A method and device for predicting illegal behavior
By obtaining and analyzing user's fund transaction and behavioral characteristics data, monitoring the flow of cash information, and using the violation prediction model, the problem of violation prediction during the cash transfer process is solved, and automatic monitoring and accurate prediction of the cash transfer process is achieved. , ensure the security of funds.
Patent Information
- Application Number
- CN202110482255.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2021-04-30
- Publication Date
- 2024-11-19
- Estimated Expiration
- 2041-04-30
AI Technical Summary
During the cash transfer process, it is difficult for the existing technology to effectively predict and monitor violations, resulting in possible capital loss and illegal operations.
By obtaining user's fund transaction data and behavioral characteristics data, analyzing the risks of cash withdrawal operations, monitoring the flow of cash information, and using the violation prediction model to predict whether there are any violations in the cash transfer process from one user to another.
Automatic violation prediction of the cash transfer process is realized, the prediction accuracy is improved, unnecessary prediction times are reduced, and funds are ensured.
Smart Images

Figure CN113052683B_ABST
Abstract
Description
Technical Field
[0001] The present application belongs to the field of data processing technology, and in particular, relates to a method and device for predicting illegal behavior. Background Art
[0002] Currently, corporate users can obtain preferential loans from banks, and banks can issue certain cash to corporate users' accounts to assist corporate development. Cash can be transferred from corporate users' capital accounts to other users' capital accounts, such as to corporate employees' accounts or accounts with relationships with corporate users. However, there may be violations during the cash transfer process, so it is necessary to predict whether there are any violations. Summary of the invention
[0003] The present application provides a method and device for predicting illegal behavior.
[0004] On the one hand, the present application provides a method for predicting illegal behavior, the method comprising:
[0005] In response to a cash withdrawal instruction, first fund transaction data of a first user and first behavior characteristic data of the first user are acquired, wherein the first fund transaction data is used to indicate the loan status and withdrawal status of the first user, and the first behavior characteristic data is used to indicate the behavior of the first user other than the use of funds;
[0006] Based on the first fund transaction data and the first behavior characteristic data, determining whether a cash withdrawal operation corresponding to the cash withdrawal instruction has a risk;
[0007] In response to the instruction that the cash withdrawal operation is risky, monitoring the cash information extracted by the cash withdrawal operation;
[0008] If the cash information is monitored again, second fund transaction data of a second user and second behavior characteristic data of the second user are obtained, and the second user is the owner of the account into which the cash information flows;
[0009] Based on the first funds transaction data, the first behavior characteristic data, the second funds transaction data and the second behavior characteristic data, the possibility of the cash information being transferred from the first user to the second user is predicted, and the possibility is used to indicate whether there is any violation in the cash transfer process corresponding to the cash information.
[0010] Optionally, the determining, based on the first fund transaction data and the first behavior characteristic data, whether there is a risk in the cash withdrawal operation corresponding to the cash withdrawal instruction includes:
[0011] Determine, from the historical data of multiple historical users stored in the historical data set, target historical data that matches the first fund transaction data and the first behavior characteristic data;
[0012] Obtaining the historical withdrawal risk probability of the target historical data;
[0013] Based on the target historical data and the historical withdrawal risk probability, a withdrawal risk probability of the first user when performing the cash withdrawal operation is predicted, and the withdrawal risk probability is used to indicate whether the cash withdrawal operation has a risk.
[0014] Optionally, the predicting, based on the target historical data and the historical withdrawal risk probability, the withdrawal risk probability of the first user when performing the cash withdrawal operation includes:
[0015] based on Predict the withdrawal risk probability P(Yi|Z) of the first user during the cash withdrawal operation, where Aj is the jth target historical data, n is the total number of target historical data, and P(Aj|Yi) is the conditional probability of Aj after the occurrence of Yi. is the prior probability of Aj, Y i Is there a risk or no risk?
[0016] Among them, the comparison between the withdrawal risk probability and the threshold value can determine whether the cash withdrawal operation is risky.
[0017] Optionally, the predicting, based on the first fund transaction data, the first behavior characteristic data, the second fund transaction data, and the second behavior characteristic data, the possibility of the cash information being transferred from the first user to the second user includes:
[0018] calling a violation prediction model, processing the first fund transaction data, the first behavior characteristic data, the second fund transaction data, and the second behavior characteristic data, and obtaining a violation probability output by the violation prediction model, wherein the violation probability is used as the possibility to indicate whether there is a violation;
[0019] Among them, the violation prediction model is trained using the historical first fund transaction data, the historical first behavior feature data, the historical second fund transaction data and the historical second behavior feature data, so that the violation prediction model can predict the probability of violation when transferring from one user to another.
[0020] Optionally, the method further includes:
[0021] In response to the fund monitoring instruction, monitoring the loan usage data pointed to by the fund monitoring instruction, wherein the loan usage data is used to indicate the transfer of the first user's loan between different accounts;
[0022] If the loan usage data indicates that there is abnormal use of funds, a prompt message is output.
[0023] Optionally, the method further includes:
[0024] In response to the violation instruction, a data upload instruction is sent to the first user;
[0025] receiving data corresponding to the data upload instruction, wherein the data is used to indicate the usage of the cash withdrawn by the first user;
[0026] The data corresponding to the data upload instruction is analyzed to determine the direction of the violation.
