DeepAssociate: A deep learning model exploring sequential influence and history-candidate association for sequence recommendation
作者：
Highlights：
• History-candidate association extraction benefits preference modeling.
• Integrating history-candidate association in recommendation improves performance.
• Adopting auxiliary training in sequential recommendation reduces model complexity.
• Weighted objective function accelerates model convergence.
• Incorporating training loss from preference modeling enhances model performance.
摘要
•History-candidate association extraction benefits preference modeling.•Integrating history-candidate association in recommendation improves performance.•Adopting auxiliary training in sequential recommendation reduces model complexity.•Weighted objective function accelerates model convergence.•Incorporating training loss from preference modeling enhances model performance.
论文关键词：Sequential recommendation,User preference,Deep learning,Attention mechanism,Recurrent neural network
论文评审过程：Received 2 September 2020, Revised 18 June 2021, Accepted 8 July 2021, Available online 13 July 2021, Version of Record 29 July 2021.
论文官网地址：https://doi.org/10.1016/j.eswa.2021.115587
