Resource transfer method and apparatus, computer device, and storage medium

Abstract

The embodiment of the application discloses a resource transfer method and device, computer equipment and a storage medium, and belongs to the technical field of computers. The application adopts different interval prediction methods to process historical data through multiple interval prediction models to predict the resource transfer situation of a first object in a target time period, and then fuses the predicted multiple numerical intervals. In this way, the historical data can be processed according to multiple attention angles, so that the multiple possible situations of resource transfer of the object in the future can be fully considered, the numerical interval obtained by fusion can accurately represent the future resource transfer situation of the object, the accuracy of the second numerical interval can be improved, and then the resource transfer request for the first object is processed based on the second numerical interval, so that the safety of the resource transfer for the first object can be ensured.

Description

Technical Field

[0001] This application relates to the field of computer technology, and in particular to a resource transfer method, apparatus, computer device, and storage medium. Background Technology

[0002] With the development of computer technology, online payment functions are becoming increasingly popular among users. For online payment platforms, merchants are the primary target of their services. Platforms typically predict the future resource transfer ranges of merchants to manage these transfers and ensure their security. However, the predicted resource transfer ranges obtained using this method are still not accurate enough. Summary of the Invention

[0003] This application provides a resource transfer method, apparatus, computer equipment, and storage medium, which can improve the accuracy of the numerical range of resource transfer. The technical solution is as follows:

[0004] On the one hand, a resource transfer method is provided, the method comprising:

[0005] Obtain historical data of a first object, wherein the historical data is used to describe the historical resource transfer behavior associated with the first object;

[0006] The historical data is processed by m interval prediction models according to their respective interval prediction methods to obtain m first numerical intervals. The first numerical interval is the resource transfer numerical interval corresponding to the resource transfer behavior of the first object within the target time period. The target time period is the time period after the current time. Different interval prediction models correspond to different interval prediction methods. m is an integer greater than 1.

[0007] The m first numerical intervals are merged to obtain the second numerical interval;

[0008] In response to a resource transfer request for the first object, the resource transfer request is processed based on the second numerical range.

[0009] On the other hand, a resource transfer device is provided, the device comprising:

[0010] The acquisition module is used to acquire historical data of a first object, wherein the historical data is used to describe the historical resource transfer behavior associated with the first object;

[0011] The processing module is used to process the historical data using m interval prediction models and according to the corresponding interval prediction methods to obtain m first numerical intervals. The first numerical interval is the resource transfer numerical interval corresponding to the resource transfer behavior of the first object within the target time period. The target time period is the time period after the current time. Different interval prediction models correspond to different interval prediction methods. m is an integer greater than 1.

[0012] A fusion module is used to fuse the m first numerical intervals to obtain a second numerical interval;

[0013] The processing module is further configured to process the resource transfer request based on the second numerical range in response to the resource transfer request for the first object.

[0014] In one possible implementation, the historical data includes data for n first statistical dimensions, where n is an integer greater than 1; the processing module is used to process the data for each first statistical dimension according to the corresponding interval prediction method using the interval prediction model to obtain a fifth numerical interval corresponding to each first statistical dimension; and to fuse the fifth numerical intervals corresponding to the n first statistical dimensions using the interval prediction model to obtain a first numerical interval corresponding to the interval prediction model.

[0015] In another possible implementation, the processing module is used to process the historical data for each interval prediction model according to the corresponding interval prediction method to obtain the first numerical intervals corresponding to p second statistical dimensions, where p is an integer greater than 1.

[0016] The fusion module is used to fuse the m interval prediction models into the first numerical intervals corresponding to the same second statistical dimension to obtain the second numerical intervals corresponding to p second statistical dimensions.

[0017] In another possible implementation, the m interval prediction models include a first prediction model;

[0018] The processing module is further configured to divide the historical data into q first data groups with a preset time period, each first data group including data within a historical period, where q is an integer greater than 1;

[0019] The processing module is configured to: extract features from each first data group using the first prediction model to obtain data features corresponding to each first data group; transform the data features corresponding to each first data group using the first prediction model to obtain a first value corresponding to each first data group, wherein the first value is a resource transfer value corresponding to the resource transfer behavior of the first object within the target time period; and determine a first value range corresponding to the first prediction model based on the first values ​​corresponding to the q first data groups using the first prediction model.

[0020] In another possible implementation, the processing module is used to sort the first values ​​corresponding to the q first data groups using the first prediction model, and determine the first value at the preset position and the minimum value among the first values ​​corresponding to the q first data groups as the maximum and minimum values ​​of the first value interval, respectively.

[0021] In another possible implementation, the m interval prediction models include a second prediction model; the historical data includes data corresponding to multiple historical resource transfer behaviors; the processing module is further configured to sort the data corresponding to the multiple historical resource transfer behaviors in order of their occurrence time from first to last or from last to first, to obtain sequence information;

[0022] The processing module is used to process the sequence information through the second prediction model to obtain a first numerical range corresponding to the second prediction model.

[0023] In another possible implementation, the m interval prediction models include a third prediction model; the processing module is further configured to divide the historical data into r second data groups based on the region to which the second object belongs, as indicated by the historical resource transfer behavior, wherein the resource transfer behavior has occurred between the second object and the first object, and the regions to which the objects indicated by the same second data group belong are the same, and r is an integer greater than 1;

[0024] The processing module is used to process each second data group according to the order of occurrence of the historical resource transfer behavior, either from first to last or from last to first, using the third prediction model to obtain r third numerical intervals; update the r third numerical intervals based on a numerical update strategy, wherein the numerical update strategy indicates that the numerical intervals are updated according to the level of the region; and determine the first numerical interval corresponding to the third prediction model based on the updated r third numerical intervals.

[0025] In another possible implementation, the processing module is configured to determine a second value from each of the updated third numerical intervals based on a preset ratio, wherein the preset ratio is the proportion of the numerical interval corresponding to the second value to the updated third numerical interval, and the numerical interval corresponding to the second value is the numerical interval between the second value and the minimum value in the updated third numerical interval; and to determine the minimum value in the updated r third numerical intervals and the maximum value in the updated r second values ​​as the minimum value and maximum value of the first numerical interval, respectively.

[0026] In another possible implementation, the m interval prediction models include a fourth prediction model; the historical data includes resource transfer data of s third objects, where the s third objects are objects associated with the first object, and s is an integer greater than 1; the processing module is used to process the resource transfer data of each third object through the fourth prediction model to obtain a fourth numerical interval corresponding to each third object; and to fuse the fourth numerical intervals corresponding to the s third objects through the fourth prediction model to obtain a first numerical interval corresponding to the fourth prediction model.

[0027] In another possible implementation, the processing module is configured to, in response to a resource transfer request for the first object, validate the resource transfer request based on a second numerical range; respond to the resource transfer request if the resource transfer value indicated by the resource transfer request meets the second numerical range; and refuse to respond to the resource transfer request if the resource transfer value indicated by the resource transfer request does not meet the second numerical range.

[0028] On the other hand, a computer device is provided, the computer device including a processor and a memory, the memory storing at least one computer program, the at least one computer program being loaded and executed by the processor to perform the operations performed by the resource transfer method as described above.

[0029] On the other hand, a computer-readable storage medium is provided that stores at least one computer program, which is loaded and executed by a processor to perform the operations performed by the resource transfer method as described above.

[0030] In another aspect, a computer program product is provided, comprising a computer program that, when executed by a processor, performs the operations performed by the resource transfer method described above.

[0031] The solution provided in this application uses multiple interval prediction models and different interval prediction methods to process historical data in order to predict the resource transfer situation of a first object within a target time period. Then, the predicted multiple numerical intervals are fused. Since different interval prediction methods have different perspectives, historical data can be processed according to multiple perspectives to fully consider various possible situations of future resource transfer of the object. This allows the fused numerical interval to accurately represent the future resource transfer situation of the object, improves the accuracy of the second numerical interval, and then processes the resource transfer request for the first object based on the second numerical interval, thus ensuring the security of the resource transfer for the first object. Attached Figure Description

[0032] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0033] Figure 1 This is a structural block diagram of a computer system provided in an embodiment of this application;

[0034] Figure 2 This is a flowchart of a resource transfer method provided in an embodiment of this application;

[0035] Figure 3 This is a flowchart of a resource transfer method provided in an embodiment of this application;

[0036] Figure 4 This is a schematic diagram of the structure of an interval fusion model provided in an embodiment of this application;

[0037] Figure 5 This is a flowchart of a resource transfer method provided in an embodiment of this application;

[0038] Figure 6 This is a flowchart of a resource transfer method provided in an embodiment of this application;

[0039] Figure 7 This is a flowchart of a resource transfer method provided in an embodiment of this application;

[0040] Figure 8 This is a schematic diagram of a popular area compensation method provided in an embodiment of this application;

[0041] Figure 9 This is a schematic diagram illustrating an association relationship between objects provided in an embodiment of this application;

[0042] Figure 10 This is a flowchart of a resource transfer method provided in an embodiment of this application;

[0043] Figure 11 This is a schematic diagram illustrating the prediction results of a prediction model for different intervals provided in an embodiment of this application;

[0044] Figure 12 This is a schematic diagram of an evaluation result provided in an embodiment of this application;

[0045] Figure 13 This is a schematic diagram of the structure of a resource transfer device provided in an embodiment of this application;

[0046] Figure 14 This is a schematic diagram of the structure of a terminal provided in an embodiment of this application;

[0047] Figure 15 This is a schematic diagram of the structure of a server provided in an embodiment of this application. Detailed Implementation

[0048] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the implementation methods of this application will be further described in detail below with reference to the accompanying drawings.

[0049] The terms “first,” “second,” “third,” “fourth,” “fifth,” etc., used in this application may be used to describe various concepts, but unless otherwise stated, these concepts are not limited by these terms. These terms are only used to distinguish one concept from another. For example, without departing from the scope of this application, a first numerical interval may be referred to as a second numerical interval, and similarly, a second numerical interval may be referred to as a first numerical interval.

[0050] As used in this application, the terms "at least one," "multiple," "each," and "any" have different meanings: at least one includes one, two, or more; multiple includes two or more; each refers to each of the corresponding multiple; and any refers to any one of the multiple. For example, multiple interval prediction models include three interval prediction models, where each refers to each of the three interval prediction models, and any refers to any one of the three interval prediction models, which could be the first interval prediction model, the second interval prediction model, or the third interval prediction model.

[0051] It should be noted that all information (including but not limited to user device information, user personal information, etc.), data (including but not limited to data used for analysis, stored data, displayed data, etc.), and signals involved in this application have been authorized by the user or fully authorized by all parties, and the collection, use, and processing of related data must comply with the relevant laws, regulations, and standards of the relevant countries and regions. For example, historical data and numerical ranges involved in this application were obtained with full authorization.

[0052] The resource transfer method provided in this application is executed by a computer device. Optionally, the computer device is a terminal or a server. Optionally, the server is an independent physical server, or a server cluster consisting of multiple physical servers. Optionally, the terminal is a smartphone, tablet computer, laptop computer, desktop computer, smart speaker, smartwatch, smart voice interaction device, smart home appliance, and vehicle terminal, etc., but is not limited to these.

[0053] Figure 1 This is a structural block diagram of a computer system provided in an embodiment of this application. See also... Figure 1 The computer system includes a terminal 101 and a server 102.

[0054] Server 102 is used to predict a second numerical range for the first object using historical data of the first object. Terminal 101 is used to trigger a resource transfer request for the first object and send the resource transfer request to server 102, and server 102 is used to process the resource transfer request based on the second numerical range.

[0055] In some embodiments, terminal 101 installs and runs application 111, which is provided by server 102. Optionally, application 111 can be any type of application, such as a game application, instant messaging application, or shopping application. Terminal 101 is a terminal used by a user, who uses terminal 101 to log in to application 111 based on an object. Terminal 101 is used to trigger a resource transfer request for a first object through application 111, send the resource transfer request to server 102, and receive the processing result of the resource transfer request returned by server 102 through application 111.

[0056] Terminal 101 is connected to server 102 via a wireless or wired network. Server 102 provides background services for application 111. For example, server 102 includes a processor and a memory. The memory further includes a receiving module and a sending module. The receiving module receives requests sent by terminal 101, such as resource transfer requests. The processor responds to the received requests. The sending module sends responses to terminal 101, such as sending processing results. Optionally, server 102 performs the primary computational work, and terminal 101 performs secondary computational work; or, server 102 performs secondary computational work, and terminal 101 performs the primary computational work.

