This research aims to conduct sentiment analysis on reviews of the film "Dirty Vote" from various sources, such as social media, film review websites, and online forums, using a fine-tuned BERT model. This approach includes review data collection, data pre-processing, BERT model refinement, and model performance evaluation. The research results show that the BERT model achieves a high level of performance with accuracy, precision, recall, and F1-score exceeding the threshold of 0.8 on the validation dataset. Sentiment analysis from various sources revealed variations in public opinion toward the film “Dirty Vote,” with significant differences in sentiment expressed via social media such as Twitter and Facebook compared to reviews from dedicated websites or online forums. In addition, discussion analysis of sentiment findings revealed people's preferences for certain aspects of films, such as visual effects and music. Sentiment analysis findings revealed that visual effects and music received the highest ratings from the public, while the cast and director received lower ratings. This information can be used by filmmakers to improve unsatisfactory aspects in subsequent film production.
Copyrights © 2024