Comparison of Naive Bayes Multinomial Algorithm and Decision Tree on Tweet Emotion Classification

Abstract

Social media has become a place where everyone can express their feelings and thoughts without limitations. One of the most widely used social media platforms is Twitter. Twitter has 238 million active users and provides access to search for information through specific tweets. Therefore, Twitter can be used as a source of information to analyze a person's emotions based on their writings/tweets. In analyzing the feelings of a tweet, a method is needed to classify tweets into appropriate emotion classes. The classification of tweet emotions aims to group tweets into predetermined emotion classes such as anger, joy, fear, love, and sadness. The algorithms used to build machine learning models for emotion classification are Multinomial Naive Bayes and Decision Trees. This research aims to determine which classification algorithm, Multinomial Naive Bayes or Decision Tree, is better by comparing the accuracy values of these classification algorithms. This study applies feature extraction using Bag of Words, Bigram and the SMOTE method. The research results show that the Multinomial Naive Bayes classification model, which involves feature extraction using Bag of Words and the SMOTE method, has the highest accuracy value of 67.15%.