False comments network identification and feature discovery
To identify the increasingly specialized fake shopping reviews in bulk from the perspective of net-work structure characteristics,this paper first identifies similar reviewer networks through a comprehensive suspicious score of network behavior and cluster analysis based on group strainer.It then compares the char-acteristics of real reviews and fake reviews across four types of products from the perspective of consumer micro-decisions to refine the identification rules for fake reviews.The results verify that there is a typical group structure in the fake review network.Unlike real reviews,fake reviews are very closely connected with the tar-get product as the medium.In the network topology formed by"reviewer-product-review",the product type is at the core of the fake review network structure and determines the attack pattern of the fake review group.The final heterogeneity of fake reviews is also related to product types,which means that designing differentiated fake review identification strategies according to different product types can improve the accuracy rate.This paper provides a valuable reference for regulating the negative impact of fake reviews on online shopping.
万方数据
维普
