- System based on distance between users (distances: cosine, TBA)
- Implicit ALS recommendation (only x86 arch., cold start with high computation cost)
- Non-negative Matrix Factorization (NMF) recommendation
Training data is converted from ratings to binary with rating threshold.
Download and unpack data:
kaggle datasets download -d hernan4444/anime-recommendation-database-2020
unzip anime-recommendation-database-2020.zip
Start server with command: python server.py < server.txt
Interfaces: HTTP requests, Android application.
Example of HTTP request with curl: curl -X POST -d 'similarity;Girls & Panzer;Shaman King;Death Note' localhost:8080
Copy server/anime_titles.txt
(generated after server start) to {ANDROID_APP_PREFIX}/res/raw
for anime title suggestions inside Android application
Security and network warning for Android application: server address is set up to localhost with security check bypass, do not use server and application in production