Analisis Sentimen Text Dengan Metode CNN Study KasusTempat Wisata Makassar

Abstract

This research aims to evaluate and determine the extent to which the CNN (Convolutional
Neural Network) method can produce accurate sentiment predictions for reviews of Makassar
tourist attractions. This sentiment analysis method uses review data collected from the Google
Maps platform. In this research, a preprocessing stage was carried out to clean the data, such
as cleaning, transform cases, tokenizing, stopwords and stemming. Next, the dataset was
divided into training data and test data with scenarios 90 : 10, 80 : 20 and 70 : 30 to train and
test the model with three review categories, namely positive, negative and neutral. The results
of sentiment analysis show that the CNN method has good abilities in predicting positive,
negative and neutral sentiment in reviews of Makassar tourist attractions. A high level of
accuracy at the training stage shows that the model is able to learn well from the dataset
provided. Although the accuracy rate at the validation stage was slightly lower, it still reached an
adequate figure, indicating that the model has quite good generalization ability in classifying
sentiment in these reviews. The highest accuracy results were obtained with Training Accuracy
which increased to obtain a training accuracy value of 95%, and Validation Accuracy obtained a
value of 73%.