PrivaSEE is a new app that allows users to get privacy recommendations on other apps simply by uploading their terms & conditions agreements.
The app is split into three components: source control and data prep, machine learning and user interaction.
The machine learning components include a fine-tuned model to help identify and classify privacy issues within the annotated text data of the ToS;DR website.
The app lets users upload a PDF of a service’s terms and conditions and receive a list of identified privacy attributes with a weighted score adjusted by severity and importance by category.
Users can also ask for app recommendations, which filters results based on a variety of criteria and uses a weighted ranking system to find the top rated app.
The API and front-end for the app have been implemented using React, Javascript and a FastAPI to create the backend RESTful APIs that handle frontend communication
Deployment was automated on a GCP host using Ansible and Kubernetes were used to handle scalability.
The team would like to expand their app to support other types of agreements and reach a larger audience.