Hate Speech Detection on Twitter API

Overview

This project provides a Dockerized Flask service for detecting hate speech in tweets using the cardiffnlp/twitter-roberta-base-hate-latest model. The API accepts a tweet in the request body and returns 0 for non-hate speech and 1 for hate speech.

Features

• Flask backend for serving the hate speech detection model.
• Preprocessing to clean and standardize tweets before analysis.
• Truncation Handling: Tweets longer than 512 tokens are automatically truncated.
• Docker Support: Easily deployable as a container.

Requirements

• Docker (for containerized execution)
• Alternatively, Python 3.9+ with dependencies installed

Installation & Setup

Note: Downloading dependencies can take up to 20 minutes, depending on your internet speed.

Docker Installation

1. Clone this repository:

   git clone https://github.com/wolffbe/twitter-hate-speech-api.git
   cd twitter-hate-speech-api
   

2. Build the Docker image:

   docker build -t twitter-hate-detector .
   

3. Run the container:

   docker run -p 8000:8000 twitter-hate-detector
   

Local Installation (without Docker)

1. Install dependencies:

   pip install -r requirements.txt
   

2. Start the Flask server:

   python main.py
   

API Usage

Endpoint: POST /detect

Request Body:

{
  "tweet": "Your tweet text here"
}

Response:

{
  "hate_speech": 0  // or 1 for hate speech
}

Example Requests

Hate Speech Example

An example of a hate speech detection request:

Linux/macOS:

curl -X POST http://localhost:8000/detect -H "Content-Type: application/json" -d '{"tweet": "Illegal immigrants steal!"}'

Windows Command Prompt:

curl -X POST http://localhost:8000/detect -H "Content-Type: application/json" -d "{\"tweet\":\"Illegal immigrants steal!\"}"

Expected Response

{
  "hate_speech": 1
}

Non-Hate Speech Example

An example of a non-hate speech detection request:

Linux/macOS:

curl -X POST http://localhost:8000/detect -H "Content-Type: application/json" -d '{"tweet": "I hope you have a wonderful day!"}'

Windows Command Prompt:

curl -X POST http://localhost:8000/detect -H "Content-Type: application/json" -d "{\"tweet\":\"I hope you have a wonderful day!\"}"

Expected Response

{
  "hate_speech": 0
}