Current information of our Google Nest Mini Offer can be found here.

Logic behind recommended songs


Logic behind recommended songs


As an artist I am interested in the logic behind recommended songs that is generated for each playlist. Reason I am asking is because of course as an artist I would love to find my own tunes in situations where the playlist (in my opinion) contains so similar tunes that my own tracks would be excellent match. So far I've noticed that the recommended tunes tend to range only a handful of artists depending on the situation.


Is it totally based on listener actions or is it based on for example data or such? How often is it recalculated? One of my main artists have gained quite a bit of popularity and has even surpassed some of my earlier "idols" so to speak, all of whom are well represented in the recommended songs, however even now I can't seem to find my artist tunes there, even when the actual listening stats should show that my artist is "more popular" than the ones being part of the recommendations. Hence it would feel right if at some point the recommendation logic would realise that there is a new popular artist out there that should belong to this recommendation cycle.

1 Reply

Re: Logic behind recommended songs

Not applicable

3 Types Of Recommendation Models : All Used Before And Rejuvenated


- Collaborating Filtering : Analyzation of user behavier

- (NLP) Narutal Language Processing : Text analyzation

- Audio : Analyzation of single tracks


Manual Curation was used by many companies before, including LAST.FM, SONGZA, THE ECHO NEST, PANDORA and many more.

All types of manual curation/algorithm were tested and found to be very useful when it came to creating recommendations for a user based on gigantic date (of course) and mostly by analyzing these 3 options.


Hope this brief answer shed some light.

If i can help or answer any other questions i'll be more than happy to!