This dissertation consists of three chapters, aiming to analyze, understand and discuss how Turkish public opinion fluctuates on social media based on governmental actions and how that fluctuation affects the society and politics in Turkey. Using textual data from social media, I combined natural language processing techniques with statistical methods, to study how Turkish public opinion is shaped by governmental actions in various scenarios. In the first chapter, I created a social network of Twitter users to detect the differences in the extent of political polarization between pro-government and opposition voters during the June 2019 Istanbul mayoral election. The second chapter focuses on the stigmatization of a social/religious group in Turkey by government-driven labeling and terrorism designation. Word embeddings are used to pinpoint the offensive language and the hate campaign against the group, considering the labels that are used to identify the group. Finally, the third chapter examines the rally-around-the-flag effect during highly inciting moments like cross-border military operations. A corpus of tweets for each of the two Turkish cross-border military operations is analyzed using topic modeling and sentiment analysis to get a grasp of the rally effect and how the governments can benefit in internal matters from the changes in public discourse because of this rally effect.

No description available.