Modelling TikTok video category transitions using Markov chains

Сажетак

In this paper, we develop a discrete-time Markov chain framework to model transitions between video content categories on TikTok based on observed sequential viewing behavior. Using a dataset of 65 consecutively viewed videos classified into six categories;motivational, dance, comedy, lifestyle, educational, and food, we estimate empirical transition probabilities and construct the associated transition matrix. We analyze the structural properties of the resulting Markov chain, establishing irreducibility, aperiodicity, and ergodicity, which guarantee the existence and uniqueness of a stationary distribution. The stationary distribution is computed explicitly and interpreted as the long-run exposure pattern across content categories under stable viewing dynamics. While the dataset is limited in size, the study is positioned as a methodological proof-of-concept demon-
strating how classical Markov chain theory can be applied to analyze category-level navigation on short-form video platforms. The proposed framework is transparent, reproducible, and readily extensible to larger datasets and more complex sequential models.