Mineração de dados e Hermenêutica na era do Big Data: limites interpretativos da análise computacional da política

Abstract

Esta pesquisa apresenta impacto científico e metodológico ao integrar hermenêutica crítica, Big Data e análise computacional da comunicação, oferecendo um modelo replicável para o estudo da mediação midiática de crises. No campo social e educacional, contribui para o fortalecimento da literacia midiática, da compreensão pública sobre desinformação e dos efeitos simbólicos da comunicação digital na democracia. Seu potencial inovador reside no uso combinado de Google Trends, mineração de notícias e análise de sentimentos para revelar padrões estruturais de governança simbólica. A pesquisa possui relevância nacional e internacional, com aplicabilidade em diferentes contextos sociopolíticos, dialogando com agendas globais sobre regulação das plataformas, saúde informacional e desenvolvimento sustentável. Ao evidenciar os limites democráticos da estabilização narrativa das crises, o estudo contribui para políticas públicas, práticas jornalísticas e debates acadêmicos voltados à promoção de uma esfera pública mais crítica, plural e socialmente responsável.

This thesis demonstrates how contemporary media systems function as powerful devices of symbolic governance, metabolizing crises into narratives of stability. Through a mixed and comparative methodology, it investigated two major axes: the COVID-19 infodemic, analyzed via search patterns on Google Trends, and the digital news coverage of three critical political events—the São Paulo municipal elections (2024), the public conflict between Elon Musk and the Brazilian Supreme Court (STF), and the assassination attempt on Donald Trump (2024)—using news mining (web scraping) and sentiment analysis with the NRC lexicon. The results of the four empirical studies converge on a central conclusion: there is a cross-cutting operational logic in crisis mediation. Whether in a global health emergency, a localized electoral dispute, a clash over digital sovereignty, or an episode of political violence, media coverage follows a predictable stabilization script. This script is supported by three pillars: 1) emotional management, with the statistical predominance of the sentiment of "trust" that calms public anxiety; 2) institutional framing, which converts complex conflicts into legal, procedural, or moral issues manageable by existing institutions; and 3) temporal control, which organizes the narrative into phases (emergence, climax, resolution), absorbing the disruptive event as a delimited and surmountable episode. While this mechanism is functional for maintaining order and institutional legitimacy in the short term, the thesis highlights its profound democratic cost. The efficiency in producing symbolic stability is achieved through the systematic displacement and silencing of the structural dimensions of conflicts—inequalities, material power disputes, geopolitical asymmetries, and the political nature of digital infrastructures. Thus, the media does not merely report crises but symbolically governs them, offering narrative resolution at the expense of political confrontation. The final contribution is a warning: by prioritizing affective governability, the prevailing communication system may paradoxically be undermining the conditions for a robust and transformative public debate, essential for a substantive democracy.