Online assortment optimization problem with position effect for e-commerce platforms

Based on the dynamic characteristics of E-commerce platforms, this paper studies a new assortment optimization problem. Assuming that consumers’ choice follows the multinominal logit model, when heterogeneous consumers arrive successively, with the goal to maximize the expected revenue, the platform needs to select a subset from the given product set and decide the display positions of products to provide to consumers, on the condition that the inventory constraints of products are satisfied. The models established by previous studies generally assumed that the arrival sequence of consumer types was known or followed some random distribution, however, in practice, the arrival sequence of consumer types is often highly uncertain and cannot be observed randomly. Under the condition that the future arrival sequence of consumer types is completely unknown, this paper constructs an online display model that considers limited inventory, assortment optimization and location effect simultaneously to maximize the revenue of the e-commerce platform, and develops a corresponding display strategy based on the online theory and competitive analysis framework. This strategy is easy to solve and an effective display scheme could be quickly obtained. The competitive performance ratio of the strategy is proved and the upper bound of the competitive performance ratio of the problem is analyzed theoretically. The results show that the strategy realizes a superior competitive performance. © 2023 Systems Engineering Society of China. All rights reserved.