Spotting review spammer groups: A cosine pattern and network basedmethod

摘要

Nowadays, online product reviews strongly influence the purchase decision of consumers inrne-commerce platforms. Driven by the immense financial profits, review spammers deliberatelyrnpost fake reviews to promote or demote their target products. Some spammers are even organizedrnas groups to work together and try to take total control of the sentiment on their targetrnproducts. To detect such spammer groups, most previous works exploit frequent itemset miningrn(FIM) to find spammer group candidates and then use unsupervised spamicity ranking methodsrnto identify real spammer groups. However, these methods usually suffer from the problem ofrnthreshold setting, ie, high support value finding fewer groups while low support value leadingrnto more coincidentally generated groups and computational inefficiency. Moreover, the unsupervisedrnmethods are not able to make good use of labeled instances which are actually obtainablernin practice. In this paper, we propose CONSGD, a cosine pattern and heterogeneous informationrnnetwork–based spammer group detecting method. Specifically, the CONSGD uses cosine patternrnmining (CPM) to discover tight spammer group candidates with a respective low supportrnvalue, where the cosine threshold is utilized to avoid coincidentally generated groups. Moreover,rnCONSGD employs heterogeneous information network classification to identify the real spammerrngroups, which could utilize the labeled instances and do not rely to the assumption of independentrninstances. Experiments on real-life dataset show that our proposedCONSGDis effectivernand outperforms the state-of-the-art spammer group detection methods.