[0027] On the other hand, the present application provides a violation behavior prediction device, the device comprising:
[0028] a first acquisition unit, configured to respond to a cash withdrawal instruction and acquire first fund transaction data of a first user and first behavior characteristic data of the first user, wherein the first fund transaction data is used to indicate a loan status and a withdrawal status of the first user, and the first behavior characteristic data is used to indicate a behavior of the first user other than fund use;
[0029] a determination unit, configured to determine whether there is a risk in the cash withdrawal operation corresponding to the cash withdrawal instruction based on the first fund transaction data and the first behavior characteristic data;
[0030] A monitoring unit, configured to respond to an instruction indicating that the cash withdrawal operation is risky and monitor information of cash withdrawn by the cash withdrawal operation;
[0031] A second acquisition unit is used to obtain second fund transaction data of a second user and second behavior characteristic data of the second user if the cash information is monitored again, the second user being the owner of the account into which the cash information flows;
[0032] A prediction unit is used to predict the possibility of transferring the cash information from the first user to the second user based on the first fund transaction data, the first behavior characteristic data, the second fund transaction data and the second behavior characteristic data, wherein the possibility is used to indicate whether there is any violation in the cash transfer process corresponding to the cash information.
[0033] Optionally, the determination unit is used to determine target historical data that matches the first fund transaction data and the first behavioral characteristic data from the historical data of multiple historical users stored in the historical data set; obtain the historical withdrawal risk probability of the target historical data; and predict the withdrawal risk probability of the first user when performing the cash withdrawal operation based on the target historical data and the historical withdrawal risk probability, wherein the withdrawal risk probability is used to indicate whether the cash withdrawal operation has a risk.
[0034] Optionally, the determining unit is used based on Predict the withdrawal risk probability P(Yi|Z) of the first user during the cash withdrawal operation, where Aj is the jth target historical data, n is the total number of target historical data, and P(Aj|Yi) is the conditional probability of Aj after the occurrence of Yi. Yes A j The prior probability, Y i Is there a risk or no risk?
[0035] Among them, the comparison between the withdrawal risk probability and the threshold value can determine whether the cash withdrawal operation is risky.
[0036] In another aspect, the present application provides a device, comprising:
[0037] processor;
[0038] a memory for storing instructions executable by the processor;
[0039] The processor is configured to execute the instructions to implement the above-mentioned violation prediction method.
[0040] On the other hand, the present application provides a storage medium having computer program code stored thereon, and the computer program code implements the above-mentioned violation prediction method when executed.
[0041] The above-mentioned illegal behavior prediction method and device, wherein the method includes: responding to a cash withdrawal instruction, obtaining the first fund transaction data of the first user and the first behavior characteristic data of the first user, the first fund transaction data is used to indicate the loan situation and withdrawal situation of the first user, and the first behavior characteristic data is used to indicate the behavior of the first user other than the use of funds; based on the first fund transaction data and the first behavior characteristic data, determining whether the cash withdrawal operation corresponding to the cash withdrawal instruction has risks; responding to the instruction that the cash withdrawal operation has risks, monitoring the cash information extracted by the cash withdrawal operation; if the cash information is monitored again, obtaining the second fund transaction data of the second user and the second behavior characteristic data of the second user, the second user is the owner of the account into which the cash information flows; based on the first fund transaction data, the first behavior characteristic data, the second fund transaction data, and the second behavior characteristic data, predicting the possibility of the cash information being transferred from the first user to the second user, the possibility being used to indicate whether there is an illegal behavior in the cash transfer process corresponding to the cash information, thereby realizing automatic prediction of illegal behavior in cash transfer. When it is determined that there is a risk in the cash withdrawal operation, the cash information is triggered to be monitored, and illegal behaviors are predicted through the cash information. In this way, the prediction of illegal behaviors can be omitted when there is no risk in the cash withdrawal operation, and the cash withdrawal operation can be screened based on whether there is a risk, so that illegal behaviors can be predicted for cash withdrawal operations with risks, thereby improving the prediction accuracy and reducing the number of predictions. BRIEF DESCRIPTION OF THE DRAWINGS
[0042] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.
[0043] Figure 1 is a flow chart of a method for predicting illegal behavior provided in an embodiment of the present application;
[0044] Figure 2 is a flow chart of another method for predicting illegal behavior provided in an embodiment of the present application;
[0045] Figure 3 is a flow chart of another violation prediction method provided in an embodiment of the present application;
[0046] Figure 4 It is a structural diagram of a violation prediction device provided in an embodiment of the present application. DETAILED DESCRIPTION
[0047] In order to make the purpose, technical solution and advantages of the embodiments of the present application clearer, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.
[0048] See also Figure 1 , which shows an optional process of a violation behavior prediction method provided by an embodiment of the present application, which may include:
[0049] 101: In response to a cash withdrawal instruction, first fund transaction data of a first user and first behavior characteristic data of the first user are obtained, where the first fund transaction data is used to indicate the loan status and withdrawal status of the first user, and the first behavior characteristic data is used to indicate the behavior of the first user other than fund use.
[0050] Among them, the cash withdrawal instruction is used to indicate the existence of a cash withdrawal operation. The cash withdrawal operation is that the first user withdraws cash from the bank terminal. The first user may be an enterprise or individual who applies for a preferential loan from the bank. For these users, if the bank terminal monitors their cash withdrawal, it triggers the cash withdrawal instruction. After full authorization by the first user, the relevant data of the first user can be obtained by responding to the cash withdrawal instruction.
