Analysis of concept drift in fake reviews detection

作者：

Highlights：

• Study the concept drift in fake reviews data.

• We evaluate four different concept drift detectors.

• Tests compared in terms of accuracy, evaluation time and a number of drifts.

• Using real world reviews to evaluate the classifiers.

• Concept drift negatively affects the prediction model performance in fake reviews detection.

摘要

•Study the concept drift in fake reviews data.•We evaluate four different concept drift detectors.•Tests compared in terms of accuracy, evaluation time and a number of drifts.•Using real world reviews to evaluate the classifiers.•Concept drift negatively affects the prediction model performance in fake reviews detection.

论文关键词：Fake reviews,Concept drift detectors,Fake reviews detection techniques,Time-changing data

论文评审过程：Received 17 June 2020, Revised 12 November 2020, Accepted 12 November 2020, Available online 18 November 2020, Version of Record 10 February 2021.

论文官网地址：https://doi.org/10.1016/j.eswa.2020.114318