Analysis of the Effect of Feature Extraction on Sentiment Analysis using BiLSTM: Monkeypox Case Study on X/Twitter

The monkeypox outbreak has again become a global concern due to its 
widespread spread in various countries. Information related to the disease is 
widely shared through social media, especially Twitter which is a major source 
of public opinion. However, the complexity of language and the diverse 
viewpoints of users often pose challenges in accurately analyzing sentiment. 
Therefore, sentiment analysis of tweets about monkeypox is important to 
understand public perception and its impact on the dissemination of health 
information. This research contributes to identifying the most effective word 
embedding-based feature extraction method for sentiment analysis of health 
issues on social media. The purpose of this study is to compare the 
performance of word embedding methods namely Word2Vec, GloVe, and 
FastText in sentiment analysis of tweets about monkeypox using the BiLSTM 
model. Data totaling 1511 tweets were collected through a crawling process 
using the Twitter API. After the data is collected, manual labeling is done into 
three sentiment categories, namely positive, negative, and neutral. 
Furthermore, the data is processed through a preprocessing stage which 
includes data cleaning, case folding, tokenization, stopword removal, and 
stemming. The evaluation results show that FastText with BiLSTM produces 
the highest accuracy of 90%, followed by Word2Vec at 89%, and GloVe at 
87%. FastText proved to be more effective in reducing classification errors, 
especially in distinguishing between negative and positive sentiments due to 
its ability to capture subword information and broader context. These findings 
suggest that the use of FastText can improve the accuracy of sentiment 
analysis, especially on health issues that develop on social media, so that it 
can support data-driven decision making by relevant parties in handling 
information dissemination.