[0051] The relevant data of the first user include: first fund transaction data and first behavior characteristic data. The first fund transaction data is related to funds, and the first fund transaction data is used to indicate the behavior of the first user in fund use. For example, the first fund transaction data is used to indicate loan status and withdrawal status. For example, the first fund transaction data includes: loan amount and withdrawal amount, and the loan amount is used to indicate the loan status, and the withdrawal amount is used to indicate the withdrawal status. The first behavior characteristic data is behavior data in other aspects except fund use, so as to indicate the behavior of the first user in other aspects except fund use. For example, the first behavior characteristic data can indicate the basic information of the first user, the online behavior of the first user, etc. The first behavior characteristic data may include: the customer level of the first user, the occupation of the first user, and the web browsing data of the first user, etc. The customer level of the first user indicates the customer type of the first user at the bank, such as one of a senior customer, an ordinary customer, etc.
[0052] The first fund transaction data and the first behavior characteristic data can be collected when responding to a cash withdrawal instruction. For example, the withdrawal amount is collected using a bank terminal when responding to a cash withdrawal instruction. Other data may have been collected before responding to the cash withdrawal instruction and can be obtained when responding to the cash withdrawal instruction. This embodiment does not limit the timing of obtaining the first fund transaction data and the first behavior characteristic data. The first behavior characteristic data may be collected by a terminal other than a bank terminal and sent to the bank terminal for use.
[0053] 102: Based on the first fund transaction data and the first behavior characteristic data, determine whether there is a risk in the cash withdrawal operation corresponding to the cash withdrawal instruction.
[0054] In this embodiment, the existence of risk in the cash withdrawal operation indicates that the first user's withdrawal of cash does not meet the cash withdrawal requirements, such as the cash withdrawal requirements can indicate the maximum amount of cash that the first user can withdraw and whether the first user can withdraw cash. If the withdrawal amount in the first fund transaction data is greater than the maximum amount, and / or the first behavior characteristic data indicates that the first user cannot withdraw cash, it is determined that there is a risk; if the withdrawal amount in the first fund transaction data is not greater than the maximum amount and the first behavior characteristic data indicates that the first user can withdraw cash, it is determined that there is no risk. The maximum amount can be the total amount of the loan, and the use of the cash withdrawal requirements to determine whether there is a risk and the description of the cash withdrawal requirements are only an example. This embodiment can also use other methods to determine whether there is a risk. One other method is as follows:
[0055] Determine target historical data that matches the first fund transaction data and the first behavior characteristic data from historical data sets stored in a historical data set; obtain a historical withdrawal risk probability of the target historical data; and predict the withdrawal risk probability of the first user when performing a cash withdrawal operation based on the target historical data and the historical withdrawal risk probability, wherein the withdrawal risk probability is used to indicate whether the cash withdrawal operation has a risk.
[0056] Let x be the historical data set, x={A 1 ,A 2 ...A m}, A i is the i-th historical data in the historical data set, so as to represent the characteristics of the withdrawal business through the historical data set. iis the characteristic attribute of x, which at least represents the withdrawal attribute of a historical user. The historical data set can store the historical data of different historical users. One historical data includes historical fund transaction data and historical behavior characteristic data. The historical fund data is related to the loan amount and the withdrawal amount. For example, the historical fund data includes the difference between the loan amount and the withdrawal amount and the historical withdrawal amount. The historical behavior characteristic data includes: the customer level of the historical user, the web browsing data of the historical user, and the occupation of the historical user, so as to indicate the characteristics of the corresponding historical user in fund transaction and web browsing behaviors through one historical data.
[0057] Correspondingly, a category set C = (Y 1 ,Y 2 ), Y 1 Indicates that there is a risk of withdrawal, Y 2 It means that there is no withdrawal risk. For historical data, whether there is a withdrawal risk can be expressed by the historical withdrawal risk probability. If the historical withdrawal risk probability is greater than the threshold, it is determined that there is a withdrawal risk; if the historical withdrawal risk probability is less than or equal to the threshold, it is determined that there is no withdrawal risk. The historical withdrawal risk probability corresponding to each historical data can be determined based on the withdrawal data of historical users (that is, the historical withdrawal risk probability corresponding to the historical data is known).
[0058] In this embodiment, the target historical data can be determined based on historical fund transaction data, historical behavior characteristic data, first fund transaction data and first behavior characteristic data. If the historical fund transaction data matches the first fund transaction data and the historical behavior characteristic data matches the first behavior characteristic data, it is determined to be the target historical data that matches the first user, so as to predict the withdrawal risk probability of the first user through the historical withdrawal risk probability of the matched target historical data.
[0059] One way to predict the first user's withdrawal risk probability by matching the historical withdrawal risk probability of the target historical data is as follows:
[0060] based on Predict the withdrawal risk probability P(Yi|Z) of the first user during the cash withdrawal operation, where Aj is the jth target historical data, n is the total number of target historical data, and P(Aj|Yi) is the conditional probability of Aj after Yi occurs. is the prior probability of Aj, and Yi is the existence or non-existence of risk. Among them, the comparison between the withdrawal risk probability and the threshold value can determine whether there is a risk in the cash withdrawal operation.
[0061] That is to say, after obtaining the target historical data, the conditional probability estimate of at least one category under the above two categories is statistically analyzed, such as: according to P(A 1 |Y 1 )P(A 2 |Y1 )…P(A m |Y 1 ), P(A 1 |Y 2 )P(A 2 |Y 2 )…P(A m |Y 2 ) obtains the probability of the withdrawal attribute pointed to by the target historical data with and without risk.
