A carregar...
Publicação
Rating predictions from movie reviews leveraging BERT: exploring the impact of sentiment polarity and review length
| Nome: | Descrição: | Tamanho: | Formato: | |
|---|---|---|---|---|
| 1.4 MB | Adobe PDF |
Autores
Orientador(es)
Resumo(s)
This study is part of a broader research initiative investigating key factors influencing movie
success including box office performance, release strategies, and audience engagement. Within this
context, this study focuses on rating predictions from textual reviews, leveraging BERT to examine
the impact of sentiment polarity and review length. Using a large Rotten Tomatoes dataset of over
1 million reviews, fine-tuned regression and classification models reveal that sentiment polarity
enhances classification performance for extreme ratings, while review length has no significant
effect. These findings provide insights for improving rating prediction models and optimizing
audience feedback analysis in the film industry.
Descrição
Palavras-chave
BERT Movie reviews Rating pprediction Sentiment polarity Review length Sentiment analysis