This thesis presents the design, implementation, and application of a computational data pipeline for analyzing Russian-language war discourse on Telegram during the Russia-Ukraine conflict (February 2022 - April 2026). Telegram, a messaging platform with minimal content moderation, has emerged as the primary medium for Russian wartime narrative production, yet systematic research linking specific military operations to measurable shifts in discourse remains limited. The pipeline (v2.5) automates the extraction, pre-processing, thematic scoring, and event- proximity labeling of messages from 54 Russian-language Telegram channels organized into seven actor categories: milbloggers, state media, nationalists, independent media, analysts, official government channels, and Ukrainian sources. Each message is scored against seven predefined narrative themes Conflict Terrain, Tactical Weaponry, Official Narrative, Mobi- lization, Victory, Dehumanization, Casualties, and classified relative to 18 major operational events using a temporal window methodology adapted from financial event-study analysis. The resulting dataset comprises 595,507 messages from 48 active channels. Post-event analysis of the 102,715 messages falling in the [+0, +14]-window reveals three principal find- ings: (1) discourse volume is event-sensitive but actor-category rankings remain stable, with state media consistently dominating output; (2) dominant narrative themes undergo temporal succession, shifting from territorial framing (2022) to weaponry-focused discourse (2023- 2025) to patriotic celebration (2025 peace summit); and (3) actor categories maintain dis- tinct thematic identities despite shared exposure to the same events. The dataset and pipeline constitute a structured resource for advanced NLP analyses.
Questa tesi presenta la progettazione, l'implementazione e l'applicazione di un pipeline computazionale per l'analisi del discorso bellico in lingua russa su Telegram durante il conflitto Russia-Ucraina (febbraio 2022 – aprile 2026). Il pipeline (v2.5) automatizza l'estrazione, la pre-elaborazione, l'assegnazione di punteggi tematici e l'etichettatura di prossimità agli eventi dei messaggi provenienti da 54 canali Telegram in lingua russa, organizzati in sette categorie di attori: milblogger, media statali, nazionalisti, media indipendenti, analisti, canali governativi ufficiali e fonti ucraine. Ogni messaggio viene valutato rispetto a sette temi narrativi predefiniti e classificato in relazione a 18 eventi operativi principali, utilizzando una metodologia a finestra temporale adattata dall'analisi event-study della finanza. Il dataset risultante comprende 595.507 messaggi provenienti da 48 canali attivi. L'analisi post-evento dei 102.715 messaggi nella finestra [+0, +14] giorni rivela tre risultati principali: (1) il volume del discorso è sensibile agli eventi, ma la classifica delle categorie di attori rimane stabile, con i media statali costantemente dominanti; (2) i temi narrativi dominanti subiscono una successione temporale, passando da un inquadramento territoriale (2022) a un discorso incentrato sugli armamenti (2023–2025) fino alla celebrazione patriottica (summit di pace 2025); e (3) le categorie di attori mantengono identità tematiche distinte nonostante l'esposizione condivisa agli stessi eventi.
TELEGRAM AS A BATTLEFIELD: COMPUTATIONAL ANALYSIS OF RUSSIAN WAR DISCOURSE ACROSS 18 OPERATIONAL EVENTS (2022–2026)
MORKA, MIHRETAB WORKU
2025/2026
Abstract
This thesis presents the design, implementation, and application of a computational data pipeline for analyzing Russian-language war discourse on Telegram during the Russia-Ukraine conflict (February 2022 - April 2026). Telegram, a messaging platform with minimal content moderation, has emerged as the primary medium for Russian wartime narrative production, yet systematic research linking specific military operations to measurable shifts in discourse remains limited. The pipeline (v2.5) automates the extraction, pre-processing, thematic scoring, and event- proximity labeling of messages from 54 Russian-language Telegram channels organized into seven actor categories: milbloggers, state media, nationalists, independent media, analysts, official government channels, and Ukrainian sources. Each message is scored against seven predefined narrative themes Conflict Terrain, Tactical Weaponry, Official Narrative, Mobi- lization, Victory, Dehumanization, Casualties, and classified relative to 18 major operational events using a temporal window methodology adapted from financial event-study analysis. The resulting dataset comprises 595,507 messages from 48 active channels. Post-event analysis of the 102,715 messages falling in the [+0, +14]-window reveals three principal find- ings: (1) discourse volume is event-sensitive but actor-category rankings remain stable, with state media consistently dominating output; (2) dominant narrative themes undergo temporal succession, shifting from territorial framing (2022) to weaponry-focused discourse (2023- 2025) to patriotic celebration (2025 peace summit); and (3) actor categories maintain dis- tinct thematic identities despite shared exposure to the same events. The dataset and pipeline constitute a structured resource for advanced NLP analyses.| File | Dimensione | Formato | |
|---|---|---|---|
|
Final_thesis_project_Mihretab_Worku_Morka (2) (1)_signed.pdf
accesso aperto
Descrizione: Laurea thesis presenting a computational data pipeline (v2.5) for analyzing Russian-language war discourse on Telegram. The pipeline processes 595,507 messages from 48 channels across 7 actor categories, with keyword-based thematic scoring and event-proxi
Dimensione
3.24 MB
Formato
Adobe PDF
|
3.24 MB | Adobe PDF | Visualizza/Apri |
I documenti in UNITESI sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/20.500.12075/27478