[0062] According to Bayes' theorem, we can deduce as follows:
[0063] P(Y i |x)=P(x|Y i )P(Y i ) / P(x), because the denominator P(x|Y i )P(Y i ) is a constant for all categories, so we can find the numerator. By normalizing the numerator, we can get the corresponding probability:
[0064] P(x|Y i )*P(Y i )=P(A 1 |Y i )*P(A 2 |Y i )*…*P(A m |Y i )*P(Y i ).
[0065] The corresponding P(Y 1 |Z)=P(Z|Y 1 )P(Y 1 )=P(A 1 |Y 1 )*P(A 2 |Y 1 )*…*P(A m |Y 1 )*P(Y 1 );
[0066] P(Y 2 |Z)=P(Z|Y 2 )P(Y 2 )=P(A 1 |Y 2 )*P(A 2 |Y 2 )*…*P(A m |Y 2 )*P
[0067] (Y 2 );
[0068] If in the historical data set, the probability that the difference between the loan amount and the withdrawal amount is zero exceeds 80% under risk conditions, the threshold can be set to 80%. If the predicted P(Y 1 | Z) is greater than 80%, it is determined that there is a risk; this embodiment can also use P (Y 2 |Z) Determine whether there is a risk, which will not be discussed here.
[0069] 103: In response to the instruction that there is a risk in the cash withdrawal operation, the cash information extracted by the cash withdrawal operation is monitored. The cash withdrawal operation represents that the first user withdraws cash. The cash is provided with a serial number so that the serial number uniquely points to a banknote. Therefore, the cash information in this embodiment may be the serial number of the banknote extracted by the cash withdrawal operation, and the serial number is used to monitor the circulation of the cash.
[0070] 104: If the cash information is monitored again, the second fund transaction data of the second user and the second behavior characteristic data of the second user are obtained, and the second user is the owner of the account into which the cash information flows.
[0071] If cash information is monitored again, it is determined that the banknotes are transferred from the first user to the bank, for example, to the second user's account. In this case, the second fund transaction data of the second user and the second behavior characteristic data of the second user are obtained. For the description of the second fund transaction data and the second behavior characteristic data, please refer to the above-mentioned first fund transaction data and first behavior characteristic data.
[0072] 105: Based on the first fund transaction data, the first behavior characteristic data, the second fund transaction data and the second behavior characteristic data, predict the possibility of the cash information being transferred from the first user to the second user, where the possibility is used to indicate whether there is any violation in the cash transfer process corresponding to the cash information.
[0073] It can be seen from the circulation of cash between different users that there are certain rules for the circulation of cash. For example, student loans will be issued to personal accounts through school accounts, and then returned from personal accounts to school accounts at intervals of a certain period of time (such as three to five years); small and micro business loans will be returned to the bank through multiple affiliated companies of the lender. Therefore, this embodiment can use the cash circulation between multiple historical users to obtain the circulation rules between different types of users, such as the circulation direction of cash between different types of users (such as from which type of user to which type of user). If the circulation rules obtained based on the first capital transaction data, the first behavior characteristic data, the second capital transaction data, and the second behavior characteristic data are different from the circulation rules summarized above, it is determined that there is a violation, otherwise it is determined that there is no violation.
[0074] The circulation rule obtained based on the first fund transaction data, the first behavior characteristic data, the second fund transaction data and the second behavior characteristic data can obtain the types of two users based on the first behavior characteristic data and the second behavior characteristic data, and then obtain the circulation amount and circulation time, etc. based on the first fund transaction data and the second fund transaction data. The corresponding first fund transaction data and the second fund transaction data include the corresponding transaction time, such as the first fund transaction data includes the loan time, and the second fund transaction data includes the fund inflow time; the type, circulation amount and circulation time of the two users are used to obtain the circulation direction of this cash circulation, such as transferring a certain amount from the first user to the second user after a certain period of time.
[0075] In this embodiment, the prediction of violation behavior can also be obtained through a violation prediction model. For example, the violation prediction model is called to process the first fund transaction data, the first behavior characteristic data, the second fund transaction data and the second behavior characteristic data to obtain the violation probability output by the violation prediction model. The violation probability is used as a possibility to indicate whether there is a violation.
[0076] Among them, the violation prediction model is trained using the historical first capital transaction data, the historical first behavior feature data, the historical second capital transaction data and the historical second behavior feature data, so that the violation prediction model can predict the probability of violation when transferring from one user to another.
[0077] It can be understood that the violation prediction model can predict whether there is a violation in the transfer from one user to another. The violation prediction model can use the historical first fund transaction data, the historical first behavior feature data, the historical second fund transaction data and the historical second behavior feature data to train the violation prediction model to predict the circulation pattern, and can predict whether there is a violation after obtaining the circulation pattern. For example, the historical data of cash circulation under different loans such as student loans and small and micro business loans can be used for training, so that the violation prediction model has the ability to predict the circulation pattern and whether there is a violation in the circulation pattern.
[0078] During the training process, the violation prediction model can use the paired historical first fund transaction data, historical first behavior feature data, historical second fund transaction data, and historical second behavior feature data to adjust the model parameters of the violation prediction model. The paired historical first fund transaction data, historical first behavior feature data, historical second fund transaction data, and historical second behavior feature data are data between two users with a circulation pattern. These two users may be users with or without violations, so that the model parameters can be adjusted and verified through the historical data of different types of users.
