ANALISIS SENTIMEN ULASAN APLIKASI CANVA MENGGUNAKAN BIDIRECTIONAL LONG SHORT-TERM MEMORY UNTUK MENINGKATKAN PENGALAMAN PENGGUNA

SENTIMENT ANALYSIS OF CANVA APP REVIEWS USING BIDIRECTIONAL LONG SHORT-TERM MEMORY TO IMPROVE USER EXPERIENCE

Abstrak

In the digital era, graphic design has become an essential skill in various fields, and the Canva app makes it easy to create creative designs online. However, user experiences with Canva vary and can be analyzed through reviews available on platforms such as the Google Play Store. This study aims to analyze the sentiment of Canva app user reviews using Long Short-Term Memory (LSTM) and Bidirectional LSTM (BiLSTM) algorithms, and compare the performance of both methods in classifying positive, negative, and neutral sentiment. A total of 10,000 review data sets were obtained through web scraping. After preprocessing, 5,514 data sets remained, divided into training and testing data sets with an 80:20 ratio. Classification models were built using both deep learning algorithms and evaluated using accuracy, precision, recall, and F1-score metrics. The results showed that BiLSTM performed slightly better with an accuracy of 92.38%, a precision of 93.05%, a recall of 90.39%, and an F1 score of 91.53%, compared to LSTM with an accuracy of 92.29%, a precision of 91.48%, a recall of 91.39%, and an F1 score of 91.44%. This study shows that BiLSTM is more effective in analyzing sentiment due to its ability to capture sentence context bidirectionally.