多Agent系统驱动的在线交易动态信任与信用管理模型

108 浏览量更新于2024-08-28 收藏 1.53MB PDF 举报

本文是一篇发表在《智能自动化与软计算》(Intelligent Automation & Soft Computing)期刊上的研究论文，标题为“基于多Agent系统的在线交易动态信任计算模型与信用管理机制”。作者包括W.J. Jiang、Y.S. Xu、H.Guo、C.C. Wang 和 L.M. Zhang，该研究发表于2016年9月26日，卷22，第4期，页面范围为639-649。论文主要探讨了在在线交易环境中，多Agent系统如何实现动态的信任评估与信用管理体系。在当前日益复杂的电子商务背景下，多Agent技术被广泛应用于处理分布式、异构的信息共享和决策制定。多Agent系统由多个独立的智能实体（Agent）组成，它们能够在网络中协同工作，模拟人类社会中的合作与竞争关系。动态信任计算模型的核心在于，它考虑了交易过程中时间、行为和环境的变化，实时更新和调整每个Agent之间的信任值。这涉及到信任度的评估因素，如历史交易记录、履行承诺的能力、信息透明度等。通过量化这些因素，模型能够动态反映交易双方的信任状况，有助于降低风险，提高交易效率。信用管理机制部分，论文提出了一套基于多Agent系统的策略，用于监控、维护和恢复信任，确保系统的公平性和可靠性。它可能包括信用评级系统、违约惩罚机制以及信用修复路径。这些机制旨在建立一个可信赖的交易环境，鼓励Agent之间的长期合作，减少欺诈行为。文章引用了DOI:10.1080/10798587.2016.1152778，读者可以通过此链接访问原文或查看相关的交叉标记数据。截至报告日期，该文章已被阅读28次，显示出其在该领域内的关注程度。这篇论文对推动在线交易环境中多Agent系统的信任管理理论和实践具有重要意义，对于理解复杂网络中的信任动态演变，以及设计更有效的信用管理系统提供了有价值的参考。如果你是研究者或者对这一领域感兴趣，深入研究这篇论文将有助于你了解如何在实际应用中构建高效且可靠的在线交易平台。

640 W. J. JIANG ET AL.

practices recommended due to (Edieal, Abraham, & Yaniv,

2003; iResearch, 2006; Zhang, Xie, et al., 2011).

For the credibility index accumulation method, some schol-

ars prevent or detect fraud by designing a reliable mechanism

(McCole) (Cripps, Mailath, & Samuelson, 2004; Peng, Lv, &

Hu, 2004; Resnick & Zeckhauser, 2000). Such a mechanism

design can be divided into two types: One mechanism, how

to make the seller truthfully announce and release product

quality and other information, such as List Fee (Zhang, Yang,

& Wang, 2007) mechanism and GWH (Dellarocas, 2003).

Another way is to promote counterparty credit transactions,

such as TTP (Dellarocas, 2002) and RD & TSD mechanism

(Ba, Whinston, & Zhang, 2003). However, these methods still

have some drawbacks, like List Fee and GWH, although it can

prevent sellers to publish false quality information, but cannot

hedge against the fraud buyer; TTP to assume that a third-

party certication body contains information global trading

parties, but there is no such organization, this strategy is not

practical; RD & TSD assume opponents can accurately predict

the auction transaction volume, is expected auction price is

still not yet complex issues to be resolved, so this mechanism

is also dicult to use.

To solve the problem of trust in the credibility of the cur-

rent network online trading mechanism exists, this paper

presents a Online Transactions Dynamic Trust Computing

Model, OTDTC based on a Multi-Agent Systems, established

online auction trust model calculation methods and credibil-

ity achieve the advance prevention, coordinating things aer-

wards punishment trust transaction management mechanism

of three-in-one. e mechanism put forward the concept of

online trading reputation index, users join the network once

the transaction is approved auction on the online auction has

the same reputation, and both parties resolved this problem is

dicult to identify the initial credit. By publishing removed

and blacklist information to other sites, eectively preventing

the members of this network transactions using fraudulent

exit strategy aer the transaction, and then add other sites to

continue fraudulent phenomenon. Meanwhile, network trans-

actions in order to protect their own reputation and earnings

will restrict speculative seller’s fraud, thereby improving net-

work integrity trading mechanism. e use of “credible threat”

concept (Tian & Guo, 2004), OTDTC mechanism constructed

aerwards credible threat of punishment, according to the

theory of repeated games, OTDTC mechanisms in long-term

transactions are eective, research-C2C online auction trust

model to provide some reference and reference.