[0057] Those skilled in the art will understand that the number of terminals described above can be more or less. For example, there may be only one terminal, or there may be dozens or hundreds of terminals, or even more. This application does not limit the number of terminals or the type of device.

[0058] Figure 2 This is a flowchart of a resource transfer method provided in an embodiment of this application, wherein the method is executed by a computer device, such as... Figure 2 As shown, the method includes:

[0059] 201. The computer device acquires historical data of the first object, which is used to describe the historical resource transfer behavior associated with the first object.

[0060] In this embodiment of the application, considering that the resource transfer situation of the first object may have a certain correlation in different time periods, and that the historical data of the first object can reflect the historical resource transfer behavior associated with the first object, that is, can reflect the historical resource transfer situation associated with the first object, therefore, the resource transfer situation of the first object in the target time period is predicted by combining the historical data of the first object, so as to obtain the resource transfer value range in the target time period of the first object, so as to use the predicted value range to verify the resource transfer behavior of the first object, so as to ensure the security of the resource transfer of the first object.

[0061] The first object can be any type of object, such as a merchant account or user account. Historical resource transfer behavior indicates the historical resource transfer activities related to the first object. Historical data can indicate the historical resource transfer activities associated with the first object. For example, historical data indicates instances where other objects transferred virtual resources to the first object. For instance, the transfer of virtual resources from other objects to the first object constitutes a resource transfer behavior. For example, historical resource transfer behavior associated with the first object indicates that other objects transferred resources to the first object, or indicates that the first object transferred resources to other objects, or indicates that other objects transferred resources to associated objects of the first object, or indicates that associated objects of the first object transferred resources to other objects.

[0062] 202. The computer equipment processes historical data using m interval prediction models and according to the corresponding interval prediction methods to obtain m first value intervals. The first value interval is the resource transfer value interval corresponding to the resource transfer behavior of the first object within the target time period. The target time period is the time period after the current time. Different interval prediction models correspond to different interval prediction methods. m is an integer greater than 1.

[0063] In the embodiments of this application, each interval prediction model has an interval prediction method. Each interval prediction model can process historical data according to the corresponding interval prediction method in order to predict the resource transfer value range of the first object. In this way, the resource transfer situation of the first object can be predicted from multiple perspectives, which can enrich the content of the value range and ensure the accuracy of the value range.

[0064] The interval prediction model can be any type of neural network model, such as a regression prediction model or a time series prediction model. The interval prediction method refers to the way historical data is processed, indicating the processing of historical data according to a specific focus. Different interval prediction methods focus on different aspects. For example, any interval prediction method might indicate predicting the numerical interval of the first object from a statistical perspective of historical data; or, it might indicate predicting the numerical interval of the first object based on the chronological order of historical resource transfers indicated by historical data; or, it might indicate predicting the numerical interval of the first object based on the region to which objects that have transferred virtual resources to the first object belong; or, it might indicate predicting the numerical interval of the first object based on the historical resource transfer data of other objects associated with the first object. There is a one-to-one correspondence between the m first numerical intervals and the m interval prediction models, meaning that each interval prediction model can predict a first numerical interval using historical data. The target time period can be any length of time, such as a day, a week, or a month. For example, the target time period could be one day after the current time, or one week after the current time, or one month after the current time.

[0065] For example, the first numerical range is used to represent the range of resource transfer values ​​for the first object within one week after the current time, such as the first numerical range being [100, 1000].

[0066] For example, the first numerical range indicates the range of the number of virtual resources that can be transferred by any resource transfer behavior within the target time period; or, the first numerical range indicates the range of the total number of virtual resources that each object can transfer to the first object within the target time period; or, the first numerical range indicates the range of the number of virtual resources that any object can transfer to the first object within the target time period.

[0067] 203. The computer equipment merges the m first numerical intervals to obtain the second numerical interval.

[0068] In this embodiment of the application, since the m first numerical intervals are predicted by different interval prediction methods, and each first numerical interval is a resource transfer numerical interval corresponding to the resource transfer behavior of the target within the target time period, the m first numerical intervals are fused to ensure that the fused second numerical interval can accurately reflect the resource transfer situation of the target within the future time period, so as to ensure the accuracy of the second numerical interval.

[0069] The second numerical range is the range of resource transfer values ​​corresponding to the resource transfer behavior of the first object within the target time period. Compared with the first numerical range, the second numerical range is more accurate.

[0070] For example, the second numerical range indicates the range of the number of virtual resources that can be transferred by any resource transfer behavior within the target time period; or, the second numerical range indicates the range of the total number of virtual resources that each object can transfer to the first object within the target time period; or, the second numerical range indicates the range of the number of virtual resources that any object can transfer to the first object within the target time period.

[0071] 204. In response to a resource transfer request for a first object, the computer device processes the resource transfer request based on a second numerical range.

[0072] In this embodiment of the application, when the second numerical range is obtained, in response to a resource transfer request for the first object, the resource transfer request can be processed based on the second numerical range to ensure that the resource transfer value for the first object meets the second numerical range, thereby ensuring the security of the resource transfer for the first object.

[0073] The solution provided in this application uses multiple interval prediction models and different interval prediction methods to process historical data to predict the resource transfer situation of a first object within a target time period. Then, the predicted multiple numerical intervals are fused. Since different interval prediction methods have different perspectives, historical data can be processed according to multiple perspectives to fully consider various possible scenarios of future resource transfer of the object. This allows the fused numerical interval to accurately represent the future resource transfer situation of the object, improving the accuracy of the second numerical interval. Based on the second numerical interval, resource transfer requests for the first object are processed, ensuring the security of resource transfer for the first object.

[0074] exist Figure 2 Based on the illustrated embodiment, the historical data in this application embodiment includes data of n first statistical dimensions, and the numerical range is predicted based on the data of n first statistical dimensions. For details of the process, please refer to the following embodiment.

[0075] Figure 3 This is a flowchart of a resource transfer method provided in an embodiment of this application, wherein the method is executed by a computer device, such as... Figure 3 As shown, the method includes:

[0076] 301. The computer device acquires historical data of the first object. The historical data is used to describe the historical resource transfer behavior associated with the first object. The historical data includes data of n first statistical dimensions, where n is an integer greater than 1.

[0077] In this embodiment of the application, the historical data includes data in n first statistical dimensions. The data in n first statistical dimensions can describe the historical resource transfer behavior associated with the first object from different perspectives. By describing the historical resource transfer behavior associated with the first object with data in n first statistical dimensions, the content contained in the historical data is enriched.

[0078] The first statistical dimension can be any dimension. For example, the n first statistical dimensions are the historical total amount of resource transfer, the number of objects that have transferred virtual resources to the first object in the past, the total number of resource transfers per month, the number of resource transfers, the resource transfer scenario, the resource transfer time period, the region to which the object that transferred virtual resources to the first object belongs, and the data of other objects associated with the object. The resource transfer scenario includes online resource transfer scenarios or offline resource transfer scenarios.

[0079] In one possible implementation, the historical data of the first object is also used to describe the first object; that is, the historical data of the first object includes data on various historical resource transfer behaviors and descriptive information of the first object.

[0080] The descriptive information is used to describe the first object from multiple dimensions. For example, the descriptive information includes the first object's basic characteristics, risk tags, transaction statistics, relationship statistics, relationship attributes, and risk statistics. Taking a merchant as an example, the descriptive information of the first object also includes the merchant's image, the merchant's product text content, transaction funds (deposits / withdrawals), and basic tags.

[0081] It should be noted that the embodiments of this application are only illustrated by taking the data including various historical resource transfer behaviors and the description information of the first object as an example. In another embodiment, the historical data includes data on historical resource transfer behaviors, relationship node graphs, images, text and other multimodal data.

[0082] 302. For each of the m interval prediction models, the computer equipment processes the data of each first statistical dimension according to the corresponding interval prediction method to obtain the fifth numerical interval corresponding to each first statistical dimension. Different interval prediction models correspond to different interval prediction methods, where m is an integer greater than 1.

[0083] In this embodiment, the historical data includes data of n first statistical dimensions. That is, the historical data can describe the historical resource transfer behavior associated with the first object using n first statistical dimensions. Therefore, prediction is first performed based on the data of different first statistical dimensions to obtain the fifth numerical interval corresponding to each first statistical dimension. Then, the fifth numerical intervals corresponding to the n first statistical dimensions are merged to obtain the first numerical interval corresponding to the interval prediction model. In this way, the content of the historical data is enriched to ensure that the predicted first numerical interval can conform to the historical resource transfer situation of the first object and improve the accuracy of the first numerical interval.

[0084] Each fifth numerical interval can serve as the resource transfer numerical interval corresponding to the resource transfer behavior of the first object within the target time period. Since different fifth numerical intervals are obtained based on data from different first statistical dimensions, the different fifth numerical intervals may be different.

[0085] In one possible implementation, the process of obtaining any fifth numerical interval includes: inputting data of any first statistical dimension into an interval prediction model, which processes the data of the first statistical dimension and outputs the fifth numerical interval of the first statistical dimension.

[0086] In this embodiment of the application, data of any first statistical dimension is input into the interval prediction model. The interval prediction model extracts features from the data and transforms the extracted features to output the fifth numerical interval of the first statistical dimension.

[0087] 303. The computer equipment uses an interval prediction model to fuse the fifth numerical intervals corresponding to n first statistical dimensions to obtain the first numerical interval corresponding to the interval prediction model. The first numerical interval is the resource transfer numerical interval corresponding to the resource transfer behavior of the first object within the target time period.

[0088] In this embodiment of the application, when the fifth numerical interval corresponding to n first statistical dimensions is obtained through the interval prediction model, the fifth numerical interval corresponding to n first statistical dimensions is fused, and the fused numerical interval is the first numerical interval corresponding to the interval prediction model.

[0089] In one possible implementation, step 303 includes: using an interval prediction model, weighted fusion of the fifth numerical intervals corresponding to the n first statistical dimensions to obtain the first numerical interval corresponding to the interval prediction model.

[0090] In this embodiment of the application, each first statistical dimension has a corresponding weight. The weight corresponding to the first statistical dimension can reflect the degree of influence of the data of the first statistical dimension on the resource transfer request for the first object. According to the weights corresponding to the n first statistical dimensions, the fifth numerical intervals corresponding to the n first statistical dimensions are weighted and fused to fully consider the influence of the data of each first statistical dimension on the resource transfer request for the first object, thereby improving the accuracy of the first numerical interval.

[0091] In one possible implementation, step 303 includes: determining a third value from the fifth numerical interval corresponding to each first statistical dimension using an interval prediction model; determining the maximum value among the n third values ​​as the maximum value of the first numerical interval; and determining the minimum value among the fifth numerical intervals corresponding to the n first statistical dimensions as the minimum value of the first numerical interval.

[0092] Specifically, for any fifth numerical interval corresponding to any first statistical dimension, determining a third value from this fifth numerical interval results in n third values ​​being determined from the n fifth numerical intervals corresponding to the first statistical dimensions. For any fifth numerical interval, the proportion of the numerical interval between the third value and the minimum value within that interval to the total fifth numerical interval is a preset proportion. This preset proportion can be any value, for example, 0.99. This preset proportion can be different for different objects.

[0093] In this embodiment of the application, when the maximum and minimum values ​​of the first numerical interval are determined, the first numerical interval is formed by the minimum value and the maximum value. For example, if the minimum value of the first numerical interval is a and the maximum value of the first numerical interval is b, then the first numerical interval is represented as [a, b].

[0094] In this embodiment of the application, by determining the minimum value among the n fifth numerical intervals as the minimum value among the first numerical intervals, and determining the maximum value among the n third values ​​determined according to a preset ratio from the n fifth numerical intervals as the maximum value of the first numerical interval, the impact of sudden changes in resource transfer values ​​can be avoided, the first numerical interval can be made to cover a larger range as much as possible, and the situation where the first numerical interval is too large and affects the security of resource transfer for the first object can be avoided, thereby improving the accuracy of the first numerical interval.

