Resource request result prediction method and device, electronic equipment and storage medium

Abstract

The invention provides a resource request result prediction method and device, electronic equipment and a storage medium, and the method comprises the steps: building a sparse matrix according to at least two historical resource requests; predicting and filling element null values in the sparse matrix to obtain a filling matrix; obtaining an element value corresponding to a to-be-tested resource request in the filling matrix, and taking the element value as a prediction request value; and determining a request result of the to-be-tested resource request according to the prediction request value and a preset threshold value. Therefore, the resource request result is automatically predicted by utilizing the existing historical resource request and identifying the possible result of the unknown to-be-detected resource request, the dependence on manual intervention can be reduced, the cost can be reduced, and the prediction reliability and efficiency can be remarkably improved due to high accuracy.

Description

Technical Field

[0001] Embodiments of the present disclosure relate to the field of information processing technology, and in particular to a resource request result prediction method, device, electronic device, and storage medium. Background Art

[0002] For institutions that provide resource services, some institutions will directly provide the corresponding resources, and they are called resource providers; while some institutions will forward resource requests to other institutions, and they are called resource forwarders. Regardless of whether they are resource providers or resource forwarders, before providing resources, they need to verify the resource requester who issued the resource request, and only provide or forward resources after the verification is passed; if the verification fails, they will refuse to provide or forward resources.

[0003] As a verification agency that provides resource request verification services, it may receive verification requests from resource providers or resource forwarders for the same resource requester. Currently, the verification agency mainly distinguishes whether the institution that issues the verification request is the resource provider or the resource forwarder through manual work, which has low efficiency and accuracy, and high labor costs. Summary of the invention

[0004] The embodiments of the present disclosure provide a resource request result prediction method, device, electronic device and storage medium.

[0005] In a first aspect, an embodiment of the present disclosure provides a resource request result prediction method, the method comprising:

[0006] Establishing a sparse matrix according to at least two historical resource requests, wherein a first dimension and a second dimension of the sparse matrix respectively correspond to a requester and a requested party included in each of the historical resource requests, and elements in the sparse matrix are used to represent request results in the historical resource requests;

[0007] Predicting and filling empty values ​​of elements in the sparse matrix to obtain a filled matrix;

[0008] Obtaining the value of the element corresponding to the resource request to be tested in the filling matrix as the predicted request value;

[0009] The request result of the resource request to be tested is determined according to the predicted request value and a preset threshold.

[0010] In some optional implementations, the step of establishing a sparse matrix according to at least two historical resource requests includes:

[0011] Acquire at least two historical resource requests, wherein the historical resource requests include a requester, a requested party, and a request result, and the request result is used to represent the result of the requester requesting the resource from the requested party;

[0012] Generate an initial matrix based on the at least two historical resource requests, wherein the first dimension and the second dimension of the initial matrix correspond to the requesting party and the requested party included in each of the historical resource requests, respectively, and the element values ​​of the initial matrix are all null values;

[0013] For each historical resource request, do the following:

[0014] For the target elements of the requester and the requested party corresponding to the historical resource request in the initial matrix, if the request result of the historical resource request indicates providing resources, the target element is set to the first parameter, and if the request result of the historical resource request indicates forwarding the resource request, the target element is set to the second parameter.

[0015] In some optional implementations, the predicting and filling of the empty values ​​of the elements in the sparse matrix to obtain a filled matrix includes:

[0016] Decomposing the sparse matrix into two decomposition matrices;

[0017] Based on the two decomposed matrices, a dense matrix is ​​obtained, and the dense matrix and the sparse matrix are the same type of matrices;

[0018] The empty values ​​of elements in the sparse matrix are replaced with the values ​​of elements at corresponding positions in the dense matrix to obtain a filled matrix.

[0019] In some optional implementations, the above decomposing the sparse matrix into two decomposition matrices includes:

[0020] Decomposing the sparse matrix into a first decomposition matrix and a second decomposition matrix, wherein the first dimension of the sparse matrix is ​​the first dimension of the first decomposition matrix, the second dimension of the sparse matrix is ​​the second dimension of the second decomposition matrix, and the second dimension of the first decomposition matrix and the first dimension of the second decomposition matrix are preset hidden layer dimensions;

[0021] Calculate a loss function between a dot product result of the first decomposition matrix and the second decomposition matrix and the sparse matrix;

[0022] The first decomposition matrix and / or the second decomposition matrix are updated according to the loss function.

