EUVIMEDEUVIMED entdecken
Crossref

Exploiting Hierarchical Category Information to Improve Next Point-of-Interest Recommendation Via Hyperbolic Graph Convolution Network

Zhuoxuan Huang, Hongyu Zhang, Tianyu Zhou, Hua Ma, Haibin Zhu

IEEE Transactions on Systems, Man, and Cybernetics: Systems · 2026-04 · 56 (4) · 2907-2923

DOI: 10.1109/tsmc.2026.3651345 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEEE Transactions on Systems, Man, and Cybernetics: Systems, 2026-04. Verfasst von Zhuoxuan Huang und 4 weiteren Autor:innen. Die Arbeit wurde bisher 3-mal zitiert (Crossref) und nennt 65 Literaturangaben. Gefördert von: National Natural Science Foundation of China, Natural Science Foundation of Hunan Province, Major Scientific and Technological Innovation Platform Project of Hunan Province.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Next point-of-interest (POI) recommendation is an essential part of location-aware services, aiming to provide users with personalized destinations. The key to improving recommendation performance lies in extracting valuable information from POIs’ features to obtain informative embeddings. However, previous studies usually focus on exploiting the spatiotemporal information of POIs, failing to comprehensively utilize their category information. A novel approach to next POI recommendation is proposed via a hyperbolic graph convolution network (HGCN). In this approach, the category information of POIs is extended to a hierarchical structure for constructing a hierarchical graph. A hierarchy-aware embedding propagation layer is devised based on an HGCN to capture both high-order collaborative signals and the inherent tree-like structure of the graph. Furthermore, an encoder–decoder architecture is employed to integrate the refined POI embeddings to model user preferences and recommend future destinations accurately. Extensive experiments on public datasets demonstrate that the approach outperforms all the baseline ones in most cases, improving accuracy by up to at most 12.52%. Meanwhile, the qualitative analysis based on the embedding visualization further verifies the effectiveness of the proposed approach.

Förderung

  • National Natural Science Foundation of China
  • Natural Science Foundation of Hunan Province
  • Major Scientific and Technological Innovation Platform Project of Hunan Province
  • Changsha Natural Science Foundation of China

Zitieren

APA

Huang, Z., Zhang, H., Zhou, T., Ma, H., & Zhu, H. (2026). Exploiting Hierarchical Category Information to Improve Next Point-of-Interest Recommendation Via Hyperbolic Graph Convolution Network. IEEE Transactions on Systems, Man, and Cybernetics: Systems, 56(4), 2907-2923. https://doi.org/10.1109/tsmc.2026.3651345

Vancouver

Huang Z, Zhang H, Zhou T, Ma H, Zhu H. Exploiting Hierarchical Category Information to Improve Next Point-of-Interest Recommendation Via Hyperbolic Graph Convolution Network. IEEE Transactions on Systems, Man, and Cybernetics: Systems. 2026;56(4):2907-2923. doi:10.1109/tsmc.2026.3651345

BibTeX
@article{Huang2026,
  author = {Huang, Zhuoxuan and Zhang, Hongyu and Zhou, Tianyu and Ma, Hua and Zhu, Haibin},
  title = {Exploiting Hierarchical Category Information to Improve Next Point-of-Interest Recommendation Via Hyperbolic Graph Convolution Network},
  journal = {IEEE Transactions on Systems, Man, and Cybernetics: Systems},
  year = {2026},
  volume = {56},
  number = {4},
  pages = {2907--2923},
  doi = {10.1109/tsmc.2026.3651345}
}
Kostenlos starten →