One of the social media platforms that has seen the most success in Indonesia is Twitter. Because of the enormous number of users and the intensity with which it is utilized, Twitter may also be used to perform sentiment analysis to search for information relating to a product or topic. On Twitter, movie reviews are consistently one of the most popular topics of conversation. Because everyone's thoughts on movie reviews might refer to a variety of elements, aspect-based sentiment analysis is a method that can be used to discover the opinions that Twitter users have regarding movie reviews based on those characteristics. As a result, this method can produce more accurate and useful results. The research developed an aspect-based sentiment analysis system by using datasets of movie reviews written in Indonesian. These datasets consisted of three aspects: the plot, the acting, and the director. The classification model makes use of the Reccurent Neural Networks (RNN) approach, which extracts features using TF-IDF and expands on features using FastText. Unbalanced data is handled with SMOTE. The research found that plot elements had an accuracy of 77.24 percent and an F1 Score of 77.19 percent, while acting aspects had an accuracy of 96.59 percent and an F1 Score of 96.59 percent, and director aspects had a 97.75 percent accuracy and an F1 Score of 97.74 percent.