A unified framework for detecting author spamicity by modeling review deviation

作者：

Highlights：

• Dealing with review spammer detection problem.

• We propose an unified framework for detecting author spamicity.

• A set of abnormity signals were proposed from deviation angle.

• An aspect-based deviation was designed to model latent content deviation.

• Experimental results suggest that our approach is appropriate for this task.

摘要

•Dealing with review spammer detection problem.•We propose an unified framework for detecting author spamicity.•A set of abnormity signals were proposed from deviation angle.•An aspect-based deviation was designed to model latent content deviation.•Experimental results suggest that our approach is appropriate for this task.

论文关键词：Review spam,Spam detection techniques,Fake reviews,Bidirectional LSTM,Review deviation

论文评审过程：Received 16 December 2017, Revised 7 April 2018, Accepted 10 June 2018, Available online 14 June 2018, Version of Record 26 June 2018.

论文官网地址：https://doi.org/10.1016/j.eswa.2018.06.028