EUVIMEDEUVIMED entdecken
Crossref

GDB-TR: Graph-Based Double-Layer Bidirectional Model for Query-Based Trip Recommendation

Xueyao Wang, Wenjing Luan, Liang Qi, Guanjun Liu

IEEE Transactions on Computational Social Systems · 2026-06 · 13 (3) · 3629-3642

DOI: 10.1109/tcss.2026.3659699 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEEE Transactions on Computational Social Systems, 2026-06. Verfasst von Xueyao Wang und 3 weiteren Autor:innen. Die Arbeit nennt 47 Literaturangaben. Gefördert von: Natural Science Foundation of Shandong Province.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Query-based trip recommendation is an important task in location-based services (LBS), which aims to provide users with a sequence of points of interest (POIs) based on their queries. In trip recommendation, the effect of the visited POIs on the following decisions of users, called a forward effect, is mined by the existing studies. However, the effect of the following POIs on the previously visited ones, called a reverse effect, receives no attention. Therefore, this work proposes a graph-based double-layer bidirectional model for trip recommendation (GDB-TR), which is designed to mine both forward and reverse effects through bidirectional computation. Forward computation explores the influence of user’s history preferences on the next choice; reverse computation explores the influence of future goals on the current decision. Specifically, the model uses a heterogeneous graph to model users’ check-in trajectories with spatial and temporal information. Subgraphs are extracted from the heterogeneous graph, and an adjacency matrix is built for each subgraph. Vector representations of POIs and POI categories are obtained by fusing matrices based on a neural network. The double-layer bidirectional neural network is used to recommend a trip based on the user query, with one layer mining users’ preferences for POIs and the other layer mining the preferences for POI categories. Bidirectional computation is performed between the initial and destination nodes in each layer, capturing both forward and reverse effects. Specifically, a forward computation mines the influence of preceding POIs or POI categories on following ones, while a reverse computation does that reversely. Finally, experiments are conducted on five popular real data sets. The results show that GDB-TR outperforms all baseline models on F 1 and pairs-F 1 values, which validates the effectiveness of the proposed approach.

Förderung

  • Natural Science Foundation of Shandong Province

Zitieren

APA

Wang, X., Luan, W., Qi, L., & Liu, G. (2026). GDB-TR: Graph-Based Double-Layer Bidirectional Model for Query-Based Trip Recommendation. IEEE Transactions on Computational Social Systems, 13(3), 3629-3642. https://doi.org/10.1109/tcss.2026.3659699

Vancouver

Wang X, Luan W, Qi L, Liu G. GDB-TR: Graph-Based Double-Layer Bidirectional Model for Query-Based Trip Recommendation. IEEE Transactions on Computational Social Systems. 2026;13(3):3629-3642. doi:10.1109/tcss.2026.3659699

BibTeX
@article{Wang2026,
  author = {Wang, Xueyao and Luan, Wenjing and Qi, Liang and Liu, Guanjun},
  title = {GDB-TR: Graph-Based Double-Layer Bidirectional Model for Query-Based Trip Recommendation},
  journal = {IEEE Transactions on Computational Social Systems},
  year = {2026},
  volume = {13},
  number = {3},
  pages = {3629--3642},
  doi = {10.1109/tcss.2026.3659699}
}
Kostenlos starten →