An Adaptive Distributed Consumer Trust Model for Social Commerce

Figures & data

Figure 1. Structure of social commerce trust model.

Table 1. Consumer agent attributes.

Attribute	Description	Value
subjectivityList	List of consumer’s subjectivity	[−0.4 -0.2 0 0.2 0.4]
eaLevel	Emotional attachment level	μ = 0.65, σ = 0.2, [0.3, 1]
activityLevel	Purchase activity level	μ = 0.625, σ = 0.2, [0.25, 1]
purchaseNumber	Number of purchases	RT
utilityGain	Total utility gain	RT
acquaintanceRepo	Repository of direct acquaintance	RT
ownRatingRepo	Repository of evaluation by oneself	RT
adviserRatingRepo	Repository of evaluation by adviser	RT
adviserLevelRepo	Repository of adviser level	RT
providerHonestyRepo	Repository of provider honesty	RT
blackList	Blacklist of inferior product provider	RT

1. The element values of subjectivityList correspond to the $\mathit{\theta }$ in Equation (21)(21) $R_{\mathit{sub}}=\left\{\begin{array}{c}{R}_{\mathit{obj}}+\mathit{\theta }\times \left(R_{\mathit{obj}}-R_{\min }\right),\mathit{\theta }<0\\ R_{\mathit{obj}}+\mathit{\theta }\times \left(R_{\mathit{max}}-R_{\mathrm{obj}}\right),\mathit{\theta }\ge 0\end{array}\right.$(21) . 2. RT indicates that the attribute value is dynamically generated during model runtime.

Table 2. Provider product utility classification.

Type	Ratio	Utility
good	(1 - IPR) * 0.2	μ = 7.5, σ = 1, [5, 10]
ordinary	(1 - IPR) * 0.8	μ = 2.5, σ = 2, [0, 5]
bad	IPR *0.9	μ=-5, σ = 2, [−10, 0]
unstable	IPR *0.1	[−10, 10]

IPR value is in Table 5.

Table 3. Provider honesty classification.

Type	Ratio	Honesty
veryHonest	1 - DPR	1
sometimesDishonest	DPR *0.8	μ = 0.5, σ = 0.1, [0.3, 0.7]
veryDishonest	DPR *0.2	0

DPR value is in Table 5.

Table 4. Provider agent attributes.

Parameter	Description	Value
type	Product utility type	See Table 2
utility	Product utility value	RT
characteristic	Product characteristic	0, 1, 2, 3, 4
judgerRatingRepo	Repository of evaluation by judger	RT
honesty	Provider honesty	See Table 3

RT stands for runtime.

Table 5. Main experimental parameters.

