Shoah Narratives on Social Media: an Analysis of the Italian "Proteggi I Fatti" Holocaust Distortion Campaign

This thesis analyses the discursive dynamics generated by the Italian Facebook campaign “Proteggi I Fatti”, which is part of the European initiative “Deconstruct: deconstructing and countering Holocaust distortion via campaign, education and training”. Through online communication campaigns, educational materials and learning activities, the initiative’s objective is to disseminate historically accurate information and provide critical thinking tools to combat Holocaust distortion and denial. The analysis is based on a corpus of about 10,178 replies to nine posts published between December 2024 and May 2025. This study offers insight into how the audience interacted with the Facebook content and which attitudes toward the Holocaust were expressed. The methodology behind the study combines NLP (Natural Language Processing) techniques with LLMs (Large Language Models). The NLP techniques include word frequency analysis, bi-gram identification, sentence length statistics, sentiment analysis, toxicity detection, bias measurement, network analysis and word/sentence embeddings. To capture a deeper understanding of the discursive patterns across the comments LLMs were applied via the Gemini API. The findings reveal a significant shift of the historical memory of the Shoah toward the current Israeli-Palestinian conflict. The analysis outlines a prevailing negative sentiment and recurring antisemitic narratives that involve the Holocaust distortion, victim-perpetrator reversal and the accusations of using the Holocaust remembrance as an instrument to legitimate military or political actions.