[0095] It should be noted that the embodiments of this application take any interval prediction model as an example. First, the fifth numerical interval corresponding to each first statistical dimension is obtained, and then the first numerical interval corresponding to the interval prediction model is obtained by fusion. In another embodiment, for each interval prediction model, the first numerical interval corresponding to each interval prediction model can be obtained by following the above steps 302-303, that is, m first numerical intervals can be obtained.

[0096] It should be noted that the embodiments of this application are illustrated using historical data including data of n first statistical dimensions. In another embodiment, it is not necessary to perform the above steps 302-303. Instead, other methods are adopted to process the historical data through m interval prediction models according to the corresponding interval prediction methods to obtain m first numerical intervals.

[0097] 304. The computer equipment merges the m first numerical intervals to obtain the second numerical interval.

[0098] In the embodiments of this application, any fusion method can be adopted to fuse m first numerical intervals, such as averaging, weighted summation, or fusion through an interval fusion model.

[0099] For example, when merging m first numerical intervals by averaging, the minimum value in each first numerical interval is called the fourth value, and the maximum value in each first numerical interval is called the fifth value. The average of the m fourth values ​​and the average of the m fifth values ​​are respectively determined as the minimum and maximum values ​​of the second numerical interval, thus obtaining the second numerical interval.

[0100] For example, when fusing m first numerical intervals using a weighted summation method, each interval prediction model has a corresponding weight. The weight of the interval prediction model is used to represent the importance of the numerical interval predicted by the interval prediction model, and can also reflect the accuracy of the numerical interval predicted by the interval prediction model. The minimum value in each first numerical interval is called the fourth value, and the maximum value in each first numerical interval is called the fifth value. According to the weights corresponding to the m interval prediction models, the m fourth values ​​are weighted and summed to obtain the minimum value of the second numerical interval. According to the weights corresponding to the m interval prediction models, the m fifth values ​​are weighted and summed to obtain the maximum value of the second numerical interval.

[0101] In one possible implementation, step 304 includes: fusing m first numerical intervals using an interval fusion model to obtain a second numerical interval.

[0102] Among them, the interval fusion model is used to fuse multiple numerical intervals. The interval fusion model can be any neural network model. For example, the interval fusion model can be a time series fusion model, such as a neural network model that uses prediction algorithms such as LSTM (Long Short-Term Memory) multivariate or Timenet (deep time series network prediction model), or a neural network model that uses other time series prediction algorithms.

[0103] In this embodiment of the application, an interval fusion model is used to fully consider the accuracy of each first numerical interval and then fuse the m first numerical intervals to improve the accuracy of the fused second numerical interval.

[0104] Optionally, the process of obtaining the second numerical interval through the interval fusion model includes: using the interval fusion model, extracting features from each first numerical interval at multiple scales to obtain multi-scale features for each first numerical interval; concatenating the multi-scale features of m first numerical intervals to obtain concatenated features; performing convolution on the concatenated features; and decoding the convolutional features to obtain the second numerical interval.

[0105] For example, taking the TimeNet model as an example of the interval fusion model, after obtaining the splicing features through the interval fusion model, the gated recurrent unit in the interval fusion model is used, according to... Figure 4 As shown, the splicing features are processed through multiple iterations to obtain the second numerical interval. Each gated loop unit has parameters U, V, and W. The input to the first gated loop unit is the splicing unit, and the first gated loop unit outputs the time step result O and the hidden state S. The input to other gated loop units besides the first gated loop unit is the hidden state S output by the previous gated loop unit.

[0106] 305. In response to a resource transfer request for a first object, the computer device verifies the resource transfer request based on a second numerical range.

[0107] In this embodiment of the application, after determining the second numerical range corresponding to the first object, a resource transfer request for the first object will be received within the target time period. The resource transfer request will be verified based on the second numerical range so that the resource transfer request can be processed according to the verification result, thereby ensuring the security of the resource transfer for the first object.

[0108] The resource transfer request is a request from any object to transfer resources to the first object, and the resource transfer request carries the resource transfer value to be transferred to the first object.

[0109] In this embodiment of the application, if a resource transfer request for a first object is received within a target time period, the resource transfer value carried by the resource transfer request is checked to see if it meets the second value range based on the second value range.

[0110] 306. The computer device responds to the resource transfer request if the resource transfer value indicated in the resource transfer request meets the second value range.

[0111] In this embodiment of the application, if the resource transfer value indicated by the resource transfer request meets the second value range, it means that the resource transfer request has passed the verification and can be responded to so as to complete the resource transfer according to the content of the resource transfer indication.

[0112] For example, the resource transfer request instructs object A to transfer 100 virtual resources to the first object, that is, the resource transfer value is 100. If the resource transfer value meets the second value range, according to the resource transfer request, 100 is deducted from the virtual resources owned by object A and 100 is added to the virtual resources owned by the first object, so as to realize the transfer of 100 virtual resources from object A to the first object.

[0113] 307. If the resource transfer value indicated in the resource transfer request does not meet the second value range, the computer device shall refuse to respond to the resource transfer request.

[0114] In this embodiment of the application, if the resource transfer value indicated by the resource transfer request does not meet the second value range, it indicates that the resource transfer request is abnormal, and the response to the resource transfer request is refused, so as to ensure the security of resource transfer for the first object.

[0115] In one possible implementation, step 307 includes: if the resource transfer value indicated in the resource transfer request does not meet the second value range, sending a rejection notification to the terminal corresponding to the fourth object, the rejection notification indicating that the resource transfer request cannot be responded to.

[0116] The fourth object is the object to which the resource transfer request indicates the transfer of virtual resources to the first object. The rejection notification can be of any type. The terminal corresponding to the fourth object is also equivalent to a terminal logged in based on the fourth object.

[0117] In this embodiment of the application, if the resource transfer value indicated by the resource transfer request does not meet the second value range, it is determined that the resource transfer request cannot be responded to, so as to send a rejection notification to the terminal corresponding to the fourth object to notify the fourth object that the resource transfer has failed.

[0118] In one possible implementation, whether the resource transfer value satisfies the second numerical range includes the following three methods.

[0119] The first approach: If the second numerical range indicates the range of virtual resources that can be transferred by any resource transfer action within the target time period, then if the resource transfer value indicated by the resource transfer request is within the second numerical range, the resource transfer value indicated by the resource transfer request satisfies the second numerical range; if the resource transfer value indicated by the resource transfer request is not within the second numerical range, the resource transfer value indicated by the resource transfer request does not satisfy the second numerical range.

[0120] In this embodiment, the second numerical range corresponds to the range of the number of virtual resources that any object can transfer in a single transfer to the first object. Thus, by verifying whether the resource transfer value indicated by the resource transfer request is within the second numerical range, it can be determined whether the resource transfer value indicated by the resource transfer request satisfies the second numerical range.

[0121] The second approach: If the second numerical range indicates the range of the total number of virtual resources that each object can transfer to the first object within the target time period, in response to receiving a resource transfer request, the sum of the resource transfer value already received by the first object within the target time period and the resource transfer value indicated by the resource transfer request is determined; if the sum is within the second numerical range, the resource transfer value indicated by the resource transfer request satisfies the second numerical range; if the sum is not within the second numerical range, the resource transfer value indicated by the resource transfer request does not satisfy the second numerical range.

[0122] In this embodiment of the application, the second numerical range is equivalent to the value of virtual resources that the first object can receive during the target time period. Therefore, it is necessary to determine the total value of virtual resources that have been transferred to the first object during the target time period. It is only necessary to verify whether the value of the resource transfer indicated by the currently received resource transfer request and the sum of the received resource transfer values ​​are within the second numerical range to determine whether the value of the resource transfer indicated by the resource transfer request satisfies the second numerical range.

[0123] The third approach: If the second numerical interval indicates the range of virtual resources that any object can transfer to the first object within the target time period, in response to receiving a resource transfer request, determine the sum of the number of virtual resources that the object sending the resource transfer request has already transferred to the first object within the target time period and the resource transfer value indicated by the resource transfer request; if the sum is within the second numerical interval, the resource transfer value indicated by the resource transfer request satisfies the second numerical interval; if the sum is not within the second numerical interval, the resource transfer value indicated by the resource transfer request does not satisfy the second numerical interval.

[0124] In this embodiment of the application, the second numerical range is equivalent to the range of virtual resource values ​​transferred by each object within the target time period. That is, an object may transfer virtual resources to the first object once or multiple times within the target time period. When the resource transfer request instructs any object to transfer virtual resources to the first object, it is necessary to determine the value of the resources that the object has transferred to the first object within the target time period, and verify whether the sum of the transferred resource value and the resource value indicated by the resource transfer request is within the second numerical range. In this way, it can be determined whether the resource transfer value indicated by the resource transfer request satisfies the second numerical range.

[0125] It should be noted that in this embodiment, the resource transfer request is first verified based on the second numerical range, and then the resource transfer request is processed according to the verification result. In another embodiment, it is not necessary to perform the above steps 305-307. Instead, other methods are adopted to process the resource transfer request based on the second numerical range in response to the resource transfer request for the first object.

[0126] The solution provided in this application uses multiple interval prediction models and different interval prediction methods to process historical data to predict the resource transfer situation of a first object within a target time period. Then, the predicted multiple numerical intervals are fused. Since different interval prediction methods have different perspectives, historical data can be processed according to multiple perspectives to fully consider various possible scenarios of future resource transfer of the object. This allows the fused numerical interval to accurately represent the future resource transfer situation of the object, improving the accuracy of the second numerical interval. Based on the second numerical interval, resource transfer requests for the first object are processed, ensuring the security of resource transfer for the first object.

[0127] In the above Figure 2 Based on the embodiments shown, this application embodiment predicts p second numerical intervals of second statistical dimensions, and then processes resource transfer requests for the first object based on the p second numerical intervals. For details, please refer to the following embodiments.

[0128] Figure 5 This is a flowchart of a resource transfer method provided in an embodiment of this application. Taking the method executed by a computer device as an example, as follows... Figure 5 As shown, the method includes:

[0129] 501. The computer device acquires historical data of the first object, which is used to describe the historical resource transfer behavior associated with the first object.

[0130] This step is the same as step 201 above, and will not be repeated here.

[0131] 502. For each of the m interval prediction models, the computer equipment processes the historical data according to the corresponding interval prediction method to obtain the first numerical intervals corresponding to the second statistical dimension, where p is an integer greater than 1. The first numerical interval is the resource transfer numerical interval corresponding to the resource transfer behavior of the first object within the target time period. Different interval prediction models correspond to different interval prediction methods, and m is an integer greater than 1.

[0132] In this embodiment, when each interval prediction model processes historical data according to the corresponding interval prediction method, it can obtain the first numerical interval corresponding to p second statistical dimensions. Then, the m interval prediction models are fused together to obtain the second numerical interval corresponding to the same second statistical dimension. In this way, the resource transfer numerical interval corresponding to the resource transfer behavior of the first object within the target time period can be described from multiple second statistical dimensions, so as to enrich the content of the second numerical interval and ensure that the subsequent processing of the resource transfer request for the first object is based on the p second numerical intervals. This allows for verification of the resource transfer situation of the first object from multiple perspectives, further improving the security of the resource transfer of the first object.

[0133] The second statistical dimension can be any dimension. For example, the first numerical intervals corresponding to p second statistical dimensions are the daily maximum and minimum numerical intervals, the cumulative resource transfer numerical intervals within the target time period, and the resource transfer numerical intervals indicated by each resource transfer behavior, etc.

[0134] In one possible implementation, the historical data includes data associated with each second statistical dimension. In this case, step 502 includes: processing the data associated with each second statistical dimension according to the interval prediction model and the corresponding interval prediction method to obtain the first numerical interval corresponding to each second statistical dimension.

[0135] In this embodiment of the application, the historical data includes data of p second statistical dimensions. By processing the data of each second statistical dimension, the first numerical range corresponding to the second statistical dimension can be obtained, thus ensuring the accuracy of the obtained first numerical range.

[0136] In one possible implementation, the historical data includes data of n first statistical dimensions, and each second statistical dimension is associated with at least one first statistical dimension. Then step 502 includes: for any second statistical dimension, using the interval prediction model, processing the data of the first statistical dimension associated with the second statistical dimension according to the corresponding interval prediction method to obtain the first numerical interval corresponding to the second statistical dimension.

