Vliv sentimentu a emocí na věkové rozložení doručení politické reklamy na sociálních sítích společnosti Meta

Political communication on social media is becoming an increasingly important part of election campaigns, and the key issue is not only the content of political messages, but also the users to whom these messages are actually delivered. This bachelor’s thesis examines whether the sentiment and emotions expressed in the texts of paid political posts on Meta platforms are related to the age structure of their delivery. The analysis is based on data obtained from the Meta Ad Library and includes paid posts published by selected Czech political actors between 2019 and 2025. The ad texts were first classified into sentiment and emotion categories using OpenAI’s GPT-5 nano model. Exploratory data analysis was then conducted, followed by modelling of age-specific delivery using fractional logit models. The results show that delivery varies systematically by age, even after accounting for age targeting and publisher differences. Positive texts are more frequently associated with younger audiences, while posts with a negative tone are more common among older age groups. The study thus provides a data-driven insight into the relationship between the sentiment and emotions of political ads and the age structure of their delivery.