Spotify is the largest on-demand music services in the world and has 299 million monthly users, including 138 million paying subscribers. One of the exciting features of the Spotify is Discover Weekly. It’s a custom mixtape of 30 songs they’ve never listened to before but will probably love, and it’s pretty much magic.

History of Music Curation

Songza used to curate playlists manually in the early 2000s which was not able to fulfil the music needs of the consumers as it was developed as on specific interest of the curator.

Pandora used tagging the songs manually with attributes like ‘folk’, ‘rock’ and curating playlist with similar tags using the code.

Spotify doesn’t use any single revolutionary recommendation model. Instead, they use three main types of recommendation models:

Collaborative filtering: It works by comparing people with similar tastes. If the user 1 used the product 1,2,3,4 and user 3 used 2,3. Then, the product 1,4 will be recommended to the user 3.

Netflix made collaborative filtering popular when they employed it for the recommendation system. They used the movie rating system to determine what movies to recommend to other similar users. But, Spotify doesn’t use it this way. It uses the stream count of the tracks and other data points like user added the song to the playlist or went to the artist page.

#machine-learning #big-data #spotify