[0137] In the embodiments of this application, each first statistical dimension is associated with a second statistical dimension, and each second statistical dimension is associated with at least one first statistical dimension. For example, historical data includes data of 100 first statistical dimensions, where p is 3. The first 30 first statistical dimensions are associated with the first second statistical dimension, the last 30 first statistical dimensions are associated with the second second statistical dimension, and the remaining 10 first statistical dimensions are associated with the third second statistical dimension.

[0138] 503. The computer equipment fuses the first numerical intervals corresponding to the same second statistical dimension of m interval prediction models to obtain the second numerical intervals corresponding to p second statistical dimensions.

[0139] In this embodiment of the application, each interval prediction model can obtain the first numerical interval corresponding to p second statistical dimensions by following the above step 502. For any second statistical dimension, the first numerical intervals corresponding to the m interval prediction models are fused to obtain the second numerical interval corresponding to the second statistical dimension. The m second numerical intervals corresponding to each second statistical dimension are fused in the above manner to obtain the second numerical intervals corresponding to p second statistical dimensions.

[0140] In this embodiment of the application, the process of fusing m interval prediction models into the first numerical interval corresponding to the same second statistical dimension is the same as step 304 above, and will not be repeated here.

[0141] It should be noted that the embodiments of this application are illustrated by taking the acquisition of the second numerical intervals corresponding to p second statistical dimensions as an example. In another embodiment, it is not necessary to perform the above steps 502-503. Instead, other methods are adopted, and historical data are processed by m interval prediction models according to the corresponding interval prediction methods to obtain m first numerical intervals. The m first numerical intervals are then fused to obtain the second numerical interval.

[0142] 504. In response to a resource transfer request for a first object, the computer device processes the resource transfer request based on p second numerical ranges.

[0143] In this application embodiment, p second numerical intervals describe the resource transfer numerical intervals corresponding to the resource transfer behavior of the first object within the target time period from p second statistical dimensions. After receiving a resource transfer request for the first object within the target time period, the resource transfer request can be processed based on the p second numerical intervals so as to verify the resource transfer situation of the first object from multiple perspectives and further improve the security of the resource transfer of the first object.

[0144] In one possible implementation, step 504 includes: in response to a resource transfer request for the first object, verifying the resource transfer request based on a second numerical range corresponding to p second statistical dimensions; responding to the resource transfer request if the resource transfer value indicated by the resource transfer request satisfies the second numerical range corresponding to p second statistical dimensions; and rejecting the response to the resource transfer request if the resource transfer value indicated by the resource transfer request does not satisfy the second numerical range corresponding to any of the second statistical dimensions.

[0145] In this embodiment of the application, after determining the p second numerical ranges corresponding to the first object, a resource transfer request for the first object will be received within the target time period. The resource transfer request will be verified based on the p second numerical ranges so that the resource transfer request can be processed according to the verification results. In this way, the basis for verifying the resource transfer request is enriched, which can ensure that the resource transfer value indicated by the resource transfer request for the first object is consistent with the second numerical range, and can ensure the security of the resource transfer for the first object.

[0146] For example, p is 3, the first second numerical interval indicates the resource transfer numerical interval corresponding to each resource transfer behavior, the second second numerical interval indicates the total resource value that can be transferred to the first object within the target time period, and the third second numerical interval indicates the number of virtual resources that each object can transfer to the first object within the target time period; when a resource transfer request for the first object is received within the target time period, the first sum of the resource value already received by the first object within the target time period and the resource transfer value indicated by the resource transfer request is determined; the second sum of the resource value that the object that sent the resource transfer request has transferred to the first object within the target time period and the resource transfer value indicated by the resource transfer request is determined; The resource transfer value indicated in the resource transfer request satisfies the second value range if the first sum is within the first second value range, the second sum is within the second second value range, and the third sum is within the third second value range; if the resource transfer value indicated in the resource transfer request is not within the first second value range, the resource transfer value indicated in the resource transfer request does not satisfy the second value range; if the first sum is within the second second value range, the resource transfer value indicated in the resource transfer request does not satisfy the second value range; if the second sum is within the third second value range, the resource transfer value indicated in the resource transfer request does not satisfy the second value range.

[0147] It should be noted that the embodiments of this application are illustrated by processing resource transfer requests based on p second numerical ranges. In another embodiment, it is not necessary to perform the above step 404. Instead, other methods are adopted to process resource transfer requests based on second numerical ranges in response to resource transfer requests for the first object.

[0148] The solution provided in this application uses multiple interval prediction models and different interval prediction methods to process historical data to predict the resource transfer situation of a first object within a target time period. Then, the predicted multiple numerical intervals are fused. Since different interval prediction methods have different perspectives, historical data can be processed according to multiple perspectives to fully consider various possible scenarios of future resource transfer of the object. This allows the fused numerical interval to accurately represent the future resource transfer situation of the object, improving the accuracy of the second numerical interval. Based on the second numerical interval, resource transfer requests for the first object are processed, ensuring the security of resource transfer for the first object.

[0149] It should be noted that the above Figure 3 The illustrated embodiments are compatible with Figure 5 The embodiments shown are combined, meaning that when each interval prediction model obtains the first numerical interval corresponding to any second statistical dimension, it can follow the above-described... Figure 3 The first numerical range is obtained in the illustrated embodiment. Furthermore, the above... Figure 5 The illustrated embodiment uses the processing of resource transfer requests based on the second numerical intervals corresponding to p second statistical dimensions as an example. In another embodiment, after obtaining the second numerical intervals corresponding to p second statistical dimensions, the p numerical intervals corresponding to the second statistical dimensions can be further fused to obtain a final numerical interval. Subsequently, in response to a resource transfer request for the first object, the resource transfer request is processed based on the final numerical interval. The final numerical interval is obtained according to the following relationship:

[0150] L i =T([p i F i R i G i ]}

[0151] [min(L1, L2, L3, ..., L i ), max(L1, L2, L3,…, L i )]

[0152] Among them, L i p is used to represent the second numerical interval corresponding to the i-th second statistical dimension. i F i R i G iLet L1, L2, L3, ..., L be the first numerical intervals corresponding to the i-th second statistical dimension obtained by each interval prediction model, and let T(·) represent the fusion algorithm. i Let L1, L2, L3, ..., L2 represent the second numerical intervals corresponding to each second statistical dimension; min(L1, L2, L3, ..., L2) i The expression `max(L1, L2, L3, ..., L)` represents taking the minimum value within each of the second numerical intervals, which is also equivalent to taking the minimum value within the final numerical interval. i This is used to represent taking the maximum value in each of the second numerical intervals, which is also equivalent to the maximum value in the final numerical interval.

[0153] In the above Figures 2 to 5 Based on the embodiments shown, any interval prediction model in the embodiments of this application can be any one of the first prediction model, the second prediction model, the third prediction model, or the fourth prediction model.

[0154] In one possible implementation, at least two of the first, second, third, or fourth prediction models are determined as m interval prediction models. For example, if m is 2, then the two interval prediction models are any two of the first, second, third, or fourth prediction models; or if m is 3, then the three interval prediction models are any three of the first, second, third, or fourth prediction models; or if m is 4, then the four interval prediction models are the first, second, third, and fourth prediction models.

[0155] In one possible implementation, the m interval prediction models include at least one of the first prediction model, the second prediction model, the third prediction model, or the fourth prediction model, as well as other interval prediction models.

[0156] Taking any interval prediction model as the first prediction model as an example, the process of obtaining the corresponding first numerical interval using the first prediction model is as follows: Figure 6 As shown, the method includes:

[0157] 601. Divide the historical data into q first data groups with a preset time period. Each first data group includes data within a historical period, where q is an integer greater than 1.

[0158] In this embodiment of the application, since historical data is used to describe multiple historical resource transfer behaviors associated with the first object, the historical data is divided into q first data groups, each of which includes data within a historical period, that is, the q first data groups correspond one-to-one with the q historical periods.

[0159] The preset duration can be any duration, such as one day or one week.

[0160] In one possible implementation, the historical data includes multiple data entries, each of which describes a historical resource transfer behavior associated with the first object. In this case, step 601 includes: dividing the multiple data entries in the historical data into q first data groups according to the occurrence time of the historical resource transfer behavior described by each data entry, with a preset time period as the period.

[0161] In this embodiment of the application, each piece of data can indicate the time when the historical resource transfer behavior occurred. Therefore, according to the occurrence time indicated by multiple pieces of data in the historical data, the historical period to which each piece of data belongs can be determined, and the data within the same historical period can be divided into the same data group. In this way, the accuracy of data division can be guaranteed.

[0162] 602. Using the first prediction model, feature extraction is performed on each first data group to obtain the data features corresponding to each first data group.

[0163] The data features corresponding to the first data group can be represented in any form, for example, the data features corresponding to the first data group can be represented in the form of a vector.

[0164] In this embodiment of the application, the first prediction model can adopt any feature extraction method to extract features from each first data group to obtain the data features corresponding to each first data group. For example, the data features corresponding to each first data group can be obtained by encoding.

[0165] 603. Using the first prediction model, the data features corresponding to each first data group are transformed to obtain the first value corresponding to each first data group. The first value is the resource transfer value corresponding to the resource transfer behavior of the first object within the target time period.

[0166] In this embodiment of the application, since each first data group is used to describe the historical resource transfer behavior associated with the first object within a historical period, the first prediction model is used to extract features from each first data group and transform the extracted data features to obtain the first value corresponding to each first data group, so as to represent the resource transfer value corresponding to the resource transfer behavior that the first object may have within the target time period.

[0167] In one possible implementation, step 603 includes: convolving the data features corresponding to the first data group using a first prediction model, decoding the convolved features, and obtaining the first value corresponding to the first data group.

[0168] In this embodiment of the application, the first prediction model converts the data features corresponding to the first data group into the first value corresponding to the first data group by adopting convolution and decoding methods, so as to ensure the accuracy of the obtained first value.

[0169] 604. Using the first prediction model, based on the first values ​​corresponding to q first data groups, determine the first value range corresponding to the first prediction model.

[0170] In this embodiment of the application, the first value corresponding to the q first data groups is also equivalent to the prediction result based on the q first data groups. That is, the q first values ​​are equivalent to multiple resource transfer values ​​corresponding to the resource transfer behavior of the first object within the target time period. Therefore, based on multiple resource transfer values, the first value range corresponding to the first prediction model can be determined to ensure the accuracy of the first value range.

[0171] In one possible implementation, step 604 includes: sorting the first values ​​corresponding to q first data groups using a first prediction model, and determining the first value at a preset position and the minimum value among the first values ​​corresponding to q first data groups as the maximum and minimum values ​​of the first value interval, respectively.

[0172] In this embodiment, the sorting result is obtained by arranging the q first values ​​in descending or ascending order. The preset position is any position in the sorting result. For example, if the q first values ​​are arranged in ascending order, the preset position is the 99th percentile value, meaning the first value at the preset position is the first value at the q*99th percentile. Similarly, if the q first values ​​are arranged in descending order, the preset position is the 1st percentile value, meaning the first value at the preset position is the first value at the q*1% percentile.

[0173] In this embodiment of the application, by sorting the first values ​​corresponding to the q first data groups, the first value at the preset position and the minimum value among the first values ​​corresponding to the q first data groups are respectively determined as the maximum and minimum values ​​of the first value interval. In this way, the impact of sudden changes in resource transfer values ​​can be avoided, the first value interval can be made to cover a larger range as much as possible, and the situation where the first value interval is too large and affects the security of resource transfer for the first object can be avoided, thereby improving the accuracy of the first value interval.

[0174] In this embodiment, the interval prediction method corresponding to the first prediction model indicates that historical data is processed in a periodic manner to predict the corresponding first numerical interval. First, the historical data is divided into q first data groups with a preset time period. Then, the first prediction model extracts features from each first data group and transforms the extracted data features to obtain the first value corresponding to each first data group. This value represents the resource transfer value corresponding to the resource transfer behavior that the first object may exhibit within the target time period. The first numerical interval is then determined using the multiple first values ​​obtained. In this way, by using a periodic approach to historical data, the resource transfer value corresponding to the resource transfer behavior that the first object may exhibit within the target time period is predicted. This comprehensively considers data from multiple historical periods, ensuring that the determined first numerical interval is consistent with the historical resource transfer behavior associated with the first object, thus ensuring the accuracy of the first numerical interval.