[0023] In some optional implementations, the obtaining of the value corresponding to the resource request to be tested in the filling matrix as the predicted request value includes:

[0024] If both the requester and the requested party of the resource request to be tested have appeared in the historical resource requests, then obtaining corresponding values ​​of the requester and the requested party of the resource request to be tested in the filling matrix as predicted request values;

[0025] If at least one of the requester and the requested party of the resource request to be tested has not appeared in the historical resource requests, obtain the requester in the historical resource requests with the highest similarity to the requester of the resource request to be tested as the replacement requester; obtain the corresponding values ​​of the replacement requester and the requested party of the resource request to be tested in the filling matrix as the predicted request values.

[0026] In some optional implementations, the obtaining, as the replacement requester, the requester with the highest similarity to the requester of the resource request to be tested in the historical resource requests, includes:

[0027] Obtaining a behavior sequence of each requester in the resource request to be tested and the historical resource request;

[0028] Calculating the similarity between the behavior sequence of the requester of the resource request to be tested and the behavior sequence of the requester in each historical resource request;

[0029] The requester of the historical resource request whose behavior sequence is most similar to that of the requester of the resource request to be tested is determined as the replacement requester.

[0030] In some optional implementations, determining the request result of the resource request to be tested according to the predicted request value and the preset threshold value includes:

[0031] If the predicted request value is greater than a first preset threshold, determining that the request result of the resource request to be tested is to provide resources;

[0032] If the predicted request value is less than a second preset threshold, it is determined that the request result of the resource request to be tested is a forwarding resource request.

[0033] In a second aspect, an embodiment of the present disclosure provides a resource request result prediction device, the device comprising:

[0034] A sparse matrix establishment module, used to establish a sparse matrix according to at least two historical resource requests, wherein the first dimension and the second dimension of the sparse matrix respectively correspond to the requesting party and the requested party included in each of the historical resource requests, and the elements in the sparse matrix are used to represent the request results in the historical resource requests;

[0035] A prediction and filling module, used for predicting and filling the empty values ​​of elements in the sparse matrix to obtain a filling matrix;

[0036] A predicted value acquisition module, used to obtain the element value corresponding to the resource request to be tested in the filling matrix as the predicted request value;

[0037] The prediction module is used to determine the request result of the resource request to be tested according to the predicted request value and a preset threshold.

[0038] In a third aspect, an embodiment of the present disclosure provides an electronic device, comprising: one or more processors; a storage device on which one or more programs are stored, and when the one or more programs are executed by the one or more processors, the one or more processors implement the method described in any implementation manner in the first aspect.

[0039] In a fourth aspect, an embodiment of the present disclosure provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by one or more processors, implements the method described in any implementation manner in the first aspect.

[0040] In a fifth aspect, an embodiment of the present disclosure provides a computer program product, including a computer program / instruction, which, when executed by a processor, implements the method described in any implementation manner in the first aspect.

[0041] In order to improve the efficiency and accuracy of predicting resource request results, the embodiments of the present invention provide a resource request result prediction method, device, electronic device and storage medium. A sparse matrix is ​​first established based on at least two historical resource requests, wherein the first dimension and the second dimension of the sparse matrix respectively correspond to the requester and the requested party included in each of the historical resource requests, and the elements in the sparse matrix are used to characterize the request results in the historical resource requests; then, the empty values ​​of the elements in the sparse matrix are predicted and filled to obtain a filled matrix; then, the element value corresponding to the resource request to be tested in the filled matrix is ​​obtained as the predicted request value; finally, the request result of the resource request to be tested is determined according to the predicted request value and a preset threshold.

[0042] In this way, by utilizing existing historical resource requests, identifying possible results of unknown resource requests to be tested, and realizing automatic prediction of resource request results, it can not only reduce dependence on manual intervention and reduce costs, but also significantly improve the reliability and efficiency of predictions due to its high accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Other features, objects and advantages of the present disclosure will become more apparent by reading the detailed description of non-limiting embodiments made with reference to the following drawings. The drawings are only for the purpose of illustrating specific embodiments and are not to be considered as limiting the present disclosure. In the drawings:

[0044] Figure 1 is a scene graph to which an embodiment of the present disclosure may be applied;

[0045] Figure 2 is an exemplary system architecture diagram in which an embodiment of the present disclosure may be applied;

[0046] Figure 3A is a flow chart of an embodiment of a resource request result prediction method according to the present disclosure;

