Article: Multi-modal advertising visual information fusion using dynamic heterogeneous graph neural networks Journal: International Journal of Information and Communication Technology (IJICT) 2026 Vol.27 No.63 pp.108 - 124 Abstract: To enhance the fusion performance of multimodal advertising content in visual comprehension and behaviour prediction, this paper proposes a DHGNN-based multimodal information fusion model. The model comprises modal feature embedding, dynamic node/edge attribute modelling, improved heterogeneous graph convolutions, and a multi-task output layer, optimising CTR, CVR, and semantic classification. Compared with the conventional Hetero GCN and MM Attn models, the proposed DHGNN improved the AUC by 3.8-5.9 p.p. on CTR, and the F1-score was 0.726, which was 8.8 percentage points above Homogeneous GCN. Moreover, the inference latency is reduced to 19.4 ms, while maintaining competitive predictive accuracy. Across Ali-AD and Tencent-Ad360, DHGNN achieves AUC gains of 3.8-5.9 percentage points on CTR and reaches an F1-score of 0.726 on semantic classification, providing quantitative evidence that the proposed dynamic heterogeneous graph modelling improves multimodal fusion under heterogeneous and time-varying advertising data for multimodal recommendation systems. Inderscience Publishers - linking academia, business and industry through research

Title: Multi-modal advertising visual information fusion using dynamic heterogeneous graph neural networks

Authors: Ruiqi Du

Addresses: College of Design and Creativity, Zhejiang Normal University, Jinhua 321000, China

Abstract: To enhance the fusion performance of multimodal advertising content in visual comprehension and behaviour prediction, this paper proposes a DHGNN-based multimodal information fusion model. The model comprises modal feature embedding, dynamic node/edge attribute modelling, improved heterogeneous graph convolutions, and a multi-task output layer, optimising CTR, CVR, and semantic classification. Compared with the conventional Hetero GCN and MM Attn models, the proposed DHGNN improved the AUC by 3.8-5.9 p.p. on CTR, and the F1-score was 0.726, which was 8.8 percentage points above Homogeneous GCN. Moreover, the inference latency is reduced to 19.4 ms, while maintaining competitive predictive accuracy. Across Ali-AD and Tencent-Ad360, DHGNN achieves AUC gains of 3.8-5.9 percentage points on CTR and reaches an F1-score of 0.726 on semantic classification, providing quantitative evidence that the proposed dynamic heterogeneous graph modelling improves multimodal fusion under heterogeneous and time-varying advertising data for multimodal recommendation systems.

Keywords: multimodal fusion; dynamic heterogeneous graph neural network; DHGNN; ad recommendation; semantic alignment; cross-modal modelling; temporal evolution modelling; heterogeneous graph attention; click-through rate prediction; semantic consistency learning; cold-start advertising scenario.

DOI: 10.1504/IJICT.2026.154106

International Journal of Information and Communication Technology, 2026 Vol.27 No.63, pp.108 - 124

Received: 03 Feb 2026
Accepted: 18 Mar 2026
Published online: 12 Jun 2026 *

Title: Multi-modal advertising visual information fusion using dynamic heterogeneous graph neural networks

Keep up-to-date