ANALISIS SENTIMEN MENGGUNAKAN ARSITEKTUR LONG SHORT-TERM MEMORY (LSTM) TERHADAP FENOMENA CITAYAM FASHION WEEK

Abstract

Sentiment analysis of the text aims to recognize whether a text contains positive, negative, or neutral emotions. The results of the analysis can be used as a tool for making decisions on an issue. Recently, the Citayam Fashion Week event become an issue that is extremely debated in Indonesia, especially in July 2022 on social media. The issue has motivated us to do sentiment analysis for better making decisions. In our work, the dataset is collected from Indonesian people's tweets with the keywords Citayam Fashion Week. Furthermore, each tweet will be labeled with a positive, negative, or neutral class based on the Indonesian lexical. This research produces a model based on Long Short Term Memory (LSTM) structure to predict every Indonesian tweet into the category of positive, negative, or neutral sentiment related to public views and opinions about the Citayam Fashion Week phenomenon. The model accuracy shows that the LSTM obtained good performance which is 88%.