摘 要

传统征信模式由于其存在的数据来源复杂、覆盖面窄及时效性差等问题，使其无法满足现代信用经济和社会发展的要求。因此在国务院“互联网 ”的政策引导下，产生了具有覆盖面广、评价维度丰富和时效性强等特点的“互联网 征信”模式。作为国内首个基于用户互联网行为数据的征信产品一芝麻信用评分，同传统征信评估体系相比，其具有数据来源丰富、群体覆盖面广、时效性更强和应用服务场景更加丰富等特点，但是其评分体系等仍然存在数据来源真实性与合法性不能确定、评分模型不够成熟等问题。因此研究芝麻信用评分的发展现状以及进行问题探讨，并提出相关建议，对未来我国互联网金融征信的发展具有借鉴意义。

本文研究：（1）传统征信模式和互联网金融征信模式在信息质量、时效性、信用信息共享等方面各有优势和劣势。（2）芝麻信用评分拥有五个评分维度即用户的信用历史、行为偏好、履约能力、身份特质、人际关系，其所采集的数据主要来源于：阿里巴巴旗下的电商平台以及参股或者合作电商的交易数据、蚂蚁金融服务集团采集的互联网金融数据、与阿里巴巴集团具有合作关系的外部机构提供的信息数据、用户自主提供的个人信用数据以及其他数据来源。（3）本文采用调查问卷及相关性分析方法分析芝麻信用评分的评分模型，得出芝麻信用分与用户的月平均收入、每月支付宝消费金额、蚂蚁花呗使用时间、使用淘宝时间、支付宝好友个数、淘气值有显著的相关性。（4）通过定量定性分析发现芝麻信用评分存在以下问题：数据来源合法性无法确定；数据采集维度不够全面以及所采集的数据真实性无法确定；芝麻信用评分模型不够成熟；用户个人信息和隐私存在泄漏风险；个人信用信息无法共享。（5）通过分析芝麻信用评分模型存在的问题，本文提出了相关的具有借鉴意义的建议：健全互联网个人征信机构和企业的监管体系；通过完善数据采集过程、利用科学技术等完善芝麻信用评分模型；通过立法、健全用户数据授权体系以及提升网络安全技术来加强互联网金融征信用户的隐私保障；建立互联网金融征信信息共享机制。

本文特色：（1）将基本理论与实际案例相结合，并且选取了最具有代表性的芝麻信用评分作为案例研究对象；（2）本文使用问卷调查方法收集用户与芝麻信用评分相关的数据，采用相关性分析，找出芝麻信用评分模型存在的问题。

关键词：互联网金融征信；芝麻信用评分；大数据；评分模型

Abstract

Because of its complicated data sources, narrow coverage and poor timeliness, the traditional credit collection model cannot meet the requirements of modern credit economy and social development. Therefore, under the guidance of "Internet plus" policy, the "Internet plus credit" mode came into being, it has wide coverage, strong timeliness and rich evaluation dimensions and other characteristics. As the first credit scoring product based on user internet behavior data, Sesame credit score, compared with the traditional credit evaluation system, has the characteristics of rich data sources, wide coverage, stronger timeliness and richer application service scenes, but its scoring system still has problem, such as the authenticity and legality of data sources cannot be determined, the scoring model is not mature enough and so on. Therefore, we study the status of the development of Sesame credit score and discuss its problems, and make some suggestions which can be used for developing Internet financial credit in the future.

The study found that: (1) the traditional credit mode and the Internet financial credit model have advantages and disadvantages in information quality, timeliness, credit information sharing and so on. (2) Sesame credit score has five dimensions are the user's credit history, behavior preference, performance, identity, and interpersonal relationship. The data collected mainly come from the e-commerce platform under the Alibaba and the trading data of participating or cooperative e-commerce, the Internet financial data collected by ant financial services group, information data provided by an external organization with a cooperative relationship with the Alibaba group, individual credit data provided by the user and other sources of data. (3) this paper adopts questionnaire and correlation analysis method to analyze the credit score of Sesame credit score, having found that sesame credit score with the user's average monthly income, the monthly consumption amount, Alipay ant flower chant time, time for Taobao, Alipay friends, Amoy gas value had a significant correlation. (4) through quantitative and qualitative analysis, it can be found that there are the following problems in the Sesame credit score: the legitimacy of the data sources cannot be determined, the dimension of the data collection is not comprehensive enough and the data authenticity cannot be determined; the credit scoring model of sesame is not mature enough; the personal credit and privacy have the risk of leakage; personal credit Information can't be shared. (5) through the analysis of the problems of the Sesame credit score, this paper puts forward some relevant suggestions for reference: perfecting the supervision system of the Internet personal credit agencies and enterprises, perfecting the Sesame credit scoring model by perfecting the data collection process and using science and technology; legislation and establishing user authorization system and promoting network security technology can enhance the privacy protection of Internet financial credit users; establishing the Internet financial credit information sharing mechanism.

The characteristics of this paper are: (1) combining the basic theory with the actual case, and selecting the most representative Sesame credit score as the case study; (2) this paper uses the questionnaire survey method to collect the data related to the Sesame credit score, and uses the correlation analysis to find out the question of the Sesame credit score.

Keywords: Internet financial credit; Sesame credit score; big data; scoring model