[0175] It should be noted that the process of obtaining the corresponding first numerical interval based on the first prediction model described above can be compared with the above... Figure 3 or Figure 5 The embodiments shown can be combined in any way. For example, using the above... Figure 3 and Figure 6 Taking the illustrated embodiment as an example, historical data for each first statistical dimension is statistically analyzed periodically to obtain q first data groups corresponding to each first statistical dimension. Then, following step 602 above, the data features corresponding to each first statistical dimension can be obtained. For example, the data features corresponding to each first statistical dimension include the total number of historical resource transfers, the number of historical resource transfer objects, monthly resource transfer values, total number of resource transfers, resource transfer value segments, online and offline resource transfer statistics, and statistics on multiple resource transfer scenarios, totaling hundreds of features. Then, through the statistical ranking sub-model in the first prediction model, following steps 603-604 above, the data features corresponding to each first statistical dimension are processed. For example, the data of the first object in the most recent n months is fused and ranked in multiple dimensions to obtain the ranking results of multiple first statistical dimensions, i.e., the numerical intervals corresponding to multiple first statistical dimensions. Then, following step 304, the ranking results of multiple first statistical dimensions are weighted and fused to obtain the final prediction result, i.e., the final second numerical interval. The second numerical interval satisfies the following relationship:

[0176] [min(p01, p02, p03,..., p0 i ), max(p991, p992, p993,…, p99 i )]

[0177] Among them, p01, p02, p03, ..., p0 iThis represents the value of the smallest quantile in the ranking results for each first statistical dimension; (p991, p992, p993, ..., p99...). i This represents the value of the 99th percentile in the ranking results for each first statistical dimension, min(p01, p02, p03, ..., p0). i The expression `max(p991, p992, p993, ..., p99)` is used to represent the minimum value in the second numerical interval. i ) is used to represent the maximum value in the second numerical range.

[0178] It should be noted that the above is only based on Figure 3 and Figure 6 The illustrated embodiments are described in conjunction with examples, while in another embodiment, the above... Figure 5 and Figure 6 The illustrated embodiments can be combined to obtain second numerical ranges for multiple second statistical dimensions. For example, the second numerical ranges for multiple second statistical dimensions could be the range of a single merchant resource transfer, the range of cumulative resource transfers per user, the range of total daily resource transfers, and the range of cumulative historical resource transfers. Figure 3 , Figure 5 and Figure 6 The illustrated embodiments, taken as examples, can be followed according to the above. Figure 3 and Figure 6 The illustrated embodiment combines the methods of obtaining each second statistical dimension, and then processing the subsequent resource transfer of the first object based on the second numerical range of multiple second statistical dimensions.

[0179] Furthermore, the confidence level of the first numerical interval obtained based on the first prediction model is fitted using historical data volume. For example, if data is reliable 60% of the time in historical data, then the confidence level of the first numerical interval is sum(t) / T, where sum(t) represents the sum of historical statistical data points and T represents the statistical time point. If the statistical period is 30 days historically, and data is available for 15 days, then 15 / 30 represents a 50% confidence level. Therefore, when obtaining the second numerical interval using multiple first numerical intervals, if the confidence level of a first numerical interval is too low, that first numerical interval will be discarded to ensure the accuracy of the subsequently obtained second numerical interval.

[0180] Taking any interval prediction model as the second prediction model as an example, the historical data includes data corresponding to multiple historical resource transfer behaviors; then the process of obtaining the corresponding first numerical interval using the second prediction model includes the following steps 1 to 2.

[0181] Step 1: Sort the data corresponding to multiple historical resource transfer behaviors in order of their occurrence time, either from first to last or from last to first, to obtain sequence information.

[0182] In this embodiment of the application, the historical data includes data corresponding to multiple historical resource transfer behaviors, that is, the historical data includes multiple data entries, each of which describes a historical resource transfer behavior associated with the first object. The data corresponding to the multiple historical resource transfer behaviors are sorted according to the order of their occurrence time from first to last or from last to first to obtain sequence information, so that the sequence information includes data corresponding to multiple historical resource transfer behaviors, and the data corresponding to the multiple historical resource transfer behaviors are arranged in the order of their occurrence time from first to last or from last to first.

[0183] In one possible implementation, the second prediction model is a time-series prediction model, which obtains sequence information according to the occurrence time of multiple historical resource transfer behaviors in either first-to-last or last-to-first order, so that the sequence information can reflect the occurrence time sequence of each historical resource transfer behavior, so that the subsequent time-series prediction model can capture features according to the occurrence time sequence of each historical resource transfer behavior, and then output the first numerical interval.

[0184] In one possible implementation, step 1 includes: concatenating the data corresponding to multiple historical resource transfer behaviors in order of their occurrence time from first to last or from last to first, to obtain sequence information.

[0185] In this embodiment of the application, the data corresponding to multiple historical resource transfer behaviors are spliced ​​together according to the order of their occurrence time from first to last or from last to first to obtain sequence information. This sequence information can reflect the situation of historical resource transfer behaviors associated with the first object, so that the sequence information can be used to accurately predict the range of resource transfer values ​​for the first object within the target time period.

[0186] Step 2: Process the sequence information using the second prediction model to obtain the first numerical interval corresponding to the second prediction model.

[0187] In this embodiment of the application, the second prediction model is a time series algorithm model. The second prediction model processes the sequence information so that it can combine the time series of historical resource transfer behavior associated with the first object to make predictions. It can take into account the correlation between the resource transfer behaviors associated with the first object, and thus accurately predict the resource transfer situation associated with the first object within the target time period, thereby obtaining the first numerical interval.

[0188] In this embodiment, the data corresponding to multiple historical resource transfer behaviors are arranged into sequence information according to the order of their occurrence time from first to last or from last to first. This sequence information can reflect the situation of historical resource transfer behaviors associated with the first object. The sequence information is processed by a second prediction model to take into account the correlation between the resource transfer behaviors associated with the first object, thereby accurately predicting the resource transfer situation associated with the first object within the target time period and obtaining a first numerical interval, thus ensuring the accuracy of the first numerical interval.

[0189] In this embodiment, for objects with long-term resource transfer patterns, or objects with large fluctuations in resource transfer values, using the aforementioned time-series prediction model can improve the accuracy of the predicted value range. The time-series prediction model can be any type of neural network model, such as a single-factor prediction model, a multi-factor / multivariate prediction model, or a neural network model employing other intelligent time-series prediction algorithms. For example, the single-factor prediction model could be ARIMA (Autoregressive Integrated Moving Average), and the multi-factor / multivariate prediction model could be VAR (Vector Autoregression). Compared to the single-factor prediction model, the multi-factor / multivariate prediction model is more accurate and can integrate prediction results from multiple dimensions. However, the single-factor prediction model performs better for objects with limited historical data.

[0190] It should be noted that the process of obtaining the corresponding first numerical interval based on the second prediction model described above can be consistent with the above... Figure 3 or Figure 5 The embodiments shown can be combined in any way. For example, using the above... Figure 3 and Figure 6 The illustrated embodiment, taking this as an example, uses a second prediction model to first obtain the fifth numerical intervals corresponding to each of the first statistical dimensions, and then merges multiple fifth numerical intervals to obtain a first numerical interval. The first numerical interval satisfies the following relationship:

[0191] [min(F1, F2, F3, ..., F i ), max(F1, F2, F3,…, F i )]

[0192] Where F1, F2, F3, ..., F i These represent the fifth numerical intervals corresponding to each of the first statistical dimensions; min(F1, F2, F3, ..., F iThe expression `max(F1, F2, F3, ..., F)` represents the minimum value in the first numerical interval. i ) is used to represent the maximum value in the first numerical range.

[0193] It should be noted that the above is only based on Figure 3 The illustrated embodiment is explained in conjunction with the process of obtaining the corresponding first numerical interval based on the second prediction model. In another embodiment, the above... Figure 5 The illustrated embodiment can be combined with the process of obtaining the corresponding first numerical interval based on the second prediction model, thereby obtaining second numerical intervals with multiple second statistical dimensions. As described above... Figure 3 and Figure 5 The illustrated embodiment, combined with the process of obtaining the corresponding first numerical interval based on the second prediction model, can be used as an example to illustrate the above-described method. Figure 3 The illustrated embodiment combines the process of obtaining the corresponding first numerical interval based on the second prediction model to obtain each second statistical dimension, and then processes the subsequent resource transfer of the first object based on the second numerical intervals of multiple second statistical dimensions.

[0194] Taking any interval prediction model as the third prediction model as an example, the process of obtaining the corresponding first numerical interval using the third prediction model is as follows: Figure 7 As shown, the method includes:

[0195] 701. Based on the region to which the second object belongs, indicated by historical resource transfer behavior, the historical data is divided into r second data groups. Resource transfer behavior has occurred between the second object and the first object. Objects in the same second data group belong to the same region. r is an integer greater than 1.

[0196] In this embodiment, the historical data includes data on multiple historical resource transfer behaviors. These historical resource transfer behaviors indicate that a second object is transferring virtual resources to a first object. Different historical resource transfer behaviors may indicate different second objects. Each second data group includes data on at least one historical resource transfer behavior. Based on the region to which the second object indicated by each historical behavior belongs, the data on each historical resource transfer behavior in the historical data is divided into r second data groups, so that the second objects indicated by data in the same second data group belong to the same region.

[0197] In this context, the region to which the second object belongs refers to the location of the second object. For example, if the first object is a merchant and the second object is a customer, historical resource transfer behavior refers to transactions between a customer and a merchant, and the region to which the customer belongs refers to the location of the customer.

[0198] 702. Using the third prediction model, each second data group is processed according to the order of occurrence of historical resource transfer behavior from first to last or from last to first, to obtain r third numerical intervals.

[0199] In this embodiment, each second data group includes data on at least one historical resource transfer behavior. For any second data group, the data in the second data group is processed using a third prediction model according to the order of occurrence of each historical resource transfer behavior indicated by the second data group, either from first to last or from last to first, to obtain a third numerical interval. Using the third prediction model, the third numerical interval corresponding to each second data group can be obtained in the above manner.

[0200] The third prediction model can be any neural network model, for example, a time-series algorithm model.

[0201] It should be noted that the process of processing the second data set with the third prediction model to obtain the third numerical interval is the same as the process of obtaining the corresponding first numerical interval with the second prediction model, and will not be repeated here.

[0202] 703. Based on the numerical update strategy, update r third numerical intervals. The numerical update strategy indicates that the numerical intervals are updated according to the level of the region.

[0203] In this embodiment of the application, the numerical update strategy can reflect the update method of the numerical intervals corresponding to different levels of regions. Considering that the resource transfer situation of the first object is different for different levels of regions, and the fluctuation of the resource transfer value when the object and the first object in different regions transfer resources is also different, based on the numerical update strategy, the r third numerical intervals are updated according to the level of each region so that the updated third numerical intervals can match the corresponding regions, thereby improving the accuracy of the updated third numerical intervals.

[0204] In one possible implementation, the numerical update strategy indicates the numerical interval update range corresponding to different levels. Then, the process of updating any third numerical interval includes: determining the numerical interval update range corresponding to the third numerical interval based on the numerical update strategy, and adjusting the third numerical interval based on the numerical interval update range.

[0205] The third numerical interval corresponds to a second data group. The objects indicated by each array in the second data group belong to the same region. Based on this region, the numerical interval update magnitude corresponding to the level of this region can be determined from the numerical update strategy. The determined numerical interval update magnitude is the numerical interval update magnitude corresponding to the third numerical interval.

[0206] For example, the numerical update strategy includes the numerical interval update range corresponding to 5 levels. The third numerical interval is [100, 200]. The level of the region corresponding to the third numerical interval is 3. The numerical interval update range corresponding to level 3 is "base value + 10%". Then the updated third numerical interval is [100, 220].

[0207] 704. Based on the updated r third numerical intervals, determine the first numerical interval corresponding to the third prediction model.

[0208] In this embodiment of the application, the updated r third numerical intervals are predicted numerical intervals that can accurately predict the resource transfer situation of objects in various regions to the first object. Therefore, the final first numerical interval is determined by combining the updated r third numerical intervals so that the first numerical interval can accurately reflect the resource transfer numerical interval corresponding to the resource transfer behavior of the first object within the target time period, so as to ensure the accuracy of the first numerical interval.

