Digital communication has transformed the public sphere into an ecosystem where platform design, emotional expression, and linguistic structure interact with information reliability. This thesis integrates network science, NLP, and eXAI to investigate these dynamics across four studies on Twitter, YouTube, and Reddit. The first study analyzes the COVID-19 vaccine debate on Twitter, revealing that while major vaccine approvals catalyze content production across all sources, interaction structures remain dominated by stable, credibility-based echo chambers. The second study focuses on YouTube, employing the Plutchik model and eXAI to identify distinct emotional signatures in misinformation consumption. It introduces the concept of "emotional arborescence", showing that misinformation-prone interactions converge around a narrow landscape of fear and reduced joy. Third, an analysis of Reddit news diffusion reveals that unreliable sources often function as structural bridges, amplifying narratives after their introduction by mainstream media. Finally, the thesis examines linguistic correlates across global debates finding that high user involvement consistently correlates with increased linguistic complexity. These studies demonstrate how methodological integration reveals a unified digital ecosystem where network topology, emotional dynamics, and linguistic engagement jointly shape the diffusion and persistence of information.
La comunicazione digitale ha trasformato la sfera pubblica in un ecosistema in cui platform design, emozioni e struttura linguistica interagiscono con l'affidabilità del' informazione. Questa tesi integra network science, NLP ed eXAI per indagare queste dinamiche in quattro studi su Twitter, YouTube e Reddit. Il primo studio analizza il dibattito sui vaccini su Twitter, rivelando che, mentre le principali approvazioni dei vaccini catalizzano la produzione di contenuti su tutte le fonti, le interazioni rimangono dominate da echo chamber stabili basate sulla credibilità. Il secondo studio si concentra su YouTube, utilizzando il modello di Plutchik ed eXAI per identificare tracce emotive nel consumo di disinformazione. Introduce la misura di "arborescenza emotiva", mostrando che interazioni misinformation-prone convergono su emozioni ristrette a paura e gioia ridotta. In terzo luogo, un'analisi della diffusione delle notizie su Reddit rivela che le fonti inaffidabili spesso fungono da ponti strutturali, amplificando le narrazioni dopo la loro introduzione da fonti credibili. Infine, la tesi esamina la struttura linguistica nei dibattiti globali, scoprendo che un elevato coinvolgimento degli utenti è correlato a una maggiore complessità linguistica. Questi studi mostrano come l'integrazione metodologica riveli un ecosistema digitale unificato in cui rete, le dinamiche emotive, e struttura linguistica modellano la diffusione e la persistenza delle informazioni.
Structure and Dynamics of the Online Information Ecosystem: A Computational Analysis of Networks, Emotion, and Language
SANTORO, ARNALDO
2026
Abstract
Digital communication has transformed the public sphere into an ecosystem where platform design, emotional expression, and linguistic structure interact with information reliability. This thesis integrates network science, NLP, and eXAI to investigate these dynamics across four studies on Twitter, YouTube, and Reddit. The first study analyzes the COVID-19 vaccine debate on Twitter, revealing that while major vaccine approvals catalyze content production across all sources, interaction structures remain dominated by stable, credibility-based echo chambers. The second study focuses on YouTube, employing the Plutchik model and eXAI to identify distinct emotional signatures in misinformation consumption. It introduces the concept of "emotional arborescence", showing that misinformation-prone interactions converge around a narrow landscape of fear and reduced joy. Third, an analysis of Reddit news diffusion reveals that unreliable sources often function as structural bridges, amplifying narratives after their introduction by mainstream media. Finally, the thesis examines linguistic correlates across global debates finding that high user involvement consistently correlates with increased linguistic complexity. These studies demonstrate how methodological integration reveals a unified digital ecosystem where network topology, emotional dynamics, and linguistic engagement jointly shape the diffusion and persistence of information.| File | Dimensione | Formato | |
|---|---|---|---|
|
2026_phd_thesis_arnaldo_post_antiAI_pdfa (1).pdf
embargo fino al 11/09/2027
Licenza:
Tutti i diritti riservati
Dimensione
18.88 MB
Formato
Adobe PDF
|
18.88 MB | Adobe PDF |
I documenti in UNITESI sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/20.500.14242/379489
URN:NBN:IT:UNIVE-379489