Eugene Yan • 4/26/2020

Serendipity: Accuracy’s Unpopular Best Friend in Recommenders

This article argues that accuracy alone is insufficient for effective recommendation systems. It introduces serendipity as a crucial metric for user surprise and discovery, contrasting it with standard accuracy metrics. The piece discusses why accuracy dominates, the business case for serendipity (improving user interest and assortment health), and touches on related concepts like diversity and novelty in recommendations.

0 comments

#Recommender Systems #Serendipity #Accuracy Metrics