Sign In

Communications of the ACM

ACM TechNews

'Dislike' Button Would Improve Spotify Recommendations

View as: Print Mobile App Share:
The analysis provided more precise information about the treatments potential benefit to individual patients, compared with algorithms that simply predicted average patient outcomes following treatment.

Data scientists at Pacific Northwest National Laboratory are turning their artificial intelligence tools to the study of fundamental questions about treatments for COVID.


A research team at Cornell University has developed a recommendation algorithm that could improve audio streaming and media services provider Spotify by incorporating both likes and dislikes.

The group showed that a listener is about 20% more likely to like a song if the algorithm recommending it is trained on 400,000 likes and dislikes, as opposed to an algorithm trained solely on likes.

Cornell's Sasha Stoikov said an algorithm that only keys on likes has a greater chance of recommending songs the listener dislikes.

The new Piki system picks music from a database of approximately 5 million songs and gives users $1 for every 25 songs they rate, which Stoikov said "incentivizes the user to vote truthfully."

From Cornell Chronicle
View Full Article


Abstracts Copyright © 2021 SmithBucklin, Washington, DC, USA


No entries found

Sign In for Full Access
» Forgot Password? » Create an ACM Web Account