A Novel Hierarchical Multimodal Recommender With Enhanced Global Collaborative Signals

Multimodal recommender systems leverage auxiliary item features, such as images and descriptions, to alleviate the data sparsity problem and facilitate the preference modeling process. Despite their potential, existing multimodal recommenders fail to exploit global collaborative signals and lack ins...

Full description

Saved in:
Bibliographic Details
Main Authors: Peng Yi, Lu Chen, Zhaoxian Li, Cheng Yang
Format: Article
Language:English
Published: IEEE 2025-01-01
Series:IEEE Access
Subjects:
Online Access:https://ieeexplore.ieee.org/document/11014073/
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850174465158152192
author Peng Yi
Lu Chen
Zhaoxian Li
Cheng Yang
author_facet Peng Yi
Lu Chen
Zhaoxian Li
Cheng Yang
author_sort Peng Yi
collection DOAJ
description Multimodal recommender systems leverage auxiliary item features, such as images and descriptions, to alleviate the data sparsity problem and facilitate the preference modeling process. Despite their potential, existing multimodal recommenders fail to exploit global collaborative signals and lack insights into the underlying interaction formation mechanism, resulting in suboptimal recommendation performance. To this end, we propose a hierarchical multimodal recommender named HMMGCF, which can capture crucial global collaborative signals through modality feature-enhanced hierarchical structures and a novel inter-modality alignment strategy. Specifically, modality features are first utilized to identify neighboring relationships, and similar users (items) are steadily merged together to form modality-specific hierarchical structures. Then, with the proper graph convolution operation on each hierarchy, the crucial global collaborative signals can be effectively extracted and integrated into the modality-specific user (item) embeddings. Moreover, a novel group-wise contrastive learning strategy is also proposed to align inter-modality preference information and further enhance the extraction of global collaborative signals. By conducting extensive experiments on three benchmark datasets, we empirically and theoretically demonstrate the superiority of HMMGCF, validating the importance of global collaborative signal extraction in multimodal recommender systems.
format Article
id doaj-art-c8fee5e14cf64c168d1ebbf3dfee18fe
institution OA Journals
issn 2169-3536
language English
publishDate 2025-01-01
publisher IEEE
record_format Article
series IEEE Access
spelling doaj-art-c8fee5e14cf64c168d1ebbf3dfee18fe2025-08-20T02:19:38ZengIEEEIEEE Access2169-35362025-01-0113921029211310.1109/ACCESS.2025.357320411014073A Novel Hierarchical Multimodal Recommender With Enhanced Global Collaborative SignalsPeng Yi0https://orcid.org/0009-0002-3114-6566Lu Chen1https://orcid.org/0009-0009-6253-5848Zhaoxian Li2https://orcid.org/0009-0005-7417-1739Cheng Yang3https://orcid.org/0000-0003-3873-6411State Key Laboratory of Media Convergence and Communication, Communication University of China, Beijing, ChinaState Key Laboratory of Media Convergence and Communication, Communication University of China, Beijing, ChinaFaculty of Social Sciences, University of Copenhagen, Copenhagen, DenmarkState Key Laboratory of Media Convergence and Communication, Communication University of China, Beijing, ChinaMultimodal recommender systems leverage auxiliary item features, such as images and descriptions, to alleviate the data sparsity problem and facilitate the preference modeling process. Despite their potential, existing multimodal recommenders fail to exploit global collaborative signals and lack insights into the underlying interaction formation mechanism, resulting in suboptimal recommendation performance. To this end, we propose a hierarchical multimodal recommender named HMMGCF, which can capture crucial global collaborative signals through modality feature-enhanced hierarchical structures and a novel inter-modality alignment strategy. Specifically, modality features are first utilized to identify neighboring relationships, and similar users (items) are steadily merged together to form modality-specific hierarchical structures. Then, with the proper graph convolution operation on each hierarchy, the crucial global collaborative signals can be effectively extracted and integrated into the modality-specific user (item) embeddings. Moreover, a novel group-wise contrastive learning strategy is also proposed to align inter-modality preference information and further enhance the extraction of global collaborative signals. By conducting extensive experiments on three benchmark datasets, we empirically and theoretically demonstrate the superiority of HMMGCF, validating the importance of global collaborative signal extraction in multimodal recommender systems.https://ieeexplore.ieee.org/document/11014073/Multimodal recommenderhierarchical graph convolutional networkscontrastive learning
spellingShingle Peng Yi
Lu Chen
Zhaoxian Li
Cheng Yang
A Novel Hierarchical Multimodal Recommender With Enhanced Global Collaborative Signals
IEEE Access
Multimodal recommender
hierarchical graph convolutional networks
contrastive learning
title A Novel Hierarchical Multimodal Recommender With Enhanced Global Collaborative Signals
title_full A Novel Hierarchical Multimodal Recommender With Enhanced Global Collaborative Signals
title_fullStr A Novel Hierarchical Multimodal Recommender With Enhanced Global Collaborative Signals
title_full_unstemmed A Novel Hierarchical Multimodal Recommender With Enhanced Global Collaborative Signals
title_short A Novel Hierarchical Multimodal Recommender With Enhanced Global Collaborative Signals
title_sort novel hierarchical multimodal recommender with enhanced global collaborative signals
topic Multimodal recommender
hierarchical graph convolutional networks
contrastive learning
url https://ieeexplore.ieee.org/document/11014073/
work_keys_str_mv AT pengyi anovelhierarchicalmultimodalrecommenderwithenhancedglobalcollaborativesignals
AT luchen anovelhierarchicalmultimodalrecommenderwithenhancedglobalcollaborativesignals
AT zhaoxianli anovelhierarchicalmultimodalrecommenderwithenhancedglobalcollaborativesignals
AT chengyang anovelhierarchicalmultimodalrecommenderwithenhancedglobalcollaborativesignals
AT pengyi novelhierarchicalmultimodalrecommenderwithenhancedglobalcollaborativesignals
AT luchen novelhierarchicalmultimodalrecommenderwithenhancedglobalcollaborativesignals
AT zhaoxianli novelhierarchicalmultimodalrecommenderwithenhancedglobalcollaborativesignals
AT chengyang novelhierarchicalmultimodalrecommenderwithenhancedglobalcollaborativesignals