ระบบจัดเก็บและค้นหางานวิจัยอัจฉริยะ

อ.รุ่งทิพย์ โคบาล | Gated Recurrent Unit Graph Transformer Hyperbolic Geometry

จำนวนผู้เข้าชม

Creative Commons License

is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 3.0 Thailand License.

Collection	RMUTK Research Repository (RMUTK IR) วารสารวิชาการ — e-Journal Articles
ID RMUTK Digital	RMUTK000036
Title
Alternative title	HGT4REC: HYPERBOLIC GRAPH TRANSFORMER FOR SEQUENTIAL AND SOCIAL RECOMMENDATION
Authors	ผศ.ดร.นิกร กรรณิกากลาง อ.รุ่งทิพย์ โคบาล ผู้แต่งหลัก อ.ภัททิรา แก้วเกิด Corresponding อ.ดร.จินตนา พลศรี Corresponding ผศ.ดร.ภูริวัตร คัมภีรภาพพัฒน์ ผศ.สุภษี ดวงใส ผศ.สุรเทพ แป้นเกิด ผศ.ดร.ณัฐรฐนนท์ กานต์รวีกุลธนา นางสาววาสนา ด้วงเหมือน ผศ.เจษฎาภรณ์ ยานุพรหม อ.พิรานันท์ จันทวิโรจน์ นายสุกัณยา กิตติคุณงาม ผศ.ดร.สุพัฒนา เตโชชลาลัย
Faculty	คณะบริหารธุรกิจ
Journal Title	Suranaree Journal of Science and Technology
ISSN	0858-849X
Volume / Issue / Pages	Vol.33 \| No.4 \| pp.1-14
Published	2026-09-03
Year	2569
Level	ระดับนานาชาติ (SCOPUS)
Quartile	Q4
DOI	https://doi.org/10.55766/sujst11821
Funding Source	มทร.กรุงเทพ
Abstract	Sequential behaviors and social ties jointly shape user preferences; however, most prior work models them in isolation and relies on shallow fusion in Euclidean space, which struggles to capture temporal drift and hierarchical social structure. We propose a novel framework; HGT4Rec, a Hyperbolic Graph Transformer for Sequential and Social Recommendation. A graph transformer encodes item-transition dependencies to track evolving preferences along the sequence, while a hyperbolic transformer operates on the social graph to represent long-range and hierarchical influence. We further introduce FusionGRU, an adaptive gating module that integrates the two representations into a unified preference state for Top-K prediction. Experiments on Yelp, iFashion, LastFM show that HGT4Rec delivers substantial improvements over the strongest baseline, achieving +348.22% / +314.86%, +383.21% / +210.19%, and +90.85% / +9.83% in Recall@10 and NDCG@10, respectively. Our results demonstrate the value of combining graph transformers with hyperbolic space modeling and the gated fusion for next-item recommendation in sequential and social settings (our code).
Abstract (EN)	Sequential behaviors and social ties jointly shape user preferences; however, most prior work models them in isolation and relies on shallow fusion in Euclidean space, which struggles to capture temporal drift and hierarchical social structure. We propose a novel framework; HGT4Rec, a Hyperbolic Graph Transformer for Sequential and Social Recommendation. A graph transformer encodes item-transition dependencies to track evolving preferences along the sequence, while a hyperbolic transformer operates on the social graph to represent long-range and hierarchical influence. We further introduce FusionGRU, an adaptive gating module that integrates the two representations into a unified preference state for Top-K prediction. Experiments on Yelp, iFashion, LastFM show that HGT4Rec delivers substantial improvements over the strongest baseline, achieving +348.22% / +314.86%, +383.21% / +210.19%, and +90.85% / +9.83% in Recall@10 and NDCG@10, respectively. Our results demonstrate the value of combining graph transformers with hyperbolic space modeling and the gated fusion for next-item recommendation in sequential and social settings (our code).
Keywords (TH)	Gated Recurrent Unit Graph Transformer Hyperbolic Geometry Sequential Recommendation Social Recommendation
Keywords (EN)	Gated Recurrent Unit Graph Transformer Hyperbolic Geometry Sequential Recommendation Social Recommendation
Access Level	Open Access (เปิดสาธารณะ)

ลิงก์แหล่งเผยแพร่

เอกสารฉบับเต็ม (ลิงก์ภายนอก)

https://doi.org/10.55766/sujst11821

เปิดลิงก์