Dynamic order distribution system and method based on multi-dimensional credit score, and storage medium

Abstract

The invention discloses a dynamic order distribution system based on a multi-dimensional credit score, and the system comprises a seller information collection module which is used for collecting seller identity authentication information, third-party credit score information and platform data information, and the platform data information comprises platform contract information, margin information and seller performance information; the credit scoring module is used for calculating credit scores of the sellers; the order distribution module is used for filtering sellers according to credit score threshold values set by the buyers, sorting the filtered sellers according to credit scores and order receiving quotation and then distributing orders; and the fund management and control module is used for transferring the prepaid money of the buyer into a bank deposit account, transferring the platform commission into a platform own fund account after the order is completed, transferring the settlement money of the seller into a special clearing account, and carrying out risk management and control on deposit withdrawal of the seller. According to the invention, from one-way order grabbing to credit two-way hedging, order distribution is carried out through credit scoring, the optimal balance of resource-risk-efficiency is realized, and the traditional rigid distribution logic is broken through.

Description

Technical Field

[0001] This invention relates to the fields of financial technology and e-commerce technology, and in particular to a dynamic order distribution system, method and storage medium based on multidimensional credit scoring. Background Technology

[0002] While third-party top-up platforms provide matching services for buyers placing orders and sellers fulfilling orders, they suffer from the following structural flaws: 1) The lack of a credit assessment system leads to fraud risks.

[0003] Existing platforms lack multi-dimensional credit profiling capabilities for both order-issuing buyers (such as suppliers of phone bill / membership orders) and order-fulfilling sellers (recharge channel providers), relying solely on basic transaction data (such as order success rates) for risk control. This leads to frequent risks. Buyer fraud risks: Orders originate from unknown sources, posing risks such as fraudulent orders for cash-out and money laundering; The authenticity of seller resources is out of control: Sellers using unauthorized channels (such as non-operator official APIs) are causing a surge in recharge failure rates, but the platform is unable to intercept this in real time; Fund security is not guaranteed: Prepayments and settlement funds are difficult to control, and it is impossible to find a balance between meeting sellers' real-time withdrawal needs and managing fund risks.

[0004] 2) Order distribution is inefficient and resources are misallocated.

[0005] The current platforms generally adopt a one-way "order-grabbing model," the core contradiction of which lies in: Buyers are passively waiting: High-credit buyers cannot get priority access to low-priced, high-quality channels, and high-value orders are often snatched up by low-credit sellers; Inflexible routing strategy: Order allocation is not combined with real-time credit changes (such as seller channel stability index); Cross-domain data fragmentation: Credit information is not shared among financial institutions, telecom operators, and e-commerce platforms, making it impossible to conduct preliminary credit assessments of sellers; Based on the above problems, there is an urgent need for an order distribution system that integrates dynamic credit assessment and intelligent routing algorithms to achieve dual optimization of risk control and resource efficiency. Summary of the Invention

[0006] The purpose of this invention is to provide a dynamic order distribution system, method, and storage medium based on multidimensional credit scoring to solve the above-mentioned problems.

[0007] To achieve the above objectives, the following technical solution is adopted: A dynamic order distribution system based on multidimensional credit scoring, including The seller information collection module is used to collect seller identity authentication information, third-party credit score information, and platform data information. Among them, the platform data information includes platform contract information, security deposit information, and seller performance information. The credit scoring module is used to calculate the seller's credit score; The order distribution module is used to filter sellers based on the credit score threshold set by the buyer, and then distribute orders to the filtered sellers by sorting them by credit score and order quotation. The funds management module is used to transfer buyers' prepayments to a bank custody account, transfer platform commissions to the platform's own funds account after an order is completed, transfer seller settlement funds to a dedicated clearing account, and manage the risk of seller withdrawals.

[0008] Furthermore, the credit scoring module calculates a basic credit score by accumulating three fundamental dimensions: seller identity authentication information, third-party credit score, and contract deposit. The identity authentication information includes real-name authentication score and qualification document score; the third-party credit score includes credit score obtained from third-party platforms; and the contract deposit score includes contract signing score and deposit amount score.

[0009] Furthermore, the credit scoring module will also calculate and adjust the credit score based on the seller's order success rate, order callback time, and buyer complaint rate, and add the basic credit score and the adjusted credit score to form the final credit score.

[0010] Furthermore, the order distribution module employs a reinforcement learning strategy to rank sellers based on their real-time credit score, channel stability index, and price-weighted matching degree, and automatically allocates orders according to the matching degree from high to low.

