【作者】 胡润波；

【导师】 杨德礼；

【作者基本信息】 大连理工大学 ， 管理科学与工程， 2010， 博士

【摘要】 随着现代无线通信技术的快速发展,移动商务作为一种新兴的虚拟电子交易模式,正对社会、经济的各个层面产生深远的影响。移动商务消除交易双方在时间和地域上的限制,给移动商家带来巨大的商机,但是,也造成了交易双方严重的信息不对称,给欺诈造成更多可乘之机,使得消费者对移动商务市场产生了信任缺失。移动商务交易欺诈现象以及信任缺失问题,已经成为阻碍移动商务发展的主要因素之一。因此,治理移动商务交易环境,降低信息不对称程度,规避交易风险,建立诚信交易环境,成为移动商务发展必须面对的现实问题。第三方信任评价作为一种能有效降低信息不对称,有效减少移动商务交易风险的管理工具,已受到愈来愈多的关注。构建合理、可靠的第三方信任评价方法对于构建诚信、安全、可靠的移动商务交易平台,促进移动商务发展具有重要的意义。鉴于此,本文做了以下三方面的探索性工作：(1)研究了基于推荐信息的移动商务信任评价方法。以当前国内外权威经典文献对移动商务环境下用户信任影响因素的观点为基础,通过对移动商务交易过程结构关系的理性分析,建立移动商家推荐信任评价指标体系。保证了评价指标在移动商务环境中的适应性、权威性和可靠性。通过所有推荐人评价之间偏差平方和最小的方法,降低了推荐信息之间的矛盾,避免了单个推荐人片面性带来的评价结果的偏差,保证了在推荐人数量较少的条件下仍然可以对移动商家信任度进行评价。在此基础上,根据移动终端受限等特点,采用改进的ELECTREⅢ方法刻画移动商务环境中消费者信任评价决策过程,构建了适合移动商务环境的推荐信任评价方法,保证了评价结果真实反映推荐人群体意见和消费者信任评价意图,有利于消费者对移动商家的可信任度作出快捷和准确的评价。(2)研究了基于在线商家信用反馈信息的移动商务信任评价方法。一是基于内容分析法建立了移动商家信用评价指标体系。根据当前相关研究权威经典文献的观点,在对有关专家和用户深入访谈的基础上,采用内容分析法对淘宝WAP网站中的文字信用评价进行分析,通过分类整理构建了移动商家信用评价指标体系,并对指标形成的原因进行了分析。通过检验相互认同度和Krippendorffi’s alpha系数保证了分析结果的有效性和可靠性。通过分析消费者评价时的态度倾向,解决了仅通过频率分析无法检验指标相关性的问题。在分析信用评价指标产生原因的基础上,进一步提炼出消费者最关注的3个指标,为移动商家有效地把握消费者需求,提高自身信用提供了有力指导。二是基于群体决策的思想建立了移动商家多维信用反馈评价模型。首先,根据移动商务终端设备的特性,选取了便于消费者在移动商务环境下使用的信用度量模式,有利于采集原始信用反馈数据。又在已有多维信用评价模型的基础上,根据移动商务交易特点,补充了交易金额权重函数和时间折扣函数,提高了对多维信用反馈评分集结模型常见欺诈行为的预防能力。然后,引入了两种新的赋权方法确定权重,通过专家赋权可以汲取专家的经验,通过频率统计分析赋权可以揭示整个消费群体对指标重要性的偏好。与现有在线商家信用评价指标赋权方法相比,两种新引入的赋权方法增加了组合权重计算结果的权威性和可靠性。最后,基于群体决策的思想建立了一种基于TOPSIS思想和广义最大熵的组合赋权模型,改进了现有单一赋权方法评价结果差异较大的问题。通过组合赋权,既保留了熵值赋权法等客观赋权对实际情况的真实反映,又反映了Delphi等主观赋权法所体现的消费者对指标的重视程度以及专家知识。避免了现有研究的主观赋权法无法反映客观条件变化,客观赋权法无法反映消费者和专家意见的弊端。(3)研究了基于移动商盟所提供信息的移动商务信任评价方法。借鉴传统交易环境中商会联盟的基本功能,设计了基于移动商盟管理的信任交易机制,从事前防范,事中协调,事后惩罚三个方面约束交易双方的欺诈行为。采用重复博弈理论,证明了该机制在长期交易中的有效性,能够降低交易风险,保障移动商务交易的安全性。该机制能够防止会员通过变换身份进行欺诈；降低网络交易平台仲裁的工作量。然后,依据移动商盟所提供的会员商家信任度评价信息,建立了移动商家信任度评价指标体系。在此基础上,从移动商盟信用指数和会员商家盟内信用指数两个方面建立了二维移动商家信任度评价模型,解决了难以识别新注册商家的信任度,新注册的诚信商家难以开展业务的问题。更多还原

