Analisis Sentimen Publik Melalui Social Media X Menggunakan Long Short Term Memory Studi Kasus Film Dokumenter “Ice Cold: Murder, Coffee, and Jessica Wongso”

The death of Wayan Mirna Salihin also known as the cyanide coffee case, has once again attracted public attention following the release of a documentary film titled “Ice Cold: Murder, Coffee, and Jessica Wongso”. This study aims to analyze public sentiment reflected through the social media platform X (Twitter), identify the most dominant type of sentiment, and evaluate the performance of the Long Short-term Memory (LSTM) method in conducting sentiment analysis on the case. The approach used in this study involves the LSTM algoritm with feature extraction using Word2Vec to classy sentiment into five categories: very positive, positive, neutral, negative, and very negative. Data was collected through a crawling process of 5.070 tweets discussing the documentary film. After data cleaning process was carried out to remove duplicaates, 4.006 data were used in the analysis. The reasult of the study show neutral sentiment was the most dominant with 1.917 tweets (47.8%), followed by positive sentiment with 1.723 tweets (43%). These findings indicate that the majority of the public responded relatively objectively to the documentary film. An evaluation of the LSTM model’s performance revealed a positive correlation between model complexity, the number of epoch, and accuray levels. The best result were achieved with LSTM_Units 128 and 150 epoch, yielding an accuracy of 93%. These result demonstrate that the LSTM method is sufficiently effective for analyzing sentiment on social media.