Asset transaction risk monitoring system and method

Abstract

The invention discloses an asset transaction risk monitoring system and method, and relates to the field of asset transaction security. The method comprises the following steps: acquiring transaction object types, transaction time, transaction amount and flow funds through a plurality of transaction objects of a user; calculating a transaction frequency index according to the transaction time of different transaction objects; collecting multiple groups of asset risk training data through historical transaction data of multiple users; each group of asset risk training data comprises the transaction object type, the transaction frequency index and the transaction total amount ratio of a plurality of historical transaction objects of each user, and a risk mark indicating whether each transaction object has a flow fund risk, and training to obtain an asset risk prediction model.

Description

Technical Field

[0001] This invention relates to the field of asset transaction security, and specifically to an asset transaction risk monitoring system and method. Background Technology

[0002] In the financial sector, risk monitoring systems are crucial for preventing and controlling potential risks in asset transactions. While existing risk monitoring technologies have made significant contributions to risk warning and suspicious entity identification, their systems and methods still have certain limitations. Specifically, existing technologies, such as the big data-driven financial risk warning system disclosed in patent CN120525336A, achieve sensitive and accurate detection of financial risks and rapid response to abnormal transactions. However, this system lacks in-depth analysis of user transaction behavior and personalized risk prediction when dealing with risks in users' personal asset transactions. Patent CN120013229A also provides a reinforcement learning-based method for identifying suspicious entities involved in illegal fundraising. Although this improves the accuracy and efficiency of risk identification, it still falls short in real-time monitoring of personal asset transaction risks and accurate judgment of abnormal transactions. In view of the above technical problems, this application proposes an asset transaction risk monitoring system and method, aiming to construct a personalized asset risk prediction model by deeply analyzing the characteristics of users' transaction behavior, promptly judging transaction anomalies, effectively preventing and controlling asset transaction risks, and improving the timeliness and accuracy of risk warnings. Summary of the Invention

[0003] The technical problem this invention aims to solve is that existing technologies lack in-depth analysis of user transaction behavior and personalized risk prediction, and are insufficient in real-time monitoring of personal asset transaction risks and accurate judgment of abnormal transactions. The purpose is to provide an asset transaction risk monitoring system and method that trains an asset risk prediction model by combining the transaction frequency index and total transaction amount ratio of user historical transaction data to achieve real-time prediction of user liquidity risk. When liquidity risk is predicted, the system further judges abnormal transactions, thereby achieving accurate monitoring and early warning of asset transaction risks, thus solving the aforementioned technical problems.

[0004] This invention is achieved through the following technical solution:

[0005] An asset transaction risk monitoring system includes:

[0006] Transaction data acquisition module: Collects transaction type, transaction time, transaction amount, and cash flow from multiple transaction objects of the user; calculates a transaction frequency index based on the transaction time of different transaction objects;

[0007] Asset risk prediction module: Collects multiple sets of asset risk training data through historical transaction data of multiple users; each set of asset risk training data includes the transaction object type, transaction frequency index and total transaction amount ratio of multiple historical transaction objects of each user, as well as the risk mark of whether each transaction object has liquidity risk, and trains to obtain an asset risk prediction model;

[0008] Transaction Anomaly Detection Module: Collects multiple historical transaction objects of the current user, the transaction frequency index of each historical transaction object, and the proportion of the total transaction amount; predicts whether each transaction object has liquidity risk through the asset risk prediction model; when the transaction object has liquidity risk, determines whether the transaction is abnormal based on the real-time transaction access characteristics between the current user and the transaction object; when the transaction is abnormal, prohibits the current user from trading with the transaction object.

[0009] A method for monitoring asset transaction risks, comprising:

[0010] Collect transaction type, transaction time, transaction amount, and cash flow from multiple transaction objects of the user; calculate the transaction frequency index based on the transaction time of different transaction objects;

[0011] Multiple sets of asset risk training data are collected by collecting historical transaction data from multiple users; each set of asset risk training data includes the transaction object type, transaction frequency index and transaction amount percentage of multiple historical transaction objects of each user, as well as whether the user has liquidity risk, and the asset risk prediction model is obtained by training.

[0012] Collect multiple historical transaction objects of the current user, the transaction frequency index of each historical transaction object, and the proportion of the total transaction amount; predict whether each transaction object has liquidity risk through the asset risk prediction model; when the transaction object has liquidity risk, determine whether the transaction is abnormal based on the real-time transaction access characteristics between the current user and the transaction object; when the transaction is abnormal, prohibit the current user from trading with the transaction object.

