![[2404.17592] Low-Rank Online Dynamic Assortment with Dual Contextual Information](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2404.17592] Low-Rank Online Dynamic Assortment with Dual Contextual Information
Summary
This paper presents a low-rank dynamic assortment model that improves real-time personalized recommendations in e-commerce by utilizing dual contextual information from users and items.
Why It Matters
As e-commerce continues to grow, optimizing product assortments based on user and item features is crucial for maximizing revenue. This research addresses computational challenges in high-dimensional settings, providing a new model and algorithm that enhance recommendation systems, which are essential for retail platforms.
Key Takeaways
- Introduces a low-rank model to manage dynamic assortment problems in e-commerce.
- Utilizes dual contextual information to enhance recommendation accuracy.
- Establishes a theoretical regret bound that improves upon existing literature.
- Demonstrates practical application using the Expedia hotel recommendation dataset.
- Offers an efficient algorithm for balancing exploration and exploitation in online decision-making.
Computer Science > Information Retrieval arXiv:2404.17592 (cs) [Submitted on 19 Apr 2024 (v1), last revised 12 Feb 2026 (this version, v3)] Title:Low-Rank Online Dynamic Assortment with Dual Contextual Information Authors:Seong Jin Lee, Will Wei Sun, Yufeng Liu View a PDF of the paper titled Low-Rank Online Dynamic Assortment with Dual Contextual Information, by Seong Jin Lee and 1 other authors View PDF HTML (experimental) Abstract:As e-commerce expands, delivering real-time personalized recommendations from vast catalogs poses a critical challenge for retail platforms. Maximizing revenue requires careful consideration of both individual customer characteristics and available item features to continuously optimize assortments over time. In this paper, we consider the dynamic assortment problem with dual contexts -- user and item features. In high-dimensional scenarios, the quadratic growth of dimensions complicates computation and estimation. To tackle this challenge, we introduce a new low-rank dynamic assortment model to transform this problem into a manageable scale. Then we propose an efficient algorithm that estimates the intrinsic subspaces and utilizes the upper confidence bound approach to address the exploration-exploitation trade-off in online decision making. Theoretically, we establish a regret bound of $\tilde{O}((d_1+d_2)r\sqrt{T})$, where $d_1, d_2$ represent the dimensions of the user and item features respectively, $r$ is the rank of the parameter matrix,...
