TY - GEN

T1 - Cumulative Polarization: Patterns of Accumulation of Neutral and Politicized Echo Chambers on Russian Twitter

AU - Бодрунова, Светлана Сергеевна

AU - Блеканов, Иван Станиславович

AU - Тарасов, Никита Андреевич

N1 - Conference code: 25

PY - 2023

Y1 - 2023

N2 - Opinion cumulation on social networks has been widely researched upon; however, we still lack knowledge on its dynamics. In particular, political polarization that leads to echo chambering with democratically inefficient homophily of views is mostly studied in terms of resulting echo chambers and factors shaping it, but not in terms of formation dynamics; this may misinform understandings of the nature of echo chambers. The concept of cumulative deliberation [1] implies opinion divergence and growth of clusters of posts/comments that may contain differing opinions in online discussions, where a ‘tug-of-war’ model of winning the dispute may be traced. However, in our earlier work [2], we have shown that user discourse may diverge into neutral vs. politically opinionated clusters, rather than, e.g., left vs. right or loyal vs. oppositional. Our other research [3] shows that, in case of global or nationwide events, news content spreads first, and then, with a time lag, issues of public importance raise in the discussions. Combining this knowledge, we use Twitter data from Russia to reveal the dynamics of echo chambering as cumulation of neutral vs. polarized content in time. For that, we use a multi-step research pipeline that includes, i.a., web graph analytics, k-means clustering, creation of frequency vocabularies, Granger testing, and visual mapping. We show that echo chambers of news-based and opinion-based nature form in different time and depend upon the lifestyle of their actors. We detect cumulative patterns in formation of echo chambers, including one where the spread of news leads to discourse radicalization and one called ‘the spiral of influence’ in interdependence of clusters.

AB - Opinion cumulation on social networks has been widely researched upon; however, we still lack knowledge on its dynamics. In particular, political polarization that leads to echo chambering with democratically inefficient homophily of views is mostly studied in terms of resulting echo chambers and factors shaping it, but not in terms of formation dynamics; this may misinform understandings of the nature of echo chambers. The concept of cumulative deliberation [1] implies opinion divergence and growth of clusters of posts/comments that may contain differing opinions in online discussions, where a ‘tug-of-war’ model of winning the dispute may be traced. However, in our earlier work [2], we have shown that user discourse may diverge into neutral vs. politically opinionated clusters, rather than, e.g., left vs. right or loyal vs. oppositional. Our other research [3] shows that, in case of global or nationwide events, news content spreads first, and then, with a time lag, issues of public importance raise in the discussions. Combining this knowledge, we use Twitter data from Russia to reveal the dynamics of echo chambering as cumulation of neutral vs. polarized content in time. For that, we use a multi-step research pipeline that includes, i.a., web graph analytics, k-means clustering, creation of frequency vocabularies, Granger testing, and visual mapping. We show that echo chambers of news-based and opinion-based nature form in different time and depend upon the lifestyle of their actors. We detect cumulative patterns in formation of echo chambers, including one where the spread of news leads to discourse radicalization and one called ‘the spiral of influence’ in interdependence of clusters.

KW - clustering

KW - cumulative deliberation

KW - dynamic polarization

KW - echo chambers

KW - opinions

KW - social media

UR - https://link.springer.com/chapter/10.1007/978-3-031-35915-6_13

UR - https://www.mendeley.com/catalogue/e079ebf4-7627-33fc-8917-22e7993e950d/

U2 - 10.1007/978-3-031-35915-6_13

DO - 10.1007/978-3-031-35915-6_13

M3 - Conference contribution

SN - 978-3-031-35914-9

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 167

EP - 182

BT - Social Computing and Social Media. HCII 2023

PB - Springer Nature

T2 - 25TH INTERNATIONAL CONFERENCE ON HUMAN-COMPUTER INTERACTION

Y2 - 23 July 2023 through 28 July 2023

ER -