Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Cross-channel Recommendation for Multi-channel Retail

Yijin Choi, Jongkyung Shin, Chiehyeon Lim

TL;DR

The paper tackles Cross-Channel Retail Recommendation (CCRR), where users and items partially overlap across offline and online channels. It introduces C^2Rec, a channel-wise attention model with three user embeddings (shared, offline, online) and a shared item embedding, trained jointly for recommendation and interaction classification under the objective $L = L_{on} + L_{off} + lambda_{cls} L_{cls} + lambda_{attn} L_{attn}$. Empirical results on a real-world multi-channel dataset show C^2Rec outperforms single-domain and cross-domain baselines, with ablations confirming the value of channel-specific versus shared representations, attention, and the auxiliary attention loss. The study highlights the importance of modeling channel context for CCRR and suggests directions for data filtering and more nuanced cross-channel transfer in omni-channel retail contexts.

Abstract

An increasing number of retailers are expanding their channels to the offline and online domains, transforming them into multi-channel retailers. This transition emphasizes the need for cross-channel recommendations. Given that each retail channel represents a separate domain with a unique context, this can be regarded as a cross-domain recommendation (CDR). However, existing studies on CDR did not address the scenarios where both users and items partially overlap across multi-retail channels which we define as "cross-channel retail recommendation (CCRR)". This paper introduces our original work on CCRR using a real-world dataset from a multi-channel retail store. Specifically, we study significant challenges in integrating user preferences across both channels and propose a novel model for CCRR using a channel-wise attention mechanism. We empirically validate our model's superiority in addressing CCRR over existing models. Finally, we offer implications for future research on CCRR, delving into our experiment results.

Cross-channel Recommendation for Multi-channel Retail

TL;DR

The paper tackles Cross-Channel Retail Recommendation (CCRR), where users and items partially overlap across offline and online channels. It introduces C^2Rec, a channel-wise attention model with three user embeddings (shared, offline, online) and a shared item embedding, trained jointly for recommendation and interaction classification under the objective . Empirical results on a real-world multi-channel dataset show C^2Rec outperforms single-domain and cross-domain baselines, with ablations confirming the value of channel-specific versus shared representations, attention, and the auxiliary attention loss. The study highlights the importance of modeling channel context for CCRR and suggests directions for data filtering and more nuanced cross-channel transfer in omni-channel retail contexts.

Abstract

An increasing number of retailers are expanding their channels to the offline and online domains, transforming them into multi-channel retailers. This transition emphasizes the need for cross-channel recommendations. Given that each retail channel represents a separate domain with a unique context, this can be regarded as a cross-domain recommendation (CDR). However, existing studies on CDR did not address the scenarios where both users and items partially overlap across multi-retail channels which we define as "cross-channel retail recommendation (CCRR)". This paper introduces our original work on CCRR using a real-world dataset from a multi-channel retail store. Specifically, we study significant challenges in integrating user preferences across both channels and propose a novel model for CCRR using a channel-wise attention mechanism. We empirically validate our model's superiority in addressing CCRR over existing models. Finally, we offer implications for future research on CCRR, delving into our experiment results.
Paper Structure (19 sections, 6 equations, 2 figures, 4 tables)

This paper contains 19 sections, 6 equations, 2 figures, 4 tables.

Table of Contents

  1. Introduction
  2. Dataset and Implication
  3. Methodology
  4. Problem Definition
  5. Model Architecture
  6. User and Item Embedding
  7. Attention Mechanism
  8. Preference Prediction
  9. Interaction Classification
  10. Multi-task Learning
  11. Experiment
  12. Experiment Settings
  13. Datasets
  14. Baselines and Evaluation Metrics
  15. Implementation Details
  16. ...and 4 more sections

Figures (2)

  • Figure 1: Examples from multi-channel retail stores illustrate how purchase behaviors vary based on users, items, and channels.
  • Figure 2: Overview of training mechanism of $\mathbf{C^{2}Rec}$.