Análise de emoções em transcrições de Adventure Time.

Abstract

Cartoons are among the most consumed forms of entertainment media worldwide. One of the main reasons is the fact that they do not have as many physical limitations as traditional cinema, allowing greater freedom in the themes explored and the construction of scenes. One of the most popular animations in recent times, Adventure Time (originally aired on the Cartoon Network), is a perfect example of this. Throughout its nearly three hundred episodes, the children's show addresses a variety of themes and adopts various narrative styles. However, when analyzing the storyline as a whole, this large number of episodes can make it challenging to keep up with the emotional complexity of numerous characters, which can be a hindrance when trying to understand the depth of the plot. With the assistance of Transformer models and the Hugging Face platform, this work investigates na approach of emotion detection on the transcriptions of the show’s characters’ lines. The results demonstrate that the approach is feasible for this type of analysis and identify the broad emotional range in this animated series.