多维个人信用评价特征感知模型的研究.doc

多维个人信用评价特征感知模型的研究 孙璐李广建 北京大学信息管理系 ［目的/意义］互联网金融背景下，个人信用评估的需求激增，大量个人信用行为信 息沉淀在各种互联网应用平台之上，这类信息语义丰富,iL隐藏着许多可用于对 现有评价模型进行降维的关键特征，本研究旨在解决如何快速地发现这类关键特 征的问题。［方法/过程］选取CNKI 10年来个人信用评价的研宂文献作为研宂样 本，从其内容、功能、应用及场景、组织及制度等多个维度,尽可能全面地构建评 价特征体系，并运用国外应用研究中主流的相关性建模方法来构建模型。［结果/ 结论］最终构建了一种多维度的个人信用评价特征的感知发现模型，通过模型可 以直观、动态地发现新的个人信用评价特征；同时本模型也可以用于个人信用评 价方法以及待征权重的比较研宄。 关键词： 多维度;个人信用评价特征;感知发现模型; 孙璐(0RCTD:0000-0002-7294-4663),研究 生；E-mail: sydncy_cllcn@pku. cdu. cn; 李广建(0RC1D:0000-0002-2897-6246)，系主任，教授。 2015-09-22 基金：国家社会科学基金项目“大数据环境下的计算型情报分析方法与技术研 究”(项目编号:2014ATQ005)研究成果之一 Study on Multiple Dimensions of the Perceived Discovery Model of Personal Credit Evaluation Characteristics Sun Lu Li Guangjian Department ofLnf ormat i on Management, Peking University; Abstract： [Purpose / significance]Under the background of Internet finance, with the surge of demand for personal credit assessment, a lot of personal credit information is recorded on a variety of Internet application platforms. This kind of information is rich in semantics, and there are many key features that can be used to reduce the dimension of the existing evaluation models. This study aims to address how quick the discovery of such key features can be. [Method /process] In this paper, the research on CXKI research literature based on related fields in the past ten years was made as a research samplefrom its content, function, appl ication and scene, organization and system, etc. It constructed a description feature system based on multiple dimensions. We used mainstream modeling methods in the foreign application research to carry out the study of the correlation. [Result / conclusion]Finally,a kind of multiple dimension model based on the perceived discovery of personal credit evaluation characteristics is constructed. Based on this model, the characteristics of the new personal credit evaluation can be found directly and dynamically. At the same time, this model can also be used for the comparative study of feature weights. Keywo