Graph-based review spammer group detection

Online product reviews nowadays are increasingly prevalent in E-commerce websites. People often refer to product reviews to evaluate the quality of a product before purchasing. However, there have been a large number of review spammers who often work collaboratively to promote or demote target produ...

Full description

Saved in:
Bibliographic Details
Published in:Knowledge and information systems Vol. 55; no. 3; pp. 571 - 597
Main Authors: Wang, Zhuo, Gu, Songmin, Zhao, Xiangnan, Xu, Xiaowei
Format: Journal Article
Language:English
Published: London Springer London 01.06.2018
Springer Nature B.V
Subjects:
Algorithms
Computer Science
Consumer behavior
Data Mining and Knowledge Discovery
Database Management
Electronic commerce
Graph theory
Graphs
Information Storage and Retrieval
Information Systems and Communication Service
Information Systems Applications (incl.Internet)
IT in Business
Machine learning
Product reviews
Regular Paper
Websites
Review spammer groups
Review spam
Opinion mining
Fraud detection
Bi-connected graph
Graph mining
ISSN:0219-1377, 0219-3116
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Online product reviews nowadays are increasingly prevalent in E-commerce websites. People often refer to product reviews to evaluate the quality of a product before purchasing. However, there have been a large number of review spammers who often work collaboratively to promote or demote target products, which severely harm the review system. Much previous work exploits machine learning approaches to detect suspicious reviews/reviewers. In this paper, we introduce a top-down computing framework, namely GGSpam , to detect review spammer groups by exploiting the topological structure of the underlying reviewer graph which reveals the co-review collusiveness. A novel instantiation of GGSpam , namely GSBC, is designed by modeling spammer groups as bi-connected graphs. Given a reviewer graph, GSBC identifies all the bi-connected components whose spamicity scores exceed the given spam threshold. For large unsuspicious bi-connected graphs, the minimum cut algorithm is used to split the graph, and the smaller graphs are further processed recursively. A variety of group spam indicators are designed to measure the spamicity of a spammer group. Experimental study shows that the proposed approach is both effective and efficient and outperforms several state-of-the-art baselines, including graph based and non-graph based, by a large margin.
Bibliography:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:0219-1377
0219-3116
DOI:10.1007/s10115-017-1068-7