[0079] The violation prediction model can be a BP neural network model. A three-layer BP neural network model is sufficient to complete any mapping from n-dimension to m-dimension. Therefore, the BP neural network model in this embodiment can be a three-layer neural network model. The BP neural network model includes an input layer, a hidden layer and an output layer. In the training and testing stages, the input layer can input historical data, such as the paired historical first fund transaction data, the historical first behavior feature data, the historical second fund transaction data and the historical second behavior feature data. The output layer is used to output the predicted violation probability. The model parameters of the hidden layer are adjusted by the violation probability output by the output layer and the loss value of the historical violation probability of the historical data. During the adjustment process, the number of nodes in the hidden layer can be determined by trial and error, thereby determining the model structure of the BP neural network model.
[0080] The above-mentioned violation prediction method responds to a cash withdrawal instruction, obtains first fund transaction data of a first user and first behavior characteristic data of the first user, the first fund transaction data is used to indicate the loan situation and withdrawal situation of the first user, and the first behavior characteristic data is used to indicate the behavior of the first user other than the use of funds; based on the first fund transaction data and the first behavior characteristic data, determines whether the cash withdrawal operation corresponding to the cash withdrawal instruction has risks; in response to the instruction that the cash withdrawal operation has risks, monitors the cash information extracted by the cash withdrawal operation; if the cash information is monitored again, obtains second fund transaction data of a second user and second behavior characteristic data of the second user, the second user is the owner of the account into which the cash information flows; based on the first fund transaction data, the first behavior characteristic data, the second fund transaction data, and the second behavior characteristic data, predicts the possibility of the cash information being transferred from the first user to the second user, the possibility is used to indicate whether there is a violation in the cash transfer process corresponding to the cash information, thereby realizing automatic prediction of violations in cash transfer. When it is determined that there is a risk in the cash withdrawal operation, the cash information is triggered to be monitored, and illegal behaviors are predicted through the cash information. In this way, the prediction of illegal behaviors can be omitted when there is no risk in the cash withdrawal operation, and the cash withdrawal operation can be screened based on whether there is a risk, so that illegal behaviors can be predicted for cash withdrawal operations with risks, thereby improving the prediction accuracy and reducing the number of predictions.
[0081] See also Figure 2 , which shows an optional process of another violation behavior prediction method provided by an embodiment of the present application, which may include the following steps:
[0082] 201: Responding to a fund monitoring instruction, monitoring the loan usage data pointed to by the fund monitoring instruction, where the loan usage data is used to indicate the transfer of the first user's loan between different accounts.
[0083] With the authorization of the first user, after the first user's preferential loan from the bank is approved, the fund monitoring instruction is triggered and responded to, so as to use the fund monitoring instruction to monitor the loan usage data of the first user, such as the loan usage data including the loan amount, loan time, transfer data, etc. The transfer data can indicate the user to which the funds flow from the first user's account.
[0084] 202: If the loan usage data indicates that there is abnormal use of funds, a prompt message is output.
[0085] When applying for a preferential loan, the first user can record the purpose of the loan of the first user, or determine the purpose of the loan by the type of the first user. If the loan usage data indicates that the loan is not used in accordance with the loan purpose, it is determined that the use of funds is abnormal. For example, the first user applies for a preferential loan for company operations, and the loan purpose can be raw material purchase, salary payment, factory construction, etc.; if the loan usage data is consistent with these purposes, it is determined that the use of funds is normal; if the loan usage data is different from these purposes, it is determined that the use of funds is abnormal. For example, if it is found that the operating loan is transferred to a real estate account or a stock account, etc., it is determined that the use of funds is abnormal, so that the loan can be monitored by monitoring the fund usage data, such as the flow direction can be obtained through the fund usage data.
[0086] When funds are used abnormally, a prompt message is output, such as prompting that the preferential loan applied for by the first user is used abnormally, thereby prompting the bank to monitor the funds of the first user. If there is an abnormality, certain measures can be taken, such as requiring the first user to return the loan or prohibiting the first user from applying for a preferential loan again.
[0087] 203: In response to the cash withdrawal instruction, first fund transaction data of the first user and first behavior characteristic data of the first user are obtained, where the first fund transaction data is used to indicate the loan status and withdrawal status of the first user, and the first behavior characteristic data is used to indicate the behavior of the first user other than the use of funds.
[0088] 204: Based on the first fund transaction data and the first behavior characteristic data, determine whether the cash withdrawal operation corresponding to the cash withdrawal instruction has risks.
[0089] 205: In response to the instruction that the cash withdrawal operation has risks, the cash information extracted by the cash withdrawal operation is monitored.
[0090] 206: If the cash information is monitored again, the second fund transaction data of the second user and the second behavior characteristic data of the second user are obtained, and the second user is the owner of the account into which the cash information flows.
[0091] 207: Based on the first fund transaction data, the first behavior characteristic data, the second fund transaction data and the second behavior characteristic data, predict the possibility of the cash information being transferred from the first user to the second user, where the possibility is used to indicate whether there is any violation in the cash transfer process corresponding to the cash information.
[0092] The above-mentioned violation prediction method responds to the fund monitoring instruction and monitors the loan usage data pointed to by the fund monitoring instruction. The loan usage data is used to indicate the transfer of the first user's loan between different accounts. If the loan usage data indicates that there is an abnormal use of funds, a prompt message is output, thereby enabling the monitoring of the first user's loan.
[0093] See also Figure 3 , which shows an optional process of another violation prediction method provided in an embodiment of the present application. Figure 1 The following steps can also be included:
[0094] 106: In response to the violation instruction, send a data upload instruction to the first user. If the predicted possibility indicates that there is a violation, trigger and respond to the violation instruction, so as to use the violation instruction to instruct the first user to upload the loan usage information to the bank terminal. If in response to the violation instruction, send a data upload instruction to the first user, the data upload instruction is used to instruct the first user to upload information, such as uploading loan usage information, and the uploaded information may be an invoice, etc. The data upload instruction may carry information type, information requirements, etc., so that the first user can upload information that meets the requirements according to the data upload instruction to prevent receiving some irrelevant information.
