GN-GCN: Combining Geographical Neighbor Concept with Graph Convolution Network for POI Recommendation

Fan Mo*, Hayato Yamana

*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceedingConference contribution

1 Citation (Scopus)

Abstract

Point-of-interest (POI) recommendation helps users filter information and discover their interests. In recent years, graph convolution network (GCN)–based methods have become state-of-the-art algorithms for improving recommendation performance. Especially integrating GCN with multiple information, such as geographical information, is a promising way to achieve better performance; however, it tends to increase the number of trainable parameters, resulting in the difficulty of model training to reduce the performance. In this study, we mine users’ active areas and extend the definition of neighbors in GCN, called active area neighbors. Our study is the first attempt to integrate geographic information into a GCN POI recommendation system without increasing the number of trainable parameters and maintaining the ease of training. The experimental evaluation confirms that compared with the state-of-the-art lightweight GCN models, our method improves Recall@ 10 from 0.0562 to 0.0590 (4.98%) on Yelp dataset and from 0.0865 to 0.0898 (3.82%) on Gowalla dataset.

Original languageEnglish
Title of host publicationInformation Integration and Web Intelligence - 24th International Conference, iiWAS 2022, Proceedings
EditorsEric Pardede, Pari Delir Haghighi, Ismail Khalil, Gabriele Kotsis
PublisherSpringer Science and Business Media Deutschland GmbH
Pages153-165
Number of pages13
ISBN (Print)9783031210464
DOIs
Publication statusPublished - 2022
Event24th International Conference on Information Integration and Web Intelligence, iiWAS 2022, held in conjunction with the 20th International Conference on Advances in Mobile Computing and Multimedia Intelligence, MoMM 2022 - Virtual, Online
Duration: 2022 Nov 282022 Nov 30

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume13635 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference24th International Conference on Information Integration and Web Intelligence, iiWAS 2022, held in conjunction with the 20th International Conference on Advances in Mobile Computing and Multimedia Intelligence, MoMM 2022
CityVirtual, Online
Period22/11/2822/11/30

Keywords

  • Geographical information
  • Graph convolution network
  • POI recommendation
  • Trainable parameter number-holding

ASJC Scopus subject areas

  • Theoretical Computer Science
  • Computer Science(all)

Fingerprint

Dive into the research topics of 'GN-GCN: Combining Geographical Neighbor Concept with Graph Convolution Network for POI Recommendation'. Together they form a unique fingerprint.

Cite this