[0013] The types of transaction objects include individuals or institutions, and credit scores; the transaction times include weekdays or holidays, and time periods.

[0014] When collecting multiple sets of asset risk training data through historical transaction data of multiple users, whether each transaction object has liquidity risk is determined by the following steps: the multiple sets of asset risk training data are grouped according to different months; a minimum asset threshold is calculated based on the average daily expenditure amount of each transaction object and the corresponding user in the previous month and the number of days remaining in the current month; it is determined whether the funds of each transaction object are lower than the minimum asset threshold, and if and only if they are lower, it indicates that the transaction object has liquidity risk.

[0015] The step of determining whether a transaction is abnormal based on the real-time transaction access characteristics of the current user and the transaction object specifically includes the following steps: collecting multiple sets of normal transaction feature samples and multiple sets of abnormal transaction access feature samples for training to obtain a transaction abnormality training model; inputting the real-time transaction access characteristics of the transaction object for the current user into the transaction abnormality training model to obtain the result of whether the transaction is abnormal; the normal transaction feature samples and the abnormal transaction access feature samples each include any one or more of the transaction object's login time, transaction amount, IP address, device type, and session duration.

[0016] The step of calculating the transaction frequency index based on the transaction time of different trading objects specifically includes the following steps:

[0017] The trading frequency of different trading objects is calculated based on their trading time; a time decay factor is calculated based on the most recent trading time of each trading object; the trading frequency index is calculated by introducing the time decay factor and the trading frequency; the proportion of total trading volume is calculated by the following steps: the total trading volume is calculated based on the trading amount of different trading objects, and the proportion of total trading volume of each trading object is calculated based on the total trading volume of all trading objects.

[0018] After prohibiting the current user from transacting with the transaction object when a transaction is abnormal, the method further includes the following steps: collecting multiple sets of successful transaction samples and failed transaction samples to train a transaction risk prediction model; the successful transaction samples include transaction object type, transaction time, transaction amount, working capital amount, and abnormal access characteristics; the failed transaction samples include collecting multiple sets of failed transaction object type, transaction time, transaction amount, working capital amount, and abnormal access characteristics.

[0019] The asset transaction risk monitoring method further includes: using a transaction risk prediction model to determine whether the transaction of the transaction object to be predicted is successful; and storing the transaction object IDs and transaction history records of each user's successful transactions.

[0020] Before predicting whether each of the transaction objects has liquidity risk using the asset risk prediction model, the method further includes:

[0021] Each user's transaction object is authenticated using the stored transaction object ID.

[0022] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0023] This application relates to an asset transaction risk monitoring system and method. It collects data on the transaction types, transaction times, transaction amounts, and working capital of multiple users' transaction objects, and calculates a transaction frequency index based on the transaction time of each transaction object. It also collects data on the transaction types, transaction frequency indices, and total transaction amount percentages of multiple transaction objects from multiple groups of different users, and trains this data to obtain an asset risk prediction model. This model predicts the working capital risk of different transaction events under the weighted influence of transaction frequency and amount, enabling accurate monitoring of secure transactions. When a user initiates an asset expenditure transaction, based on historical transaction objects, the transaction frequency index, and total transaction amount percentage of each historical transaction object, the asset risk prediction model predicts whether each historical transaction object has working capital risk leading to bond problems, achieving early screening and warning of the working capital safety risk of historical transaction objects. Furthermore, when a transaction object has working capital risk, the system monitors and locates the transaction object when it engages in real-time transactions with any user, and determines whether the transaction is abnormal based on real-time transaction access characteristics. If the transaction is abnormal, the system prohibits the transaction object from continuing to trade. This invention enables the prediction of liquidity risk of transaction objects based on a user's historical transactions and provides real-time monitoring of specific transaction objects with liquidity risks. This improves the accuracy of risk warnings, effectively prevents the escalation of transaction risks, and safeguards the assets of all users. By training an asset risk prediction model using the transaction frequency index and total transaction amount ratio of a user's historical transaction data, it achieves real-time prediction of the user's liquidity risk. When liquidity risk is predicted, it further identifies abnormal transactions, thereby achieving precise monitoring and early warning of asset transaction risks. This solves the problems of in-depth analysis of user transaction behavior and personalized risk prediction, and meets the requirements for real-time monitoring of personal asset transaction risks and accurate judgment of abnormal transactions. Attached Figure Description

[0024] To more clearly illustrate the technical solutions of the exemplary embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly described below. It should be understood that the following drawings only show some embodiments of the present invention and should not be considered as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort. In the drawings:

[0025] Figure 1 This is a flowchart of the asset transaction risk monitoring method in an embodiment of this application. Detailed Implementation

[0026] To make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the embodiments and accompanying drawings. The illustrative embodiments and descriptions of the present invention are only used to explain the present invention and are not intended to limit the present invention.

