Institusion
Universitas Sriwijaya
Author
HAIRUNNISA, NADIA RIZKY
Abdiansah, Abdiansah
Yusliani, Novi
Subject
QA76.9.B45 Big data. Machine learning. Quantitative research. Metaheuristics.
Datestamp
2023-05-29 04:15:39
Abstract :
Sentiment analysis is a computational study of human opinions, sentiments, emotions, and behavior toward entities or attributes expressed through written text. Sentiment analysis plays a significant role for companies and organizations because public opinion about their products and services is valuable for business strategy and evaluation. This research developed a system to classify product review sentiment using Bidirectional Long Short-Term Memory (Bi-LSTM) algorithm and Word Embedding, Word2Vec as the embedding layer. The built system uses two models, the base model, which has the same parameter configuration as the CNN model used in previous research, and the tuned model, whose parameter configuration is based on the results of hyperparameter tuning. The results showed that the second model has the best performance with an accuracy of 90.33%, precision of 99.41%, recall of 90.29%, and F1-Score of 94.61%.