Afidah, Liya Nur (2024) Klasifikasi Cyber Harassment Pada Media Sosial Twitter Menggunakan Metode Logistic Regression. Other thesis, Universitas Islam Riau.

Text Liya Nur Afidah.pdf - Submitted Version Restricted to Registered users only Download (3MB) | Request a copy

Abstract

Cyber Harassment (online harassment) is defined as intentional and repeated behavior utilizing technology such as social media, including Twitter. Twitter is one of the social media that is widely used in Indonesia. However, there are still many people who use it incorrectly, so that cases of Cyber Harassment occur on Twitter social media. This study aims to compare the performance of the model in classifying the types of Cyber Harassment on Twitter social media with the Logistic Regression method compared to the K-Nearest Neighbor (KNN) and Naïve Bayes methods. Data is taken from various tweets that contain elements of Cyber Harassment. The data used in this research was 2500 Indonesian language tweet data collected using Tweet-Harvest. The types of Cyber Harassment consist of Physical Threats, Purposeful Embarrassment, Racist, Sexual Harassment, and Neutral. Based on the research results, the success rate for Cyber Harassment classification on Twitter social media using K-fold Cross Validation with the Logistic Regression method resulted in an accuracy of 83%, precision of 82%, recall of 83% and f1-score of 82%. With the KNN method, results obtained were 76% accuracy, 76% precision, 76% recall and 76% f1-score, while with the Naïve Bayes method, 79% accuracy, 78% precision, 79% recall and 78% f1-score were obtained. The research results show that the best model performance for classifying types of Cyber Harassment in this research is the Logistic Regression method because it produces the highest accuracy, precision, recall and f1-score values.

Actions (login required)

View Item