Research And Implement Of Credit Management Platform Technology Supporting Logistics Transition

The credit management platform technology supporting logistics transition is an important aspect of "the key technologies and platforms of modern logistics commonness", which is one project of "11th Five-Year" major national scientific and technological projects. The platform aims at collecting credit information of various logistics enterprises according to their transition behaviour in the markets, researching on credit management technologies, evaluating logistics enterprises'credit and building credit management platform so as to promote current credit situations in the logistics industry.This paper mainly researches on the system architecture of the credit management platform supporting logistics transition, builds a credit evaluation index system of logistics industry, constructs adaptive credit models to analyze enterprises credit ability and describes our implement of the credit management platform.Credit management platform should include Data Acquisition Module, information dissemination module, data pre-processing module, credit evaluation module, credit service module, objection module handling, and operation management module and so on. We design the architecture of the platform and implement most modules.Credit index system is the basis and standard of credit evaluation, which compositively reflects the complicated system built by all kinds of attributes of enterprises themselves and their surroundings. We design the index selection principles after reading a mount of related materials and make credit index system after consulting some logostics specialists.Credit evaluation models are keys to credit evaluation, which use modeling method to analyze a great lot of historical credite data of enterprises, find their rlues and determine the credit status of a certain enterprise. Based on comparison with some classical credit modles, we mainly research on desition trees and neural network model to implement the platform. We also forward a new algorithm, multivariate desition trees based on regression and discriminant analysis.