.png){kind=icon}
.png){kind=icon}
Return
HGMRec: A Hypergraph-Based Medical Knowledge-Enhanced Literature Recommendation Model
.png){kind=icon}
PRE
AI
H
W
Y
DOI：10.1109/tbdata.2026.3679457.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
The rapid expansion of medical literature has intensified the challenge of providing researchers and clinicians with accurate and personalized knowledge recommendations. This paper introduces HGMRec, a hypergraph- and medical knowledge graph-enhanced recommendation model designed for heterogeneous and multimodal medical information. The model leverages UMLS (Unified Medical Language System)-based entity recognition and TransE embeddings to inject domain-specific knowledge, and constructs a heterogeneous hypergraph that jointly represents literature, entities, and images. A hypergraph attention mechanism captures higher-order relationships, while a contrastive alignment module ensures semantic consistency across modalities, and a user-side dynamic attention network adapts to evolving user interests. Experiments on two real-world datasets demonstrate that HGMRec significantly outperforms state-of-the-art baselines. HGMRec achieved an NDCG@10 of 0.442, representing a 4.25% improvement over the strongest baseline, and an F1@10 of 0.402, surpassing others by at least 4.7%.
Keywords:
Hypergraph learning
knowledge graph
multimodal fusion
contrastive learning
medical literature recommendation
Journal
I
IF:
5.7
Papers:
834
Citations:
3.0K
Organization
Cited Papers