Knowledge-Enhanced Dual-Attention Mechanism Recommendation Method
IEEE Access · 2026 · 14 · 42185-42195
DOI: 10.1109/access.2026.3674979 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Access, 2026. Verfasst von Lisi Zhang und 2 weiteren Autor:innen. Die Arbeit nennt 36 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC-ND 4.0). Gefördert von: 2024 University-Level Project “Research on Knowledge-Enhanced Recommendation Methods”, 2025 Young Innovative Talents Project of Regular Institutions of Higher Education in Guangdong Province “Research on Recommendation Methods Based on Knowledge….
Abstract (Originalsprache)
Knowledge graph (KG), when used as supplementary information in the recommendation domain, effectively alleviate the data sparsity and cold start problems inherent to collaborative filtering. However, the majority of current knowledge graph-based recommendation methods focus primarily on the effective encoding of the KG, often overlooking user-specific information. This leads to an inadequate latent semantic representation of both users and items. In this paper, we propose a knowledge-enhanced dual attention mechanism recommendation approach. This method employs a cross-compression fusion unit to capture latent features from user-specific information and interaction history, thus enhancing user feature representations. Furthermore, acknowledging that KG hold varying significance for users and items, we introduce a dual attention mechanism to pay attention to important neighbors of both users and items separately. This is designed to mine the structural and semantic information present in the knowledge graph. To validate the effectiveness of our approach, experiments were conducted on four datasets: MovieLens-1M, MovieLens-20M, Book-Crossing, and Last-FM. Experimental results show that our method significantly outperforms several convincing state-of-the-art baselines.
Förderung
- 2024 University-Level Project “Research on Knowledge-Enhanced Recommendation Methods”
- 2025 Young Innovative Talents Project of Regular Institutions of Higher Education in Guangdong Province “Research on Recommendation Methods Based on Knowledge…
Zitieren
APA
Zhang, L., Zhou, B., & Cui, W. (2026). Knowledge-Enhanced Dual-Attention Mechanism Recommendation Method. IEEE Access, 14, 42185-42195. https://doi.org/10.1109/access.2026.3674979
Vancouver
Zhang L, Zhou B, Cui W. Knowledge-Enhanced Dual-Attention Mechanism Recommendation Method. IEEE Access. 2026;14:42185-42195. doi:10.1109/access.2026.3674979
BibTeX
@article{Zhang2026,
author = {Zhang, Lisi and Zhou, Beijing and Cui, Wei},
title = {Knowledge-Enhanced Dual-Attention Mechanism Recommendation Method},
journal = {IEEE Access},
year = {2026},
volume = {14},
pages = {42185--42195},
doi = {10.1109/access.2026.3674979}
}
EUVIMED