Fanaticism Category Generation Using Tree-Based Machine Learning Method

Abstract

This research has produced categorization of political fanaticism on social media. The results of this study divide political fanaticism into 4 categories: positive, neutral, negative and very negative. The background of the research is that currently social media cannot recognize the elements of fanaticism and does not care about the spread of negative fanaticism. One of the best solutions to prevent the spread of negative fanaticism on social media is to develop a computer system which is able to recognize, categorize fanaticism and reject the entry of negative fanaticism. This research extracts features of sentiment, emotions, and expressions of hatred in opinions on social media. The results of the feature extraction are then processed into knowledge using tree-based machine learning methods, namely Random Forest and ID3. The validation result shows that the knowledge generated by the Random Forest process outperforms other techniques with an accuracy of 91.82%. Knowledge from the Random Forest can be used to categorize fanatical texts based on sentiment, emotion and hatred in further research.