[0027] Example

[0028] like Figure 1 As shown in the embodiment of this application, an asset transaction risk monitoring method is provided, including:

[0029] Collect transaction type, transaction time, transaction amount, and cash flow from multiple transaction objects of the user; calculate the transaction frequency index based on the transaction time of different transaction objects;

[0030] Multiple sets of asset risk training data are collected by collecting historical transaction data from multiple users; each set of asset risk training data includes the transaction object type, transaction frequency index and total transaction amount percentage of multiple historical transaction objects of each user, as well as the risk mark of whether each transaction object has liquidity risk, and the asset risk prediction model is obtained by training.

[0031] Collect multiple historical transaction objects of the current user, the transaction frequency index of each historical transaction object, and the proportion of the total transaction amount; predict whether each transaction object has liquidity risk through the asset risk prediction model; when the transaction object has liquidity risk, determine whether the transaction is abnormal based on the real-time transaction access characteristics between the current user and the transaction object; when the transaction is abnormal, prohibit the current user from trading with the transaction object.

[0032] Specifically, the system collects data on the types, times, amounts, and liquidity of multiple trading partners through monthly historical transaction data. The trading frequency of each partner can be calculated based on the transaction times of multiple trades, and then mathematically standardized into a trading frequency index. Training data for each asset risk group is obtained from each user's monthly historical transaction data. Multiple sets of asset risk training data are used to train an asset risk prediction model using reinforcement learning methods. This model can monitor the liquidity risk of a user's historical trading partners based on their historical trading activity. By identifying and verifying actively traded partners with high liquidity, the system improves the efficiency of transaction security management.

[0033] The types of transaction objects include individuals or institutions, and credit scores; the transaction times include weekdays or holidays, and time periods.

[0034] The transactions are categorized based on the type of individual or institution and their credit score, with the credit score expressed on a 100-point scale to reflect an individual's debt repayment performance. Transaction times are divided according to weekdays, holidays, and time periods to determine the transaction patterns of users and their counterparties.

[0035] When collecting multiple sets of asset risk training data through historical transaction data of multiple users, whether each transaction object has liquidity risk is determined by the following steps: the multiple sets of asset risk training data are grouped according to different months; a minimum asset threshold is calculated based on the average daily expenditure amount of each transaction object and the corresponding user in the previous month and the number of days remaining in the current month; it is determined whether the funds of each transaction object are lower than the minimum asset threshold, and if and only if they are lower, it indicates that the transaction object has liquidity risk.

[0036] The average daily expenditure of each trading partner is calculated based on the total transaction expenditure of each trading partner and its corresponding user in the previous month. This average daily expenditure is then used to determine whether the trading partner poses a liquidity risk based on the remaining days of the current month, and if so, prohibiting the user from continuing transactions with that trading partner. The liquidity risk of a trading partner is determined by comparing its transaction expenditure with that of the corresponding user in the previous month, taking into account the remaining days of the current month, to ensure the security of monthly transactions. One trading partner corresponds to one user. Optionally, the same trading partner can correspond to multiple users, and the total transaction expenditure of the trading partner in the previous month can be calculated based on the sum of the expenditures of all users. The remaining days of the current month can be calculated based on the number of days remaining in the month in which the liquidity risk update date falls.

[0037] The step of determining whether a transaction is abnormal based on the real-time transaction access characteristics of the current user and the transaction object specifically includes the following steps: collecting multiple sets of normal transaction feature samples and multiple sets of abnormal transaction access feature samples for training to obtain a transaction abnormality training model; inputting the real-time transaction access characteristics of the transaction object for the current user into the transaction abnormality training model to obtain the result of whether the transaction is abnormal; the normal transaction feature samples and the abnormal transaction access feature samples each include any one or more of the transaction object's login time, transaction amount, IP address, device type, and session duration.

[0038] Specifically, the system uses samples from normal and abnormal transaction accesses collected from logs for training, which are then used to determine whether a transaction is normal based on real-time transaction access characteristics monitored in real time. Login time, transaction amount, IP address, device type, and session duration are used to determine whether the transaction is within a security threshold.

