<![CDATA[Sentiment analysis has gained significant attention in recent years, with researchers employing various machine learning and deep learning techniques. While Convolutional Neural Networks (CNN) were initially designed for image processing, their effectiveness in text classification tasks has been established. This study focuses on enhancing sentiment classification performance for customer reviews of MyPertamina by utilizing a CNN model with Word2Vec embeddings. The research involves analyzing user reviews obtained from the Google Play Store for MyPertamina's mobile application. The CNN model, incorporating Word2Vec embeddings, is trained to classify these reviews into positive and negative sentiment categories. Experimental evaluations reveal that the optimal hyperparameters for the Word2Vec model are a window size of 5 and a word embedding dimension of 300. Regarding the CNN model, both the dropout rate and learning rate significantly impact classification performance. The best results are achieved with a learning rate of 0.001 and a dropout rate of 0.3. The findings demonstrate that the CNN model with Word2Vec embeddings achieves an impressive accuracy of 96.14% in classifying customer reviews within the MyPertamina application. This underscores the efficacy of employing this approach to improve sentiment classification for customer feedback. ]]>
Copyrights © 2025
