Ανάλυση Συναισθήματος και Πολλαπλή Παραγοντική Ανάλυση στη Διερεύνηση Επιστημονικών Μεταδεδομένων: Η περίπτωση των Σινο-Αμερικανικών Σχέσεων

Abstract

This study examines the combined use of Sentiment Analysis and Multiple Cor-respondence Analysis to advance the understanding of emotional expression in textual data, with a particular focus on international relations between the Unit-ed States and China. The methodological approach begins with a bibliometric analysis that yields a dataset of 203 scholarly articles, each of which is evaluated for emotional tone using the NRC emotion lexicon. The resulting sentiment data are then subjected to Multiple Correspondence Analysis, which serves to uncover underlying structures and associations among categorical variables. The findings indicate the prevalence of positive emotions, notably trust, as well as the signifi-cant presence of negative emotions, such as fear, particularly in texts addressing geopolitical tensions. The application of Multiple Correspondence Analysis facili-tates a more nuanced and multidimensional interpretation of the sentiment landscape, revealing patterns that conventional sentiment analysis techniques may overlook. The study demonstrates that this integrative approach enhances both analytical depth and interpretive clarity. It ultimately argues for the incor-poration of Multiple Correspondence Analysis into future sentiment analysis methodologies, emphasizing its potential to contribute meaningfully to the study of international relations and the emotional undercurrents that shape global political discourse.