Cyberbullying Classification Dataset

The Cyberbullying Classification Dataset is a labeled social media text dataset designed for natural language processing and machine learning projects. It contains more than 47,000 tweets that researchers can classify according to different types of cyberbullying.

The dataset uses two primary fields: tweet text and cyberbullying type. Therefore, developers can train classification models to identify the category associated with a social media post.

The dataset includes six categories: age, ethnicity, gender, religion, other cyberbullying, and not cyberbullying. Moreover, available documentation indicates that the classes contain roughly 8,000 examples each, supporting balanced multi-class classification experiments.

Key Features

• More than 47,000 labeled tweets

• Social media text suitable for NLP projects

• Two primary data fields: tweet text and cyberbullying type

• Six classification categories

• Includes both cyberbullying and non-cyberbullying content

• Suitable for multi-class text classification

• Useful for machine learning and deep learning

• Supports research into social media language patterns

Categories Included

The dataset classifies tweets into six categories:

Age

Tweets associated with cyberbullying based on age.

Ethnicity

Tweets containing cyberbullying related to ethnicity.

Gender

Tweets associated with gender-based cyberbullying.

Religion

Tweets classified as cyberbullying related to religion.

Other Cyberbullying

Cyberbullying content that does not fall into the specific age, ethnicity, gender, or religion categories.

Not Cyberbullying

Tweets that do not contain cyberbullying content.

These categories allow researchers to build models that distinguish between different forms of cyberbullying rather than simply detecting whether bullying exists.

Data Included in the Dataset

The dataset structure supports supervised learning because each text sample has an associated classification label.

Applications of the Cyberbullying Classification Dataset

Cyberbullying Detection

Train NLP models to identify potentially cyberbullying-related social media content.

Multi-Class Text Classification

Build models that classify tweets into different cyberbullying categories.

Natural Language Processing

Use the dataset for text preprocessing, tokenization, feature extraction, and NLP model development.

Social Media Analysis

Analyze language patterns associated with different types of cyberbullying.

Machine Learning

Experiment with algorithms such as logistic regression, support vector machines, random forests, and other text classification approaches.

Deep Learning

The dataset can also support neural-network-based NLP experiments, including transformer and recurrent neural network models.

Why Use the Cyberbullying Classification Dataset?

The dataset provides labeled social media text for a practical multi-class NLP problem. Instead of using only a binary bullying label, researchers can analyze several specific categories.

Moreover, the dataset contains a separate not_cyberbullying category. As a result, models can learn to distinguish cyberbullying-related content from ordinary social media posts.

Furthermore, the relatively balanced class distribution makes the dataset useful for comparing different classification approaches. However, researchers should still evaluate model performance using class-level metrics such as precision, recall, and F1-score rather than relying only on accuracy.

Who Can Use This Dataset?

The Cyberbullying Classification Dataset can be useful for:

• NLP researchers

• Machine learning developers

• Data scientists

• AI researchers

• Social media researchers

• Text classification practitioners

• Deep learning developers

• Students learning NLP

• Researchers studying online behavior

Conclusion

The Cyberbullying Classification Dataset provides more than 47,000 labeled tweets for studying cyberbullying through natural language processing and machine learning. With six classification categories, it supports both general detection and analysis of different types of cyberbullying.

Researchers and developers can use the dataset to build text classification models, analyze social media language, and experiment with NLP and deep learning approaches.

The dataset is available on Kaggle.