About the Project

In this project I used Django as backend and Django REST Framework to build the RESTful API to analyse the sentiment of a text. The API endpoint is /analyze which takes the following data as POST request payload:

{
  "text": "Simple text"
}

And return the output after analyse the sentiment of that text as following:

{
  "sentiment": "positive/negative/neutral"
}

The API can be tested at docs/ endpoint which will load the Swagger Docs.

About the Model

The pre-trained model used in this project to analyze the sentiment of a sentence/text cardiffnlp/twitter-roberta-base-sentiment-latest which can be found here.

This is Twitter-roBERTa-base for Sentiment Analysis model which updated at 2022 and trained on ~124M tweets from January 2018 to December 2021, and fine-tuned for sentiment analysis with the TweetEval benchmark.

Project Setup

The project can be setup and run by 2 ways:

• By using Docker
• Manual Setup

Using Docker

This is the easiest way to set up and run the project. For this, the Docker and docker compose plugin are need to be installed in your local computer. The installation instructions can be found here.

After that, use the following commands to run the project:

1. git clone https://github.com/shazolKh/Sentiment-Analysis.git
2. cd Sentiment-Analysis/
3. cp .env.example .env and put some random text as the value of SECRET_KEY at .env file.
4. docker compose up --build

The last command will take some time to download the model and install the required dependencies to run the project. The project will be available at localhost:8000 or http://127.0.0.1:8000. After that, the API can be tested with API testing tool or at docs/ endpoint with Swagger.

Manual Setup

Assuming you have completed the necessary setup to perform python related commands. And also assuming that you are using a linux computer or you are using WSL on windows. Use the following commands to run the project. Noted that, at least, python 3.8 is needed.

1. apt update
2. apt-get install git-lfs
3. git lfs install
4. git clone https://github.com/shazolKh/Sentiment-Analysis.git
5. cd Sentiment-Analysis/
6. python -m venv venv or python3 -m venv venv
7. Activate the virtual environment.
8. pip install -r requirements.txt
9. Change the working directory to model/ folder.
10. git lfs clone https://huggingface.co/cardiffnlp/twitter-roberta-base-sentiment-latest
11. Change the working directory back to Sentiment-Analysis/ i.e. project root folder.
12. cp .env.example .env and put some random text as the value of SECRET_KEY at .env file.
13. python manage.py migrate
14. python manage.py runserver

The project will be available at http://127.0.0.1:8000. And the API can be tested with API testing tool or at docs/ endpoint with Swagger.