Method for risk assessment and amount determination of advanced freight payment of water transportation platform

Abstract

The invention discloses a method for risk assessment and amount determination of advanced payment of a water transportation platform, and relates to the technical field of waterway transportation logistics finance. After multi-channel data acquisition preprocessing, multi-dimensional score screening, track authenticity verification, ship owner risk assessment, fusion type fraud identification, dynamic determination of a payable amount and risk control linkage are carried out; the method has the advantages that the waybill multi-dimensional quantitative index and the AIS track depth verification technology are integrated, the event chain integrity, the track continuity and the channel deviation characteristics of the transportation link are linked, and the abnormal point intelligent identification and track correction method is matched, so that the risk control efficiency is improved. The industry pain point that the authenticity of the waybill is difficult to accurately identify in the prior art is effectively solved, accurate discrimination of false transportation behaviors is realized, and platform fund loss caused by waybill counterfeiting is avoided from the source.

Description

Technical Field

[0001] This invention relates to the field of waterway transportation logistics finance technology, specifically a method for risk assessment and amount determination of advance freight withdrawal by a waterway platform. Background Technology

[0002] In the water transport logistics industry, ship owners bear operating costs such as fuel and crew wages during transportation, while cargo owners often only pay freight after the transportation is completed, creating a significant payment period. Therefore, ship owners generally have a need to withdraw a portion of freight before the cargo owner actually pays for working capital. Currently, existing domestic platforms mainly rely on manual review, static rules, or fixed percentages to determine the withdrawal amount when providing early withdrawal services. However, water transport has characteristics such as easy track drift, unstable AIS signals, and dispersed operation links, making it difficult to verify the authenticity of waybills through traditional methods. At the same time, there is a lack of effective quantitative means for ship owners' operational capabilities, payment cycles, sailing patterns, and capital usage habits, making it difficult to accurately assess the risk level. In addition, existing technology is also unable to identify fraudulent activities such as fictitious voyages, splitting orders for cash, and abnormal price adjustments, making the early withdrawal business significantly inadequate in terms of authenticity verification and risk identification.

[0003] Based on the above shortcomings, existing early withdrawal of freight charges schemes are prone to three main problems: First, the authenticity of waybills cannot be accurately identified, leading to the risk of fraudulent transportation; second, due to the lack of dynamic risk control capabilities, credit assessments may be distorted, easily resulting in excessively high or low credit limits; third, the lack of identification of complex fraudulent activities and declining shipowner operating capabilities easily leads to loan defaults. Against this backdrop, there is an urgent need for a technical solution that incorporates the characteristics of the water transport industry, capable of comprehensively and dynamically assessing the authenticity of waybills, shipowner risks, fraudulent activities, and the amount available for withdrawal, thereby achieving automated risk control and prudent lending for early withdrawal of freight charges. To this end, we propose a method for risk assessment and amount determination for early withdrawal of freight charges on water transport platforms. Summary of the Invention

[0004] The purpose of this invention is to provide a method for assessing the risk and determining the amount of freight charges that can be withdrawn in advance by a water transport platform.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a method for risk assessment and amount determination of advance freight charges by a water transport platform, the assessment method comprising the following steps:

[0006] Step 1: Collect and preprocess waybill-related data from multiple channels, and extract standardized basic variables;

[0007] Step 2: Calculate multidimensional indicators and obtain a comprehensive qualification score through a linear weighted model to screen qualified waybills;

[0008] Step 3: Identify and correct anomalies in the AIS tracks of qualified waybills, and calculate the transportation authenticity score;

[0009] Step 4: Extract and standardize the shipowner's historical operating indicators, and obtain the shipowner's risk score through a weighted model;

[0010] Step 5: Fraud detection is performed by fusing the rule layer and the model layer to generate a fraud score;

[0011] Step 6: Based on the aforementioned multiple scoring systems and basic parameters, calculate the maximum withdrawable amount and perform threshold verification;

[0012] Step 7: Write all scoring parameters and judgment results back to the business system to achieve risk control linkage and complete the risk assessment and amount determination for early withdrawal of freight charges.

