https://github.com/lionello/tidalsocial

Recommendation engine for TIDAL playlists and mixes.
https://github.com/lionello/tidalsocial

ecmascript flask machine-learning nix nodejs recommender-system tidal typescript

Last synced: about 1 month ago
JSON representation

Recommendation engine for TIDAL playlists and mixes.

• Host: GitHub
• URL: https://github.com/lionello/tidalsocial
• Owner: lionello
• License: mit
• Created: 2021-02-27T04:21:39.000Z (almost 4 years ago)
• Default Branch: master
• Last Pushed: 2023-01-04T04:03:59.000Z (about 2 years ago)
• Last Synced: 2024-11-02T13:51:52.038Z (3 months ago)
• Topics: ecmascript, flask, machine-learning, nix, nodejs, recommender-system, tidal, typescript
• Language: Python
• Homepage:
• Size: 223 KB
• Stars: 1
• Watchers: 2
• Forks: 0
• Open Issues: 3
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
# TIdalSocial ![logo](https://github.com/lionello/TIdalSocial/blob/master/static/favicon-32x32.png?raw=true)

[![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](https://opensource.org/licenses/MIT)

[![TypeScript](https://img.shields.io/badge/%3C%2F%3E-TypeScript-%230074c1.svg)](https://www.typescriptlang.org/)

[![code style: prettier](https://img.shields.io/badge/code_style-prettier-f8bc45.svg)](https://github.com/prettier/prettier)

![GitHub package.json version](https://img.shields.io/github/package-json/v/lionello/tidalsocial)

![Docker Image Size (tag)](https://img.shields.io/docker/image-size/lionello/tidalsocial/latest)

**This project is not affiliated with TIDAL or ASPIRO AB.**

TIdalSocial is my attempt at creating a playlist recommendation engine for the awesome [TIDAL](https://tidal.com/) music streaming service.

TIDAL does not currently have a way to browse or search for shared playlists, but that doesn't stop us, the fans, from creating our own platform.

TIdalSocial uses the following libraries for the recommendations:

* [implicit](https://github.com/benfred/implicit/) for the Alternating Least Squares matrix decomposition

* [hnswlib](https://github.com/nmslib/hnswlib) for the Hierarchical Navigable Small Worlds approximate nearest neighbors

For more details about the inner working of the algorithm, please check the excelent blog post by @benfred on https://www.benfrederickson.com/matrix-factorization/

The frontend is written in HTML and [TypeScript](https://www.typescriptlang.org) and uses [Vue.js](https://vuejs.org) and is served by [ExpressJS](https://expressjs.com).

The backend is also written in TypeScript, runs on [NodeJS](https://nodejs.org/), but `POST`s requests to a local Python 3 [Flask](https://flask.palletsprojects.com/en/1.1.x/) server for the actual recommendations. I'm using ECMAScript Modules (ESM) to allow the generated JavaScript to work in the browser as well as on NodeJS.

## Development

Use `direnv` with the following `.envrc`:

```sh

use flake

export PATH=$PATH:$PWD/node_modules/.bin

layout python

```

Or, without direnv:

```sh

nix-shell --pure

npm install

source .venv/bin/activate # or whichever venv you want

# For some reason CMAKE ignores the CXX from mkShell and insists on using clang

CC=$CXX pip install -r requirements.txt

```

## Test

```sh

nix-shell

npm test

python -m unittest discover -s model

```

## Run

```sh

nix-shell

npm start

```

## Publish

```sh

npm version patch # or minor, or major

npm run docker:push

```