Is diversity optimization always suitable? Toward a better understanding of diversity within recommendation approaches

作者：

Highlights：

• Different recommender systems have very different levels of diversity.

• None of the state-of-the-art recommenders provides personalized diversity for users.

• Diversity optimization with classical post-processing may indeed be detrimental.

• Diversity optimization accounting for each user’s specificity is more beneficial.

摘要

•Different recommender systems have very different levels of diversity.•None of the state-of-the-art recommenders provides personalized diversity for users.•Diversity optimization with classical post-processing may indeed be detrimental.•Diversity optimization accounting for each user’s specificity is more beneficial.

论文关键词：Recommender system,Diversity,Greedy optimization,DBpedia,Knowledge graph embedding,Deep learning

论文评审过程：Received 21 December 2020, Revised 31 July 2021, Accepted 9 August 2021, Available online 5 September 2021, Version of Record 5 September 2021.

论文官网地址：https://doi.org/10.1016/j.ipm.2021.102721