THE SENTIMENTAL FRAME OF THE MEDIA: ANALYZING THE NEWS HEADLINES ABOUT IMMIGRANTS AND REFUGEES IN ONLINE NEWS SOURCES THROUGH SENTIMENT ANALYSIS AND TOPIC MODELING

This research examines the representation of immigrants and refugees in Türkiye's online news media through published news headlines. A collection of headlines from four national news sources (TRT Haber, BirGün, Cumhuriyet, and Yeni Şafak) is examined using BERT-based sentiment analysis and BERTopic topic modeling. Negative feelings dominate news headlines across all media venues, while positive depictions are scarce. At the end of 2024 and mid-2025, time series analysis shows increased negativity. Topic modeling revealed border crossings, security, illegal immigration, mass deportation, and refugee discourses tied to national policy. Word cloud and co-occurring word pair studies show that "migrant caught", "irregular migrant", "detained”, "migrant smuggling" and "illegal" are frequently repeated. These discursive tendencies show that immigration is mostly framed as a security danger and control issue. The study uses computational social sciences to show that media content is both communicative and strategic, with social and political effects.