Digital Voter Behavior: A Text Mining Approach to Analyzing Political Party Perceptions on X

Abstract

Social media provides an important source for analyzing public opinion and societal trends. This study examines sentiment toward five political parties in Türkiye using sentiment analysis of Turkish tweets. More than 60,000 tweets collected via the X API on May 11, 12, and 14, 2023, were used to select a balanced dataset of 1,250 tweets, with 250 tweets per party. Tweets were classified as positive, neutral, or negative and analyzed using Naive Bayes and Decision Tree algorithms. The results indicate similar classification performance between the two algorithms, while sentiment distributions varied across political parties and sentiment categories. The study contributes to research on the relationship between social media and politics by demonstrating the potential of sentiment analysis for evaluating political discourse.