Detection of spam reviews through a hierarchical attention architecture with N-gram CNN and Bi-LSTM

作者：

Highlights：

• We proposed a novel hierarchical attention architecture for spam review detection.

• The Word2Sent-level captures multi-granularity and informative information.

• The Sent2Doc-level extracts comprehensive and important information.

• Extensive experiments demonstrate that our model has superior detection performance.

摘要

•We proposed a novel hierarchical attention architecture for spam review detection.•The Word2Sent-level captures multi-granularity and informative information.•The Sent2Doc-level extracts comprehensive and important information.•Extensive experiments demonstrate that our model has superior detection performance.

论文关键词：Deceptive review detection,Hierarchical attention,N-gram CNN,Document representation,Bi-LSTM

论文评审过程：Received 20 July 2020, Revised 12 July 2021, Accepted 18 July 2021, Available online 29 July 2021, Version of Record 27 September 2021.

论文官网地址：https://doi.org/10.1016/j.is.2021.101865