Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

CDAML: a cluster-based domain adaptive meta-learning model for cross domain recommendation

CDAML: a cluster-based domain adaptive meta-learning model for cross domain recommendation Recommender systems play an important role in providing users required information in a timely and effective manner. However, the cold-start problem of limited historical records of a new user in target domain makes it difficult to model user’s comprehensive preferences. This severely affects the accuracy of recommendation. Although some meta-learning based approaches have alleviated the cold-start problem by learning well-generalized initial parameters for each user, they neglect user’s information in source domains, and seem to be weak in providing each user suitable initial parameters separately. To tackle these challenges, we propose a novel cluster-based domain adaptive meta-learning model for cross-domain recommendation (CDAML). Specially, we utilize the adversarial cross-domain methods to introduce domain adaptation into the meta-learning framework, which can transfer domain-independent user preferences (i.e. intrinsic preferences) from source domains for improved recommendation in target domain via adversarial learning. Besides, we further design a soft-clustering based method to guide the globally shared parameter initialization in a finer granularity of cluster level, which not only contribute to avoid local optima, but also better transfer the shared knowledge among users with similar cross-domain preferences. Finally, comprehensive experiments are conducted on three real-world datasets to demonstrate the superior performance of CDAML compared with state-of-the-art recommendation methods. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png World Wide Web Springer Journals

CDAML: a cluster-based domain adaptive meta-learning model for cross domain recommendation

Xu, Jiajie; Song, Jiayu; Sang, Yu; Yin, Lihua
World Wide Web , Volume 26 (3) – May 1, 2023
Download PDF
15 pages

Loading next page...
 
/lp/springer-journals/cdaml-a-cluster-based-domain-adaptive-meta-learning-model-for-cross-oJ2xWGnesb

References (34)

Publisher
Springer Journals
Copyright
Copyright © The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2022
ISSN
1386-145X
eISSN
1573-1413
DOI
10.1007/s11280-022-01068-5
Publisher site
See Article on Publisher Site

Abstract

Recommender systems play an important role in providing users required information in a timely and effective manner. However, the cold-start problem of limited historical records of a new user in target domain makes it difficult to model user’s comprehensive preferences. This severely affects the accuracy of recommendation. Although some meta-learning based approaches have alleviated the cold-start problem by learning well-generalized initial parameters for each user, they neglect user’s information in source domains, and seem to be weak in providing each user suitable initial parameters separately. To tackle these challenges, we propose a novel cluster-based domain adaptive meta-learning model for cross-domain recommendation (CDAML). Specially, we utilize the adversarial cross-domain methods to introduce domain adaptation into the meta-learning framework, which can transfer domain-independent user preferences (i.e. intrinsic preferences) from source domains for improved recommendation in target domain via adversarial learning. Besides, we further design a soft-clustering based method to guide the globally shared parameter initialization in a finer granularity of cluster level, which not only contribute to avoid local optima, but also better transfer the shared knowledge among users with similar cross-domain preferences. Finally, comprehensive experiments are conducted on three real-world datasets to demonstrate the superior performance of CDAML compared with state-of-the-art recommendation methods.

Journal

World Wide Web – Springer Journals

Published: May 1, 2023

Keywords: Recommendation; Meta-learning; Cross-domain; Adversarial mechanism

There are no references for this article.