Integração de Dados de Audiência de TV Linear e Streaming: Um Modelo Computacional para Medição de Engajamento

Abstract

This research aims to investigate the integration and comparison of audience data from linear television, streaming services, and the emerging context of TV 3.0, proposing a machine learning model to interpret audience engagement. To this end, strategies for cross-domain compatibility are analyzed, transforming streaming data into structural representations compatible with aggregated linear TV data. Results show that knowledge transfer is sensitive to differences in data observation regimes, producing partial gains when structural compatibility exists, but also revealing limitations when target variables differ in nature. These findings delimit the conditions under which integra- tion between streaming and linear TV data is methodologically plausible, contributing to the development of integrated audience measurement approaches in hybrid media environments.