双向跨域推荐模型
Journal: Advances in Computer and Autonomous Intelligence Research DOI: 10.12238/acair.v3i4.17934
Abstract
用户冷启动问题是推荐系统面临的一大挑战,而跨域推荐是解决用户冷启动问题的有效手段之一。以往的跨域推荐主要依赖于用户特征的单向映射,即从源推荐领域到目标推荐领域或从目标推荐领域到源推荐领域,没有有效地融合用户在两个推荐领域上的信息。为了解决上述问题,提出了一种双向跨域推荐模型(Bidirectional Cross Domain Recommendation Model,Bi-CDRM),该模型通过训练两个用户特征映射网络,分别实现用户特征从源域到目标域和从目标域到源域的映射,以得到粗粒度的用户映射特征；引入相关性计算单元,利用注意力机制对粗粒度的用户映射特征和交互物品特征进行加权,最后使用平均池化得到更细粒度的用户全局特征。这一过程不仅提升了跨域推荐模型的性能,也增强了用户特征映射的可解释性。在Amazon数据集的三个跨域推荐场景下,Bi-CDRM在评分预测任务中,效果相较于对比模型有显著的提升。
Keywords
推荐系统；跨域推荐；用户冷启动；数据挖掘；深度学习
Full Text
PDF - Viewed/Downloaded: 0 TimesReferences
[1] LINDEN G,SMITH B, YORK J. Amazon.com recommendations: item-to-item collaborative filtering[J].IEEE Internet Compu
[2] SU M K,LIM E P,ZHU F D.A survey of recommender systems in twitter[C]//Proceedings of the 4th International Conferen
[3] WANG H W,ZHANG F Z,XIE X,et al.DKN: Deep Knowledge-
[4] CHEN C,ZHANG M, ZHANG Y,et al.Efficient heterogeneous collaborative filtering without negative sampling for recommen
[5] TANG J, BELLETTI F,JAIN S,et al. Towards neural mixture recommender for long range dependent user sequences[C]//
[6] LOPS P,DE GEMMIS M,SEMERARO G.Content-based Recommen
[7] HUANG X W,SANG J T,YU J,et al.Learning to Learn a Cold-
[8] ZANG T Z, ZHU Y, LIU H B, et al. A Survey on Cross-domain Recommendation:Taxonomies,Methods,and Future Directions[J]. ACM Transactions on Information Systems,2022,41(2):1-39.
[9] SINGH A P,GORDON G J.Relational Learning via Collect
[2] SU M K,LIM E P,ZHU F D.A survey of recommender systems in twitter[C]//Proceedings of the 4th International Conferen
[3] WANG H W,ZHANG F Z,XIE X,et al.DKN: Deep Knowledge-
[4] CHEN C,ZHANG M, ZHANG Y,et al.Efficient heterogeneous collaborative filtering without negative sampling for recommen
[5] TANG J, BELLETTI F,JAIN S,et al. Towards neural mixture recommender for long range dependent user sequences[C]//
[6] LOPS P,DE GEMMIS M,SEMERARO G.Content-based Recommen
[7] HUANG X W,SANG J T,YU J,et al.Learning to Learn a Cold-
[8] ZANG T Z, ZHU Y, LIU H B, et al. A Survey on Cross-domain Recommendation:Taxonomies,Methods,and Future Directions[J]. ACM Transactions on Information Systems,2022,41(2):1-39.
[9] SINGH A P,GORDON G J.Relational Learning via Collect
Copyright © 2025 郑键珑, 王丹阳
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License