Multi-Task Hypergraph-Attention Framework for Multimodal Sentiment Analysis

Yibing Wang; Zhutian Yang; Linhan Wang; Mingqian Liu; Yushi Chen

doi:10.1109/VTC2025-Spring65109.2025.11174446

Multi-Task Hypergraph-Attention Framework for Multimodal Sentiment Analysis

Yibing Wang^*
, Zhutian Yang^*
, Linhan Wang^*
, Mingqian Liu
, Yushi Chen^*

^*Corresponding author for this work

School of Electronics and Information Engineering, Harbin Institute of Technology
Xidian University

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

1 Scopus citations

Abstract

Multimodal sentiment analysis has emerged as a critical research area. However, existing methods face significant challenges: (1) Unimodal feature extraction techniques often fail to capture the topological structure within data, and do not effectively integrate local and global information, leading to information loss. (2) Traditional multimodal fusion methods, such as concatenation, addition, and multiplication, struggle to model modality differences and inter-modal correlations. In this paper, we propose a novel multi-task hypergraph-attention framework (MTHA) to improve feature discrimination and model performance. Experimental results demonstrate that MTHA outperforms most baseline models in both sentiment classification and regression.

Original language	English
Title of host publication	2025 IEEE 101st Vehicular Technology Conference, VTC 2025-Spring 2025 - Proceedings
Publisher	Institute of Electrical and Electronics Engineers Inc.
ISBN (Electronic)	9798331531478
DOIs	https://doi.org/10.1109/VTC2025-Spring65109.2025.11174446
State	Published - 2025
Externally published	Yes
Event	101st IEEE Vehicular Technology Conference, VTC 2025-Spring 2025 - Oslo, Norway Duration: 17 Jun 2025 → 20 Jun 2025

Publication series

Name	IEEE Vehicular Technology Conference
ISSN (Print)	1550-2252

Conference

Conference	101st IEEE Vehicular Technology Conference, VTC 2025-Spring 2025
Country/Territory	Norway
City	Oslo
Period	17/06/25 → 20/06/25

Keywords

Multimodal sentiment analysis
hypergraph learning
multi-task
self-attention

Access to Document

10.1109/VTC2025-Spring65109.2025.11174446

Cite this

Wang, Y., Yang, Z., Wang, L., Liu, M., & Chen, Y. (2025). Multi-Task Hypergraph-Attention Framework for Multimodal Sentiment Analysis. In 2025 IEEE 101st Vehicular Technology Conference, VTC 2025-Spring 2025 - Proceedings (IEEE Vehicular Technology Conference). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/VTC2025-Spring65109.2025.11174446

@inproceedings{ce5f54052ec24accb8c87902882fc854,

title = "Multi-Task Hypergraph-Attention Framework for Multimodal Sentiment Analysis",

abstract = "Multimodal sentiment analysis has emerged as a critical research area. However, existing methods face significant challenges: (1) Unimodal feature extraction techniques often fail to capture the topological structure within data, and do not effectively integrate local and global information, leading to information loss. (2) Traditional multimodal fusion methods, such as concatenation, addition, and multiplication, struggle to model modality differences and inter-modal correlations. In this paper, we propose a novel multi-task hypergraph-attention framework (MTHA) to improve feature discrimination and model performance. Experimental results demonstrate that MTHA outperforms most baseline models in both sentiment classification and regression.",

keywords = "Multimodal sentiment analysis, hypergraph learning, multi-task, self-attention",

author = "Yibing Wang and Zhutian Yang and Linhan Wang and Mingqian Liu and Yushi Chen",

note = "Publisher Copyright: {\textcopyright} 2025 IEEE.; 101st IEEE Vehicular Technology Conference, VTC 2025-Spring 2025 ; Conference date: 17-06-2025 Through 20-06-2025",

year = "2025",

doi = "10.1109/VTC2025-Spring65109.2025.11174446",

language = "英语",

series = "IEEE Vehicular Technology Conference",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

booktitle = "2025 IEEE 101st Vehicular Technology Conference, VTC 2025-Spring 2025 - Proceedings",

address = "美国",

}

Wang, Y, Yang, Z, Wang, L, Liu, M & Chen, Y 2025, Multi-Task Hypergraph-Attention Framework for Multimodal Sentiment Analysis. in 2025 IEEE 101st Vehicular Technology Conference, VTC 2025-Spring 2025 - Proceedings. IEEE Vehicular Technology Conference, Institute of Electrical and Electronics Engineers Inc., 101st IEEE Vehicular Technology Conference, VTC 2025-Spring 2025, Oslo, Norway, 17/06/25. https://doi.org/10.1109/VTC2025-Spring65109.2025.11174446

Multi-Task Hypergraph-Attention Framework for Multimodal Sentiment Analysis. / Wang, Yibing; Yang, Zhutian; Wang, Linhan et al.
2025 IEEE 101st Vehicular Technology Conference, VTC 2025-Spring 2025 - Proceedings. Institute of Electrical and Electronics Engineers Inc., 2025. (IEEE Vehicular Technology Conference).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - Multi-Task Hypergraph-Attention Framework for Multimodal Sentiment Analysis

AU - Wang, Yibing

AU - Yang, Zhutian

AU - Wang, Linhan

AU - Liu, Mingqian

AU - Chen, Yushi

PY - 2025

Y1 - 2025

N2 - Multimodal sentiment analysis has emerged as a critical research area. However, existing methods face significant challenges: (1) Unimodal feature extraction techniques often fail to capture the topological structure within data, and do not effectively integrate local and global information, leading to information loss. (2) Traditional multimodal fusion methods, such as concatenation, addition, and multiplication, struggle to model modality differences and inter-modal correlations. In this paper, we propose a novel multi-task hypergraph-attention framework (MTHA) to improve feature discrimination and model performance. Experimental results demonstrate that MTHA outperforms most baseline models in both sentiment classification and regression.

AB - Multimodal sentiment analysis has emerged as a critical research area. However, existing methods face significant challenges: (1) Unimodal feature extraction techniques often fail to capture the topological structure within data, and do not effectively integrate local and global information, leading to information loss. (2) Traditional multimodal fusion methods, such as concatenation, addition, and multiplication, struggle to model modality differences and inter-modal correlations. In this paper, we propose a novel multi-task hypergraph-attention framework (MTHA) to improve feature discrimination and model performance. Experimental results demonstrate that MTHA outperforms most baseline models in both sentiment classification and regression.

KW - Multimodal sentiment analysis

KW - hypergraph learning

KW - multi-task

KW - self-attention

UR - https://www.scopus.com/pages/publications/105019042749

U2 - 10.1109/VTC2025-Spring65109.2025.11174446

DO - 10.1109/VTC2025-Spring65109.2025.11174446

M3 - 会议稿件

AN - SCOPUS:105019042749

T3 - IEEE Vehicular Technology Conference

BT - 2025 IEEE 101st Vehicular Technology Conference, VTC 2025-Spring 2025 - Proceedings

PB - Institute of Electrical and Electronics Engineers Inc.

T2 - 101st IEEE Vehicular Technology Conference, VTC 2025-Spring 2025

Y2 - 17 June 2025 through 20 June 2025

ER -

Multi-Task Hypergraph-Attention Framework for Multimodal Sentiment Analysis

Abstract

Publication series

Conference

Keywords

Access to Document

Other files and links

Fingerprint

Cite this