Paper Recommendation Method based on Attention Mechanism and Graph Neural Network

At present, most recommendation technologies only consider text or citation information, which suffers from data sparseness and cold start problems. Therefore, an academic paper recommendation method based on attention mechanism and heterogeneous graph CAH is proposed. This method considers textual...

Full description

Saved in:
Bibliographic Details
Published inJournal of Electrical Systems Vol. 20; no. 2; pp. 88 - 95
Main Authors Li, Ailin, Jing, Rong, Guo, Qi, Wei, Bin
Format Journal Article
LanguageEnglish
Published Paris Engineering and Scientific Research Groups 18.04.2024
Subjects
Algorithms
Artificial intelligence
Graph neural networks
Keywords
Neural networks
Recommender systems
Semantics
Online AccessGet full text

Cover

More Information
Summary:At present, most recommendation technologies only consider text or citation information, which suffers from data sparseness and cold start problems. Therefore, an academic paper recommendation method based on attention mechanism and heterogeneous graph CAH is proposed. This method considers textual information and heterogeneous graph structure information to obtain a richer and more complete feature representation. Finally, cosine similarity is calculated to generate recommendations. The results show that compared with the content-based recommendation method, the accuracy rate, recall rate and f value of CAH method are increased by nearly 5.6%, 5.8% and 8.7%, respectively, which are significantly improved compared with the basic method. This method is expected to promote the in-depth application of recommendation systems in the field of artificial intelligence.
ISSN:1112-5209
DOI:10.52783/jes.1101