WO2020151211A1

WO2020151211A1 - Anti-fraud rule model construction system and method based on data variables

Info

Publication number: WO2020151211A1
Application number: PCT/CN2019/098126
Authority: WO
Inventors: 崔晶晶; 张永名; 吴祖琪
Original assignee: 集奥聚合(北京)人工智能科技有限公司
Priority date: 2019-01-23
Filing date: 2019-07-29
Publication date: 2020-07-30
Also published as: CN109784933A

Abstract

Provided are an anti-fraud rule model construction system and method based on data variables. The system comprises: a multi-user management and permission module, an anti-fraud rule model and custom configuration module, an anti-fraud business access and analysis statistics module and a rule model template release and management module. By means of the present invention, a rule model can be constructed based on self-owned data or a third-party risk control data source, and a business system provides accurate risk identification and professional information verification, thereby ensuring that a self-owned business is secure, reducing the influence of false information on later loans, and losses caused by a fraud request, etc., and reducing the occurrence of risk events to the maximum extent.

Description

An anti-fraud rule model construction system and method based on data variables

Technical field

The invention relates to the technical field of information processing, and in particular to a system for building an anti-fraud rule model based on data variables.

Background technique

In the context of innovation strongly encouraged by the country, Internet finance has rapidly expanded its market size and its participants have continued to increase. The use of online transactions for transactions, lack of credit management and analysis, the integration of the Internet and finance, and the core of Internet finance lies in risk control, risk control to prevent fraud, information screening and rule verification to ensure the safe operation of its own business and reduce the loss of false information , Which can reduce risk events.

Credit risk is the main risk faced by banks. Certain banks in the current market in my country still lack scientific and advanced credit management technology. Based on the current big data background, build a pre-lending data model and establish a comprehensive credit risk control system. Its anti-fraud system analyzes users' customer data, advances and makes pre-loan decisions and anti-fraud work before loans.

Using the access technology of the data source platform, the user's credit rating can be further improved and innovated. The era of big data has initiated major changes, using progressively developed science and technology to establish an anti-fraud model suitable for Internet finance to help the efficient and orderly development of the Internet finance industry. In addition, the use of anti-fraud rule models and custom models can effectively reduce pre-loan risks, thereby providing users with certain analysis and reference conditions, and formulating a complete comprehensive risk monitoring management system.

Summary of the invention

To achieve the above objectives, the present invention provides an anti-fraud rule model construction system based on data variables, which can build a rule model based on own data or third-party risk control data sources. The business system provides accurate risk identification, professional information verification, and guarantees There is a safe operation of business to reduce the impact of false information on subsequent loans and losses caused by fraudulent requests, and minimize risk events.

An anti-fraud rule model construction system based on data variables includes: multi-user management and authority module, anti-fraud rule model and custom configuration module, anti-fraud business input and analysis and statistics module, rule model template release and management module.

The multi-user management and authority module is oriented to business personnel and business systems. It needs to provide functions such as multi-user registration, authority setting, business isolation, etc. It supports Web client multi-user registration and management, authority division and role setting to satisfy users The management and security level allocation requirements for system users.

Among them, the system defaults to the three default roles of super administrator, administrator, and ordinary user, which can use the business system and lower the threshold for using the system. At the same time, there are special requirements. The system provides a custom role function. Users can set new roles through the super administrator and assign corresponding access permissions.

The anti-fraud rule model and custom configuration are also the core functions of the system. It provides anti-fraud rule model configuration for business personnel and model use for business systems. The system provides a rule set on the Web client interface. Define the configuration. According to actual business requirements, business personnel configure rule models in different scenarios to provide external services.

The rule model is constructed by the rule set-rule-condition-field variable to construct the operating logic, which supports parameter setting from the rule set to the custom field variable level. The field level supports the calculation of different types of parameters such as numeric value, character, time and date, covering The vast majority of computing models for actual business scenarios.

The said anti-fraud business input and analysis and statistics module, the system provides external services for API business input. Business personnel can conduct statistical analysis on the business input by accessing the Web terminal to fully understand the business level, business trend, risk interception, etc. Detailed statistical indicators such as the acceptance rate and rejection rate.

The detailed input indicators of the above multiple dimensions can comprehensively reflect the effect of the business model. With the help of the index analysis results, users can make corresponding adjustments to optimize the business model. Including optimization measures such as increasing or reducing the business pass/interception ratio, increasing the level of business input, and reducing invalid data applications, ultimately improving the efficiency of the business system and ensuring the safe operation of the business system.

