Predikce ukazatelů interakce čtenářů se zprávami pomocí strojového učení

In the digital environment, where a large number of news articles are published every day, understanding the factors influencing user engagement is crucial for publishers, digital platforms, and advertisers. The level of engagement, expressed for example through the number of comments, shares, or clicks, reflects not only the attractiveness of the content but also its emotional impact and relevance. Predicting this engagement based on textual features allows for estimating the potential success of an article even before its publication. At the same time, it helps publishers find a balance between creating compelling content and maintaining credibility and responsibility towards users. This bachelor’s thesis focuses on predicting user engagement with news using machine learning methods and natural language processing (NLP) techniques. The goal is to develop a model that can predict the expected level of interaction with an article based on textual elements, particularly headlines and teasers. The analysis is based on articles published on the iDNES.cz portal. The thesis combines a theoretical overview of current research with the practical implementation of a machine learning model. In addition to the model’s design and training, the work also evaluates its results and suggests possibilities for further development in the field of user engagement prediction.