[0047] Figure 3B is a flowchart of step 302 of an embodiment of a resource request result prediction method according to the present disclosure;

[0048] Figure 4 is a structural diagram of an embodiment of a resource request result prediction device according to the present disclosure;

[0049] Figure 5 It is a structural diagram of a computer system of an electronic device suitable for implementing an embodiment of the present disclosure. DETAILED DESCRIPTION

[0050] The present disclosure is further described in detail below in conjunction with the accompanying drawings and embodiments. It is understood that the specific embodiments described herein are only used to explain the relevant invention, rather than to limit the invention. It is also necessary to explain that, for ease of description, only the parts related to the relevant invention are shown in the accompanying drawings.

[0051] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present disclosure can be combined with each other. The information, data and signals involved in the present disclosure are all authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data comply with the relevant laws, regulations and standards of relevant countries and regions. The present disclosure will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0052] Figure 1 FIG. 2 shows a scene graph in which an embodiment of the present disclosure can be applied. Figure 1 As shown, the resource requester submits a resource request application to the requested party A, and A then contacts and verifies the relevant conditions of the requester D to verify whether they meet the requirements.

[0053] Once the verification is successful, A may transfer the requesting party's application materials to the requested party B, and this process may continue until the requested party C finally provides resources to the requesting party. In this chain, the verification agency D will receive verification requests from multiple requested parties.

[0054] For the verification agency D, it is usually a challenge to accurately distinguish which requests are initiated by the requested party that forwards the resource request (i.e., the resource forwarder) and which are initiated by the requested party that actually provides the resources (i.e., the resource provider).

[0055] At present, verification agency D usually roughly infers which requested parties are actually providing resources by manually analyzing the historical average number of applications of each requested party, the sudden surge in the number of applications within a certain period of time, and the order of the application time of each requested party. This judgment based on manual business rules is very inefficient and inaccurate.

[0056] For example, in the financial field, the resource requester, i.e. the client, may need to submit a loan application to a bank or other financial institution (requested party A). When processing loan applications, these financial institutions will require clients to provide a series of financial and personal information for credit assessment. At this point, the bank or financial institution (requested party A) will contact a credit assessment agency (verification agency D) to verify the client's credit record and repayment ability. If the credit assessment results meet the loan conditions, requested party A may transfer the client's application materials to a partner (requested party B), such as other banks or financial institutions. The above verification process is repeated until the bank or financial institution (requested party C) finally grants the client a loan.

[0057] In the field of data sharing, the resource requester, i.e. the application developer, may need to submit an application for data acquisition to the data service provider (requestee A) to enrich the functions of their application. Before obtaining the data, they need to ensure that they have the right to access the data, which usually involves keys, tokens or other forms of authorization. The verification agency D, which may be a dedicated authorization server or management platform, is responsible for verifying the authorization of the application and ensuring that they comply with data usage rules and privacy rules. Once the verification is passed, the requested party A may transfer the application developer's application for data acquisition to other data service providers (requestees B and C). The above verification process is repeated until the data service provider (requestee C) finally provides the required data to the application developer.

[0058] In addition, the embodiments of the present disclosure may also be applied to other fields, which are not limited here.

[0059] Figure 2 An exemplary system architecture 200 is shown to which embodiments of the resource request result prediction method, apparatus, electronic device, and storage medium of the present disclosure can be applied.

[0060] like Figure 2As shown, the system architecture 200 may include terminal devices 201, 202, 203, a network 204 and a server 205. The network 204 is used to provide a medium for communication links between the terminal devices 201, 202, 203 and the server 205. The network 204 may include various connection types, such as wired, wireless communication links or optical fiber cables, etc.

[0061] The user can use the terminal devices 201, 202, 203 to interact with the server 205 through the network 204 to receive or send messages, etc. Various communication client applications can be installed on the terminal devices 201, 202, 203, such as financial applications, voice recognition applications, short video social applications, audio and video conferencing applications, video live broadcast applications, document editing applications, input method applications, web browser applications, shopping applications, search applications, instant messaging tools, email clients, social platform software, etc.

