The field of music recommendation is moving towards a more nuanced understanding of user behavior and preferences. Researchers are developing new methods to analyze and model user interactions with music streaming platforms, taking into account factors such as uncertainty, repetition, and discovery patterns. These advances have the potential to improve the accuracy and relevance of music recommendations, as well as provide new insights into user behavior. Notable papers in this area include:

• Uncertainty in Repeated Implicit Feedback as a Measure of Reliability, which proposes a Bayesian model for implicit listening feedback and demonstrates its effectiveness in improving recommendation accuracy.
• Modeling Musical Genre Trajectories through Pathlet Learning, which introduces a new framework for modeling user trajectories across different musical genres and releases a novel dataset.
• CoCoB: Adaptive Collaborative Combinatorial Bandits for Online Recommendation, which presents an innovative two-sided bandit architecture for adaptive collaborative filtering.