Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

A Unified Framework for Cross-Domain Recommendation

Jiangxia Cao, Shen Wang, Gaode Chen, Rui Huang, Shuang Yang, Zhaojie Liu, Guorui Zhou

TL;DR

To coherently adapt to various scenarios, and drawing inspiration from the concept of domain-invariant transfer learning, the former SOTA model UniCDR is extended in five different aspects, named as UniCDR+.

Abstract

In addressing the persistent challenges of data-sparsity and cold-start issues in domain-expert recommender systems, Cross-Domain Recommendation (CDR) emerges as a promising methodology. CDR aims at enhancing prediction performance in the target domain by leveraging interaction knowledge from related source domains, particularly through users or items that span across multiple domains (e.g., Short-Video and Living-Room). For academic research purposes, there are a number of distinct aspects to guide CDR method designing, including the auxiliary domain number, domain-overlapped element, user-item interaction types, and downstream tasks. With so many different CDR combination scenario settings, the proposed scenario-expert approaches are tailored to address a specific vertical CDR scenario, and often lack the capacity to adapt to multiple horizontal scenarios. In an effect to coherently adapt to various scenarios, and drawing inspiration from the concept of domain-invariant transfer learning, we extend the former SOTA model UniCDR in five different aspects, named as UniCDR+. Our work was successfully deployed on the Kuaishou Living-Room RecSys.

A Unified Framework for Cross-Domain Recommendation

TL;DR

To coherently adapt to various scenarios, and drawing inspiration from the concept of domain-invariant transfer learning, the former SOTA model UniCDR is extended in five different aspects, named as UniCDR+.

Abstract

In addressing the persistent challenges of data-sparsity and cold-start issues in domain-expert recommender systems, Cross-Domain Recommendation (CDR) emerges as a promising methodology. CDR aims at enhancing prediction performance in the target domain by leveraging interaction knowledge from related source domains, particularly through users or items that span across multiple domains (e.g., Short-Video and Living-Room). For academic research purposes, there are a number of distinct aspects to guide CDR method designing, including the auxiliary domain number, domain-overlapped element, user-item interaction types, and downstream tasks. With so many different CDR combination scenario settings, the proposed scenario-expert approaches are tailored to address a specific vertical CDR scenario, and often lack the capacity to adapt to multiple horizontal scenarios. In an effect to coherently adapt to various scenarios, and drawing inspiration from the concept of domain-invariant transfer learning, we extend the former SOTA model UniCDR in five different aspects, named as UniCDR+. Our work was successfully deployed on the Kuaishou Living-Room RecSys.
Paper Structure (35 sections, 16 equations, 5 figures, 11 tables)

This paper contains 35 sections, 16 equations, 5 figures, 11 tables.

Table of Contents

  1. Introduction
  2. Preliminary: CDR Scenarios Statement
  3. Preliminary: UniCDR Workflow
  4. Embedding Module
  5. User/Item ID Embedding
  6. Aggregators
  7. Aggregators details
  8. Domain-specific & domain-shared representations
  9. Mask Module
  10. Mechanism details
  11. Contrastive Module
  12. Hard contrastive objective
  13. Model Training
  14. Single Action Interaction Prediction
  15. Loss Function
  16. ...and 20 more sections

Figures (5)

  • Figure 1: Intra-/Inter- downstream tasks. Yellow user is overlapped user who has interaction in multiple domains, while purple/green users denote corresponding domain users.
  • Figure 2: Domain shift and adaptation of transfer learning.
  • Figure 3: A practical roadmap for unified CDR.
  • Figure 4: The UniCDR framework.
  • Figure 5: The UniCDR+ training procedure in Kuaishou