Navigating Extremes with Transformers and Fuzzy Logic: Polarization in the Spanish Presidential Ejections 2023

In this communication we present a case study of the Spanish General Elections (2023) in which we have measured the affective polarization in the digital debate. The official electoral campaign took place after the unexpected call of the Spanish president on July 23. By employing transformer models, we assessed various topics and conducted sentiment analysis within the political discourse of the elections to gauge the emotional tone of the discussions. Our findings revealed that the most frequently debated subjects were Candidates (n1 = 17170) and Opposition (n3 = 15327), both exhibiting significant levels of polarization. Utilizing an affective polarization framework, we implemented a polarization metric (JDJ) rooted in a fuzzy set theory. Among the topics analyzed, activism showed the highest polarization, while voting had the lowest. Using transformers, we have detected topics and applied sentiment analysis in the political debate during the elections to measure the emotional valence of the debate. We found that the topics that occupy most of the debate are Candidates (n1 = 17170) and Opposition (n3 = 15327). These topics also present the highest typical polarization deviances. Based on the affective polarization, we applied a polarization measure (JDJ) based on the fuzzy sets. The topic activism has the highest polarization value and topic voting the lowest. Thanks to this analysis we can identify the presence of a dichotomy that defines the Spanish political reality, the positive image of traditional political participation and engagement juxtaposed against an aversion to collective action initiatives.