Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Behavioral Feature Boosting via Substitute Relationships for E-commerce Search

Chaosheng Dong, Michinari Momma, Yijia Wang, Yan Gao, Yi Sun

TL;DR

Experiments show that BFS significantly improves search relevance and product discovery for cold-start products, and is scalable and practical, improving user experience while increasing exposure for newly launched items in E-commerce search.

Abstract

On E-commerce platforms, new products often suffer from the cold-start problem: limited interaction data reduces their search visibility and hurts relevance ranking. To address this, we propose a simple yet effective behavior feature boosting method that leverages substitute relationships among products (BFS). BFS identifies substitutes-products that satisfy similar user needs-and aggregates their behavioral signals (e.g., clicks, add-to-carts, purchases, and ratings) to provide a warm start for new items. Incorporating these enriched signals into ranking models mitigates cold-start effects and improves relevance and competitiveness. Experiments on a large E-commerce platform, both offline and online, show that BFS significantly improves search relevance and product discovery for cold-start products. BFS is scalable and practical, improving user experience while increasing exposure for newly launched items in E-commerce search. The BFS-enhanced ranking model has been launched in production and has served customers since 2025.

Behavioral Feature Boosting via Substitute Relationships for E-commerce Search

TL;DR

Experiments show that BFS significantly improves search relevance and product discovery for cold-start products, and is scalable and practical, improving user experience while increasing exposure for newly launched items in E-commerce search.

Abstract

On E-commerce platforms, new products often suffer from the cold-start problem: limited interaction data reduces their search visibility and hurts relevance ranking. To address this, we propose a simple yet effective behavior feature boosting method that leverages substitute relationships among products (BFS). BFS identifies substitutes-products that satisfy similar user needs-and aggregates their behavioral signals (e.g., clicks, add-to-carts, purchases, and ratings) to provide a warm start for new items. Incorporating these enriched signals into ranking models mitigates cold-start effects and improves relevance and competitiveness. Experiments on a large E-commerce platform, both offline and online, show that BFS significantly improves search relevance and product discovery for cold-start products. BFS is scalable and practical, improving user experience while increasing exposure for newly launched items in E-commerce search. The BFS-enhanced ranking model has been launched in production and has served customers since 2025.
Paper Structure (13 sections, 3 equations, 5 figures, 2 tables)

This paper contains 13 sections, 3 equations, 5 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Methodology
  4. Definition of Behavioral Feature
  5. Substitute Identification
  6. Behavioral Feature Boosting Design
  7. Integration BFS into Ranking Model
  8. Experiments
  9. Experimental Setup
  10. Offline Evaluation
  11. Online Evaluation on a large E-commerce platform
  12. Qualitative Analysis
  13. Conclusion

Figures (5)

  • Figure 1: Diagram of the substitute identification system.
  • Figure 2: Example substitute products. The seed item is a cartoon laptop-skin sticker.
  • Figure 3: Distribution of SV and SV_Subs. Blue denotes SV, orange denotes SV_Subs, and brown indicates the overlap between the two histograms.
  • Figure 4: Relevance score as a function of the feature value.
  • Figure 5: Top-8 search results for the query pilot explorer fountain pen during the online A/B test. The first, second, and third rows show the ranked lists produced by the production ranker, T1, and T2, respectively.