In the rule model template publishing and management module, the system provides the rule model template market and management functions for business personnel. New users can directly import the business rule template to quickly build and use the rule model, reducing the user's use threshold. At the same time, mature business rule models can be published as templates, which can be quickly imported and used for the same type of business, reducing the time cost of business model construction and improving the efficiency of business personnel.

Benefits:

The present invention provides a risk platform, which enables users to set, analyze and view risk models in multiple dimensions, and perform multi-dimensional analysis and display of incoming parts. For various business systems in the fields of marketing, financial credit investigation, information security, etc., provide them with pre-risk control prevention, information verification, anti-fraud identification and other services. Users can base on their own data or third-party risk control data sources Build a rule model, the business system provides accurate risk identification, professional information verification, to ensure the safe conduct of its own business, reduce the impact of false information on subsequent loans, and losses caused by fraudulent requests, and minimize risk events.

Description of the drawings

Figure 1 is a schematic diagram of an anti-fraud rule model construction system based on data variables provided by the present invention;

Fig. 2 is a flowchart of a method for building an anti-fraud rule model based on data variables provided by the present invention.

detailed description

As shown in Figure 1, the data variable-based anti-fraud rule model construction system of the present invention is characterized by: multi-user management and authority module, anti-fraud rule model and custom configuration module, anti-fraud business input and analysis Statistical module, rule model template release and management module; among them,

The multi-user management and authority module supports multi-user registration and management of the Web client, and uses authority division and role setting to meet the user's management and security level assignment requirements for system users;

The anti-fraud rule model and custom configuration module provide the configuration of anti-fraud rule models for business personnel, and provide custom configuration of rule models for business systems. Business personnel can configure rule models in different scenarios according to actual business needs. Provide external services;

In the anti-fraud business input and analysis and statistics module, the business personnel conduct statistical analysis on the business input by accessing the Web client, comprehensively understand the detailed input indicators of multiple dimensions, and adjust the corresponding business model with the help of the indicator analysis results. Optimize and improve the efficiency of the business system;

The rule model template publishing and management module provides a rule model template market, where new users using the system can directly import business rule model templates to quickly build and use rule models; mature business rule models can be published as templates for the same type Services can be quickly imported and used.

Further, in the multi-user management and authority module, the system defaults to three default roles of super administrator, administrator, and ordinary user. At the same time, the system also provides a custom role function, and users can set new roles by themselves through the super administrator , And assign corresponding access permissions.

Further, in the anti-fraud rule model and custom configuration module, the system provides a rule set and custom configuration of the rule model on the Web client interface, and the rule model is constructed and operated by the rule set-rule-condition-field variable Logic, supports parameter setting from rule set to custom field variable level, and field level supports calculation of different types of parameters such as numeric value, character, time and date.

Further, in the anti-fraud business input and analysis and statistics module, detailed input indicators in multiple dimensions include: business volume, business trend, risk interception, input passing and rejection ratio; optimization measures include: increasing or Reduce the business pass/intercept ratio, increase the level of business input, and reduce the use of invalid data.

For the above-mentioned new rule set, there are two types of rule set creation rules: fixed template and custom setting. In the rule set, you can create new conditions and new associated fields. Users can choose what they want. The activation of the rule set must be approved and passed, and it can be used after activation.

The rule set is composed of rules, which in turn are composed of conditions, and conditions are composed of fields. Layers of correlation form a set of rules. Through the reference rules contained in the rule set, users can perform logical operations based on parameter settings to verify whether they meet the user's pre-credit needs.

The rules are analyzed and compared. If the rule is hit, it means that the rule is a lack of reference for the lender’s credit. Constructing a risk model and analyzing and calculating the information of lenders, this information processing method is used for customers, which brings great convenience. Input known data, low cost, system possibilities have been greatly expanded.

The system has built-in operator data, bank data, loan data, identity verification, location and other multi-dimensional indicator variables. Relying on comprehensive indicators, it builds a rule model that meets the actual scenario, and treats illegal business data such as potential fraud risks and false users from multiple aspects. Troubleshoot. Such as mobile phone online time, status, identity verification, work and residence verification, etc.

Anti-fraud system, providing system configuration and management services. The anti-fraud rule engine accesses data interface services of the authenticity interface platform, basic platform, and log platform. The external data of the anti-fraud business system is accessed and managed through the data source layer, providing massive data to support the operation of the anti-fraud engine before lending.

As shown in Figure 2, an anti-fraud rule model construction method based on data variables includes the following steps:

(1) The user requests access after registration and generates a data source request log;

(2) Collect user information requesting access and create a new rule set; collected user information includes regular data and real-time data, collect user information and implement distributed computing by running a computing framework.