[0095] The data upload instruction can be sent to the user terminal corresponding to the mobile phone number reserved by the first user, so that the first user can receive the data upload instruction in time. In addition, the violation instruction can also be triggered when the loan usage data indicates that there is abnormal use of funds, which will not be elaborated in this embodiment.
[0096] 107: Receive data corresponding to the data upload instruction, the data is used to point to the usage of the cash taken by the first user, such as the information of the second user to which the cash flows, so as to determine the usage of the cash through the information of the second user. For example, taking the above-mentioned company operating loan as an example, the loan is withdrawn during the company operating loan process, and the second user's information can be used to determine that it is transferred to an account of a real estate company, thereby determining that the usage of the cash is transferred to the real estate field.
[0097] 108: Analyze the data corresponding to the data upload instruction to determine the direction of violation. The direction of violation can indicate the existence of illegal fund usage data. For example, taking the above-mentioned company operating loan as an example, the loan is withdrawn during the company operating loan process. Through the information of the second user, it can be determined that it is transferred to an account of a real estate company, thereby determining that the use of the banknotes is transferred to the real estate field, so as to focus on monitoring the first user and users similar to the first user through the direction of violation.
[0098] For the aforementioned method embodiments, for the sake of simplicity, they are all described as a series of action combinations, but those skilled in the art should be aware that the present application is not limited by the order of the actions described, because according to the present application, some steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the present application.
[0099] Corresponding to the above method embodiment, the present application embodiment also provides a violation behavior prediction device, whose optional structure is as follows: Figure 4 As shown, it may include: a first acquisition unit 10 , a determination unit 20 , a monitoring unit 30 , a second acquisition unit 40 and a prediction unit 50 .
[0100] The first acquisition unit 10 is used to respond to a cash withdrawal instruction to acquire first fund transaction data of a first user and first behavior characteristic data of the first user, wherein the first fund transaction data is used to indicate the loan status and withdrawal status of the first user, and the first behavior characteristic data is used to indicate the behavior of the first user other than the use of funds.
[0101] Among them, the cash withdrawal instruction is used to indicate the existence of a cash withdrawal operation. The cash withdrawal operation is that the first user withdraws cash from the bank terminal. The first user may be an enterprise or individual who applies for a preferential loan from the bank. For these users, if the bank terminal monitors their cash withdrawal, it triggers the cash withdrawal instruction. After full authorization by the first user, the relevant data of the first user can be obtained by responding to the cash withdrawal instruction.
[0102] The relevant data of the first user includes: first fund transaction data and first behavior characteristic data. For the description of the first fund transaction data and the first behavior characteristic data, please refer to the above embodiment.
[0103] The determination unit 20 is used to determine whether there is a risk in the cash withdrawal operation corresponding to the cash withdrawal instruction based on the first fund transaction data and the first behavior characteristic data. In this embodiment, the existence of a risk in the cash withdrawal operation indicates that the first user's withdrawal of cash does not meet the cash withdrawal requirements, such as the cash withdrawal requirements can indicate the maximum amount of cash that the first user can withdraw and whether the first user can withdraw cash. If the withdrawal amount in the first fund transaction data is greater than the maximum amount, and / or the first behavior characteristic data indicates that the first user cannot withdraw cash, it is determined that there is a risk; if the withdrawal amount in the first fund transaction data is not greater than the maximum amount and the first behavior characteristic data indicates that the first user can withdraw cash, it is determined that there is no risk. The maximum amount can be the total amount of the loan, and the use of the cash withdrawal requirements to determine whether there is a risk and the description of the cash withdrawal requirements are only an example. This embodiment can also use other methods to determine whether there is a risk. One other method is as follows:
[0104] Determine target historical data that matches the first fund transaction data and the first behavior characteristic data from historical data sets stored in a historical data set; obtain a historical withdrawal risk probability of the target historical data; and predict the withdrawal risk probability of the first user when performing a cash withdrawal operation based on the target historical data and the historical withdrawal risk probability, wherein the withdrawal risk probability is used to indicate whether the cash withdrawal operation has a risk.
[0105] For example, based on Predict the withdrawal risk probability P(Yi|Z) of the first user during the cash withdrawal operation, where Aj is the jth target historical data, n is the total number of target historical data, and P(Aj|Yi) is the conditional probability of Aj after Yi occurs. is the prior probability of Aj, and Yi is the existence or non-existence of risk. Among them, the comparison between the withdrawal risk probability and the threshold value can determine whether there is a risk in the cash withdrawal operation.
[0106] The monitoring unit 30 is used to respond to the instruction that there is a risk in the cash withdrawal operation and monitor the cash information extracted by the cash withdrawal operation. The cash withdrawal operation represents that the first user withdraws cash. The cash is provided with a serial number so that the serial number uniquely points to a banknote. Therefore, the cash information in this embodiment can be the serial number of the banknote extracted by the cash withdrawal operation, and the serial number is used to monitor the circulation of the cash.
[0107] The second acquisition unit 40 is used to obtain second fund transaction data of a second user and second behavior characteristic data of the second user if the cash information is monitored again, where the second user is the owner of the account into which the cash information flows.
