“User-User Collaborative-Filtering Algorithm Leads to Three Phases of Behavior and Identifies Region for Personalized Recommendations without Filter Bubbles”

https://pubmed.ncbi.nlm.nih.gov/38115540

This study finds that a user-user collaborative-filtering algorithm can lead to three phases of user behavior – disorder, consensus, and polarization – and identifies a region where personalized recommendations can be made without causing filter bubbles, offering potential for developing less polarizing recommendation algorithms.