(3) Check interface parameters: enter the user’s rules, conditions, and field information. If it is normal, execute (4). Otherwise, the result containing the reason for the abnormality will be directly returned to the business system, and the access will fail; parse the user’s basic information, including: identity, Residential information, to obtain user-checking rules, conditions, and field information, which is used to check the conditions of parameter rules when inbound.

(4) According to the parameter settings in the newly created rule set, obtain the calculation request interface list to establish a rule template, and perform logical operations with the rule template of the external interface data; the rule model is used for parameter verification, and the access of the model is used as the verification standard .

(5) Analyze the field results, analyze and compare the rules, the hit rule is the reference condition, return the result to the business system, and the connection is successful; otherwise, the result containing the abnormal reason is returned to the business system, and the connection fails, and the rule model is constructed. If the rule is hit, the risk level of the rule is used as the reference condition.

In addition, it should be understood that although this specification is described in accordance with the implementation manners, not each implementation manner only includes an independent technical solution. This narration in the specification is only for clarity, and those skilled in the art should consider the specification as a whole The technical solutions in each embodiment can also be appropriately combined to form other implementations that can be understood by those skilled in the art.

Claims

An anti-fraud rule model construction system based on data variables is characterized in that it includes:

Multi-user management and authority module, anti-fraud rule model and custom configuration module, anti-fraud business input and analysis statistics module, rule model template release and management module; among them,

The multi-user management and authority module supports multi-user registration and management of the Web client, and uses authority division and role setting to meet the user's management and security level assignment requirements for system users;

The anti-fraud rule model and custom configuration module provide the configuration of anti-fraud rule models for business personnel, and provide custom configuration of rule models for business systems. Business personnel can configure rule models in different scenarios according to actual business needs. Provide external services;

In the anti-fraud business input and analysis and statistics module, the business personnel conduct statistical analysis on the business input by accessing the Web client, comprehensively understand the detailed input indicators of multiple dimensions, and adjust the corresponding business model with the help of the indicator analysis results. Optimize and improve the efficiency of the business system;

The rule model template publishing and management module provides a rule model template market, where new users using the system can directly import business rule model templates to quickly build and use rule models; mature business rule models can be published as templates for the same type Services can be quickly imported and used.
An anti-fraud rule model construction system based on data variables according to claim 1, characterized in that, in the multi-user management and authority module, the system defaults to three default roles of super administrator, administrator and ordinary user, At the same time, the system also provides a function of custom roles. Users can set new roles through the super administrator and assign corresponding access permissions.
An anti-fraud rule model construction system based on data variables according to claim 1, characterized in that, in the anti-fraud rule model and custom configuration module, the system provides a rule set on the Web client interface, and the rule model The custom configuration of the rule model is constructed by the rule set-rule-condition-field variable, and it supports parameter setting from the rule set to the custom field variable level. The field level supports different types of values, characters, time and date. Operation of parameters.
An anti-fraud rule model construction system based on data variables according to claim 1, wherein in the anti-fraud business input and analysis and statistics module, detailed input indicators of multiple dimensions include: business volume level , Business trends, risk interception volume, incoming piece passing and rejection ratios; optimization measures include: increasing or reducing the business pass/intercept ratio, increasing the level of business incoming pieces, and reducing invalid data applications.
A method for constructing an anti-fraud rule model based on data variables is characterized in that it comprises the following steps:

(1) The user requests access after registration and generates a data source request log;

(2) Collect user information requesting access and create a new rule set;

(3) Check interface parameters: enter the user's rules, conditions, and field information. If it is normal, execute (4), otherwise the result containing the abnormal reason will be directly returned to the business system, and the access will fail;

(4) According to the parameter settings in the newly created rule set, obtain the calculation request interface list to establish a rule template, and perform logical operations with the rule template of the external interface data;

(5) Analyze the field results, analyze and compare the rules, the hit rule is the reference condition, return the result to the business system, and the connection is successful; otherwise, the result containing the abnormal reason is returned to the business system, and the connection fails, and the rule model is constructed.
The method for building an anti-fraud rule model based on data variables according to claim 5, characterized in that: in the step (2), the collected user information includes regular data and real-time data, and the user information is collected and implemented by running a computing framework Distributed Computing.
The method for constructing an anti-fraud rule model based on data variables according to claim 5, characterized in that: in the step (3), the basic user information is analyzed, including: identity, residence information, and rules, conditions, The field information is used to check the conditions of the parameter rules when inbound.
The method for constructing an anti-fraud rule model based on data variables according to claim 5, characterized in that: in the step (4), the rule model is used during parameter verification, and the access of the model is used as the verification standard.
The method for building an anti-fraud rule model based on data variables according to claim 5, characterized in that: in the step (5), the rule hits, and the risk degree of the rule is used as a reference condition.