[0011] Furthermore, the fund management module's risk control for seller withdrawals specifically includes: When a single seller's daily withdrawal count exceeds a preset threshold, facial recognition verification is triggered. Seller funds will be automatically frozen when the seller's daily cumulative withdrawal limit is reached or when abnormal login is detected. When a seller's credit score falls below a preset threshold, settlement will be processed on a T+1 basis.

[0012] A dynamic order distribution method based on multidimensional credit scoring includes the following steps: S1: Collect seller identity verification information, third-party credit score information and platform data information, and calculate the seller's credit score. Among them, the platform data information includes platform contract information, security deposit information and seller performance information. S2: Buyers post orders with prepayment and can choose to set a seller credit score threshold; S3: After the platform receives an order, if the buyer has set a credit score threshold, it will filter out sellers that are not lower than the threshold, and calculate the weighted matching degree of the filtered sellers according to their credit score and the order price, and sort them from high to low. If the buyer has not set a credit score threshold, all sellers will be retained. S4: The platform pushes orders to the sorted sellers in sequence. If the current seller fails to accept the order, the platform continues to push the order to the next seller until a seller successfully accepts the order or there are no sellers to accept the order. S5: After the order is completed, the buyer's prepayment will be transferred from the bank custody account to the seller's clearing account, and the corresponding platform commission will be transferred to the platform's own funds account.

[0013] Furthermore, in step S1, the adjustment credit score will be calculated based on the seller's order success rate, order callback time, and buyer complaint rate, and the basic credit score and the adjustment credit score will be added together to form the final credit score.

[0014] Furthermore, in step S5, when a seller applies for withdrawal, if the seller's credit score is lower than the platform's set threshold, settlement will be carried out on a T+1 basis; if the seller's number of withdrawals in a day exceeds the platform's threshold, facial recognition verification will be performed; if the cumulative number of withdrawals in a day reaches the upper limit or abnormal login is detected, the seller's account and funds will be frozen.

[0015] A computer-readable storage medium storing a computer program adapted to be loaded and executed by a processor to cause a computer device having the processor to perform the methods described above.

[0016] By adopting the above solution, the beneficial effects of the present invention are: 1) Credit assessment dimension: From static one-way to dynamic integration, breaking through the traditional one-dimensional assessment model, constructing a cross-domain dynamic profile, and solving the pain points of channel fraud and resource authenticity unique to the recharge industry; 2) Order distribution mechanism: From one-way order grabbing to two-way credit hedging, order distribution is carried out through credit scoring, which achieves the optimal balance of resources, risks and efficiency, breaking through the traditional rigid distribution logic; 3) Fund security control: Fund risks are dispersed through a three-account separation scheme, ensuring security throughout the entire fund chain; 4) Industry adaptability advantage: Customized innovation for recharge scenarios, with special order allocation logic designed for the unique risk points of recharge such as phone bill / coupon redemption, and improved overall business compliance through risk hedging. Attached Figure Description

[0017] Figure 1 This is a flowchart of the method of the present invention; Figure 2 This is a flowchart of the seller credit score calculation process of the present invention; Figure 3 This is a flowchart of the order distribution process of the present invention; Figure 4 This is a flowchart illustrating the financial risk management of this invention. Detailed Implementation

[0018] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments.

[0019] Reference Figures 1 to 4 As shown, the present invention provides a dynamic order distribution system based on multidimensional credit scoring. In one embodiment, it includes... The seller information collection module is used to collect seller identity authentication information, third-party credit score information, and platform data information. Among them, the platform data information includes platform contract information, security deposit information, and seller performance information. The credit scoring module is used to calculate the seller's credit score; The order distribution module is used to filter sellers based on the credit score threshold set by the buyer, and then distribute orders to the filtered sellers by sorting them by credit score and order quotation. The funds management module is used to transfer buyers' prepayments to a bank custody account, transfer platform commissions to the platform's own funds account after an order is completed, transfer seller settlement funds to a dedicated clearing account, and manage the risk of seller withdrawals.

[0020] In this embodiment, the seller information collection module extracts and standardizes three types of data sources: identity authentication information (real name and qualifications), third-party credit scores (operators / e-commerce / financial institutions), and platform data information (contracts, security deposits, performance records). The data can be organized using a "fact table + dimension table," and master data alignment is completed with the seller ID as the primary key. Specifically: On the identity authentication side: OCR recognizes business licenses / authorization documents, and combines liveness detection with document consistency verification before storing the data in the database; Third-party credit side: When calling external interfaces, only the final score or necessary statistics are retrieved to avoid collecting raw sensitive details; the interface response can be hashed and optionally stored on the blockchain for later verification. On the platform's data side: Contract status (valid / invalid / pending renewal), deposit amount, and performance events (success rate, callback delay, complaint records) are continuously entered into the database, providing real-time features for scoring and routing.

