Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

A Unified Search and Recommendation Framework Based on Multi-Scenario Learning for Ranking in E-commerce

Jinhan Liu, Qiyu Chen, Junjie Xu, Junjie Li, Baoli Li, Sulong Xu

TL;DR

This work tackles the challenge of jointly modeling search and recommendation in e-commerce by addressing limitations of traditional multi-scenario models that fail to capture scenario-specific representations and cross-domain label information. It introduces the Unified Search and Recommendation (USR) framework, featuring S&R Views User Interest Extractor (IE), S&R Views Feature Generator (FG), and a Global Label Space Multi-Task Layer (GLMT) that leverages global labels to jointly train main and auxiliary tasks. The approach yields substantial offline improvements across multiple multi-scenario baselines and demonstrates clear online gains in A/B tests, validating its practicality for real-world deployments (e.g., the 7Fresh app). The combination of fine-grained user-interest representations, scenario-agnostic feature scaling, and global-label supervision constitutes a robust, generalizable pathway for enhancing ranking in multi-scenario S&R systems.

Abstract

Search and recommendation (S&R) are the two most important scenarios in e-commerce. The majority of users typically interact with products in S&R scenarios, indicating the need and potential for joint modeling. Traditional multi-scenario models use shared parameters to learn the similarity of multiple tasks, and task-specific parameters to learn the divergence of individual tasks. This coarse-grained modeling approach does not effectively capture the differences between S&R scenarios. Furthermore, this approach does not sufficiently exploit the information across the global label space. These issues can result in the suboptimal performance of multi-scenario models in handling both S&R scenarios. To address these issues, we propose an effective and universal framework for Unified Search and Recommendation (USR), designed with S&R Views User Interest Extractor Layer (IE) and S&R Views Feature Generator Layer (FG) to separately generate user interests and scenario-agnostic feature representations for S&R. Next, we introduce a Global Label Space Multi-Task Layer (GLMT) that uses global labels as supervised signals of auxiliary tasks and jointly models the main task and auxiliary tasks using conditional probability. Extensive experimental evaluations on real-world industrial datasets show that USR can be applied to various multi-scenario models and significantly improve their performance. Online A/B testing also indicates substantial performance gains across multiple metrics. Currently, USR has been successfully deployed in the 7Fresh App.

A Unified Search and Recommendation Framework Based on Multi-Scenario Learning for Ranking in E-commerce

TL;DR

This work tackles the challenge of jointly modeling search and recommendation in e-commerce by addressing limitations of traditional multi-scenario models that fail to capture scenario-specific representations and cross-domain label information. It introduces the Unified Search and Recommendation (USR) framework, featuring S&R Views User Interest Extractor (IE), S&R Views Feature Generator (FG), and a Global Label Space Multi-Task Layer (GLMT) that leverages global labels to jointly train main and auxiliary tasks. The approach yields substantial offline improvements across multiple multi-scenario baselines and demonstrates clear online gains in A/B tests, validating its practicality for real-world deployments (e.g., the 7Fresh app). The combination of fine-grained user-interest representations, scenario-agnostic feature scaling, and global-label supervision constitutes a robust, generalizable pathway for enhancing ranking in multi-scenario S&R systems.

Abstract

Search and recommendation (S&R) are the two most important scenarios in e-commerce. The majority of users typically interact with products in S&R scenarios, indicating the need and potential for joint modeling. Traditional multi-scenario models use shared parameters to learn the similarity of multiple tasks, and task-specific parameters to learn the divergence of individual tasks. This coarse-grained modeling approach does not effectively capture the differences between S&R scenarios. Furthermore, this approach does not sufficiently exploit the information across the global label space. These issues can result in the suboptimal performance of multi-scenario models in handling both S&R scenarios. To address these issues, we propose an effective and universal framework for Unified Search and Recommendation (USR), designed with S&R Views User Interest Extractor Layer (IE) and S&R Views Feature Generator Layer (FG) to separately generate user interests and scenario-agnostic feature representations for S&R. Next, we introduce a Global Label Space Multi-Task Layer (GLMT) that uses global labels as supervised signals of auxiliary tasks and jointly models the main task and auxiliary tasks using conditional probability. Extensive experimental evaluations on real-world industrial datasets show that USR can be applied to various multi-scenario models and significantly improve their performance. Online A/B testing also indicates substantial performance gains across multiple metrics. Currently, USR has been successfully deployed in the 7Fresh App.
Paper Structure (19 sections, 15 equations, 2 figures, 1 table)

This paper contains 19 sections, 15 equations, 2 figures, 1 table.

Table of Contents

  1. Introduction
  2. PROBLEM FORMULATION
  3. METHOD
  4. The Input Network
  5. S&R Views User Interest Extractor Layer
  6. S&R Views Feature Generator Layer
  7. Global Label Space Multi-Task Layer
  8. Model Optimization
  9. EXPERIMENT
  10. Experimental Setup
  11. Dataset.
  12. Baselines.
  13. Evaluation metrics and hyper-parameters.
  14. Results and Discussion
  15. Overall performance.
  16. ...and 4 more sections

Figures (2)

  • Figure 1: Overall structure of USR.
  • Figure 2: (a) Visualizes the representations from S&R views user interests and scenario-agnostic features. (b) Shows the ablation study of USR. (c) Shows the effect of dimension $d$.