[0209] In one possible implementation, step 704 includes: determining a second value from each of the updated third value intervals based on a preset ratio, wherein the preset ratio is the proportion of the value interval corresponding to the second value to the updated third value interval, and the value interval corresponding to the second value is the value interval between the second value and the minimum value in the updated third value interval; and determining the minimum value in the updated r third value intervals and the maximum value in the updated r second values ​​as the minimum value and maximum value in the first value interval, respectively.

[0210] The preset ratio can be any value, for example, the preset ratio is 0.99.

[0211] In this embodiment of the application, by determining the minimum value among the updated r third numerical intervals as the minimum value among the first numerical intervals, and determining the maximum value among the r second values ​​determined according to a preset ratio among the updated r third numerical intervals as the maximum value of the first numerical interval, the impact of sudden changes in resource transfer values ​​can be avoided, the first numerical interval can be made to cover a larger range as much as possible, and the situation where the first numerical interval is too large and affects the security of resource transfer for the first object can be avoided, thereby improving the accuracy of the first numerical interval.

[0212] In this embodiment of the application, when the maximum and minimum values ​​of the first numerical interval are determined, the first numerical interval is formed by the minimum value and the maximum value. For example, if the minimum value of the first numerical interval is a and the maximum value of the first numerical interval is b, then the first numerical interval is represented as [a, b].

[0213] In this embodiment, considering the differences in resource transfer between objects in different regions and the first object, the historical data is divided into r second data groups according to the region to which the second object that has historically transferred resources with the first object belongs. Then, through the second prediction model, the third numerical interval corresponding to each region is predicted for the second data group corresponding to each region. Then, based on the numerical update strategy, the r third numerical intervals are updated according to the level of the region, so that the first numerical interval corresponding to the third prediction model is determined based on the updated r third numerical intervals. In this way, the resource transfer between objects in different regions and the first object can be taken into account, enriching the basis for predicting the numerical interval and ensuring that the obtained first numerical interval can be applied to the resource transfer between objects in different regions and the first object, thus improving the accuracy of the first numerical interval.

[0214] In this embodiment, the ranking optimization of the region model of the object to which the resource transfer is performed for the first object is added, which can better correct the prediction results, better interpret the results, and make differentiated regional applications in marketing and other scenarios.

[0215] First, aggregate the regions of customers transferring merchant resources to obtain the merchants' frequently used transaction areas. If multiple regions exist, use the top N most frequently used regions to predict their transactions, and then take the maximum and minimum values. Next, sort the transactions for each region separately to obtain the ranking of popular consumption regions. Figure 8 As shown, the system statistically analyzes the locations corresponding to merchant orders to determine their positions; it also determines merchant transaction locations and frequently used user locations; it sorts merchant order amounts to obtain merchant popularity positions; and further sorts order amounts to obtain numerical range correction compensation. Based on the above determinations, it identifies the tiers of each region and the corresponding compensation ranges for each tier, thus segmenting merchant transaction regions into different tiers for differentiated compensation. The compensation method involves overlaying the regional consumption hotspot ranking results onto a baseline value; that is, adjusting each numerical region based on a numerical adjustment strategy.

[0216] It should be noted that the above Figure 7 The embodiments shown are compatible with the above-described embodiments. Figure 3 or Figure 5 The embodiments shown can be combined in any way. For example, using the above... Figure 3 and Figure 7 The illustrated embodiment, taking this as an example, uses a third prediction model to first obtain the fifth numerical intervals corresponding to each of the first statistical dimensions, and then merges multiple fifth numerical intervals to obtain a first numerical interval. The first numerical interval satisfies the following relationship:

[0217] [min(R1,R2,R3,…,Ri),max(R1,R2,R3,…,R i )]

[0218] Where R1, R2, R3, ..., R i These represent the fifth numerical intervals corresponding to each of the first statistical dimensions; min(R1, R2, R3, ..., R i The expression `max(R1, R2, R3, ..., R)` is used to represent the minimum value in the first numerical interval. i ) is used to represent the maximum value in the first numerical range.

[0219] It should be noted that the above is only based on Figure 3 and Figure 7 The illustrated embodiments are described in conjunction with examples, while in another embodiment, the above... Figure 5 and Figure 7 The illustrated embodiments can be combined to obtain multiple second numerical intervals for second statistical dimensions. As described above... Figure 3 , Figure 5 and Figure 7 The illustrated embodiments, taken as examples, can be followed according to the above. Figure 3 and Figure 7 The illustrated embodiment combines the methods of obtaining each second statistical dimension, and then processing the subsequent resource transfer of the first object based on the second numerical range of multiple second statistical dimensions.

[0220] Taking any interval prediction model as the fourth prediction model as an example, the historical data includes resource transfer data of s third objects, where s third objects are objects associated with the first object, and s is an integer greater than 1. The process of obtaining the corresponding first numerical interval using the fourth prediction model includes the following steps 3 to 4.

[0221] Step 3: Using the fourth prediction model, process the resource transfer data of each third object to obtain the fourth numerical range corresponding to each third object.

[0222] In this embodiment of the application, the first object is associated with s third objects. The resource transfer between the associated objects may have a certain correlation. Therefore, the resource transfer data used to represent the historical resource transfer of the third objects can be used to predict the resource transfer situation of the first object for the target time period.

[0223] The resource transfer data of the third object is equivalent to the historical resource transfer data of the third object, used to represent the historical transfer situation from other objects to the third object. The first object and the third object can be associated in any way. For example, if both the first and third objects are merchants, and their legal representatives are the same person, then the first object and the third object are associated; or, if the first object and the third object are merchants under the same group, then the first object and the third object are associated; or, if the first object and the third object are merchants managed by the same manager, then the first object and the third object are associated. The fourth prediction model can be any type of model; for example, the fourth prediction model can be a diffusion model.

[0224] In this embodiment of the application, taking the example that both the first object and the third object are merchants, the first object is a newly created merchant. Since the historical resource transfer data of the first object is too small, the historical resource transfer data of other merchants associated with the first object can be used to predict the resource transfer situation of the first object within the target time period.

[0225] In this embodiment of the application, the resource transfer data of any third object is input into the fourth prediction model so that the fourth prediction model processes the resource transfer data of the third object and obtains the fourth numerical interval corresponding to the third object. In this way, the fourth numerical interval corresponding to each third object can be obtained through the fourth prediction model.

[0226] It should be noted that the process of obtaining the fourth numerical interval corresponding to each third object through the fourth prediction model is the same as the process of predicting the numerical interval by the first, second or third prediction model, and will not be repeated here.

[0227] Step 4: Using the fourth prediction model, fuse the fourth numerical intervals corresponding to the s third objects to obtain the first numerical interval corresponding to the fourth prediction model.

[0228] In this embodiment of the application, since all s third objects are associated with the first object, when the fourth numerical intervals corresponding to the s third objects are obtained, the fourth prediction model is used to fuse the fourth numerical intervals corresponding to the s third objects so that the fused numerical interval can accurately predict the resource transfer situation of the first object within the target time period, thereby ensuring the accuracy of the first numerical interval.

[0229] In one possible implementation, step 4 includes: using a fourth prediction model to perform weighted fusion on the fourth numerical intervals corresponding to s third objects to obtain the first numerical interval corresponding to the fourth prediction model.

[0230] In this embodiment of the application, each third object has a corresponding weight. Based on the weight of each third object, the s fourth numerical intervals are weighted and fused to ensure the accuracy of the first numerical interval.

[0231] Optionally, the process of determining the weight of the third object includes: determining the weight of the third object based on the association between the third object and the first object.

[0232] In this embodiment, the association relationships between different third objects and the first object may be different. Each association relationship corresponds to a weight, and different association relationships correspond to different weights. Therefore, based on the association relationship between the third object and the first object, the weight of the third object can be obtained by querying the correspondence between the association relationship and the weight. The correspondence between the association relationship and the weight includes at least one association relationship and the weight of each association relationship.

[0233] In this embodiment, considering that the resource transfer between related objects may have a certain correlation, the resource transfer data of s third objects associated with the first object are obtained. In order to use the resource transfer data representing the historical resource transfer of the third objects, the resource transfer situation of the first object corresponding to the target time period is predicted, that is, the first numerical interval is predicted, so as to ensure the applicability of this prediction method. Even if the first object may not have had resource transfers with other objects, that is, the data representing the historical resource transfer of the first object cannot be obtained, the resource transfer situation of the first object in the target time period can still be predicted, ensuring the accuracy of the predicted first numerical interval and improving the applicability of this interval prediction method.

[0234] Based on the embodiments described above, for many merchants without historical transaction information or newly joined merchants, revenue transaction results are supplemented through diffusion, following the prediction method of the fourth prediction model described above. When determining the second object associated with the first object, as... Figure 9 As shown, diffusion dimensions typically include: subject, legal person, actual manager of merchant account, group merchant number, etc. For merchants with abundant data, the first numerical interval can be obtained without following the prediction method of the fourth prediction model described above. In this embodiment, the fourth prediction model is also equivalent to a diffusion model, which can expand the data related to the first object, thereby obtaining the first numerical interval of the first object. The first numerical interval satisfies the following relationship:

[0235] [min(G1, G2, G3, ..., G i ),max(G1, G2, G3, ..., G i )]

[0236] Among them, G1, G2, G3, ..., Gi These represent the fourth numerical intervals corresponding to each of the third objects; min(G1,G2,G3,…,G…) i The expression `max(G1,G2,G3,…,G)` represents the minimum value within the fourth numerical interval corresponding to each third object. i This is used to represent the maximum value in the fourth numerical interval corresponding to each third object.

[0237] It should be noted that the above-mentioned multiple optional embodiments can be combined in any way, and this application will not elaborate on them one by one here.

[0238] Figure 10 This is a flowchart of a resource transfer method provided in an embodiment of this application, such as... Figure 10 As shown, the method includes:

[0239] Step 1: Obtain merchant transaction behavior characteristics.

[0240] In this embodiment, merchant transaction behavior characteristics include transaction statistics, transaction scenario characteristics, transaction time period distribution, location of merchant customer groups, and various related data of merchants. Merchant-level characteristics include: basic features, risk tags, transaction statistics, relationship statistics, relationship attributes, and risk statistics, as well as offline features such as merchant images, merchant product text content, transaction funds (deposits / withdrawals), and basic tags. Detailed statistics on merchant transaction behavior characteristics are employed across multiple periods and scenarios.

[0241] The basic characteristics of merchants can be divided into two parts based on application scenarios and implementation methods: real-time and offline. The real-time part includes importing offline results into key-value (KV) databases, performing real-time statistical requests on transactions, onboarding, and interventions, which is directly applied to real-time countermeasure strategies and provides real-time analysis services. The offline part is based on payment and onboarding transaction data, transaction data, and intervention data, statistically analyzing thousands of features across multiple dimensions such as merchant fundamentals, funds, and tags over time periods such as years, months, and days. For example, the basic characteristics of merchants are shown in Table 1.

[0242] Table 1

[0243]

[0244]

[0245] In this application embodiment, any data platform can be used to store merchant transaction behavior characteristics, for example, based on various big data analysis storage architectures.

[0246] Step 2: Use multiple interval prediction models to predict the revenue range of merchants.

[0247] In this embodiment, multiple interval prediction models are established by combining data from different dimensions to predict the revenue range of merchants. These multiple interval prediction models include: statistical prediction models, transaction time-series prediction models, transaction region ranking models, and merchant multi-relationship diffusion models, etc. Furthermore, the final revenue range prediction result is obtained by fusing and overlaying these multiple models. The prediction results for different interval prediction models are as follows: Figure 11 As shown.

[0248] Step 3: Evaluate the error of the prediction results. In this embodiment, multiple evaluation methods are used for the merchant prediction results, including: amount error evaluation, account error (evaluation of the overall impact on merchants), top merchant error, and comprehensive error weighting. The final comprehensive weighted evaluation result can be viewed, and different evaluation methods can be selected for different scenarios. A multi-model fusion approach based on statistics and time series is used to predict merchant transactions in layers (e.g., daily, monthly, historical), ultimately fusing the results to provide single transaction amount ranges, daily revenue ranges, and cumulative revenue ranges. Multiple dimensions are used to verify the evaluation effect, including: amount error evaluation, account error evaluation, and top merchant verification.

