Knowledge-Flow Contrastive Learning for recommendation

Wenhe Xing

doi:10.1016/j.inffus.2025.103477

Knowledge-Flow Contrastive Learning for recommendation

2025-07-09

0

PRE

AI

DOI：10.1016/j.inffus.2025.103477

Original

Original request for help

Save

Abstract

En 中文

• KFCL can mitigate the noise introduced by CL in a KG-aware recommender system. • User preference is enhanced by knowledge fusion and multi-views feature fusion. • The complementary roles of contrastive learning and feature fusion are explored. • Analyzed and demonstrated the source of noise in the KG-aware recommender system.

Keywords:

Knowledge graph

Contrastive learning

Feature fusion

Graph neural networks

Recommender system

Knowledge-Flow Contrastive Learning for recommendation

Abstract

Journal

Organization