Combating Cyberbullying on Social Media - A Machine Learning Approach with Text Analysis on Twitter

Aug 4, 2026

Upload by Rowan

The work addresses the rapid growth of cyberbullying, toxicity, and harassment on social media and its harmful effects on individuals and communities, motivating the need for automatic detection. It applies active machine learning models—Logistic Regression, Multinomial Naive Bayes, K-Nearest Neighbour, and Extreme Gradient Boosting—to Twitter textual data. Using Bag of Words and TF-IDF feature extraction, the study converts text to numerical representations and evaluates performance, reporting the strongest results with XGBoost and Logistic Regression.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages6

Reading Time15 min

AccessFree to download

Introduction

Related Works

Data Sets and Preparation

Feature Extraction Methods

Model Implementation (Traditional Machine Learning)

Conclusion

University of Central Florida STARS Data Science and Data Mining Fall 2024 Combating Cyberbullying on Social Media: A Machine Learning Approach with Text Analysis on Twitter Amir Alipour Yengejeh University of Central Florida, [amir.alipouryengejeh@ucf.edu](amir.alipouryengejeh@ucf.edu) Part of the Data Science Commons Find similar works at: [https://stars.library.ucf.edu/data-science-mining](https://stars.library.ucf.edu/data-science-mining) University of Central Florida Libraries [http://library.ucf.edu](http://library.ucf.edu) This Article is brought to you for free and open access by STARS. It has been accepted for inclusion in Data Science and Data Mining by an authorized administrator of STARS. For more information, please [contact STARS@ucf.edu](contact STARS@ucf.edu). STARS Citation Alipour Yengejeh, Amir, "Combating Cyberbullying on Social Media: A Machine Learning Approach with Text Analysis on Twitter" (2024) . Data Science and Data Mining. 15. [https://stars.library.ucf.edu/data-science-mining/15](https://stars.library.ucf.edu/data-science-mining/15) Combating Cyberbullying on Social Media: A Machine Learning Approach with Text Analysis on Twitter Amir Alipour Yengejeh dept. Statistics and Data Science University of Central Florida Orlando, United States [amir.alipouryengejeh@ucf.edu](amir.alipouryengejeh@ucf.edu) Abstract—The popularity of the electronic mobile devices along with social media as well as networking websites have been tremendously increased in the recent year. Most people around the world daily engage in the variety of cyberspace additives. Even though the users can take most advantages of these system such as exchange the idea and information, being sociable, and enjoyments, they might be faced with such adverse behaviours such as toxicity, bullying, extremism, and cruelty. The recent statistics reports that such mentioned behaviours has been noticeably grown on the cyberspace such that can threaten the individuals and even any community. Thus, it is drastically demand to invent a device to detect cyberbullying automatically. To do so, most studies are using the idea of the classifcation and then machine learning algorithms to build such a device. In this study, therefore, we employed some active machine learning models like Logistic Regression(LR), Multinomial Naive Bayes(MNB), K-Nearest Neighbour (KNN), and Extreme Gradient Boosting(XGbost) on Twitter textual dataset to detect the quality of cyberbullying related to ethnicity and religion. Since the data is contextual, we used some feature extraction techniques like Bag of Words and TFIDF to convert the texts into numerical sets. According the computational results, we saw XGbost and LR achieves the highest performance. Index Terms—cyberbullying, social media, machine learning, classifcation, feature extraction I. INTRODUCTION Bullying is considered as an intentional and most frequent adverse and aggressive behavior can be conducted by an individual or group of people to attack and insult another one as a victim [1] [2] . There are various types of bullying like verbally, physically, psychologically etc. One evolution form of the bullying is called cyberbullying can be carried out on a cyberspace [2] . Recently, the prevalence of cyberbullying has been a complicated and signifcant phenomenon in the online social medias like Twitter, Facebook, Instagram, and to name but a few. This is because that statistics shows that cyberbullying can lead to adverse consequences such as anxiety, depression, harassment, and even suicide among the individuals in particular youths. Therefore, researches has been conducted to cope with this horrible incident in these platforms so much so that to control users’ activities and detect bullying or harassment language automatically. To do so, studies try to employ statistical and computational algorithms such machine learning. To apply these methods, however, the collec

FAQs

What is the main goal of the study?

The study aims to detect cyberbullying automatically from Twitter text and compare machine learning approaches for performance.

Which machine learning models are used?

It evaluates Logistic Regression, Multinomial Naive Bayes, K-Nearest Neighbour, and Extreme Gradient Boosting on a Twitter textual dataset.

How are tweets converted into model inputs?

Tweets are transformed into numerical features using feature extraction methods such as Bag of Words and TF-IDF, enabling traditional classifiers to learn patterns from text.

Aug 4, 2026

Upload by Rowan