【Abstract】 With the rapid development of wireless communication technology, mobile commerce (called m-commerce as a brief) becomes a newly virtual electronic transaction mode, and has a far-reaching influence on every aspect of economy and society. By removing restricts of time and regions, m-commerce can bring enormous business opportunities. However, because serious information asymmetry between two parties results in more opportunity for fraud, consumer loses confidence. The fraud phenomenon and trust minsing problem in m-comerce become one of the main factors that obstruct the development of m-commerce. So, governing transaction environment, reducing information asymmetries, avoiding transaction risk and establishing credit transaction environment become practical problems must be faced. To construct a reasonable and reliable third-party trust evaluation method is of great importance to establish the credibility, safe, and reliable trading platform. The major work of this paper is as follows:(1) The paper studies the trust evaluation method based on recommended information in m-commerce.Based on Trust factors occurred in the typical documents from the international authorities, by analyzing elements relation in the structure of m-commerce transaction process, the paper constructs m-seller’s evaluation indexes based the multiple portfolio analysis that include qualitative analysis and rational analysis. It ensures that indicators is authority and reliability, and fits m-commerce environment. The method of minimizing the sum of recommended results deviation square reduces conflict among recommended results, avoids one-sided bias from one referrer to be brought to the result, and ensures even though number of referrer’s number is small the seller’s trust level can be gotten also. Furthermore, based on characteristic of MT, recommended trust mechanism fit m-commerce environment is built by using improved ELECTREⅢto describe the formation of consumer trust. It ensures evaluation result reflects referrers’group minds and consumer’s evaluation which can help consumer evaluate m-seller’s trust leve fastly and accurately. (2) The paper builds the trust evaluation method based on seller’s feedback credit information in m-commerce.Based on content analysis method, the article builds credit evaluation indxes of m-seller and analyzes the reason. That checked degree of mutual recognition and Krippendorffi’s alph coefficient ensures the validity and reliability of analysis result. That analyzed consumer attitude toward soloves the problem that correlation can’t be tested only using frequency analysis. The paper extrats three indicators most concerned by consumer which can help m-seller grasps consumer demand effectively and improve their own credit.The article buids multi-dimensional credit feedback evaluation model. Firstly, according to characteristic of MT, the paper selects credit metrics model that is favorable for gathering original credit data. Secondly, that added weight function of goods prices and time attenuation function improve improves the credit fraud prevention ability. Thirdly, the frequency statistical analysis weighting reveals most consumer preference, so increases the authority and reliability of calculation results from combination weighting. Fianly, the paper establishs the combination determining weights method based on TOPSIS and generalized maximum entropy which partly soloves the problem that results computed by different sigle weighting method is always quite different. It retains not only objective weighting method reflecting the true situation such as entroy weighting method, but also subjective weighting method which reflects indicators attached by consumers and expertise such as Delphi.(3) The paper presents the trust evaluation method based on information coming from mobile marketing alliance in m-commerce.The mechanism makes the m-commerce transaction safe by the way of preventing beforehand, coordinating among the event and punishment afterwards. According to Game Theory, it can reduces transaction risk, and make m-commerce safe. Meanwhile, it can prevent members from changing identity to cheat and reduce the workload of business platform arbitration. Furthermore, based on information coming from mobile marketing alliance, the paper buids trust level evaluation indexes. And, two dimension decision model is set from views of the mobile marketing alliance’s credit and the member’s credit from its alliance. The model solve the problem in m-commerce effectively:few person wants to trade with newly registered users.更多还原