[0108] If cash information is monitored again, it is determined that the banknotes are transferred from the first user to the bank, for example, to the second user's account. In this case, the second fund transaction data of the second user and the second behavior characteristic data of the second user are obtained. For the description of the second fund transaction data and the second behavior characteristic data, please refer to the above-mentioned first fund transaction data and first behavior characteristic data.
[0109] The prediction unit 50 is used to predict the possibility of transferring cash information from a first user to a second user based on the first fund transaction data, the first behavior characteristic data, the second fund transaction data and the second behavior characteristic data, and the possibility is used to indicate whether there is any violation in the cash transfer process corresponding to the cash information.
[0110] It can be seen from the circulation of cash between different users that there are certain rules for the circulation of cash. For example, student loans will be issued to personal accounts through school accounts, and then returned from personal accounts to school accounts at intervals of a certain period of time (such as three to five years); small and micro business loans will be returned to the bank through multiple affiliated companies of the lender. Therefore, this embodiment can use the cash circulation between multiple historical users to obtain the circulation rules between different types of users, such as the circulation direction of cash between different types of users (such as from which type of user to which type of user). If the circulation rules obtained based on the first capital transaction data, the first behavior characteristic data, the second capital transaction data, and the second behavior characteristic data are different from the circulation rules summarized above, it is determined that there is a violation, otherwise it is determined that there is no violation.
[0111] The circulation rule obtained based on the first fund transaction data, the first behavior characteristic data, the second fund transaction data and the second behavior characteristic data can obtain the types of two users based on the first behavior characteristic data and the second behavior characteristic data, and then obtain the circulation amount and circulation time, etc. based on the first fund transaction data and the second fund transaction data. The corresponding first fund transaction data and the second fund transaction data include the corresponding transaction time, such as the first fund transaction data includes the loan time, and the second fund transaction data includes the fund inflow time; the type, circulation amount and circulation time of the two users are used to obtain the circulation direction of this cash circulation, such as transferring a certain amount from the first user to the second user after a certain period of time.
[0112] In this embodiment, the violation prediction can also be obtained through a violation prediction model. For example, the prediction unit 50 calls the violation prediction model, processes the first fund transaction data, the first behavior characteristic data, the second fund transaction data and the second behavior characteristic data, and obtains the violation probability output by the violation prediction model. The violation probability is used as a possibility to indicate whether there is a violation.
[0113] Among them, the violation prediction model is trained using the historical first fund transaction data, the historical first behavior feature data, the historical second fund transaction data and the historical second behavior feature data, so that the violation prediction model can predict the probability of violation when transferring from one user to another. For the description of the violation prediction model, please refer to the above embodiment.
[0114] The above-mentioned violation prediction device may also include: a monitoring unit and an output unit; the monitoring unit is used to respond to the fund monitoring instruction and monitor the loan usage data pointed to by the fund monitoring instruction, and the loan usage data is used to indicate the transfer of the first user's loan between different accounts. The output unit is used to output a prompt message if the loan usage data indicates that there is abnormal fund usage.
[0115] With the authorization of the first user, after the first user's preferential loan from the bank is approved, the fund monitoring instruction is triggered and responded to, so as to use the fund monitoring instruction to monitor the loan usage data of the first user, such as the loan usage data including the loan amount, loan time, transfer data, etc. The transfer data can indicate the user to which the funds flow from the first user's account.
[0116] The above-mentioned violation prediction device may also include: a sending unit, a receiving unit and an analyzing unit. The sending unit is used to respond to the violation instruction and send a data upload instruction to the first user; the receiving unit is used to receive data corresponding to the data upload instruction, and the data is used to point to the use of the cash withdrawn by the first user. The analyzing unit is used to analyze the data corresponding to the data upload instruction to determine the direction of the violation. The direction of the violation can indicate the existence of illegal fund usage data. For example, taking the above-mentioned company operating loan as an example, the loan is withdrawn during the company operating loan process. Through the information of the second user, it can be determined that it is transferred to the account of a real estate company, thereby determining that the use of the cash is transferred to the real estate field, so as to focus on monitoring the first user and users similar to the first user through the direction of the violation.
[0117] In addition, an embodiment of the present application further provides a device, comprising: a processor and a memory for storing instructions executable by the processor, wherein the processor is configured to execute the instructions to implement the above-mentioned violation behavior prediction method.
[0118] The embodiment of the present application also provides a storage medium having computer program code stored thereon, and the computer program code implements the above-mentioned violation behavior prediction method when executed.
[0119] It should be noted that the various embodiments in this specification can be described in a progressive manner, and the features recorded in the various embodiments in this specification can be replaced or combined with each other. Each embodiment focuses on the differences from other embodiments, and the same and similar parts between the various embodiments can be referred to each other. For the device embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.
[0120] Finally, it should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the presence of other identical elements in the process, method, article or device including the elements.
[0121] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.
[0122] The above is only a preferred implementation of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present application. These improvements and modifications should also be regarded as the scope of protection of the present application.
