Purpose. The study's main objective is to analyze publications on transmedia and narrative in the Scopus database, identifying the evolution of terms by year, to propose a classification according to the broad themes found. Methodology. We applied an unsupervised machine learning algorithm, called topic modeling, to identify the main themes addressed. Then, we apply open and axial coding from grounded theory to identify important concepts, aggregate them, and propose higher-order themes that can be used to classify the knowledge generated on transmedia and narrative. Results and conclusions. Five research themes are revealed, displayed in an evolutionary model: Convergence between platforms: games, television, and movies; Intermediality, authorship, and semiotic universes; Semiotic analysis and multimodality in transmedia narratives; Political and cultural interpretations in transmedia j...