City-level User Location Detection from Social Media and Analysis of Political Participation

The social media platform X (Twitter), regularly used by hundreds of thousands of users in Türkiye, serves as a space where users share content, express opinions on current issues, and interact with each other. Some of the agenda on social media reflects events that occurred on the same day, allowing users to follow closely events happening in their vicinity through the platform. The locations of these users in different cities of Türkiye can be inferred from the metadata of the tweets, but there is currently no system specifically designed for Türkiye. We first present an approach to automatically infer the user locations at the city-level. Subsequently, the validation of these inferences examined by analyzing volume of activities obtained when a politician visiting a particular city. Finally, the relationship between voting percentages at the election and changes observed in the quantity of social media posts is investigated.