🔮 Sentiment Analysis in Portuguese

This is a simple, but complete, application with back-end and front-end for an artificial intelligence model that analyzes and predicts the sentiment of a text written in Brazilian Portuguese.

BERT (Bidirectional Encoder Representations from Transformers) was used to create this model, transferring the learning through a dataset extracted from the Play Store. The training was carried out in the Colaboratory and the files generated are at /ai folder.

The model was put into production from an back-end API built using FastAPI and Uvicorn at /server folder. The front-end is a React/Next.js application built with the Context API for state management and the generated data is saved in local storage, which can be manipulated through the user interface and is located at /client folder.

This app was built with

How to start

To try this project just run this lines in your terminal:

git clone https://github.com/mkuchak/sentiment-analysis.git
cd sentiment-analysis
docker-compose up -d

So then you will be able to access the project on localhost:3000.

😃😐😡