Coarse-to-fine: A dual-view attention network for click-through rate prediction

作者：

Highlights：

• Extracting features from different views can add effective prior information.

• Attention is used based on the user’s differential preferences and the item’s similarities.

• Using sigmoid to model the relation can accommodate the problem of a large number of inputs.

• Remove noisy behaviors to refine behavior list with the help of attention scores.

摘要

•Extracting features from different views can add effective prior information.•Attention is used based on the user’s differential preferences and the item’s similarities.•Using sigmoid to model the relation can accommodate the problem of a large number of inputs.•Remove noisy behaviors to refine behavior list with the help of attention scores.

论文关键词：Click-through rate prediction,Attention,Selection mechanism

论文评审过程：Received 23 January 2020, Revised 6 January 2021, Accepted 7 January 2021, Available online 21 January 2021, Version of Record 6 February 2021.

论文官网地址：https://doi.org/10.1016/j.knosys.2021.106767