[0013] As a further aspect of the present invention: In step one, based on data collected from web pages, mobile applications, port operation systems, AIS latitude and longitude and speed data, original freight amounts and change records, and APP location data, after removing invalid noise data, five types of standardized basic variables are extracted, namely the actual total number of events such as loading, departure, arrival, unloading, and receipt in the transportation process. Total number of trajectory records collected Valid number of trajectory records The shortest distance between each valid trajectory sampling point and the standard waterway Original amount of shipping fee With change amount .

[0014] As a further aspect of the present invention: in step two, the event chain integrity is calculated based on the basic variables from step one. Continuity of flight path Normalized value of track deviation index and freight abnormality coefficient The overall qualification score is calculated using a linear weighted scoring model, and the calculation formula is as follows:

[0015] ;

[0016] in, , , , The data was obtained through training on a set of historical real waybill samples, with a set threshold. , ≥ The waybill must be qualified for subsequent risk control verification; otherwise, the application for early withdrawal will be rejected.

[0017] As a further aspect of the present invention: In step three, the trajectory points are arranged in chronological order to form a sequence, and the deviation sequence from the standard waterway is calculated based on the actual displacement. With theoretical displacement Comparison, satisfaction The time point is marked as an anomaly. The value is set to 1.5-2.5. Neighborhood smoothing, linear interpolation, and path reconstruction are used to correct the trajectory, and the trajectory offset score is calculated. Outlier ratio score Trajectory continuity score ,pass Obtain a transportation authenticity score. The weights are obtained through training.

[0018] As a further aspect of the present invention: in step four, the repayment success rate of the shipowner over the past six months is extracted. Average payment cycle Continuous valid sailing days Average load factor abnormal navigation percentage Capital utilization volatility coefficient The metrics are standardized using z-scores. Obtain the shipowner's risk score. ~ The weights are obtained by training on historical default samples, and the sum of the weights is 1.

[0019] As a further aspect of the present invention: In step five, the rule layer presets rules for navigation anomalies, missing operational events, abnormal freight rate changes, account and equipment anomalies, and abnormal fund flows. Triggering any one of these rules... ,otherwise The model layer constructs feature vectors of waybills, trajectories, and shipowner behavior, which are then input into the logistic regression model. Calculate the probability of fraud, set ,pass Receive the final fraud rating. , ≥ Early withdrawal is prohibited.

[0020] As a further aspect of the present invention: in step six, a comprehensive qualification score is obtained. Transportation authenticity score Shipowner risk rating Fraudulent ratings Original freight amount on the waybill and basic withdrawal ratio ,pass Calculate the withdrawable percentage, and then obtain Set a minimum withdrawal ratio threshold , Then you are not eligible for early withdrawal; otherwise, you are allowed to withdraw in advance. Withdrawals are permitted within the specified scope.

[0021] As a further aspect of the present invention: in step seven, the maximum withdrawable amount Comprehensive qualification score Transportation authenticity score Shipowner risk rating Fraudulent ratings The judgment results are written back to the water transport platform business system for loan interface amount verification, risk control audit and result backtracking, post-withdrawal credit limit rolling management and risk tracking and anomaly analysis, realizing a closed loop of risk control throughout the entire process.

[0022] Compared with the prior art, the beneficial effects of the present invention by adopting the above technical solution are as follows:

[0023] 1. This invention integrates multi-dimensional quantitative indicators of waybills with AIS trajectory deep verification technology, links the integrity of the event chain in the transportation process, the continuity of the flight path and the characteristics of the airway deviation, and combines anomaly point intelligent identification and trajectory correction methods to effectively solve the industry pain point that existing technologies cannot accurately identify the authenticity of waybills, achieve accurate identification of fraudulent transportation behavior, and avoid platform financial losses caused by waybill fraud from the source.

[0024] 2. This invention constructs a quantitative evaluation system for shipowners' full-dimensional business behavior, comprehensively covering core dimensions such as repayment performance records, operational stability, and the health of the capital chain. Combined with a dynamic credit limit calculation mechanism that links multiple risk factors, it solves the problems of distorted credit assessment and rigid withdrawal limits in existing technologies. It achieves precise matching between withdrawal limits and the actual risk level of shipowners, eliminating the risk of default caused by excessive credit limits for high-risk shipowners, while fully meeting the reasonable cash flow needs of low-risk shipowners.

