Refining Contrastive Learning and Homography Relations for Multi-Modal Recommendation

TL;DR

This paper introduces REARM, a novel framework that enhances multi-modal recommendation by refining contrastive learning with meta-networks and orthogonal constraints, and by exploring homography relations through multiple graph structures.

Contribution

The paper proposes a new framework that improves multi-modal recommendation by refining contrastive learning and exploring homography relations with multiple graphs, addressing data sparsity and feature noise.

Findings

REARM outperforms state-of-the-art baselines on three real-world datasets.

The framework effectively filters noise and retains valuable modal-specific information.

Visualization shows improved distinction between shared and unique modal features.

Abstract

Multi-modal recommender system focuses on utilizing rich modal information ( i.e., images and textual descriptions) of items to improve recommendation performance. The current methods have achieved remarkable success with the powerful structure modeling capability of graph neural networks. However, these methods are often hindered by sparse data in real-world scenarios. Although contrastive learning and homography ( i.e., homogeneous graphs) are employed to address the data sparsity challenge, existing methods still suffer two main limitations: 1) Simple multi-modal feature contrasts fail to produce effective representations, causing noisy modal-shared features and loss of valuable information in modal-unique features; 2) The lack of exploration of the homograph relations between user interests and item co-occurrence results in incomplete mining of user-item interplay. To address the above limitations, we propose a novel framework for \textbf{R}\textbf{E}fining multi-mod\textbf{A}l cont\textbf{R}astive learning and ho\textbf{M}ography relations (\textbf{REARM}). Specifically, we complement multi-modal contrastive learning by employing meta-network and orthogonal constraint strategies, which filter out noise in modal-shared features and retain recommendation-relevant information in modal-unique features. To mine homogeneous relationships effectively, we integrate a newly constructed user interest graph and an item co-occurrence graph with the existing user co-occurrence and item semantic graphs for graph learning. The extensive experiments on three real-world datasets demonstrate the superiority of REARM to various state-of-the-art baselines. Our visualization further shows an improvement made by REARM in distinguishing between modal-shared and modal-unique features. Code is available \href{https://github.com/MrShouxingMa/REARM}{here}.