With the era of digitalization, movie-watching has gained immense popularity, with platforms like Disney+ offering easy access to a variety of films. After watching, users frequently share their opinions on social media platforms such as Twitter, because of it is freedom of expression. With numerous movies available, users frequently encounter challenges in deciding what to watch. To address this, a recommendation system is proposed to streamline the decision-making process for users. Collaborative Filtering (CF), Content-Based Filtering (CBF), and Hybrid Filtering are common techniques used in recommendation systems. However, CF and CBF techniques face issues like cold start, sparse data, and overspecialization. To overcome these, this research constructs a Hybrid Filtering recommendation system, with a weight-based of CF-CBF coupled with Long Short-Term Memory (LSTM) classification. The classification uses various optimizers, including Adam, SGD, Nadam, RMSprop, and Adamax. Dataset is sourced from Kaggle website, which includes movie-related tweets linked to the Disney+ platform. The results indicate that Weight-Based Hybrid Filtering utilizing Adamax optimizer in LSTM classification yields superior performance metrics, by having 78% Precision, 79% Recall, 79% Accuracy, and 77% F1-Score value.