The analysis of opinions on social networks is a fundamental task to understand perceptions and trends within online communities. This work presents an automated methodology based on transformers, a deep learning architecture that has proven highly effective in natural language processing (NLP) tasks. Automation using transformers allows for faster and more accurate analysis of large volumes of textual data, providing valuable insights for decision-making. The main challenges are discussed, and experimental results are presented to demonstrate the effectiveness of the proposed approach.