AdvaithNair / clickbait

Trending YouTube Video Recommender

Clickbait

Clickbait is a project where I will recommend trending YouTube videos to a user. I plan to use a bit of NLP to play around with content-based recommenders, microservices, Docker, and CI/CD.

Technology Overview

It utilizes the following technologies:

• Web
  • VueJS
  • NuxtJS
• Recommender
  • Python
  • Flask
  • SciKit Learn
• User
  • TypeScript
  • NodeJS + Express
  • PassportJS

All code in this repository is constructed using Vue, Python, and TypeScript with a simple microservice architecture. They are built using Docker, so you will need Docker to run Clickbait. It is also composed together using Nginx.

Prerequisites

Setup

DISCLAIMER: Make sure you're on Node v12.20.0 and Python 3.8+

Upgrade/Downgrade Node

npm i -g n

THEN

n 12.20.0

(You can skip the first step if you have n installed already)

Set Up Virtual Environment

cd recommender
virtualenv -p python3 env
source env/bin/activate
pip install -r requirements.txt

Download Data

Please download the USvideos.csv from this link. Rename this file to raw.csv and place it in /recommender/data

Preprocess Data

yarn preprocess

OR

Run preprocess.ipynb in recommender/data

Commands

Web

Run Web (Developer Mode)

yarn web

Recommender

Run Recommender

yarn recommender

Runs on localhost:5002

Stop Recommender

When you run the previous command, the last line (before the ✨ Done in xy.zzs. line) will be a docker_id.

docker stop [docker_id]

Server Maintainance

Run the following commands in the server directory

Install New Dependencies

source env/bin/activate
pip install [dependency]
pip freeze > requirements.txt

Install All Dependencies

pip install -r requirements.txt

Deactivate Python Environment (if active)

deactivate

NOTE: You probably want to run web and recommender in seperate terminals. They use hot reloading.

Contributors

• Advaith Nair

  • Full Stack Developer
  • Website