Text Mining Applications to Brands’ Social Media Messages Pre-, During and Post-Pandemic
(R2024-122611)

AUTHORS

Mustafa Cerit, Turk Alman Universitesi; Elif Karaosmanoglu, Istanbul Technical University

ABSTRACT

By applying text-mining techniques to tweets of companies in Turkey, we aim to examine and to contrast the alterations in the businesses' communication from before to after the Covid-19 pandemic. By using Structural Topic Model, we found that online transactions and communication campaigns emerged as prominent marketing-related topics. We broaden our investigation by analyzing the emotional patterns of the identified themes and found notable variations in emotional patterns across communication campaigns, depending on the time frame. However, we did not see any major variances in the discourses linked to online transactions. Furthermore, while trust, anticipation, and joy are consistently desired emotions in brand messages regardless of the situation, fear, sadness, anger, disgust, and surprise vary depending on the time of the pandemic.