? First, the construction of user credit portrait
Speaking of the role of user credit portrait construction in the whole risk control system, there is no doubt that different financial platforms can build their own user portraits according to their own business scenarios and capabilities. After all, the data of some portraits are not available, and it is difficult to obtain from other three-party platforms, so the construction should be based on your own business scenarios and company conditions.
The composition system of user credit portrait includes but is not limited to the following points: user identity information, social data of marriage and love, sesame credit, user authentication data, consumption income and expenditure data, user behavior data, people's bank credit report, black and gray list of mutual funds and banks, equipment-related data, etc.
1. 1 user identity information
This information is relatively easy to obtain in the nine major data, including the user's three-body or four-body data, residence, marital status, children's situation, work unit, position, real estate, income, contact information and other data. These data are mostly used for credit evaluation before lending and authentication before entering the user platform.
1.2 Social data of marriage and love
What role does social data of marriage and love play in the whole credit system?
Other data types can show a person's credit level, repayment ability, consumption level, user behavior and so on. However, the evaluation of personality quality is slightly weak. The data of a person's family responsibility, social speech, life planning, social circle and other behaviors can better reflect what role a mainstream personality quality category belongs to.
In the social situation of marriage and love, several data are particularly important: family situation, real estate situation, education, life schedule, love planning, love account level, social credit of marriage and love, social network and so on.
1.3 Sesame Credit
Sesame credit also occupies a certain position in the existing credit system, and many credit certifications involving funds will list sesame credit as one of the evaluation criteria.
It mainly includes credit score, industry attention list, application fraud score, fraud information verification, fraud attention list and enterprise credit score.
1.4 user authentication data
At present, there are several pieces of user authentication data, including provident fund social security, operator communication, academic data (Xuexin.com) and professional data (authentication data of professional recruitment platforms such as pulse, hunting and boss).
1.5 consumption revenue and expenditure data
Consumption data is one of the most important data, which is related to personal economic level, repayment ability and judgment of consumption behavior. Consumption revenue and expenditure data mainly include the following points: online e-commerce and offline consumption, UnionPay consumption, bank card revenue and expenditure, air travel data, etc.
1.6 user behavior data
User behavior data can take the filling time of the application form and the residence time of the loan agreement page as one of the reference data.
1.7 PBC credit report
Needless to say, the data of the People's Bank of China's credit information is one of the important bases of financial credit information, including the user's loan information, credit transaction information, personal public information and so on.
1.8 List of Mutual Gold and Bank Black Ash
The black and gray lists of mutual funds and banks can be used as an important basis for establishing black and gray lists in risk control. Black-gray list includes credit overdue list, bad judicial list, multi-head application and multi-head debt list and team fraud list.
1.9 Equipment related data
There are several dimensions of device-related data, such as fingerprint and facial recognition of the device, hardware information of the device, GPS positioning, and APP data installed by the device.
Second, anti-fraud services.
Different financial application scenarios have different business processes and links, so it is necessary to design different risk inspection links and risk control strategies, and build scenarios, events and rules-driven fraud risk judgment services. Through flexible configuration, the ability to judge fraud risk in different scenarios and different business links can be met.
Financial application: activation event, registration event, login event, authentication event, binding card event, activity time and other scenarios;
Loan application: registration event, login event, card binding event, recharge event, credit granting event, withdrawal event, etc.
The customer's credit is determined by anti-fraud model, user behavior analysis, risk information base and anti-fraud rule base.
2. 1 evaluation dimension
(1) identity evaluation module
Equipment abnormality evaluation: equipment binding authentication, equipment similarity evaluation, and inter-agency equipment registration.
Geographical location assessment: GPS ip assessment, unfamiliar transaction area assessment, cross-regional and inter-agency assessment, and GPS positioning does not match the application address;
Abnormal customer environment: IOS jailbreak, Android ROOT, public WIFI, suspected Trojan application;
Habit similarity evaluation: case habits, reading time, customer input method behavior;
Customer authentication: name, ID card, mobile phone number, human body live verification.
(2) Credit transaction behavior evaluation
Trading behavior: trading time habit, trading frequency, short-term trading quantity and trading amount;
(3) Credit evaluation
Fraud information database: equipment fraud database, IP fraud database, account fraud database;
Information base of dishonesty: credit overdue list, bad judicial list, multi-head application multi-head debt list, gang fraud list;
Fraud correlation diagram: fraud correlation diagram.
2.2 Anti-fraud strategy
(1) Seven Strategies
(2) the whole process of credit
User registration: three elements: authentication, client environment detection, and whether the registered device is associated with multiple users;
Login: abnormal device login detection, abnormal login location detection, abnormal login IP detection, abnormal login face recognition.
Account opening binding card: four elements: verification, face recognition, whether the device is bound to multiple bank cards to open an account, and whether the bank cards involve fraud;
Credit application: comparison of fraud list, dishonesty list, multi-head loan application list, multi-head overdue debt list and related party fraud list;
Confirmation of credit: false credit effect and credit behavior record;
Withdrawal: same card monitoring.
Third, finally.
This paper is some integration of the author's learning content. I have been studying the topic of risk control recently, and I feel it. If you have any good suggestions, please comment below and discuss with each other.