[0039] The step of calculating the transaction frequency index based on the transaction time of different trading objects specifically includes the following steps:

[0040] The trading frequency of different trading objects is calculated based on their trading time; a time decay factor is calculated based on the most recent trading time of each trading object; the trading frequency index is calculated by introducing the time decay factor and the trading frequency; the proportion of total trading volume is calculated by the following steps: the total trading volume is calculated based on the trading amount of different trading objects, and the proportion of total trading volume of each trading object is calculated based on the total trading volume of all trading objects.

[0041] The trading frequency for the month is calculated based on transaction time and standardized as a trading frequency factor. The frequency of transactions occurring multiple times a day or once every few days is converted into the smallest time unit, such as one hour or two hours, and then converted to a value between 0 and 1. Combining the most recent transaction time for each transaction, the influence of the time dimension gradually decreases as the time distance increases, resulting in a smaller time decay factor value (0-1). Optionally, the time decay factor can be calculated as the percentage of the remaining days of the month after the most recent transaction, allowing for dynamic updates based on changes in collected data and dates, enabling rapid capture of daily trading activity changes. When the trading frequency factor and time decay factor are both 0, they can be ignored. First, the total transaction amount for all transactions of the user is calculated. Then, based on the percentage of each transaction's total transaction amount for the month to the total transaction amount of all transactions for the month, the percentage of each transaction's total transaction amount is calculated.

[0042] After prohibiting the current user from transacting with the transaction object when a transaction is abnormal, the method further includes the following steps: collecting multiple sets of successful transaction samples and failed transaction samples to train a transaction risk prediction model; the successful transaction samples include transaction object type, transaction time, transaction amount, working capital amount, and abnormal access characteristics; the failed transaction samples include collecting multiple sets of failed transaction object type, transaction time, transaction amount, working capital amount, and abnormal access characteristics.

[0043] By training multiple samples of successful and failed transactions from various users, a transaction risk prediction model is obtained that combines the transaction object type, transaction time, transaction amount, liquid funds amount, and transaction access characteristics of each user. This model can then quickly screen out transaction objects with transaction risks by combining asset risk and access anomalies.

[0044] The asset transaction risk monitoring method further includes: using a transaction risk prediction model to determine whether the transaction of the transaction object to be predicted is successful; and storing the transaction object IDs and transaction history records of each user's successful transactions.

[0045] Before predicting whether each of the transaction objects has liquidity risk using the asset risk prediction model, the method further includes:

[0046] Each user's transaction object is authenticated using the stored transaction object ID.

[0047] When determining the success of a transaction with a predicted trading object based on the transaction risk prediction model, the transaction object ID and transaction history are stored for easy retrieval by the user. The transaction history can include basic information such as the transaction object type, transaction amount, and transaction time.

[0048] In summary, the embodiments of this application provide an asset transaction risk monitoring system and method:

[0049] By collecting data on the transaction types, times, amounts, and liquidity of multiple users' transaction objects, a transaction frequency index is calculated based on the transaction time of each object. Furthermore, by collecting data on the transaction types, frequency indices, and total transaction amount percentages of multiple transaction objects from different users, an asset risk prediction model is trained. This model predicts the liquidity risk of different transaction events under the weighted influence of transaction frequency and amount, enabling accurate monitoring of secure transactions. When a user initiates an asset expenditure transaction, the model predicts whether each historical transaction object has liquidity risk leading to bond issues, based on historical transaction objects, their transaction frequency indices, and total transaction amount percentages. This allows for early screening and warning of the financial security risks of historical transaction objects. Moreover, when a transaction object has liquidity risk, the model monitors and locates the object's real-time transactions with any user, determining whether the transaction is abnormal based on real-time transaction access characteristics. If an abnormal transaction occurs, the object is prohibited from continuing to trade. This invention enables the prediction of liquidity risk for users' historical transactions and real-time monitoring of specific transaction objects with financial risks, improving the accuracy of risk warnings, effectively preventing the expansion of transaction risks, and ensuring the asset security of all users.

[0050] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. An asset transaction risk monitoring system, characterized in that, include: Transaction data acquisition module: Collects transaction data such as transaction type, transaction time, transaction amount, and cash flow from multiple transaction objects of the user; Calculate the transaction frequency index based on the transaction time of different trading objects; Asset risk prediction module: Collects multiple sets of asset risk training data through historical transaction data of multiple users; each set of asset risk training data includes the transaction object type, transaction frequency index and total transaction amount ratio of multiple historical transaction objects of each user, as well as the risk mark of whether each transaction object has liquidity risk, and trains to obtain an asset risk prediction model; Transaction Anomaly Detection Module: Collects multiple historical transaction objects of the current user, the transaction frequency index of each historical transaction object, and the proportion of the total transaction amount; predicts whether each transaction object has liquidity risk through the asset risk prediction model; when the transaction object has liquidity risk, determines whether the transaction is abnormal based on the real-time transaction access characteristics between the current user and the transaction object; when the transaction is abnormal, prohibits the current user from trading with the transaction object.