[0025] 3. This invention utilizes a dual fraud detection mechanism that integrates rapid interception at the rule layer and deep identification at the model layer. This mechanism not only efficiently covers explicit fraudulent behavior but also accurately captures hidden and multi-factor-related fraud risks. Furthermore, it links with a dynamic assessment module for shipowners' operational capabilities to compensate for the shortcomings of existing technologies in identifying complex fraud and declining operational capabilities. This significantly improves the comprehensiveness and timeliness of risk warnings and substantially reduces the probability of loan defaults. Attached Figure Description

[0026] Figure 1 This is a schematic diagram of the method steps in an embodiment of the present invention. Detailed Implementation

[0027] The specific embodiments of the present invention will be further described below with reference to the accompanying drawings. It should be noted that the description of these embodiments is for the purpose of helping to understand the present invention, but does not constitute a limitation of the present invention.

[0028] Furthermore, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0029] Please see the appendix Figure 1 This invention discloses a method for risk assessment and amount determination of advance payment of freight charges by a water transport platform. The assessment method includes the following steps:

[0030] Step 1: Collect and preprocess waybill-related data from multiple channels, and extract standardized basic variables;

[0031] Step 2: Calculate multidimensional indicators and obtain a comprehensive qualification score through a linear weighted model to screen qualified waybills;

[0032] Step 3: Identify and correct anomalies in the AIS tracks of qualified waybills, and calculate the transportation authenticity score;

[0033] Step 4: Extract and standardize the shipowner's historical operating indicators, and obtain the shipowner's risk score through a weighted model;

[0034] Step 5: Fraud detection is performed by fusing the rule layer and the model layer to generate a fraud score;

[0035] Step 6: Based on the aforementioned multiple scoring systems and basic parameters, calculate the maximum withdrawable amount and perform threshold verification;

[0036] Step 7: Write all scoring parameters and judgment results back to the business system to achieve risk control linkage and complete the risk assessment and amount determination for early withdrawal of freight charges.

[0037] Example 1

[0038] Step 1: Basic Data Collection and Preprocessing of Waybills: Waybill data from an inland waterway shipping owner is collected via web, mobile, and port operation systems. The original amount of the waybill is... Yuan, amount of change =0 yuan; Actual total number of events in the transportation process =5 (completely includes loading, departure, arrival, unloading, and receipt); AIS equipment collects trajectory data at a frequency of 3 minutes, with a total number of trajectory records collected. =240 valid trajectories =232; the average shortest distance between valid trajectory sampling points and standard waterways is 0.8km, and the maximum distance is 2.3km.

[0039] Step 2, Waybill Screening and Eligibility Determination: Calculating Event Chain Completeness =5 / 5=1.0, track continuity =232 / 240≈0.967, Track Deviation Index =0.8km, after normalization =(0.8−0) / (5−0)=0.16, Freight Anomaly Coefficient FAS=0 / 100000=0, Substitute into the comprehensive qualification scoring formula: =0.3×1.0+0.25×0.967+0.25×(1−0.16)+0.2×(1−0)=0.3+0.242+0.21+0.2=0.952, ≥0.75, passed the screening.

[0040] Step 3: Verification of Transportation Authenticity: After comparing the actual displacement with the theoretical displacement, identify the number of outliers. =4; Trajectory offset score =1−0.8 / 5=0.84, outlier percentage score =1−4 / 240≈0.983, Trajectory Continuity Score =232 / 240≈0.967; Transportation authenticity score =0.4×0.84+0.3×0.983+0.3×0.967≈0.336+0.295+0.29=0.921.

[0041] Step 4: Shipowner Risk Assessment: Extract the shipowner's indicators for the past 6 months: =98%=0.98, =30 days =160 days =85%=0.85, =3%=0.03, =0.12; Risk score calculated after standardization: =0.3×0.98+0.15×(1 / 30)+0.2×(160 / 180)+0.15×0.85+0.1×(1−0.03)+0.1×(1−0.12)≈0.294+0.005+0.178+0.128+0.097+0.088=0.79.

[0042] Step 5, Fraud Detection: No fraud rules were triggered at the rule layer. =0; After the model layer inputs the feature vector, it calculates the fraud probability. =0.12; Final fraud score =0.4×0+0.6×0.12=0.072<0.6, no risk of fraud.

