Abstract
The 2022 Brazilian presidential election was characterized by deep ideological polarization, hyper-partisan mobilization, and unprecedented levels of digital engagement. This study investigates the architecture and dynamics of Twitter (now X) discourse during the critical runoff phase between Luiz Inácio Lula da Silva and Jair Bolsonaro. Utilizing a multi-method computational framework, we conducted a quantitative content analysis and network topology evaluation of 2.8 million election-related tweets collected between October 2 and October 30, 2022. By integrating community detection algorithms with supervised text classification and sentiment scoring, we mapped the structural fragmentation of online public spheres and assessed the degree of affective polarization across distinct political clusters. Our findings reveal a bifurcated network structure marked by profound ideological segregation, with homophily coefficients exceeding 0.82 in core communities. Discursive frames within the pro-Bolsonaro cluster disproportionately leveraged systemic distrust, anti-establishment rhetoric, and electoral integrity skepticism, whereas the pro-Lula cluster emphasized institutional preservation, welfare, and democratic defense. Furthermore, peripheral bridge nodes were systematically delegitimized by in-group partisans, effectively dampening inter-cluster deliberation. These empirical observations elucidate how algorithmic affordances interact with political cleavages in the Global South, contributing to persistent affective echo chambers that both reflect and exacerbate electoral instability in digital democracies.
