Learning in Position-Aware Multinomial Logit Bandits: From Multiplicative to General Position Effects
Researchers have developed new algorithms for optimizing product assortment and display position under a Multinomial Logit choice framework. These algorithms address both multiplicative and general position effects models, aiming to improve decision-making on modern platforms. The proposed methods, P2MLE-UCB and GP2-UCB, achieve regret-optimal characterizations and outperform existing benchmarks in numerical experiments. AI
IMPACT Introduces novel algorithms for optimizing product selection and positioning, potentially improving recommendation systems and e-commerce platforms.