Free and Open-Source BERT Multilabel Classifier for Hate Speech - Accurately Identify Multiple Categories of Hate Speech

Hate Speech Multilabel Classification With Bert

Developed by wesleyacheng

A BERT-based multilabel hate speech classifier designed to identify categories of hate speech in text, including race, religion, origin, gender, sexual orientation, age, and disability.

Text Classification

Transformers

EnglishOpen Source License:Apache-2.0 #Multilabel Hate Speech Classification #Social Media Content Moderation #Data Perspectivism Application

Downloads 91

Release Time : 6/21/2023

Model Overview

This model is trained on the hate speech dataset from UC Berkeley's D-Lab, capable of performing multilabel classification on hate speech in social media comments.

Model Features

Multilabel Classification

Capable of identifying multiple categories of hate speech in text simultaneously, such as race, religion, origin, etc.

Data Perspectivism

The dataset includes annotators' background information, such as ideology, income, race, etc., helping to understand how annotators' perspectives influence hate speech judgments.

Transfer Learning

Utilizes BERT for transfer learning, leveraging pre-trained models to enhance classification performance.

Model Capabilities

Hate Speech Detection

Multilabel Text Classification

Social Media Comment Analysis

Use Cases

Social Media Content Moderation

Hate Speech Filtering

Automatically identifies and filters hate speech content on social media platforms.

Improves content moderation efficiency and reduces manual review workload.

Academic Research

Hate Speech Analysis

Used to study the distribution and characteristics of hate speech, aiding in understanding its social impact.

Provides data support for social psychology and communication studies.

🚀 Hate Speech MultiLabel Classifier

This project is a hate speech multi - label classifier that can classify independent targets such as race, religion, origin, gender, sexuality, age, and disability. It utilizes transfer learning on BERT with the UC Berkeley D - Lab's Hate Speech Dataset.

🚀 Quick Start

First posted on Kaggle.

✨ Features

Dataset Features

I've recently stumbled upon a very comprehensive dataset on measurement of hate speech from my alma mater, UC Berkeley. It aggregated social media comments from Youtube, Reddit, and Twitter. The interesting thing about the dataset is that the annotator's profile with attributes such as ideology, income, race, etc., is included in the dataset. This is, unfortunately, uncommon in most social media datasets and so I was intrigued about this.

Data Perspectivism

I also find the paper's idea of data perspectivism interesting as it argues that the disagreement of the annotator on the attributes of a hate speech is informative, rather than throwing it away.

Model Functionality

Here I made a Hate Speech MultiLabel Classifier to classify independent targets of race, religion, origin, gender, sexuality, age, disability by doing transfer learning on BERT with the UC Berkeley D - Lab's Hate Speech Dataset from the paper The Measuring Hate Speech Corpus: Leveraging Rasch Measurement Theory for Data Perspectivism.

📚 Documentation

Dataset

Property	Details
Model Type	Text Classification
Training Data	ucberkeley - dlab/measuring - hate - speech
Metrics	F1, Accuracy

Widget Examples

Example Title	Text
Race Hate Example	Yellow peril.
Religion Hate Example	Nietzsche said 'God is dead'.
Origin Hate Example	Go back to where you came from.
Gender Hate Example	You're being emotional.
Sexuality Hate Example	I identify as a sandwich.
Age Hate Example	Old fart.
Disability Hate Example	Confined to a wheelchair.

Suggestion

💡 Usage Tip

Although we are not leveraging the annotator's information in this model, I encourage you to explore the dataset and maybe leverage the annotator's information to make variations of this classifier.

📄 License

This project is licensed under the Apache 2.0 License.

Featured Recommended AI Models

Empowering the Future, Your AI Solution Knowledge Base

English 简体中文繁體中文にほんご