Główne pojęcia
Optimizing two-sided assortments through adaptivity and approximation.
Streszczenie
The article discusses the challenges faced by two-sided matching platforms due to choice congestion among popular options. It introduces a framework for assortment optimization to maximize matches by designing assortments and their presentation order. The study compares different policy classes, showing adaptivity gaps between them. Results include approximation algorithms for various policy classes under different choice models.
Statystyki
Gap between one-sided static and adaptive policies is 1 - 1/e.
Gap between one-sided adaptive and fully adaptive policies is 1/2.
Approximation factor of 0.066 for multinomial-logit models under fully static policies.
Cytaty
"There exists a polynomial time policy that achieves a 1/4 approximation factor within the class of policies that adaptively show assortments to agents one by one."
"The worst policies are those who simultaneously show assortments to all the agents."
"Balancing relevant options with reducing choice congestion leads to better market outcomes."