【作者】 张晓丽；

【导师】 王东；

【作者基本信息】 上海交通大学 ， 软件工程， 2011， 硕士

【摘要】 随着电子商务的快速发展,电子商务过程中物流的重要性日益凸显,通过物流及时、准确地完成商品的空间转移成为电子商务成功的关键。由于我国物流发展水平等因素的限制和相关标准的不完善,电子商务过程中物流时间长、商品丢失或破损、服务质量差等现象屡见不鲜,这使得物流纠纷成为电子商务纠纷中的主体。电子商务企业大多面临无法确定物流企业和物流个人信用状况的困扰,这影响到其物流合作伙伴的选择,因而电子商务企业对准确了解物流企业和物流个人基本信息及信用信息的需求越来越强,但目前还没有完整的面向电子商务的物流征信系统。本文对现有的征信体系和信用评估算法进行了深入研究和分析,结合电子商务物流特点和电子商务企业对物流信用的需求,提出了面向电子商务的物流征信系统。该系统主要包括物流信用评估指标体系、物流信用评分模型和信用评估算法。在物流信用评估指标体系方面,从电子商务物流的特点出发,根据物流主体的不同,按照指标体系的设计原则,使用德尔菲方法分别设计了物流企业和物流个人信用评估指标体系;在物流信用评分模型方面,结合电子商务物流的特点,综合考虑评价主体给出的评分和评价主体信用状况、交易金额、交易次数、评价时间等其他影响因素,提出了改进的物流信用评分模型,并通过实验验证了改进模型的有效性;在信用评估算法方面,深入研究了聚类算法(Clustering Algorithm)和模糊支持向量机(Fussy Support Vector Machine, FSVM),结合面向电子商务的物流征信系统对评估方法的分类性能和时间性能的要求,实现了基于聚类的模糊多分类支持向量机算法(Fussy Multi-class Support Vector Machine based on Clustering),该算法在模糊多分类支持向量机中引入K-WMeans聚类算法和基于相似度的层次聚类算法以提高算法性能,通过实验验证了改进算法的有效性,并进行了信用等级设计。结合电子商务物流征信的需求,系统架构分为C/S子系统、数据库层和B/S子系统,C/S子系统主要用于物流企业和物流个人的基础信息管理及信用评估管理;数据库层主要用于存储相关基础信息和信用信息;B/S子系统主要用于电子商务企业及相关个人进行物流信息录入、信用信息查询及展示物流企业和物流个人的信用状况。本文首先对现有征信技术进行综述,分析和比较了国内外学者和企业在征信方面取得的研究成果;其次,总结面向电子商务的物流征信系统的功能需求及现有物流征信体系中存在的问题;然后,对面向电子商务的物流征信系统中的关键技术进行设计,结合电子商务物流的特点,构建了物流信用评估指标体系,对现有的信用评分模型改进后提出了改进的物流信用评分模型,实现了基于聚类的模糊多分类支持向量机以提高信用评估结果的可靠性和时间性能,并进行信用等级设计;最后,本文描述了面向电子商务的物流征信系统的功能架构、实现技术并进行了系统演示。更多还原

【Abstract】 With the rapid development of E-commerce, the importance of logistics has become increasingly evident. To complete the space transfer of goods timely and accurately is the key to success of E-commerce. Because of the limitation of domestic logistics level and lack of relevant standards and such factors, long distribution time, lost or damaged goods and poor quality of logistics services are common occurrence, making logistics disputes as the main disputes in E-commerce. Most of E-commerce enterprises are facing the problems that have to choose logistics partner without grasping the credit of logistics enterprises and logistics persons, affected their choices. And the demand to know the basic information and credit information of logistics enterprises and logistics person are growing, but there are no such complete logistics credit evaluation system for E-commerce at present.In this paper, with in-depth research and analysis on existing credit evaluation system and credit evaluation models, considering the characteristics of E-commerce logistics and the logistics credit requirement of E-commerce enterprises, we propose the logistics credit evaluation system for E-commerce. It is composed of credit evaluation index system, credit scoring models and credit evaluation methods of logistics. The credit index system of logistics is designed using Delphi method according the principle of index design. From the characteristics of E-commerce logistics, the index system is divided into logistics enterprise credit evaluation index system and logistics person credit evaluation index system. The improved credit scoring models not only consider the given score but also the credit of rater, transaction amount, transaction times, evaluation time and other factors. And the validity of improved credit scoring models is proved by experiments. With in-depth research on clustering algorithm and Fuzzy Support Vector Machin(eFSVM) and also the requirement of logistics credit evaluation system for E-commerce about accuracy and time performance, we realized Fussy Multi-class Support Vector Machine (FMSVM) based on Clustering, in which K-WMeans clustering algorithm and hierarchical clustering algorithm based on similarity is introduced to FMSVM in order to improve the performance of proposed algorithm, and the experimental results proves this. Also credit degrees are designed. With the needs of logistics credit evaluation system for E-commerce, system architecture is divided into C/S subsystem, database layer and B/S subsystem. C/S subsystem is mainly used for management of basic information and credit information about logistics enterprises and logistics persons. Database layer is used to store related basic information and credit information. And B/S subsystem is mainly used to accept E-commerce logistics information input by E-commerce enterprises , search and shows logistics credit information of logistics enterprises and logistics persons.The paper firstly summarizes the existing technology of credit evaluation, analyzes and compares the achieved results of domestic and foreign scholars and enterprises. Then sum up the functional requirements of logistics credit evaluation system for E-commerce and existing problems. After that, combined with the characteristics of E-commerce logistics, design credit evaluation index system of logistics, improve the existing credit scoring models and propose fuzzy multi-class support vector machine based on clustering to improve the reliability of the results of credit evaluation and time performance, and also credit degrees are designed. Finally, this paper describes the architecture and implementation of logistics credit evaluation system for E-commerce, and then presents the system.更多还原