IMPLEMENTASI BERT UNTUK ANALISIS SENTIMEN KOMENTAR PUBLIK TERHADAP MULTI LANE FREE FLOW (MLFF) GERBANG TOL

The implementation of MLFF still raises various opinions among the public. This research aims to analyze the sentiment of public comments regarding toll gate MLFF using the fine-tuned Bidirectional Encoder Representations from Transformers (BERT) model. BERT is used because of its high ability to understand sentence context and produce accurate predictions in Natural Language Processing (NLP) tasks. The research results show that the fine-tuned BERT model is able to classify public comment sentiment with high accuracy. Factors that influence public sentiment towards MLFF include aspects of comfort, security and efficiency. These findings provide insight for policy makers to improve the socialization and implementation of MLFF so that it is more accepted by the community. This research contributes to the development of a sentiment analysis model using BERT and provides an overview of public perception of MLFF technology in Indonesia with an accuracy rate of 83%, 75% macro avg and 83% and weighted avg. In addition, it is hoped that this research can become a reference for studies next in the same field.
Keywords: Sentiment Analysis, BERT, Multi Lane Free Flow, Toll Gate