Good Netizens cyberbullying classification model API

This is the API for our machine learning model used to classify For FIT5120.

It uses FastAPI to serve the API, a logistic regression model (using scikit-learn) to classify text as cyberbullying or not, and an RNN (using Tensorflow) to further categorise the text as toxic, aggressive, or attacking.

Usage

To run this, you must have Docker and Docker Compose installed. You can find instructions on how to install these in the Docker docs.

First, clone the repository and cd into it:

$ git clone https://github.com/mattjamesdev/goodnetizensau-model-pipeline.git
$ cd goodnetizensau-model-pipeline

Now build and run the container using Docker Compose:

$ docker compose up

# Or if you want to run it in detached mode
$ docker compose up -d

If you make any changes to the files, make sure to add the --build flag to rebuild the image.

$ docker compose up --build

You can now access the API from http://localhost:8000/.

To see the FastAPI auto-generated docs, head to http://localhost:8000/docs.