English

Contextual-Bandit Based Personalized Recommendation with Time-Varying User Interests

Machine Learning 2020-03-03 v1 Artificial Intelligence Machine Learning

Abstract

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 different items vary differently over time. In the disjoint payoff model, the reward of playing an arm is determined by an arm-specific preference vector, which is piecewise-stationary with asynchronous and distinct changes across different arms. An efficient learning algorithm that is adaptive to abrupt reward changes is proposed and theoretical regret analysis is provided to show that a sublinear scaling of regret in the time length TT is achieved. The algorithm is further extended to a more general setting with hybrid payoffs where the reward of playing an arm is determined by both an arm-specific preference vector and a joint coefficient vector shared by all arms. Empirical experiments are conducted on real-world datasets to verify the advantages of the proposed learning algorithms against baseline ones in both settings.

Keywords

Cite

@article{arxiv.2003.00359,
  title  = {Contextual-Bandit Based Personalized Recommendation with Time-Varying User Interests},
  author = {Xiao Xu and Fang Dong and Yanghua Li and Shaojian He and Xin Li},
  journal= {arXiv preprint arXiv:2003.00359},
  year   = {2020}
}

Comments

Accepted by AAAI 20

R2 v1 2026-06-23T13:59:00.589Z