Comparison Of Machine Learning Algorithms On Sentiment Analysis Of Elsagate Content

This study responds to the increasing phenomenon of elsagate content on various platforms, especially on YouTube Kids and YouTube, which are often accessed by children. Elsagate content contains sensitive elements for children such as horror, sexuality, and violence. The community’s response to this content varies, from positive to negative, to neutral, especially on platforms like YouTube. The main purpose of this research is to understand the opinions of the YouTube community regarding children’s content with unclear meanings or containing elsagate elements. Using 2452 data, this study applies five machine learning algorithms to classify sentiment: Naive Bayes Classifier (NBC), Random Forest (RF), Support Vector Machine (SVM), K-Nearest Neighbors (K-NN), and Logistic Regression (LR). The research results show that data division with a 90:10 ratio provides the best performance. The Support Vector Machine algorithm achieves the highest accuracy of 63%, with precision of 61%, recall of 63%, and an F1-score of 60%. On the other hand, the K-Nearest Neighbors algorithm shows the lowest performance with an accuracy of 56%, precision of 55%, recall of 56%, and an F1-score of 55%. Thus, besides aiming to provide insights into elsagate, this research also highlights the performance of Support Vector Machine in analyzing sentiment towards elsagate content.