Analisis Ulasan Pengguna Aplikasi Diagnosa Tanaman Di Play Store Menggunakan Naïve Bayes

Abstract

Recognizing plant diseases requires very deep literacy so novice farmers feel reluctant to study agriculture. Agriculture 4.0 has been implemented in several countries, so beginners to farming don't need to worry anymore about agriculture 4.0. The plant disease diagnostic application can be downloaded on the Google Play Store and many reviews and comments from users. With so many reviews from existing comments, it becomes difficult to process them manually, even though there are ratings. In general, ratings are not necessarily in accordance with the contents of user reviews. Therefore, it is necessary to process the results of user reviews and be able to see user tendencies towards the application. The method used is Naïve Bayes. For data labeling, an Indonesian language expert is required who is labeled manually based on Indonesian knowledge and KBBI. Labeling is Positive, Negative and Neutral. The dataset obtained as many as 252 reviews. From the average test, it gets an accuracy value of 79%. Meanwhile, the Precision value is 100% positive sentiment, 76% for neutral sentiment and 77% for negative sentiment. The Recall value for positive sentiment is 37%, neutral sentiment is 100% and for negative sentiment is 100%. And the F1-Score itself has a positive sentiment of 56%, a neutral sentiment of 85% and a negative sentiment of 87%.