[0021] The credit scoring module calculates a basic credit score by accumulating three fundamental dimensions: seller identity verification information, third-party credit score, and contract deposit. The identity verification information includes real-name authentication score and qualification document score; the third-party credit score includes credit score obtained from third-party platforms; and the contract deposit score includes contract signing score and deposit amount score. In addition, the credit scoring module also calculates a regulatory credit score based on the seller's order success rate, order callback time, and buyer complaint rate, and then adds the basic credit score and the regulatory credit score to form the final credit score.

[0022] The scoring system consists of a "basic credit score + a regulatory credit score." The basic credit score is accumulated from three dimensions (real-name registration / qualification, third-party credit, and contract and security deposit), with a maximum score of 600. The regulatory credit score is adjusted based on daily performance (success rate, callback time, and complaints) to reflect timeliness and behavioral feedback. Specifically: Dimension 1's static data score accounts for 30% of the total score. Completing real-name authentication adds 100 points. For each channel agent authorization document / business license uploaded and recognized by OCR, 20 points are added. The total score for Dimension 1 is capped at 180 points. Dimension 2, the third-party credit score, accounts for 40% of the total score. It obtains the user's credit score on other platforms through supported third-party data interfaces, such as WeChat credit score, Alipay credit score, and financial institution scores. The total score for Dimension 2 is capped at 240 points. Dimension 3, including contracts and security deposits, accounts for 30% of the total score. Signing a transaction contract provided by the platform adds 80 points; paying a platform security deposit adds 1 point for every 100 yuan paid, up to a maximum of 100 points; the total score for Dimension 3 is capped at 180 points. The calculation formula is as follows: Dimension 1 = Real-name authentication score + min(number of qualification certificates × 20, 80); Dimension 2 = (WeChat Credit Score + Alipay Credit Score + Financial Institution Credit Score) × 10%; Dimension 3 = Contract signing score + min(deposit payment × 0.1, 100); The above is the basic credit score calculation logic. Dimension 4 is used to calculate the regulatory credit score. The script calculates the seller's daily order success rate. A success rate of 80% earns 20 points, and for every 5% increase (decrease) in the success rate, 20 points are added (subtracted), with a total maximum of 100 points. Calculate the average callback time for seller orders, with 20 points based on a two-minute timeframe. For every second the callback time decreases (increases), add (subtract) 1 point, up to a total maximum of 100 points. The number of buyer complaints is counted. 100 points are added for no complaints and 20 points are deducted for each complaint. The maximum increase or decrease is 100 points.

[0023] For the order distribution module, preferably, a reinforcement learning strategy can be adopted to sort sellers according to their real-time credit score, channel stability index and price weighted matching degree, and automatically allocate orders according to the matching degree from high to low.

[0024] This module first performs credit threshold filtering, then calculates a "weighted matching score" for candidate sellers to rank and push them. If the current seller fails to accept an order or times out, the next seller is polled in sequence. The weighted matching score uses a linear or segmented weighted model based on the seller's real-time credit score, channel stability index, and price, taking into account "risk, efficiency, and cost." Specifically: Threshold filtering: If the buyer sets a threshold (e.g., ≥700 points), the system will only retain qualified sellers; if no threshold is set, the platform's minimum threshold strategy (e.g., ≥ platform default score) can be applied. Stability Index: An index can be constructed based on the success rate, timeout rate, and average callback duration of the last N orders, and is updated in real time as the order cycle closes. Weighted matching degree (example): Score = α·Norm (credit score) + β·Norm (stability index) − γ·Norm (quote), α, β, γ ∈ [0,1] and sum to 1; the parameters can be gradually converged through historical backtesting / online A / B testing; Push and rollback: Push orders in order of priority; if the order fails (rejection / timeout / abnormality), it will automatically roll back to the next candidate until it succeeds or there are no available sellers.

[0025] For the fund management module, its risk control measures for seller withdrawals specifically include: When a single seller's daily withdrawal count exceeds a preset threshold, facial recognition verification is triggered. Seller funds will be automatically frozen when the seller's daily cumulative withdrawal limit is reached or when abnormal login is detected. When a seller's credit score falls below a preset threshold, settlement will be processed on a T+1 basis.