Claims
1. A method for predicting illegal behavior, characterized in that: The method comprises: In response to a cash withdrawal instruction, first fund transaction data of a first user and first behavior characteristic data of the first user are acquired, wherein the first fund transaction data is used to indicate the loan status and withdrawal status of the first user, and the first behavior characteristic data is used to indicate the behavior of the first user other than the use of funds; Based on the first fund transaction data and the first behavior characteristic data, determining whether a cash withdrawal operation corresponding to the cash withdrawal instruction has a risk; In response to the instruction that the cash withdrawal operation is risky, monitoring the cash information extracted by the cash withdrawal operation; If the cash information is monitored again, second fund transaction data of a second user and second behavior characteristic data of the second user are obtained, and the second user is the owner of the account into which the cash information flows; Based on the first fund transaction data, the first behavior characteristic data, the second fund transaction data and the second behavior characteristic data, predicting the possibility of the cash information being transferred from the first user to the second user, the possibility being used to indicate whether there is any violation in the cash transfer process corresponding to the cash information; Wherein, determining whether there is a risk in the cash withdrawal operation corresponding to the cash withdrawal instruction based on the first fund transaction data and the first behavior characteristic data includes: Determine, from the historical data of multiple historical users stored in the historical data set, target historical data that matches the first fund transaction data and the first behavior characteristic data; Obtaining the historical withdrawal risk probability of the target historical data; Based on the target historical data and the historical withdrawal risk probability, predicting the withdrawal risk probability of the first user when performing the cash withdrawal operation, the withdrawal risk probability being used to indicate whether the cash withdrawal operation has a risk; Wherein, predicting the possibility of the cash information being transferred from the first user to the second user based on the first fund transaction data, the first behavior characteristic data, the second fund transaction data, and the second behavior characteristic data includes: Calling a violation prediction model to process the first fund transaction data, the first behavior feature data, the second fund transaction data, and the second behavior feature data to obtain a violation probability output by the violation prediction model, wherein the violation probability is used as the possibility to indicate whether there is a violation; Among them, the violation prediction model is trained using historical first fund transaction data, historical first behavior feature data, historical second fund transaction data and historical second behavior feature data, so that the violation prediction model can predict the probability of violation when transferring from one user to another.
2. The method according to claim 1, characterized in that The predicting, based on the target historical data and the historical withdrawal risk probability, the withdrawal risk probability of the first user when performing the cash withdrawal operation comprises: based on Predict the first user's withdrawal risk probability P(Y i |Z), A j is the jth target historical data, n is the total number of target historical data, P(A j |Y i ) is the known Y i After the occurrence of A j The conditional probability of Yes A j The prior probability, Y i Is there a risk or no risk? Among them, the comparison between the withdrawal risk probability and the threshold value can determine whether the cash withdrawal operation is risky.
3. The method according to claim 1 or 2, characterized in that: The method further comprises: In response to the fund monitoring instruction, monitoring the loan usage data pointed to by the fund monitoring instruction, wherein the loan usage data is used to indicate the transfer of the first user's loan between different accounts; If the loan usage data indicates that there is abnormal use of funds, a prompt message is output.
4. The method according to claim 1 or 2, characterized in that: The method further comprises: In response to the violation instruction, a data upload instruction is sent to the first user; receiving data corresponding to the data upload instruction, wherein the data is used to indicate the usage of the cash withdrawn by the first user; The data corresponding to the data upload instruction is analyzed to determine the direction of the violation.
5. A device for predicting illegal behavior, characterized in that: The device comprises: a first acquisition unit, configured to respond to a cash withdrawal instruction and acquire first fund transaction data of a first user and first behavior characteristic data of the first user, wherein the first fund transaction data is used to indicate a loan status and a withdrawal status of the first user, and the first behavior characteristic data is used to indicate a behavior of the first user other than fund use; a determination unit, configured to determine whether there is a risk in the cash withdrawal operation corresponding to the cash withdrawal instruction based on the first fund transaction data and the first behavior characteristic data; A monitoring unit, configured to respond to an instruction indicating that the cash withdrawal operation is risky and monitor information of cash withdrawn by the cash withdrawal operation; A second acquisition unit is used to obtain second fund transaction data of a second user and second behavior characteristic data of the second user if the cash information is monitored again, the second user being the owner of the account into which the cash information flows; a prediction unit, configured to predict, based on the first fund transaction data, the first behavior characteristic data, the second fund transaction data, and the second behavior characteristic data, the possibility of the cash information being transferred from the first user to the second user, wherein the possibility is used to indicate whether there is any violation in the cash transfer process corresponding to the cash information; The determining unit is used to determine target historical data matching the first fund transaction data and the first behavior characteristic data from the historical data of multiple historical users stored in the historical data set; obtain the historical withdrawal risk probability of the target historical data; and predict the withdrawal risk probability of the first user when performing the cash withdrawal operation based on the target historical data and the historical withdrawal risk probability, wherein the withdrawal risk probability is used to indicate whether the cash withdrawal operation has a risk; The prediction unit is used to call the violation prediction model, process the first fund transaction data, the first behavior feature data, the second fund transaction data and the second behavior feature data, and obtain the violation probability output by the violation prediction model, and the violation probability is used as the possibility to indicate whether there is a violation; Among them, the violation prediction model is trained using historical first fund transaction data, historical first behavior feature data, historical second fund transaction data and historical second behavior feature data, so that the violation prediction model can predict the probability of violation when transferring from one user to another.
6. The device according to claim 5, characterized in that The determining unit is used to Predict the first user's withdrawal risk probability P(Yi|Z) during the cash withdrawal operation, A j is the jth target historical data, n is the total number of target historical data, P(A j |Y i ) is the known Y i After the occurrence of A j The conditional probability of Yes A j The prior probability, Y i Is there a risk or no risk? Among them, the comparison between the withdrawal risk probability and the threshold value can determine whether the cash withdrawal operation is risky.
7. A device, characterized in that: include: processor; a memory for storing instructions executable by the processor; The processor is configured to execute the instructions to implement the violation prediction method according to any one of claims 1 to 4.
