Abstract
The study proposes a new model for quantitative and qualitative assessment of clickbait aimed at identifying hidden manipulative strategies in the headlines of Kazakhstani online media. The work seeks to create a tool for systematic assessment of manipulative practices in media texts. The scientific significance lies in the creation of a clickbait depth scale that combines a binary system for assessing features and a weighted interpretation of the manipulation level. The practical value lies in the possibility of further automation of the model using natural language processing (NLP) technologies. The methodology includes content analysis of 120 headlines from four Kazakhstani media resources selected by key news headings. The generative model of artificial intelligence PerplexityAI was used as an additional source of alternative analysis. The method enabled to test the assessment model in a semi-automatic mode with full manual rechecking of the interpretation results. The analysis showed differences in the use of clickbait strategies between tabloid, state and independent publications. The most common signs of clickbait and their dependence on genre and heading characteristics of news texts are identified. The work proposes an approach to assessing manipulativeness, allowing to record not only the presence of clickbait, but also its degree. The study contributes to the field of media linguistics and digital journalism, offering a conceptual and technological basis for the analysis of manipulative strategies in media texts. The results can be used in the development of automatic monitoring systems for the quality of news content, as well as in educational and editorial practice.