[0062] Terminal devices 201, 202, and 203 can be hardware or software. When terminal devices 201, 202, and 203 are hardware, they can be various electronic devices with display screens, including but not limited to smart phones, tablet computers, e-book readers, MP3 players (Moving Picture Experts Group Audio Layer III, Moving Picture Experts Compression Standard Audio Layer 3), MP4 (Moving Picture Experts Group Audio Layer IV, Moving Picture Experts Compression Standard Audio Layer 4) players, laptop computers, and desktop computers, etc. When terminal devices 201, 202, and 203 are software, they can be installed in the terminal devices listed above. It can be implemented as multiple software or software modules (for example, used to provide resource request result prediction services), or it can be implemented as a single software or software module. No specific limitation is made here.

[0063] In some cases, the resource request result prediction method provided by the present disclosure may be executed by the terminal devices 201, 202, 203, and accordingly, the resource request result prediction apparatus may be disposed in the terminal devices 201, 202, 203. In this case, the system architecture 200 may not include the server 205.

[0064] In some cases, the resource request result prediction method provided by the present disclosure can be jointly performed by the terminal devices 201, 202, 203 and the server 205. For example, the step of "establishing a sparse matrix based on at least two historical resource requests" can be performed by the terminal devices 201, 202, 203, and the steps of "predicting and filling the element null values ​​in the sparse matrix to obtain a filled matrix" can be performed by the server 205. The present disclosure does not limit this. Accordingly, the resource request result prediction device can also be respectively set in the terminal devices 201, 202, 203 and the server 205.

[0065] In some cases, the resource request result prediction method provided in the present disclosure may be executed by the server 205 , and accordingly, the resource request result prediction device may also be set in the server 205 . In this case, the system architecture 200 may not include the terminal devices 201 , 202 , and 203 .

[0066] It should be noted that the server 205 can be hardware or software. When the server 205 is hardware, it can be implemented as a distributed server cluster consisting of multiple servers, or it can be implemented as a single server. When the server 205 is software, it can be implemented as multiple software or software modules (for example, for providing distributed services), or it can be implemented as a single software or software module. No specific limitation is made here.

[0067] It should be understood that Figure 2 The number of terminal devices, networks and servers in the embodiment is only for illustration. Any number of terminal devices, networks and servers may be provided according to implementation requirements.

[0068] Continue to refer Figure 3A , which shows a process 300 of an embodiment of a resource request result prediction method according to the present disclosure, the resource request result prediction method comprises the following steps:

[0069] Step 301: Create a sparse matrix based on at least two historical resource requests.

[0070] Resource request means that the requesting party requests certain resources from the requested party, and the request result is used to characterize the result of the resource request. Both the requesting party and the requested party can be devices corresponding to various subjects such as individuals, companies, legal persons, institutions or organizations. In the embodiments of the present disclosure, various subjects are represented by subject identifiers, and subject identifiers include but are not limited to identifiers composed of strings and identifiers generated by subject information; the device corresponding to the subject can be the device logged in or used by the subject identifier corresponding to the subject. The request result can be flexibly set according to the content of the resource request, such as forwarding resources, providing resources, rejecting requests, accepting requests, etc.

[0071] Historical resource requests are known resource requests with a corresponding relationship of "requester-requestee-request result". For example, requester Zhang San requests resources from requested party A, and the request result is that requested party A forwards the resource request to requested party B.

[0072] For ease of understanding, see Table 1 below. Table 1 represents multiple historical resource requests in a table format, where each row represents a historical resource request:

[0073] Table 1

[0074] Requester Requested Party Request results Zhang San A Forward resource request to B Zhang San B Provide resources Zhang San C Forward resource request to B Li Si A Forward resource request to B Li Si B Provide resources Wang Wu B Forward resource request to C

[0075] In some example scenarios, such as in the financial field, a historical resource request can be an institution that verifies the credit status of a customer to obtain the actual loan records of some customers. The requester is the customer, the requested party is the institution that requests to verify the customer's credit status, and the request result is "loan" or "forward credit request". For another example, in the data sharing field, a historical resource request can be an authorization server to obtain the data provision record of a service provider. The requester is an application developer, the requested party is a service provider, and the request result is "provide data" or "forward data acquisition application"

[0076] According to the at least two historical resource requests, a sparse matrix may be established. The sparse matrix may be a two-dimensional matrix, wherein the first dimension corresponds to the requester and the second dimension corresponds to the requested party, and the elements in the sparse matrix are used to represent the request result.

[0077] As a possible implementation, the process of establishing a sparse matrix in step 301 may include the following steps 3011 to 3013:

[0078] Step 3011, obtaining at least two historical resource requests.

[0079] The historical resource request includes a requester, a requested party and a request result, and the request result is used to represent the result of the requester requesting the resource from the requested party.

