The Comparison of LSTM Algorithms for Twitter User Sentiment Analysis on Hospital Services During the Covid-19 Pandemic
| Main Authors: | Rolangon, Anggreiny , Weku, Axcel , Sandag, Green Arther |
|---|---|
| Format: | Article info application/pdf eJournal |
| Bahasa: | ind |
| Terbitan: |
Fakultas Teknologi Informasi - Universitas Advent Indonesia
, 2023
|
| Subjects: | |
| Online Access: |
https://jurnal.unai.edu/index.php/teika/article/view/3063 https://jurnal.unai.edu/index.php/teika/article/view/3063/2187 |
Daftar Isi:
- Sentiment analysis has become a crucial aspect in understanding people’s opinions and emotions on various issues. In this study, we conducted sentiment analysis on tweets related to hospital services during the COVID-19 pandemic using LSTM, BiLSTM, GRU, and SimpleRNN models. The data collection process was carried out using the Twitter API and resulted in 15,093 tweets. The data preprocessing process includes data cleaning, case folding, tokenization, filtering, and stemming. The dataset was divided into 80% for training and 20% for testing. The results showed that the BiLSTM model had the highest accuracy of 86%, followed by the GRU model with an accuracy of 86%, the LSTM model with an accuracy of 85%, and the SimpleRNN model with an accuracy of 75%. The BiLSTM model also had the highest MCC of 71%. The study concludes that the BiLSTM model outperformed other models in predicting the sentiment of tweets related to hospital services during the COVID-19 pandemic. This study’s findings may have significant implications for healthcare providers in enhancing their services’ quality and improving patients’ satisfaction during pandemics.
- Analisis sentimen telah menjadi aspek penting dalam memahami pendapat dan emosi masyarakat tentang berbagai isu. Dalam penelitian ini, dilakukan analisis sentimen pada tweet terkait layanan rumah sakit selama pandemi COVID-19 menggunakan model LSTM, BiLSTM, GRU, dan SimpleRNN. Proses pengumpulan data dilakukan menggunakan Twitter API dan menghasilkan 15.093 tweet. Proses preprocessing data meliputi pembersihan data, case folding, tokenisasi, filtering, dan stemming. Dataset dibagi menjadi 80% untuk pelatihan dan 20% untuk pengujian. Hasilnya menunjukkan bahwa model BiLSTM memiliki akurasi tertinggi sebesar 86%, diikuti model GRU dengan akurasi 86%, model LSTM dengan akurasi 85%, dan model SimpleRNN dengan akurasi 75%. Model BiLSTM juga memiliki MCC tertinggi sebesar 71%. Penelitian ini menyimpulkan bahwa model BiLSTM lebih unggul dibandingkan model lain dalam memprediksi sentimen tweet terkait layanan rumah sakit selama pandemi COVID-19. Temuan penelitian ini dapat memiliki implikasi signifikan bagi penyedia layanan kesehatan dalam meningkatkan kualitas layanan dan meningkatkan kepuasan pasien selama pandemi.