which users are ex-posed to information and opinions they might not choose inadvance. Despite this promising finding, the work of Yardiand boyd (2010) suggests that cross-ideological interactionsmay reinforce pre-existing in-group/out-group identities, ex-acerbating the problem of political polarization.

3.2 Content HomogeneityThe clustering described above was based only on the net-work properties of the retweet and mention graphs. An inter-esting question, therefore, is whether it has any significancein terms of the actual content of the discussions involved.To address this issue we associate each user with a profilevector containing all the hashtags in her tweets, weighted bytheir frequencies. We can then compute the cosine similari-ties between each pair of user profiles, separately for usersin the same cluster and users in different clusters. Figure 2shows that in the mention network, users placed in the samecluster are not likely to be much more similar to each otherthan users in different clusters. On the other hand, in theretweet network, users in cluster A are more likely to havevery similar profiles than users in cluster B, and users in dif-ferent clusters are the least similar to each other. As a resultthe average similarity within retweet clusters is higher thanacross clusters. Further, we note that in both mention andretweet networks, one of the clusters is more cohesive thanthe other — meaning the tag usage within one community ismore homogeneous.

Retweet MentionA$A 0.31 0.31B$B 0.20 0.22A$B 0.13 0.26 10-1

100

101

0 0.2 0.4 0.6 0.8 1

P(c

os(

a,

b))

cos(a, b)

Clusters ACluster B

Different clusters

Figure 2: Cosine similarities among user profiles. The tableon the left shows the average similarities in the retweet andmention networks for pairs of users both in cluster A, both incluster B, and for users in different clusters. All differencesare significant at the 95% confidence level. The plot on theright displays the actual distributions of cosine similaritiesfor the retweet network.

3.3 Political PolarizationGiven the communities of the retweet network identified in§ 3.1, their content homogeneity uncovered in § 3.2, andthe findings of previous studies, it is natural to investigatewhether the clusters in the retweet network correspond togroups of users of similar political alignment.

To accomplish this in a systematic, reproducible way weused a set of techniques from the social sciences knownas qualitative content analysis (Krippendorff 2004; Kolbe1991). Similar to assigning class labels to training data in su-pervised machine learning, content analysis defines a set ofpractices that enable social scientists to define reproduciblecategories for qualitative features of text. Next we outlineour annotation categories, and then explain the procedures

Demócratas

Republicanos

Conover et al. 2011


PSC$

15m$

CiU$

SI$

ERC$

Irrupción de nuevos grupos en la comunicación política

• Contexto: Retallades • Sep-Nov 2011 •  30k tweets

(M.L. Congosto data)


Huelga General 2012

27M-2A 750k tweets"300k usuarios


Diferencias entre comunidades detectadas


yohagohuelga General

consumo Tomalahuelga

Enhuelga conseguir

Fracasosindical"Trabajar"

piquete sindical"liberado"

subvención"Derecho"respetar

trabajar respetar destrozar

cobrar

conseguir apoyar

encender sumar

instaurar

Diferencias entre comunidades detectadas


Estructura de grupos en la comunicación política en redes sociales

• El estudio en Twitter permite detectar diferentes grupos en las redes sociales • Comunican de diferentes cosas (partidos, opiniones, etc.) • Se comunican en diferentes idiomas • Hablan de manera diferente • Dicen cosas diferentes • Mutan de manera diferente • Tienen diferentes líderes (influenciadores)

• Por tanto...


Twitter es un conjunto de conversaciones

charla en el congreso alice 2012

Education