[0249] Among them, the amount error assessment is based on comparing the predicted result of [T] obtained from the prediction of [T-1] with the actual transaction squared error assessment of [T], using the following absolute error relationship:

[0250]

[0251] Among them, y t p is used to represent the actual value on day t. t-1 This represents the predicted value for day t obtained from the prediction on day t-1, where N represents the sample size, and y... i，t p is used to represent the true value of the i-th sample on day t. i，t-1 MAE(·) represents the predicted value of the i-th sample on day t-1, and MAE(·) represents the mean absolute error function.

[0252] In this embodiment of the application, in order to ensure the overall accuracy of the evaluation and prediction results, multiple results such as single transaction amount, number of users, daily revenue, and historical cumulative revenue can be evaluated.

[0253] Account error assessment: The ratio of predicted results to the number of merchants with actual transaction amounts is assessed using the following relationship:

[0254]

[0255] Among them, y t p is used to represent the actual value on day t. t-1This represents the predicted value for day t obtained from the prediction on day t-1, where N represents the sample size, and y... i，t p is used to represent the true value of the i-th sample on day t. i，t-1 The function is used to represent the predicted value of the i-th sample on day t-1, sgn(·) is used to represent the sign function, max(·) is used to represent the maximization function, and Err(·) is used to represent the error evaluation function.

[0256] In this embodiment of the application, when the actual value on day t is greater than the predicted value on day t, it will affect the normal transactions of the merchants.

[0257] Among them, verification by leading merchants: combining case studies of leading merchants can better guarantee effectiveness in real-world scenarios, and the following relationships are used for evaluation:

[0258]

[0259] Among them, y t p is used to represent the actual value on day t. t-1 Used to represent the predicted value for day t obtained from the prediction on day t-1, K is used to represent the sample size, and y i，t p is used to represent the true value of the i-th sample on day t. i，t-1 MAE(·) represents the predicted value of the i-th sample on day t-1, and MAE(·) represents the mean absolute error function.

[0260] Multiple evaluations revealed that the deep temporal fusion model can achieve an accuracy of <10% error across multiple standards. The evaluation results are as follows: Figure 12 As shown in the figure. Here, error represents the difference between the calculated prediction and the actual value; the better the error, the more accurate the prediction. Therefore, based on... Figure 12 The content is restrained, and the method provided in this application for real-time prediction results is accurate enough.

[0261] Step 4: Optimize the prediction results.

[0262] In this embodiment, the prediction results require iterative optimization, with multiple feedback channels to verify the actual effect. These include: post-transaction feedback from merchants, complaint feedback regarding interventions in risk control strategies, and manual operational observation and optimization by leading merchants. Based on the feedback information, various interval prediction models can be trained to improve their accuracy, thereby optimizing the prediction results. Prediction results may not be accurate under certain unforeseen events, requiring iterative optimization based on feedback. When significant discrepancies are found in the results through feedback, the time-series model and interval fusion model will be readjusted and retrained iteratively. Outliers can also be excluded from the statistical models of some merchants.

[0263] In this embodiment, through intelligent multi-model fusion revenue prediction and identification, and various comprehensive evaluation methods, the true revenue range of merchants can be identified more accurately and effectively, and the prediction results can be quickly applied to real-world scenarios. By fusing these multi-models, a more accurate identification of multi-dimensional revenue results for merchants is achieved, enabling more comprehensive and effective differentiation between normal and abnormal transactions. Normal and high-quality transaction requests are protected, while abnormal transactions can be identified and intervened with enhanced risk strategies to ensure the safety of users and the platform. Simultaneously, this revenue range can also be used for the daily operational management of merchants.

[0264] The solution provided in this application, combining multi-dimensional data from merchants, establishes various merchant revenue range prediction models, such as statistical prediction models, transaction time-series prediction models, transaction area ranking models, and merchant multi-relationship diffusion models. Using interval fusion models can improve recognition accuracy. Furthermore, the predicted business results are multi-dimensional, including the merchant's daily maximum and minimum revenue range, cumulative revenue range, and single transaction amount range. This makes the solution more flexible in different scenarios and can be applied to multiple situations, such as risk control and marketing operations for merchants.

[0265] The accuracy of prediction results using the interval fusion model can reach 90%+ under multi-dimensional evaluation, and it can more accurately distinguish between merchant risk and normal transactions, especially for new merchants or merchants with few historical transactions.

[0266] Figure 13 This is a schematic diagram of the structure of a resource transfer device provided in an embodiment of this application, as shown below. Figure 13 As shown, the device includes:

[0267] The acquisition module 1301 is used to acquire historical data of the first object, which describes the historical resource transfer behavior associated with the first object.

[0268] The processing module 1302 is used to process historical data through m interval prediction models and according to the corresponding interval prediction methods to obtain m first value intervals. The first value interval is the resource transfer value interval corresponding to the resource transfer behavior of the first object within the target time period. The target time period is the time period after the current time. Different interval prediction models correspond to different interval prediction methods. m is an integer greater than 1.

[0269] The fusion module 1303 is used to fuse m first numerical intervals to obtain a second numerical interval;

[0270] The processing module 1302 is also configured to process the resource transfer request based on a second numerical range in response to the resource transfer request for the first object.

[0271] In one possible implementation, the historical data includes data for n first statistical dimensions, where n is an integer greater than 1; the processing module 1302 is used to process the data for each first statistical dimension according to the corresponding interval prediction method through the interval prediction model to obtain the fifth numerical interval corresponding to each first statistical dimension; and to fuse the fifth numerical intervals corresponding to the n first statistical dimensions through the interval prediction model to obtain the first numerical interval corresponding to the interval prediction model.

[0272] In another possible implementation, the processing module 1302 is used to process historical data for each interval prediction model according to the corresponding interval prediction method to obtain the first numerical intervals corresponding to p second statistical dimensions, where p is an integer greater than 1.

[0273] The fusion module 1303 is used to fuse the first numerical intervals corresponding to the same second statistical dimension of m interval prediction models to obtain the second numerical intervals corresponding to p second statistical dimensions.

[0274] In another possible implementation, the m interval prediction models include the first prediction model;

[0275] The processing module 1302 is also used to divide the historical data into q first data groups with a preset duration as the period, each first data group including data within a historical period, where q is an integer greater than 1;

[0276] The processing module 1302 is used to extract features from each first data group using a first prediction model to obtain data features corresponding to each first data group; to transform the data features corresponding to each first data group using the first prediction model to obtain a first value corresponding to each first data group, wherein the first value is the resource transfer value corresponding to the resource transfer behavior of the first object within the target time period; and to determine the first value range corresponding to the first prediction model based on the first values ​​corresponding to q first data groups using the first prediction model.

[0277] In another possible implementation, the processing module 1302 is used to sort the first values ​​corresponding to the q first data groups through the first prediction model, and determine the first value in the preset position and the minimum value among the first values ​​corresponding to the q first data groups as the maximum and minimum values ​​of the first value interval, respectively.

[0278] In another possible implementation, the m interval prediction models include a second prediction model; the historical data includes data corresponding to multiple historical resource transfer behaviors; the processing module 1302 is also used to sort the data corresponding to multiple historical resource transfer behaviors in order of their occurrence time from first to last or from last to first, to obtain sequence information;

[0279] The processing module 1302 is used to process the sequence information through the second prediction model to obtain the first numerical interval corresponding to the second prediction model.

[0280] In another possible implementation, the m interval prediction models include a third prediction model; the processing module 1302 is also used to divide the historical data into r second data groups based on the region to which the second object belongs, indicated by the historical resource transfer behavior, wherein the second object and the first object have had resource transfer behavior, and the regions to which the objects indicated by the same second data group belong are the same, and r is an integer greater than 1.

[0281] The processing module 1302 is used to process each second data group according to the order of occurrence of historical resource transfer behavior from first to last or from last to first, using the third prediction model, to obtain r third numerical intervals; update the r third numerical intervals based on the numerical update strategy, which indicates that the numerical intervals are updated according to the level of the region; and determine the first numerical interval corresponding to the third prediction model based on the updated r third numerical intervals.

[0282] In another possible implementation, the processing module 1302 is used to determine the second value from each of the updated third value intervals based on a preset ratio, wherein the preset ratio is the proportion of the value interval corresponding to the second value to the updated third value interval, and the value interval corresponding to the second value is the value interval between the second value and the minimum value in the updated third value interval; and the minimum value in the updated r third value intervals and the maximum value in the r second values ​​are determined as the minimum value and maximum value of the first value interval, respectively.

[0283] In another possible implementation, the m interval prediction models include a fourth prediction model; the historical data includes resource transfer data of s third objects, where s third objects are objects associated with the first object, and s is an integer greater than 1; the processing module 1302 is used to process the resource transfer data of each third object through the fourth prediction model to obtain the fourth numerical interval corresponding to each third object; and to fuse the fourth numerical intervals corresponding to the s third objects through the fourth prediction model to obtain the first numerical interval corresponding to the fourth prediction model.

[0284] In another possible implementation, the processing module 1302 is configured to, in response to a resource transfer request for the first object, validate the resource transfer request based on a second numerical range; respond to the resource transfer request if the resource transfer value indicated in the resource transfer request meets the second numerical range; and refuse to respond to the resource transfer request if the resource transfer value indicated in the resource transfer request does not meet the second numerical range.

[0285] It should be noted that the resource transfer device provided in the above embodiments is only an example of the division of the above functional modules. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the computer device can be divided into different functional modules to complete all or part of the functions described above. In addition, the resource transfer device and the resource transfer method embodiments provided in the above embodiments belong to the same concept, and their specific implementation process can be found in the method embodiments, which will not be repeated here.

[0286] This application also provides a computer device, which includes a processor and a memory. The memory stores at least one computer program, which is loaded and executed by the processor to perform the operations performed by the resource transfer method of the above embodiments.

[0287] Optionally, the computer device is provided as a terminal. Figure 14 A structural block diagram of a terminal 1400 provided in an exemplary embodiment of this application is shown. The terminal 1400 includes a processor 1401 and a memory 1402.

[0288] Processor 1401 may include one or more processing cores, such as a quad-core processor, an octa-core processor, etc. Processor 1401 may be implemented using at least one hardware form selected from DSP (Digital Signal Processing), FPGA (Field-Programmable Gate Array), and PLA (Programmable Logic Array). Processor 1401 may also include a main processor and a coprocessor. The main processor, also known as a CPU (Central Processing Unit), is used to process data in the wake-up state; the coprocessor is a low-power processor used to process data in the standby state. In some embodiments, processor 1401 may integrate a GPU (Graphics Processing Unit), which is responsible for rendering and drawing the content required to be displayed on the screen. In some embodiments, processor 1401 may also include an AI (Artificial Intelligence) processor, which is used to handle computational operations related to machine learning.

[0289] The memory 1402 may include one or more computer-readable storage media, which may be non-transitory. The memory 1402 may also include high-speed random access memory and non-volatile memory, such as one or more disk storage devices or flash memory devices. In some embodiments, the non-transitory computer-readable storage media in the memory 1402 are used to store at least one computer program, which is executed by the processor 1401 to implement the resource transfer method provided in the method embodiments of this application.

[0290] In some embodiments, the terminal 1400 may also optionally include a peripheral device interface 1403 and at least one peripheral device. The processor 1401, memory 1402, and peripheral device interface 1403 can be connected via a bus or signal line. Each peripheral device can be connected to the peripheral device interface 1403 via a bus, signal line, or circuit board. Specifically, the peripheral device includes at least one of the following: a radio frequency circuit 1404, a display screen 1405, a camera assembly 1406, an audio circuit 1407, and a power supply 1408.

[0291] Peripheral device interface 1403 can be used to connect at least one I / O (Input / Output) related peripheral device to processor 1401 and memory 1402. In some embodiments, processor 1401, memory 1402 and peripheral device interface 1403 are integrated on the same chip or circuit board; in some other embodiments, any one or two of processor 1401, memory 1402 and peripheral device interface 1403 can be implemented on separate chips or circuit boards, which is not limited in this embodiment.

