A Hierarchical Attention Network for Cross-Domain Group Recommendation
IEEE Transactions on Neural Networks and Learning Systems · 2024-03 · 35 (3) · 3859-3873
DOI: 10.1109/tnnls.2022.3200480 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Transactions on Neural Networks and Learning Systems, 2024-03. Verfasst von Ruxia Liang und 3 weiteren Autor:innen. Die Arbeit wurde bisher 33-mal zitiert (Crossref) und nennt 55 Literaturangaben. Gefördert von: Australian Research Council (ARC) under the Australian Laureate Fellowship, National Natural Science Foundation of China, China Scholarship Council.
Abstract (Originalsprache)
Many online services allow users to participate in various group activities such as online meeting or group buying, and thus need to provide user groups with services that they are interested. The group recommender systems (GRSs) emerge as required and provide personalized services for various online user groups. Data sparsity is an important issue in GRSs, since even fewer group-item interactions are observed. Moreover, the group and the group members have complex and mutual relationships with each other, which exacerbates the difficulty in modeling the preferences of both a group and its members for recommendation. The cross-domain recommender system (CDRS) is a solution to alleviate data sparsity and assist preference modeling by transferring knowledge from a source domain which has relatively dense data to another. The existing CDRSs are usually developed for individual users and cannot be directly applied for group recommendation. To alleviate the data sparsity issue in GRSs, we first study the cross-domain group recommendation problem and propose a hierarchical attention network-based cross-domain group recommendation method, called HAN-CDGR. HAN-CDGR takes the advantage of data from a source domain to benefit recommendation generation for both the individual users and groups in the target domain which has data sparsity and cannot generate accurate recommendation. In HAN-CDGR, a hierarchical attention network is constructed to learn and model individual and group preferences, with consideration of both group members' interactions and dynamic weights and the complex relationships between individuals and groups. Adversarial learning is used to effectively transfer knowledge from a source domain to the target domain. Extensive experiments, which demonstrate the effectiveness and superiority of our proposal, providing accurate recommendation for both individual users and groups, are conducted on three tasks.
Förderung
- Australian Research Council (ARC) under the Australian Laureate Fellowship
- National Natural Science Foundation of China
- China Scholarship Council
Zitieren
APA
Liang, R., Zhang, Q., Wang, J., & Lu, J. (2024). A Hierarchical Attention Network for Cross-Domain Group Recommendation. IEEE Transactions on Neural Networks and Learning Systems, 35(3), 3859-3873. https://doi.org/10.1109/tnnls.2022.3200480
Vancouver
Liang R, Zhang Q, Wang J, Lu J. A Hierarchical Attention Network for Cross-Domain Group Recommendation. IEEE Transactions on Neural Networks and Learning Systems. 2024;35(3):3859-3873. doi:10.1109/tnnls.2022.3200480
BibTeX
@article{Liang2024,
author = {Liang, Ruxia and Zhang, Qian and Wang, Jianqiang and Lu, Jie},
title = {A Hierarchical Attention Network for Cross-Domain Group Recommendation},
journal = {IEEE Transactions on Neural Networks and Learning Systems},
year = {2024},
volume = {35},
number = {3},
pages = {3859--3873},
doi = {10.1109/tnnls.2022.3200480}
}
EUVIMED