Parameter	Description	Value
maxSimTime	Number of experiment executions	10
maxTick	Number of rounds per experiment	200
consumerNumber	Number of consumer agents per group	500
providerNumber	Number of provider agents	100
acquaintanceCircle	Circle of consumer agents’ direct acquaintances	9
consumerVision	Vision of consumer agents	12
breadthThold	Breadth threshold for consumers to search advisers	3
depthThold	Depth threshold for consumers to search advisers	2
dishonestProviderRatio	Ratio of dishonest providers	0.6
inferiorProviderRatio	Ratio of inferior product providers	0.5
maxUtility	Maximum utility of the product	10
minUtility	Minimum utility of the product	−10
maxEaLevel	Maximum level of emotional attachment	1
minEaLevel	Minimum level of emotional attachment	0.3
utilityChangeProb	Utility change probability of stable product	0.1
utilityChangeValue	Utility change value of stable product	RN: [−1, 1]
providerTypeChangeProb	Probability of providers changing product utility type	0.01
providerMoveProb	Probability of providers moving in the market	0.1
providerExitProb	Probability of providers exiting from the market	0.01
recencyScalingFactor	Recency scaling factor (γ in Equation (3)(3) $\mathit{\omega }\left(\mathit{t}\right)=e^{\frac{-\mathrm{\Delta }t\left(r_i\right)}{\mathit{\gamma }}}$(3) )	−5/ln(0.5)
eaLvlThold	Threshold of emotional attachment level (Ethold in Equation (4)(4) $P_{\mathit{fa}}=\left\{\begin{array}{c}{e}^{\frac{-(E_1+E_2)/2}{\mathit{\phi }}},(E_1+E_2)/2\le E_{\mathit{thold}}\\ 0,(E_1+E_2)/2>E_{\mathit{thold}}\end{array}\right.$(4) )	0.8
deceptProbScalingFactor	Deception probability scaling factor (φ in Equation (4)(4) $P_{\mathit{fa}}=\left\{\begin{array}{c}{e}^{\frac{-(E_1+E_2)/2}{\mathit{\phi }}},(E_1+E_2)/2\le E_{\mathit{thold}}\\ 0,(E_1+E_2)/2>E_{\mathit{thold}}\end{array}\right.$(4) )	0.2
maxRating	Maximum rating (Rmax in Equation (5)(5) $R_f=\left\{\begin{array}{c}{R}_{\mathit{og}}-\mathit{\beta }\times \left(R_{\mathit{og}}-R_{\min }\right),R_{\mathit{og}}>\frac{({\mathit{R}}_{\mathit{max}}+R_{\min })}{2}\\ R_{\mathit{og}}+\mathit{\beta }\times \left(R_{\mathit{max}}-R_{\mathrm{og}}\right),R_{\mathit{og}}\le \frac{({\mathit{R}}_{\mathit{max}}+R_{\min })}{2}\end{array}\right.$(5) )	1
minRating	Minimum rating (Rmin in Equation (5)(5) $R_f=\left\{\begin{array}{c}{R}_{\mathit{og}}-\mathit{\beta }\times \left(R_{\mathit{og}}-R_{\min }\right),R_{\mathit{og}}>\frac{({\mathit{R}}_{\mathit{max}}+R_{\min })}{2}\\ R_{\mathit{og}}+\mathit{\beta }\times \left(R_{\mathit{max}}-R_{\mathrm{og}}\right),R_{\mathit{og}}\le \frac{({\mathit{R}}_{\mathit{max}}+R_{\min })}{2}\end{array}\right.$(5) )	0
manipulationExtent	Manipulation extent of adviser rating (β in Equation (5)(5) $R_f=\left\{\begin{array}{c}{R}_{\mathit{og}}-\mathit{\beta }\times \left(R_{\mathit{og}}-R_{\min }\right),R_{\mathit{og}}>\frac{({\mathit{R}}_{\mathit{max}}+R_{\min })}{2}\\ R_{\mathit{og}}+\mathit{\beta }\times \left(R_{\mathit{max}}-R_{\mathrm{og}}\right),R_{\mathit{og}}\le \frac{({\mathit{R}}_{\mathit{max}}+R_{\min })}{2}\end{array}\right.$(5) )	RN: 0.8 or 1
ownTrustWeight	Importance of own trust (Wo in Equation (7)(7) $R_{\mathit{oa}}\left(\mathit{c},\mathit{p},\mathit{d}\right)=\frac{\sum _{\mathit{s}\in \{\mathit{o},\mathit{a},\mathit{j}\}}{R}_s\left(\mathit{c},\mathit{p},\mathit{d}\right)\times W_s\times {\rho }_s}{\sum _{\mathit{s}\in \{\mathit{o},\mathit{a},\mathit{j}\}}{W}_s\times {\rho }_s}$(7) )	1
adviserTrustWeightDefault	Default importance of adviser trust (Wad in Equations (8(8) $W_s=\left\{\begin{array}{c}\frac{\sum _{\mathit{i}=1}^n\mathit{w}{t}_{s_i}\times \mathit{c}{r}_{s_i}}{\sum _{\mathit{i}=1}^n\mathit{w}{t}_{s_i}}\\ W_{\mathit{sd}}\end{array},\mathit{s}\in \left\{\mathit{a},\mathit{j}\right\}\right.$(8) ))	0.5
judgerTrustWeightDefault	Default importance of judger trust (Wjd in Equations (8(8) $W_s=\left\{\begin{array}{c}\frac{\sum _{\mathit{i}=1}^n\mathit{w}{t}_{s_i}\times \mathit{c}{r}_{s_i}}{\sum _{\mathit{i}=1}^n\mathit{w}{t}_{s_i}}\\ W_{\mathit{sd}}\end{array},\mathit{s}\in \left\{\mathit{a},\mathit{j}\right\}\right.$(8) ))	0.25
timeIntervalElastParam	Elasticity parameter of time interval (λ1 in Equation (17)(17) $\mathit{Wl}=\prod _{\mathit{i}=1}^n\mathit{w}{l}_i^{{\lambda }_i},\mathit{n}\ge 1$(17) )	0.5
quantityElastParam	Elasticity parameter of quantity (λ2 in Equation (17)(17) $\mathit{Wl}=\prod _{\mathit{i}=1}^n\mathit{w}{l}_i^{{\lambda }_i},\mathit{n}\ge 1$(17) )	0.5
quantitySlopeFactor	Quantity slope factor (αq in Equation (18)(18) $\mathit{w}{l}_q=1-e^{-{\alpha }_q\times \mathit{q}}$(18) )	-ln(0.5)
acceptableRatingThreshold	Threshold of acceptable rating (ARthold in Equation (19)(19) $P_{\mathit{bl}}\left(\mathit{c},\mathit{p}\right)=1-e^{\left(-{\mathit{F}}_{\mathit{bl}}\right)}$(19) )	0.6

RN indicates that the random number.

Figure 2. Performance based on four scenarios.

Figure 3. Performance based on four scenarios with changes in the dishonest providers ratio.

Figure 4. Performance based on four scenarios with changes in the inferior product providers ratio.

Figure 5. Performance based on four scenarios with changes in the ratio of deceptive behavior of advisers.

Figure 6. Performance based on four scenarios with changes in the number of providers.

Figure 7. Performance based on four scenarios with changes in the dynamics of providers.

Data Availability Statement

The authors confirm that the data supporting the findings of this study are available within the article. For more data, contact the corresponding author.