(MEF Üniversitesi, Fen Bilimleri Enstitüsü, 2017) Ceran, Serkan; Akpınar, Ezgit
During the last years, the importance of social media is increasing in an amazing way. In this paper, we looked at one such popular microblog platform called Twitter and build models for classifying “tweets” into some specific emotion. We used Turkey’s twitter data in order to explore the change in emotions over time using sentiment analysis. Using LIWC dictionary database, we conducted an emotion analysis of approximately 2.2 million tweets. We tracked how emotions evolve over time based on the prominent events in and or related to Turkey. Our results showed that there is a significant relationship between emotions and prominent events. We also analyzed the correlation between these emotions and the dollar exchange and made a predictive modeling experiment.