2. Network transactions trust model based on mas

rough the above analysis of the existing trust model, draw-

ing on the theoretical basis of the results of research on trust,

we combine the characteristics of C2C online transactions,

the establishment of a C2C online auction dynamic trust

computing model. is model includes feedback evaluation,

transaction value, Ratings trust users, time weight, recent

trust, community contribution of six factors (Abdul-Rahman

& Hailes, 2000; Buyya et al., 2011; He, 2010; Jsang, Ismail, &

Boyd, 2007; Lackermair, 2011; Li, 2009, 2010; Marsh, 1994;

Mui, Mohtashemi, & Halberstadt, 2002; Sinclaire et al., 2010;

Tian & Lin, 2007; Xu & Zhang, 2009; Wu et al., 2010; Yaghoubi

et al., 2011; Yu & Munindar, 2002; Zacharia & Maes, 2000;

Zhang, 2010a; Zhang, 2010b; Zimmer et al., 2010; Zhu, 2011).

2.1. Online trading dynamic trust model

Denition 1 Let C＝{c

…,c

} is the key factors of credit set,

represents a key factor in the i-th credit, specically auction

items quality or price. For a user u, when the auction transac-

tion is successful, the user v credibility with its trading feedback

to the user u evaluates f(v, u) is c

, c

…, c

n-dimensional

vector, i.e. f(v, u)=(

(v, u),

(v, u), …,

(v, u)),

(v, u) ∈

[-1,1] is the user V to user u is reputation feedback evaluation

scores, When the key factors of credit is c

(Jiang, Zhong, Ji, &

Wu, 2011).

Denition 2 Given a user u, set N(u) for user u trading

partners (Jiang, Zhang, & Wang, 2009a).

Denition 3 Given a user u, in [t-1, t] time domain, assum-

ing that x is the user u trading partner, that x∈N(u), t

∈[t-1, t]

represents the time that user x and u auction trading, said the

ρ(t

, t) is x to u gives reputation feedback evaluation scores of

f(x, u) time discount function.

Denition 4 Given a user u, in [t-1, t] time domain, assum-

ing that x is the user u trading partner, called p(x, u) for the

user x and u the value of the transaction.

Given time t, for user u, its trust computational model is

(Jiang, Zhang & Wang, 2009b; Jiang, Zhong, Zhang & Shi, 2013):

where N(u) is a set of user transactions, i.e. the set of [t − 1,t]

in the time domain, the user u of the user transaction. R

rep-

resents t − 1, the user u trust. For N

represents t − 1, the

user u trust. W[p(x,u)] is the weight function of the transaction

value, which can be written as:

where p(x,u) in [t − 1,t] time domain, the value, which the user

x and u the transaction. μ is the network system (auction) the

minimum value of transactions, the transaction value to meet

the insurance claim.

At present, eBay just for the transaction value of not less than

$200 deal with insurance. According to the 2009 i Research

“China Online Auction Research Report”, now the average

trading price of the auction items traded at around $ 200.

According to this report, we set μ=200, i.e.

[p(x, u)] =

p(x,u)

200

Cr[τ

t−1

(x)] is the weight of user trust value x, said the cred-

ibility of the user u transaction object x given feedback evalu-

ation scores of credibility.

ρ(t

,t) is the time discounting function, said time weight of

reputation feedback evaluation score, when t

leaned closer to

t, the weights are user Agent x to user Agent u gives reputation

feedback evaluation score is higher, we put it into

𝜌

, t)=𝜌

t−t

0<ρ<1,which ρ is the time weight factor.



f (x

is average reputation feedback evaluation scores,

which Agent x score for Agent u, aer the end of the sale,

where, |C| is the base consisting of set of the credibility of the

key factors.

𝜏

(u)=

⎧

⎪

⎨

⎪

⎩

𝛼𝜏

t−1

(u)+𝛽e

∑

k∈N(u)

w[p(x,u)]Cr[𝜏

t−1

(x)]𝜌(t

)

⋅



f (x, u)+𝛾cf (u)) N(u) ≠ 0

𝜏

t−1

(u) N(u)=0

(2)

[p(x, u)] =

p(x, u)

𝜇

(3)



f (x, u)=

�C�

∑

i=1

𝜔

(x, u)

�

∑

i=1

𝜔

剩余11页未读，继续阅读

weixin_38501810

粉丝: 2
资源: 942

多Agent系统驱动的在线交易动态信任与信用管理模型

移动agent系统安全问题的动态信任计算模型.pdf

基于Agent的分布式系统信任模型仿真.pdf

基于移动Agent的移动智能商务交易安全机制研究.

基于MAS的移动互联网中移动商务交易安全机制研究

基于强化学习的机器学习欺诈检测技术探索

信贷风险评估的机器学习革命：理论与实践

【机器学习到深度学习】：金融预测技术的突破与演进

springboot网页时装购物系统

开题报告Python游戏辅助和内容更新系统.docx

（全新整理）能源研究专题数据资料大全

最新资源