APPLICATION OF SOCIOPHYSICS FOR SENTIMENT ANALYSIS REGARDING THE USE OF QRIS AS A DIGITAL PAYMENT INSTRUMENT USING NAÏVE-BAYES AND MAXIMUM ENTROPY METHODS

• 主要作者: Dhiya Ulhaq Mulia, Syakura
• 格式: Final Project
• 語言: Indonesia
• 在線閱讀: https://digilib.itb.ac.id/gdl/view/81440
• 機構: Institut Teknologi Bandung

Sociophysics is a cross- disciplinary field that uses physics methods to understand human behaviour. Sociophysics can be used to conduct sentiment analysis on social phenomena with social media as the data. Sentiment analysis is a process of processing data such as text to extract sentiment information that is contained in it. In this research, sentiment analysis is carried out with a classification process using two machine learning algorithms, which are Naïve-Bayes and Maximum Entropy. The research was conducted to produce a representation of Twitter sentiment data (X) related to the use of QRIS as a digital payment tool before (period I) and after (period II) the imposition of the 0.3% MDR for micro merchants, and compare the accuracy of the two methods used. The analysis process is conducted by data crawling, data processing, labelling, weighting, and sentiment classification. Based on the classification results, it is found that QRIS transactions have more negative sentiment for both periods. Meanwhile, based on the model evaluation results, Maximum Entropy has a higher accuracy value of 81.82% in period I and 71.96% in period II, while Naïve-Bayes has an accuracy of 74.13% for period I and 69.16% for period II. Then, by varying the amount of training data, it is obtained that there is an increase that is not significant enough and tends to be constant in the accuracy value when the amount of training data is added.