Skip to main content

Table 4 Parameter settings

From: Detecting fake reviewers in heterogeneous networks of buyers and sellers: a collaborative training-based spammer group algorithm

Method

Parameter

Description

Value

SGDCTH (Ours)

\(\delta_{{\widetilde{p}}}\)

Target product filtration threshold

0.5–0.7

ϵ

Neighborhood radius threshold

0.4–0.8

\(\phi\)

Minimum number of sample points threshold

2, 3, 4

\(L\)

Number of GNN layers

2

\(lr\)

Learning rate

0.01

\(\lambda\)

Learnable parameter or fixed scalar

0

\(t\)

Number of epochs

50

\(d\)

Embedding dimension of nodes

128

GSDB

\(\delta_{p}\)

Target product filtration threshold

0.1

\(\delta_{I}\)

Individual spamicity threshold

0.43

\(\delta_{G}\)

Group spamicity threshold

0.57

GSBC

\(\tau\)

Co-review time window size

30

\(\delta\)

Edge weight threshold

0.1

\(MP\)

User-specified parameter

1000

\(\delta_{G}\)

Group spamicity threshold

0.53

GroupStrainer

\(\delta_{p}\)

Target product filtration threshold

0.75

HoloScope

\(b\)

Scaling base

32