Contextual-Bandit Based Personalized Recommendation with Time-Varying User Interests
A contextual bandit problem is studied in a highly non-stationary environment, which is ubiquitous in various recommender systems due to the time-varying interests of users. Two models with disjoint and hybrid payoffs are considered to characterize the phenomenon that users' preferences towards diff...
Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Published: |
2020
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2003.00359 http://arxiv.org/abs/2003.00359 |
| Tags: |
No Tags, Be the first to tag this record!
|