PERFORMANCE COMPARISON OF SENTIMENT ANALYSIS ON NATIONAL MONUMENT (MONAS) REVIEWS USING BERT, SUPPORT VECTOR MACHINE, LOGISTIC REGRESSION, AND MULTINOMIAL NAÏVE BAYES

Riki Daniel Tanebeth — Thu, 23 Apr 2026 03:06:25 +0000

The National Monument (Monas), as an icon of Indonesian tourism, faces challenges in maintaining visitor satisfaction in the digital era. Online reviews on Google Maps serve as a crucial data source for understanding public perception. However, the large volume of data and the informal nature of review language hinder manual analysis. This study aims to analyze Monas visitor sentiment and compare the performance of conventional Machine Learning models with modern Deep Learning approaches. The method involves comparing the Multinomial Naïve Bayes algorithm using TF-IDF feature extraction against the IndoBERT (Bidirectional Encoder Representations from Transformers) model based on fine-tuning. The dataset consists of 1,110 visitor reviews from the 2023-2024 period. Experimental results demonstrate that the IndoBERT model significantly outperforms Naïve Bayes, achieving an accuracy of 93.5% and an F1-Score of 93.0%, while Naïve Bayes only reached 49.1% accuracy. Further aspect-based analysis reveals that although positive sentiment is dominant (49%), there are critical complaints regarding the digital ticketing system and elevator queues. This study recommends the implementation of transformer-based models for analyzing Indonesian tourism reviews and suggests improvements in queue management for Monas management.

RICE PRICE PREDICTION IN EAST SUMBA REGENCY USING THE NEURAL NETWORK ALGORITHM

Renol Bulu Manggal, Arini Aha Pekuwali, Raynesta Mikaela Indri Malo — Fri, 24 Apr 2026 01:29:47 +0000

Fluctuations in rice prices in East Sumba Regency are an important issue that directly affects farmers, traders, and consumers. Unstable price changes are influenced by weather conditions, supply availability, distribution, and market dynamics. Therefore, a prediction method is needed that can provide accurate estimates of rice prices as a basis for decision making. This study aims to predict the price of medium rice in East Sumba Regency using the Neural Network algorithm, specifically Long Short-Term Memory (LSTM), which is effective in modeling time series data. The data used are monthly rice price data for the period January 2021 to December 2025 obtained from Perum BULOG Waingapu Branch Office, with data processing and analysis carried out after all 2025 data became available. The research stages include data collection, data preprocessing, normalization using Min-Max Scaling, time series dataset formation, division of training and testing data, LSTM model training, and model performance evaluation. The evaluation was carried out using the Root Mean Square Error (RMSE) metric. The results show that the LSTM model is able to predict rice prices with an RMSE value of 360.91 Rp/Kg or around 3.35% of the average rice price. This value indicates that the prediction error of the model is relatively small, so the model can be said to have good prediction performance. Therefore, the developed LSTM model is considered feasible to be used as a tool for predicting rice prices and is expected to help farmers and traders in planning sales and become a consideration for the local government in maintaining rice price stability in East Sumba Regency.

J-Icon : Jurnal Komputer dan Informatika

PERFORMANCE COMPARISON OF SENTIMENT ANALYSIS ON NATIONAL MONUMENT (MONAS) REVIEWS USING BERT, SUPPORT VECTOR MACHINE, LOGISTIC REGRESSION, AND MULTINOMIAL NAÏVE BAYES

RICE PRICE PREDICTION IN EAST SUMBA REGENCY USING THE NEURAL NETWORK ALGORITHM