[0043] Step Six: Determining the Withdrawal Amount: Basic Withdrawal Ratio =70%, withdrawable percentage =70%×0.952×0.921×0.79×(1−0.072)≈70%×0.952×0.921×0.79×0.928≈0.468; Maximum withdrawable amount =100000×0.468=46800 yuan, 46800≥100000×20%=20000, therefore, the applicant is eligible for early withdrawal and can withdraw within the range of 46800 yuan.

[0044] Step 7: Result Write-back and Risk Control Linkage: Write the above score and the withdrawable amount of 46,800 yuan back to the business system to complete the risk control loop.

[0045] Example 2

[0046] The difference from Example 1 is that the original amount on the waybill... =80,000 yuan, changed amount =5000 yuan; Total number of trajectory records collected =200 valid trajectories =185 ships; shipowners in the last 6 months =92%=0.92, =130 days; the rule layer triggered the "significant price adjustment of shipping costs in a short period of time" rule. =1, fraud probability at the model layer =0.35.

[0047] Calculation process: =5000 / 80000=0.0625, =0.3×1.0+0.25×(185 / 200)+0.25×(1−0.2)+0.2×(1−0.0625)=0.3+0.231+0.2+0.1875=0.9185;

[0048] =0.4×(1−1.2 / 5)+0.3×(1−8 / 200)+0.3×(185 / 200)=0.4×0.76+0.3×0.96+0.3×0.925=0.304+0.288+0.2775=0.8695;

[0049] =0.3×0.92+0.15×(1 / 28)+0.2×(130 / 180)+0.15×0.82+0.1×(1−0.05)+0.1×(1−0.15)=0.276+0.005+0.144+0.123+0.095+0.085=0.728;

[0050] =0.4×1+0.6×0.35=0.4+0.21=0.61≥0.6, early withdrawal is prohibited.

[0051] Example 3

[0052] The difference from Example 1 is: basic withdrawal ratio =65%, minimum withdrawal ratio threshold =15%; Original amount on waybill =50,000 yuan, M=4 (missing "signed" event); shipowner =45 days =78%.

[0053] Calculation process:

[0054] =4 / 5=0.8, =0.3×0.8+0.25×0.95+0.25×0.85+0.2×1.0=0.24+0.2375+0.2125+0.2=0.89;

[0055] =0.4×0.88+0.3×0.97+0.3×0.94=0.352+0.291+0.282=0.925;

[0056] =0.3×0.96+0.15×(1 / 45)+0.2×150 / 180+0.15×0.78+0.1×0.94+0.1×0.88=0.288+0.003+0.167+0.117+0.094+0.088=0.757;

[0057] =0.08, R=65%×0.89×0.925×0.757×(1−0.08)≈65%×0.89×0.925×0.757×0.92≈0.347;

[0058] =50000×0.347=17350 yuan, 17350≥50000×15%=7500, thus qualifying for early withdrawal.

[0059] Comparative Example 1

[0060] The difference from Example 1 is that step five, fraud detection, is missing; it is directly based on... , , The withdrawal amount is calculated, and the remaining steps and parameters are the same as in Example 1.

[0061] Calculation results: withdrawable proportion =70%×0.952×0.921×0.79≈0.504, =100000×0.504=50400 yuan; subsequent investigation revealed that the shipowner had engaged in concealed fraudulent activities of splitting orders to obtain cash, leading to the default on the loan.

[0062] Comparative Example 2

[0063] The difference from Example 1 is that no trajectory correction is performed in step three; the original trajectory is used directly to calculate TRV. The remaining steps and parameters are the same as in Example 1.

[0064] Calculation result: Outliers were not corrected, leading to =1−1.5 / 5=0.7, =1−4 / 240≈0.983, =0.4×0.7+0.3×0.983+0.3×0.967≈0.28+0.295+0.29=0.865;

[0065] Available ratio =70%×0.952×0.865×0.79×0.928≈0.432, =43,200 yuan, due to deviations in the assessment of the authenticity of the trajectory, resulting in a decrease in the suitability of the credit limit.