[0080] Step 3012: Generate an initial matrix based on the at least two historical resource requests.

[0081] The first dimension and the second dimension of the initial matrix correspond to the requester and the requested party included in each historical resource request, and the elements of the initial matrix are all null values. For example, to obtain the requesters "Zhang San", "Li Si", and "Wang Wu" of all historical resource requests in Table 1; all requested parties "A", "B", and "C", a 3×3 initial matrix with all values ​​​​taken as 0 is established.

[0082] Step 3013: for each historical resource request, perform the following operations:

[0083] For the target elements of the requester and the requested party corresponding to the historical resource request in the initial matrix, if the request result of the historical resource request indicates providing resources, the target element is set to the first parameter; if the request result of the historical resource request indicates forwarding the resource request, the target element is set to the second parameter.

[0084] The first parameter and the second parameter can be set as needed and are not limited here.

[0085] For example, if a sparse matrix is ​​established based on the historical resource requests in Table 1, the sparse matrix may be as shown in Table 2 below:

[0086] Table 2

[0087] A B C Zhang San 0 1 0 Li Si 0 1 NaN Wang Wu NaN 0 NaN

[0088] The sparse matrix in the above table is a 3×3 matrix, with the first dimension corresponding to multiple requesters and the second dimension corresponding to multiple requested parties. Each element represents the request result of the requester-requestee corresponding to the row and column. If the element value is 0, it means that the request result is a forwarding resource request; if the element value is 1, it means that the request result is to provide resources; if the element value is NaN, it represents a null value, that is, the historical resource request does not record the request result of the requester-requestee corresponding to the element.

[0089] In summary, according to steps 3011 to 3013, a sparse matrix can be automatically established according to at least two historical resource requests, thereby improving the efficiency of matrix generation.

[0090] Step 302, predict and fill in the empty values ​​of the elements in the sparse matrix to obtain a filled matrix.

[0091] There are many ways to predict and fill in empty values ​​of elements in sparse matrices, including simple filling (such as filling with mean, median, mode), model-based filling (such as linear regression, random forest), proximity-based filling (such as filling using k-nearest neighbor algorithm), time series methods (such as filling using recurrent neural networks, long short-term memory networks), matrix decomposition (such as alternating least squares (ALS) matrix decomposition), graph neural networks, Bayesian interpolation, interpolation method and machine learning-based interpolation. You can choose the appropriate method based on data characteristics and business needs.

[0092] As a possible implementation, the process of predicting and filling the empty values ​​of elements in the sparse matrix in step 302 may include the following steps: Figure 3B Steps 3021 to 3023 shown:

[0093] Step 3021, decompose the sparse matrix into two decomposition matrices.

[0094] Matrix decomposition is the process of decomposing a matrix into two or more smaller matrices, the product of which can approximate or completely reconstruct the original matrix. Using matrix decomposition technology, a sparse matrix can be decomposed into two decomposition matrices in order to reduce the dimension of the original data while retaining the main features of the data.

[0095] Since the implementation of the above step 3021 directly affects the accuracy of the entire prediction, two more accurate decomposition matrices can be obtained through multiple iterations of the following steps 30211 to 30213:

[0096] Step 30211, decompose the sparse matrix into a first decomposition matrix and a second decomposition matrix.

[0097] Among them, the first dimension of the sparse matrix is ​​the first dimension of the first decomposition matrix, the second dimension of the sparse matrix is ​​the second dimension of the second decomposition matrix, and the second dimension of the first decomposition matrix and the first dimension of the second decomposition matrix are the preset hidden layer dimensions. In other words, the sparse matrix M with dimension a×b is decomposed into decomposition matrix A with dimension a×k and decomposition matrix B with dimension k×b. Among them, k is the preset hidden layer dimension, and the value can be set as needed, generally between 2-10. The larger k is, the closer the reconstructed matrix is ​​to the original matrix, and the greater the amount of calculation.

[0098] Step 30212, calculate the loss function between the dot product result of the first decomposition matrix and the second decomposition matrix and the sparse matrix.

[0099] The calculation method of the loss function is not limited here, and any loss function such as cosine similarity loss, Manhattan distance loss, Wasserstein distance loss, etc. can be used.

[0100] Step 30213, update the first decomposition matrix and / or the second decomposition matrix according to the loss function.