[0026] The fund management module follows the principle of separation of three accounts: buyer prepayment is deposited into a bank custody account (refundable if the order is not completed); after the order is completed, the platform commission is deposited into its own account and the seller's payment is deposited into a clearing account; during the withdrawal stage, thresholds, identity verification and abnormal freezing are introduced to isolate bad debts and the risk of theft.

[0027] In addition, a dynamic order distribution method based on multidimensional credit scoring is also provided, including the following steps: S1: Collect seller identity verification information, third-party credit score information, and platform data information, and calculate the seller's credit score. The platform data information includes platform contract information, security deposit information, and seller performance information.

[0028] In step S1, the adjustment credit score is calculated based on the seller's order success rate, order callback time, and buyer complaint rate, and the basic credit score and the adjustment credit score are added together to form the final credit score.

[0029] In this process, sellers first register an account on the platform and complete initial information collection, with static data on real-name registration being mandatory. Next, they obtain user authorization to access third-party credit data, which is used to calculate the platform's credit score proportionally. Finally, sellers can improve their credit score by signing a contract with the platform and paying a deposit, with the contract and deposit amounts proportionally used to calculate the platform's credit score. Except for real-name registration, which is mandatory, the other two methods are optional. After completing the basic credit score calculation, sellers can begin transaction activities. During these activities, seller behavior is continuously monitored, with transaction metrics (success rate, recharge efficiency, etc.) set to assess seller fulfillment rates. Meeting daily targets increases the seller's credit score, while failing to meet them decreases it. Through continuous monitoring and assessment, the seller's credit score is updated in real time.

[0030] Then, the order distribution stage begins, namely: S2: Buyers post orders with prepayment and can choose to set a seller credit score threshold; S3: After the platform receives an order, if the buyer has set a credit score threshold, it will filter out sellers that are not lower than the threshold, and calculate the weighted matching degree of the filtered sellers according to their credit score and the order price, and sort them from high to low. If the buyer has not set a credit score threshold, all sellers will be retained. S4: The platform pushes orders to the sorted sellers in sequence. If the current seller fails to accept the order, it continues to push the order to the next seller until a seller successfully accepts the order or there are no sellers to accept the order.

[0031] In this stage, buyers first post prepaid orders on the platform. Buyers can choose a credit score threshold to increase the success rate. If a buyer sets a credit score threshold, the platform filters sellers based on this threshold after the order enters the platform. Sellers with insufficient credit scores are filtered out, and the remaining sellers are ranked according to their credit score and order price. Orders are distributed to the corresponding sellers from highest to lowest score. During this process, the platform checks the seller's order acceptance status and removes sellers not currently accepting orders from the ranking. Once a seller accepts an order, the process ends if successful; otherwise, the platform rules cycle the orders and push them to the next seller in the ranking. Alternatively, if the buyer does not set a credit score threshold, the platform checks if a minimum credit score threshold is set. If so, refer to the steps above; if not, sellers are not removed, and the remaining logic for order distribution is the same as described above.

[0032] S5: After the order is completed, the buyer's prepayment will be transferred from the bank custody account to the seller's dedicated clearing account, and the corresponding platform commission will be transferred to the platform's own funds account. Meanwhile, in step S5, when the seller applies for withdrawal, if the seller's credit score is lower than the platform's set threshold, settlement will be processed on a T+1 basis; if the seller's daily withdrawal count exceeds the platform's threshold, facial recognition verification will be performed; if the cumulative daily withdrawal count reaches the upper limit or abnormal login is detected, the seller's account and funds will be frozen.

[0033] Based on the premise that the buyer initiates a prepayment order, the prepayment is deposited into a bank escrow account to isolate the buyer's funds from the platform's funds. The order is distributed and circulated on the platform according to the above logic. If the order fails, the buyer's prepayment will be refunded to the original payment method. Based on the above, if the order is successful, the order settlement will be carried out according to the platform rules. The fund routing is determined, the platform commission is allocated to the platform's own account, and the seller's settlement funds are allocated to a special clearing account. Seller withdrawals are subject to a risk circuit breaker mechanism, and multiple assessments are conducted on the seller. Sellers with a credit score lower than the platform's set limit will be settled on a T+1 basis. If the number of withdrawals in a single day exceeds the platform's set limit, facial verification is required. If the cumulative number of withdrawals in a single day reaches the platform's set limit or an abnormal login device is detected, the seller's funds will be frozen for 24 hours.