[0292] The radio frequency (RF) circuit 1404 is used to receive and transmit RF (Radio Frequency) signals, also known as electromagnetic signals. The RF circuit 1404 communicates with communication networks and other communication devices via electromagnetic signals. The RF circuit 1404 converts electrical signals into electromagnetic signals for transmission, or converts received electromagnetic signals back into electrical signals. Optionally, the RF circuit 1404 includes: an antenna system, an RF transceiver, one or more amplifiers, a tuner, an oscillator, a digital signal processor, a codec chipset, a user identity module card, etc. The RF circuit 1404 can communicate with other terminals through at least one wireless communication protocol. This wireless communication protocol includes, but is not limited to: the World Wide Web, metropolitan area networks, intranets, various generations of mobile communication networks (2G, 3G, 4G, and 5G), wireless local area networks, and / or WiFi (Wireless Fidelity) networks. In some embodiments, the RF circuit 1404 may also include circuitry related to NFC (Near Field Communication), which is not limited in this application.

[0293] Display screen 1405 is used to display a UI (User Interface). This UI may include graphics, text, icons, videos, and any combination thereof. When display screen 1405 is a touch display screen, it also has the ability to collect touch signals on or above its surface. These touch signals can be input as control signals to processor 1401 for processing. In this case, display screen 1405 can also be used to provide virtual buttons and / or a virtual keyboard, also known as soft buttons and / or a soft keyboard. In some embodiments, there may be one display screen 1405, disposed on the front panel of terminal 1400; in other embodiments, there may be at least two display screens, disposed on different surfaces of terminal 1400 or in a folded design; in still other embodiments, display screen 1405 may be a flexible display screen, disposed on a curved or folded surface of terminal 1400. Furthermore, display screen 1405 may be configured as a non-rectangular, irregular shape, i.e., a non-rectangular screen. The display screen 1405 can be made of materials such as LCD (Liquid Crystal Display) and OLED (Organic Light-Emitting Diode).

[0294] The camera assembly 1406 is used to acquire images or videos. Optionally, the camera assembly 1406 includes a front-facing camera and a rear-facing camera. The front-facing camera is disposed on the front panel of the terminal, and the rear-facing camera is disposed on the back of the terminal. In some embodiments, there are at least two rear-facing cameras, which are any one of a main camera, a depth-sensing camera, a wide-angle camera, and a telephoto camera, to achieve background blurring by fusion of the main camera and the depth-sensing camera, panoramic shooting by fusion of the main camera and the wide-angle camera, VR (Virtual Reality) shooting, or other fusion shooting functions. In some embodiments, the camera assembly 1406 may also include a flash. The flash can be a single-color temperature flash or a dual-color temperature flash. A dual-color temperature flash refers to a combination of a warm light flash and a cool light flash, which can be used for light compensation at different color temperatures.

[0295] The audio circuit 1407 may include a microphone and a speaker. The microphone is used to collect sound waves from the user and the environment, converting the sound waves into electrical signals that are input to the processor 1401 for processing, or input to the radio frequency circuit 1404 for voice communication. For stereo sound acquisition or noise reduction purposes, multiple microphones may be used, each positioned at a different location on the terminal 1400. The microphone may also be an array microphone or an omnidirectional microphone. The speaker is used to convert electrical signals from the processor 1401 or the radio frequency circuit 1404 into sound waves. The speaker may be a conventional diaphragm speaker or a piezoelectric ceramic speaker. When the speaker is a piezoelectric ceramic speaker, it can convert electrical signals not only into audible sound waves but also into inaudible sound waves for purposes such as distance measurement. In some embodiments, the audio circuit 1407 may also include a headphone jack.

[0296] Power supply 1408 is used to power the various components in terminal 1400. Power supply 1408 can be AC ​​power, DC power, a disposable battery, or a rechargeable battery. When power supply 1408 includes a rechargeable battery, the rechargeable battery can be a wired rechargeable battery or a wireless rechargeable battery. A wired rechargeable battery is a battery that is charged via a wired line, and a wireless rechargeable battery is a battery that is charged via a wireless coil. The rechargeable battery can also be used to support fast charging technology.

[0297] Those skilled in the art will understand that Figure 14 The structure shown does not constitute a limitation on terminal 1400 and may include more or fewer components than shown, or combine certain components, or use different component arrangements.

[0298] Optionally, the computer equipment is provided as a server. Figure 15 This is a schematic diagram of a server structure provided in an embodiment of this application. The server 1500 can vary significantly due to different configurations or performance. It may include one or more Central Processing Units (CPUs) 1501 and one or more memories 1502. The memories 1502 store at least one computer program, which is loaded and executed by the processor 1501 to implement the methods provided in the above-described method embodiments. Of course, the server may also have wired or wireless network interfaces, a keyboard, and input / output interfaces for input and output. The server may also include other components for implementing device functions, which will not be elaborated upon here.

[0299] This application also provides a computer-readable storage medium storing at least one computer program, which is loaded and executed by a processor to perform the operations performed by the resource transfer method of the above embodiments.

[0300] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the operations performed by the resource transfer method of the above embodiments.

[0301] Those skilled in the art will understand that all or part of the steps of the above embodiments can be implemented by hardware or by a program instructing related hardware. The program can be stored in a computer-readable storage medium, such as a read-only memory, a disk, or an optical disk.

[0302] The above description is only an optional embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present application should be included within the protection scope of the present application.

Claims

1. A resource transfer method, characterized in that, The method includes: Obtain historical data of a first object, wherein the historical data is used to describe the historical resource transfer behavior associated with the first object; The historical data is processed by m interval prediction models according to their respective interval prediction methods to obtain m first numerical intervals. The first numerical interval is the resource transfer numerical interval corresponding to the resource transfer behavior of the first object within the target time period. The target time period is the time period after the current time. Different interval prediction models correspond to different interval prediction methods. m is an integer greater than 1. The m first numerical intervals are merged to obtain the second numerical interval; In response to a resource transfer request for the first object, the resource transfer request is processed based on the second numerical range.

2. The method according to claim 1, characterized in that, The historical data includes data from n first statistical dimensions, where n is an integer greater than 1; the historical data is processed using m interval prediction models, each according to its corresponding interval prediction method, to obtain m first numerical intervals, including: Using the interval prediction model, the data of each first statistical dimension is processed according to the corresponding interval prediction method to obtain the fifth numerical interval corresponding to each first statistical dimension; By using the interval prediction model, the fifth numerical intervals corresponding to the n first statistical dimensions are fused to obtain the first numerical interval corresponding to the interval prediction model.

3. The method according to claim 1, characterized in that, The historical data is processed using m interval prediction models, each according to its corresponding interval prediction method, to obtain m first numerical intervals, including: For each interval prediction model, the historical data is processed according to the corresponding interval prediction method to obtain the first numerical intervals corresponding to p second statistical dimensions, where p is an integer greater than 1. The process of fusing the m first numerical intervals to obtain the second numerical interval includes: The m interval prediction models are fused together to obtain p second numerical intervals corresponding to the same second statistical dimension.

4. The method according to any one of claims 1 to 3, characterized in that, The m interval prediction models include a first prediction model; before processing the historical data using the m interval prediction models according to their corresponding interval prediction methods to obtain m first numerical intervals, the method further includes: The historical data is divided into q first data groups with a preset time period as the cycle. Each first data group includes data within a historical cycle, where q is an integer greater than 1. The historical data is processed using m interval prediction models, each according to its corresponding interval prediction method, to obtain m first numerical intervals, including: Using the first prediction model, feature extraction is performed on each first data group to obtain the data features corresponding to each first data group; The data features corresponding to each first data group are transformed using the first prediction model to obtain a first value corresponding to each first data group. The first value is the resource transfer value corresponding to the resource transfer behavior of the first object within the target time period. Based on the first values ​​corresponding to the q first data groups, the first value range corresponding to the first prediction model is determined using the first prediction model.

5. The method according to claim 4, characterized in that, The step of determining the first numerical range corresponding to the first prediction model based on the first values ​​corresponding to the q first data groups using the first prediction model includes: Using the first prediction model, the first values ​​corresponding to the q first data groups are sorted, and the first value in the preset position and the minimum value among the first values ​​corresponding to the q first data groups are respectively determined as the maximum and minimum values ​​of the first value interval.

6. The method according to any one of claims 1 to 3, characterized in that, The m interval prediction models include a second prediction model; the historical data includes data corresponding to multiple historical resource transfer behaviors; before processing the historical data using the m interval prediction models according to their respective interval prediction methods to obtain m first numerical intervals, the method further includes: The data corresponding to the multiple historical resource transfer behaviors are sorted according to the order of their occurrence time, either from first to last or from last to first, to obtain sequence information; The historical data is processed using m interval prediction models, each according to its corresponding interval prediction method, to obtain m first numerical intervals, including: The sequence information is processed using the second prediction model to obtain the first numerical range corresponding to the second prediction model.

7. The method according to any one of claims 1 to 3, characterized in that, The m interval prediction models include a third prediction model; before processing the historical data using the m interval prediction models according to their corresponding interval prediction methods to obtain m first numerical intervals, the method further includes: Based on the region to which the second object belongs, indicated by the historical resource transfer behavior, the historical data is divided into r second data groups. The resource transfer behavior has occurred between the second object and the first object. The objects in the same second data group belong to the same region. r is an integer greater than 1. The historical data is processed using m interval prediction models, each according to its corresponding interval prediction method, to obtain m first numerical intervals, including: Using the third prediction model, each second data group is processed according to the order of occurrence of the historical resource transfer behavior from first to last or from last to first, to obtain r third numerical intervals; Based on a numerical update strategy, r third numerical intervals are updated, wherein the numerical update strategy indicates that the numerical intervals are updated according to the level of the region; Based on the updated r third numerical intervals, the first numerical interval corresponding to the third prediction model is determined.

8. The method according to claim 7, characterized in that, The step of determining the first numerical interval corresponding to the third prediction model based on the updated r third numerical intervals includes: Based on a preset ratio, a second value is determined from each of the updated third value intervals. The preset ratio is the proportion of the value interval corresponding to the second value to the updated third value interval. The value interval corresponding to the second value is the value interval between the second value and the minimum value in the updated third value interval. The minimum value among the updated r third numerical intervals and the maximum value among the r second numerical intervals are respectively determined as the minimum value and maximum value of the first numerical interval.

9. The method according to any one of claims 1 to 3, characterized in that, The m interval prediction models include a fourth prediction model; the historical data includes resource transfer data of s third objects, where the s third objects are objects associated with the first object, and s is an integer greater than 1; the historical data is processed using the m interval prediction models according to their respective interval prediction methods to obtain m first numerical intervals, including: The resource transfer data of each third object is processed using the fourth prediction model to obtain the fourth numerical range corresponding to each third object. The fourth prediction model is used to fuse the fourth numerical intervals corresponding to the s third objects to obtain the first numerical interval corresponding to the fourth prediction model.

10. The method according to any one of claims 1 to 3, characterized in that, The step of processing the resource transfer request in response to the resource transfer request for the first object, based on the second numerical range, includes: In response to a resource transfer request for the first object, the resource transfer request is validated based on the second numerical range; If the resource transfer value indicated in the resource transfer request meets the second value range, the resource transfer request will be responded to. If the resource transfer value indicated in the resource transfer request does not meet the second value range, the resource transfer request will not be responded to.

11. A resource transfer device, characterized in that, The device includes: The acquisition module is used to acquire historical data of a first object, wherein the historical data is used to describe the historical resource transfer behavior associated with the first object; The processing module is used to process the historical data using m interval prediction models and according to the corresponding interval prediction methods to obtain m first numerical intervals. The first numerical interval is the resource transfer numerical interval corresponding to the resource transfer behavior of the first object within the target time period. The target time period is the time period after the current time. Different interval prediction models correspond to different interval prediction methods. m is an integer greater than 1. A fusion module is used to fuse the m first numerical intervals to obtain a second numerical interval; The processing module is further configured to process the resource transfer request based on the second numerical range in response to the resource transfer request for the first object.

12. A computer device, characterized in that, The computer device includes a processor and a memory, the memory storing at least one computer program, which is loaded and executed by the processor to perform the operations of the resource transfer method as described in any one of claims 1 to 10.

13. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores at least one computer program, which is loaded and executed by a processor to perform the operations performed by the resource transfer method as described in any one of claims 1 to 10.

14. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it performs the operations of the resource transfer method as described in any one of claims 1 to 10.

Images