2. A method for monitoring asset transaction risks, characterized in that, include: Collect transaction type, transaction time, transaction amount, and cash flow from multiple transaction objects of the user; Calculate the transaction frequency index based on the transaction time of different trading objects; Multiple sets of asset risk training data are collected by collecting historical transaction data from multiple users; each set of asset risk training data includes the transaction object type, transaction frequency index and transaction amount percentage of multiple historical transaction objects of each user, as well as whether the user has liquidity risk, and the asset risk prediction model is obtained by training. Collect multiple historical transaction objects of the current user, the transaction frequency index of each historical transaction object, and the proportion of the total transaction amount; predict whether each transaction object has liquidity risk through the asset risk prediction model; when the transaction object has liquidity risk, determine whether the transaction is abnormal based on the real-time transaction access characteristics between the current user and the transaction object; when the transaction is abnormal, prohibit the current user from trading with the transaction object.

3. The asset transaction risk monitoring method according to claim 2, characterized in that, The types of transaction objects include individuals or institutions, and credit scores; the transaction times include weekdays or holidays, and time periods.

4. The asset transaction risk monitoring method according to claim 2, characterized in that, When collecting multiple sets of asset risk training data through historical transaction data of multiple users, whether each transaction object has liquidity risk is determined by the following steps: the multiple sets of asset risk training data are grouped according to different months; a minimum asset threshold is calculated based on the average daily expenditure amount of each transaction object and the corresponding user in the previous month and the number of days remaining in the current month; it is determined whether the funds of each transaction object are lower than the minimum asset threshold, and if and only if they are lower, it indicates that the transaction object has liquidity risk.

5. The asset transaction risk monitoring method according to claim 2, characterized in that, The step of determining whether a transaction is abnormal based on the real-time transaction access characteristics of the current user and the transaction object specifically includes the following steps: collecting multiple sets of normal transaction feature samples and multiple sets of abnormal transaction access feature samples for training to obtain a transaction abnormality training model; inputting the real-time transaction access characteristics of the transaction object for the current user into the transaction abnormality training model to obtain the result of whether the transaction is abnormal. Both the normal transaction feature sample and the abnormal transaction access feature sample include any one or more of the following: login time of the transaction object, transaction amount, IP address, device type, and session duration.

6. The asset transaction risk monitoring method according to claim 2, characterized in that, The step of calculating the transaction frequency index based on the transaction time of different trading objects specifically includes the following steps: The trading frequency of different trading objects is calculated based on their trading times; a time decay factor is calculated based on the most recent trading time of each trading object; and a trading frequency index is calculated by introducing the time decay factor and the trading frequency. The percentage of the total transaction amount is calculated through the following steps: the total transaction amount is calculated based on the transaction amount of different trading objects, and the percentage of the total transaction amount of each trading object is calculated based on the total transaction amount of all trading objects.

7. The asset transaction risk monitoring method according to claim 2, characterized in that, After prohibiting the current user from transacting with the transaction object when a transaction is abnormal, the method further includes the following steps: collecting multiple sets of successful transaction samples and failed transaction samples to train a transaction risk prediction model; the successful transaction samples include transaction object type, transaction time, transaction amount, working capital amount, and abnormal access characteristics; the failed transaction samples include collecting multiple sets of failed transaction object type, transaction time, transaction amount, working capital amount, and abnormal access characteristics.

8. The asset transaction risk monitoring method according to claim 7, characterized in that, Also includes: Use a transaction risk prediction model to determine whether the transaction of the target transaction object is successful; store the transaction object IDs and transaction history of each user's successful transactions; Before predicting whether each of the transaction objects has liquidity risk using the asset risk prediction model, the method further includes: Each user's transaction object is authenticated using the stored transaction object ID.

9. An electronic device comprising a memory, a processor, and a computer program running on the processor, characterized in that: When the processor executes the computer program, it implements the steps of the asset transaction risk monitoring method as described in any one of claims 2 to 8.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by the processor, it implements the steps of the asset transaction risk monitoring method as described in any one of claims 2 to 8.

Images