[0066] Test methods

[0067] To verify the effectiveness of the present invention, the following test indicators and methods were designed, and test conditions were set with reference to water transport logistics industry standards and the actual business scenarios of the platform:

[0068] Waybill authenticity identification accuracy: 1000 known authentic waybills (including 300 fake waybills) were selected to test the accuracy of the invention in identifying the authenticity of waybills. The calculation method is "number of correctly identified waybills / total number of test cases × 100%";

[0069] Risk assessment accuracy: Track the actual default situation of 1,000 loan transactions and compare the consistency between the risk score S of this invention and the actual default results. Accuracy = "Number of transactions where the risk score matches the actual default / Total number of transactions × 100%";

[0070] Fraud detection rate: 500 waybills containing fraudulent behavior (including 200 cases of concealed fraud) were selected to test the fraud detection success rate of the present invention. The detection rate = "number of successfully detected fraud / total number of fraudulent waybills × 100%";

[0071] Loan default rate: The actual default rate of all loan transactions during the statistical testing period. Default rate = "Number of defaulted loans / Total number of loans" × 100%;

[0072] Appropriateness of withdrawable amount: The evaluation of the reasonableness of the limit was collected from 500 ship owners and platform risk control personnel through questionnaires. The appropriateness = "number of evaluations that believe the limit is reasonable / total number of evaluations × 100%"; Test conditions: The test period is 6 months, covering inland waterway and coastal routes, involving 100 ship owners and 3,000 waybills. The basic withdrawal ratio of the platform is uniformly set at 70%.

[0073] Test metrics Example 1 Example 2 Example 3 Comparative Example 1 Comparative Example 2 Waybill authenticity verification accuracy rate (%) 96.8 95.5 94.2 96.5 88.7 Risk assessment accuracy (%) 93.5 92.8 91.7 93.2 90.5 Fraud detection rate (%) 92.3 94.1 91.5 65.7 90.8 Loan default rate (%) 2.1 1.8 2.5 8.7 3.8 Appropriation of withdrawable amount (%) 94.6 93.2 92.8 94.3 87.4

[0074] As can be seen from the test results, Embodiments 1-3 of the present invention are significantly superior to the comparative examples and the prior art in all indicators:

[0075] The accuracy rate of waybill authenticity identification exceeds 94%, which is more than 20% higher than the existing technology. This is thanks to the multi-dimensional indicator screening in step two and the trajectory correction technology in step three, which effectively solves the problem of identifying fake waybills.

[0076] The risk assessment accuracy rate reached 91.7%-93.5%, which is about 23 percentage points higher than the existing technology. This is attributed to the multi-dimensional shipowner operation indicator system and standardized scoring model constructed in step four, which enabled the quantitative assessment of risks.

[0077] The fraud detection rate is as high as 91.5%-94.1%, which is more than 50% higher than the existing technology. This reflects the advantages of the "rule layer + model layer" integrated identification mechanism, which can effectively cover both explicit and hidden fraud.

[0078] The loan default rate was controlled at 1.8%-2.5%, which is significantly lower than the existing technology (16.8%), verifying the effectiveness of the full-process risk control system of this invention;

[0079] The applicability of withdrawable amounts exceeds 92%, an improvement of approximately 27 percentage points compared to existing technologies, indicating that the multi-factor linkage method for calculating credit limits can achieve accurate matching between risk and credit limit.

[0080] Summarize

[0081] This invention addresses the core shortcomings of existing technologies in areas such as waybill authenticity identification, shipowner risk assessment, fraud interception, and credit limit adaptation by constructing a comprehensive technical system that integrates data collection, waybill screening, authenticity verification, risk assessment, fraud identification, credit limit determination, and risk control.

[0082] The implementation of this invention requires no new hardware equipment and can be directly deployed on existing water transport platforms. This reduces the risk of loan defaults and manual review costs on the platform, while also meeting the reasonable cash flow needs of ship owners. It significantly improves the operational efficiency and security of water transport logistics financial services and has broad application prospects and market value.

[0083] While the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the invention. Any variations and modifications can be made by those skilled in the art without departing from the spirit and scope of the invention. Therefore, any modifications, equivalent changes, and alterations made to the above embodiments based on the technical essence of the present invention, without departing from the scope of the invention, fall within the protection scope defined by the claims of the present invention.

