NRWalk2Vec-HIN: spammer group detection based on heterogeneous information network embedding over social media.

Online reviews have a significant influence on consumers' purchasing decisions. Unfortunately, many sellers exploit these reviews by employing a large number of spammers who strategically craft fake reviews to enhance their own reputation or tarnish their competitors'. Although several models have been proposed to address this issue in recent years, they often overlook the importance of considering a combination of structural and behavioural features, focusing solely on either structural or textual aspects. To overcome these limitations, we propose a novel model for detecting organized spammer groups called the Fake Reviewers Groups Detection Model. This model comprehensively considers the network structure and reviewer behavioural features to identify such groups. By extracting user, product, review time, and rating information, we construct a heterogeneous information network using a Meta-graph, which explores user relationships. Then apply the Node Ranking Walk2Vec algorithm to generate random walks within this heterogeneous information network, enabling us to obtain low-dimensional vector representations of the user nodes. Subsequently, utilize the Gaussian fuzzy Cluster Means algorithm for clustering, thereby identifying candidate groups of reviewers. The dynamic weight of each detection indicator is determined using the entropy approach, allowing us to assign an appropriate rank to reviewer groups based on their level of suspiciousness and identify them as fake reviewer groups. [ABSTRACT FROM AUTHOR]