[0034] Furthermore, a computer-readable storage medium is provided, which stores a computer program adapted to be loaded and executed by a processor, causing a computer device having the processor to perform the methods described above. This computer-readable storage medium may be an image processing module provided in any of the foregoing embodiments or an internal storage unit of the computer device, such as a hard disk or memory of the computer device. The computer-readable storage medium may also be an external storage device of the computer device, such as a plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, etc., provided on the computer device. Further, the computer-readable storage medium may include both internal storage units and external storage devices of the computer device. The computer-readable storage medium is used to store the computer program and other programs and data required by the computer device. The computer-readable storage medium can also be used to temporarily store data that has been output or will be output.

[0035] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A dynamic order distribution system based on multidimensional credit scoring, characterized in that, include The seller information collection module is used to collect seller identity authentication information, third-party credit score information, and platform data information. Among them, the platform data information includes platform contract information, security deposit information, and seller performance information. The credit scoring module is used to calculate the seller's credit score; The order distribution module is used to filter sellers based on the credit score threshold set by the buyer, and then distribute orders to the filtered sellers by sorting them by credit score and order quotation. The funds management module is used to transfer buyers' prepayments to a bank custody account, transfer platform commissions to the platform's own funds account after an order is completed, transfer seller settlement funds to a dedicated clearing account, and manage the risk of seller withdrawals.

2. The dynamic order distribution system based on multidimensional credit scoring according to claim 1, characterized in that, The credit scoring module calculates a basic credit score by accumulating three fundamental dimensions: seller identity verification information, third-party credit score, and contract deposit. The identity verification information includes real-name authentication score and qualification document score; the third-party credit score includes credit score obtained from third-party platforms; and the contract deposit score includes contract signing score and deposit amount score.

3. The dynamic order distribution system based on multidimensional credit scoring according to claim 2, characterized in that, The credit scoring module also calculates and adjusts the credit score based on the seller's order success rate, order callback time, and buyer complaint rate, and adds the basic credit score and the adjusted credit score to form the final credit score.

4. The dynamic order distribution system based on multidimensional credit scoring according to claim 1, characterized in that, The order distribution module employs a reinforcement learning strategy to rank sellers based on their real-time credit score, channel stability index, and price-weighted matching degree, and automatically allocates orders according to the matching degree from high to low.

5. The dynamic order distribution system based on multidimensional credit scoring according to claim 1, characterized in that, The fund management module specifically manages the risk of seller withdrawals, including: When a single seller's daily withdrawal count exceeds a preset threshold, facial recognition verification is triggered. Seller funds will be automatically frozen when the seller's daily cumulative withdrawal limit is reached or when abnormal login is detected. When a seller's credit score falls below a preset threshold, settlement will be processed on a T+1 basis.

6. A dynamic order distribution method based on multidimensional credit scoring, characterized in that, Includes the following steps: S1: Collect seller identity verification information, third-party credit score information and platform data information, and calculate the seller's credit score. Among them, the platform data information includes platform contract information, security deposit information and seller performance information. S2: Buyers post orders with prepayment and can choose to set a seller credit score threshold; S3: After the platform receives an order, if the buyer has set a credit score threshold, it will filter out sellers that are not lower than the threshold, and calculate the weighted matching degree of the filtered sellers according to their credit score and the order price, and sort them from high to low. If the buyer has not set a credit score threshold, all sellers will be retained. S4: The platform pushes orders to the sorted sellers in sequence. If the current seller fails to accept the order, the platform continues to push the order to the next seller until a seller successfully accepts the order or there are no sellers to accept the order. S5: After the order is completed, the buyer's prepayment will be transferred from the bank custody account to the seller's clearing account, and the corresponding platform commission will be transferred to the platform's own funds account.

7. The dynamic order distribution method based on multidimensional credit scoring according to claim 6, characterized in that, In step S1, the adjustment credit score will be calculated based on the seller's order success rate, order callback time, and buyer complaint rate, and the basic credit score and the adjustment credit score will be added together to form the final credit score.

8. The dynamic order distribution method based on multidimensional credit scoring according to claim 6, characterized in that, In step S5, when a seller applies for withdrawal, if the seller's credit score is lower than the platform's set threshold, settlement will be carried out on a T+1 basis; if the seller's number of withdrawals in a day exceeds the platform's threshold, facial recognition verification will be performed. If the daily withdrawal limit is reached or an abnormal login is detected, the seller's account and funds will be frozen.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program adapted to be loaded and executed by a processor so that a computer device having the processor performs the method according to any one of claims 6 to 8.

Images