Claims

1. A method for risk assessment and amount determination of advance freight charges for water transport platforms, characterized in that, The evaluation method includes the following steps: Step 1: Collect and preprocess waybill-related data from multiple channels, and extract standardized basic variables; Step 2: Calculate multidimensional indicators and obtain a comprehensive qualification score through a linear weighted model to screen qualified waybills; Step 3: Identify and correct anomalies in the AIS tracks of qualified waybills, and calculate the transportation authenticity score; Step 4: Extract and standardize the shipowner's historical operating indicators, and obtain the shipowner's risk score through a weighted model; Step 5: Fraud detection is performed by fusing the rule layer and the model layer to generate a fraud score; Step 6: Based on the aforementioned multiple scoring systems and basic parameters, calculate the maximum withdrawable amount and perform threshold verification; Step 7: Write all scoring parameters and judgment results back to the business system to achieve risk control linkage and complete the risk assessment and amount determination for early withdrawal of freight charges.

2. The method for risk assessment and amount determination of advance freight charges for a water transport platform according to claim 1, characterized in that: In step one, based on data collected from web pages, mobile applications, port operation systems, AIS latitude and longitude and speed data, original freight amounts and change records, and APP location data, after removing invalid noise data, five types of standardized basic variables are extracted: the actual total number of loading, departure, arrival, unloading, and receipt events in the transportation process. Total number of trajectory records collected Valid number of trajectory records The shortest distance between each valid trajectory sampling point and the standard waterway Original amount of shipping fee With change amount .

3. The method for risk assessment and amount determination of advance freight charges for a water transport platform according to claim 1, characterized in that: In step two, the event chain integrity is calculated based on the basic variables from step one. Continuity of flight path Normalized value of track deviation index and freight abnormality coefficient The overall qualification score is calculated using a linear weighted scoring model, and the calculation formula is as follows: ； in, , , , The data was obtained through training on a set of historical real waybill samples, with a set threshold. , ≥ The waybill must be qualified for subsequent risk control verification; otherwise, the application for early withdrawal will be rejected.

4. The method for risk assessment and amount determination of advance freight charges for a water transport platform according to claim 1, characterized in that: In step three, the trajectory points are arranged in chronological order to form a sequence, and the deviation sequence from the standard waterway is calculated based on the actual displacement. With theoretical displacement Comparison, satisfaction The time point is marked as an anomaly. The value is set to 1.5-2.

5. Neighborhood smoothing, linear interpolation, and path reconstruction are used to correct the trajectory, and the trajectory offset score is calculated. Outlier ratio score Trajectory continuity score ,pass Obtain a transportation authenticity score. The weights are obtained through training.

5. The method for risk assessment and amount determination of advance payment of freight charges by a water transport platform according to claim 1, characterized in that: In step four, the shipowner's repayment success rate over the past six months is extracted. Average payment cycle Continuous valid sailing days Average load factor abnormal navigation percentage Capital utilization volatility coefficient The metrics are standardized using z-scores. Obtain the shipowner's risk score. ~ The weights are obtained by training on historical default samples, and the sum of the weights is 1.

6. The method for risk assessment and amount determination of advance freight charges for a water transport platform according to claim 1, characterized in that: In step five, the rule layer presets rules for navigation anomalies, missing operational events, abnormal freight rate changes, account and equipment anomalies, and abnormal fund flows. Triggering any one of these rules will trigger a new rule. ,otherwise The model layer constructs feature vectors of waybills, trajectories, and shipowner behavior, which are then input into the logistic regression model. Calculate the probability of fraud, set ,pass Receive the final fraud rating. , ≥ Early withdrawal is prohibited.

7. The method for risk assessment and amount determination of advance payment of freight charges by a water transport platform according to claim 1, characterized in that: In step six, a comprehensive qualification score is obtained. Transportation authenticity score Shipowner risk rating Fraudulent ratings Original freight amount on the waybill and basic withdrawal ratio ,pass Calculate the withdrawable percentage, and then obtain Set a minimum withdrawal ratio threshold , Then you are not eligible for early withdrawal; otherwise, you are allowed to withdraw early. Withdrawals are permitted within the specified scope.

8. The method for risk assessment and amount determination of advance freight charges for a water transport platform according to claim 1, characterized in that: In step seven, the maximum withdrawable amount Comprehensive qualification score Transportation authenticity score Shipowner risk rating Fraudulent ratings The judgment results are written back to the water transport platform business system for loan amount verification, risk control audit and result backtracking, post-withdrawal credit limit rolling management, and risk tracking and anomaly analysis.

Images