PENGGUNAAN WORD EMBEDDING WORD2VEC DALAM PENGEMBANGAN MODEL CNN STUDY KASUS ANALISIS SENTIMEN TEMPAT WISATA MAKASSAR

Abstract

This research aims to evaluate the effect of applying the Word Embedding Word2Vec technique on the accuracy of the Convolutional Neural Network (CNN) model in sentiment analysis of tourist attraction reviews in Makassar. Sentiment analysis is the process of identifying and classifying emotions or opinions contained in text, whether positive, negative, or neutral. The research dataset consists of 4500 tourist attraction reviews taken from Google Maps. The data was then processed using the Word2Vec technique to generate vector representations of the words in the reviews. These vectors were used as input to the CNN model for sentiment classification. The study employed three data splitting scenarios, namely 90:10, 80:20, and 70:30, for training and testing the model. The results showed that the application of Word2Vec in the CNN model improved sentiment prediction accuracy. The CNN model with Word2Vec achieved an accuracy of 79%, while the CNN model without Word2Vec only reached an accuracy of 74%. This indicates that the use of Word2Vec can enhance the performance of the model in classifying sentiment in tourist attraction reviews.