[0101] When the first decomposition matrix and / or the second decomposition matrix are updated according to the loss function, the first decomposition matrix and the second decomposition matrix can be updated alternately by the alternating least squares method until n rounds of iteration or iteration until the loss function converges, and the final first decomposition matrix and the second decomposition matrix are obtained. It should be noted that the larger n is, the higher the similarity between the reconstructed matrix and the original matrix, the higher the accuracy of the subsequent prediction, the more decimal places the predicted value has, and the greater the amount of calculation.

[0102] In summary, by executing steps 30211 to 30213, the alternating least squares (ALS) matrix decomposition algorithm is used, and after multiple iterative decompositions, two more accurate decomposition matrices are obtained, which is beneficial for the subsequent steps to obtain a dense matrix that is close to the original matrix and incorporates some of the new data.

[0103] Step 3022, obtaining a dense matrix based on the two decomposed matrices.

[0104] The process of obtaining a dense matrix based on two decomposed matrices can be a process of multiplying the two decomposed matrices, so that the obtained dense matrix and the sparse matrix are the same type of matrices.

[0105] Step 3023, replace the empty values ​​of elements in the sparse matrix with the values ​​of elements at corresponding positions in the dense matrix to obtain a filled matrix.

[0106] For example, the sparse matrix before decomposition in step 3021 is shown in Table 3 below:

[0107] Table 3

[0108] Requested Party 1 Requested Party 2 Requested Party 3 Requested Party 4 Requested Party 5 Requested Party N Requester 1 1.0 0.0 0.0 NaN 0.0 ... Requester 2 0.0 NaN NaN 1.0 NaN ... Requester 3 0.0 0.0 NaN NaN 1.0 ... Requester 4 NaN NaN 1.0 NaN NaN ... Requester 5 NaN 1.0 NaN NaN NaN ... Requester N ... ... ... ... ... ...

[0109] After step 3021 to step 3023, a filling matrix as shown in Table 4 below can be obtained. The null value NaN in Table 3 is replaced by the value in the dense matrix:

[0110] Table 4

[0111] Requested Party 1 Requested Party 2 Requested Party 3 Requested Party 4 Requested Party 5 Requested Party N Requester 1 1.0 0 0 0.749858 0 ... Requester 2 0.0 0.792915 0.494838 1 0.789392 ... Requester 3 0.0 0 0.621670 0.783287 1 ... Requester 4 0.709580 0.607542 1 0.493709 0.625085 ... Requester 5 0.847463 1 0.602621 0.784698 0.854489 ... Requester N ... ... ... ... ... ...

[0112] After decomposition and reconstruction, the dense matrix is ​​highly similar to the sparse matrix. Therefore, this process can also be used to predict the empty values ​​in the original sparse matrix, replacing the empty values ​​of the elements in the sparse matrix with the values ​​of the elements in the corresponding positions in the dense matrix. It should be noted that the parts of the original sparse matrix that have been assigned values ​​will not be filled.

[0113] In summary, through steps 3021 to 3023, the prediction and filling of empty values ​​of elements in the sparse matrix are realized, with high computational efficiency, good data stability, and convenience for parallel computing. At the same time, it is beneficial to the subsequent prediction of the request results based on the filled values.

[0114] Step 303: Obtain the element value corresponding to the resource request to be tested in the filling matrix as the predicted request value.

[0115] The resource request to be tested refers to the resource request for which the request result needs to be predicted, which usually only includes the requester and the requested party. When the requester and the requested party are known, the corresponding element can be found in the filling matrix, and its value is the predicted request value, which is used to predict the request result of the resource request to be tested based on its value.

[0116] It should be noted that whether the element value related to the resource request to be tested can be found in the filling matrix depends on whether the requester and the requested party have both appeared in the historical resource request. If at least one of the requester or the requested party of the resource request to be tested has never appeared in the historical resource request, the corresponding element value cannot be determined. Therefore, in a possible implementation, the above step 303 may include the following steps 3031 to 3033:

[0117] Step 3031: If the requester and the requested party of the resource request to be tested have both appeared in historical resource requests, the corresponding values ​​of the requester and the requested party of the resource request to be tested in the filling matrix are obtained as predicted request values.

[0118] Step 3032: If at least one of the requester and the requested party of the resource request to be tested has not appeared in the historical resource requests, the requester in the historical resource requests with the highest similarity to the requester of the resource request to be tested is obtained as a replacement requester.

[0119] In the historical resource requests, when searching for the requester with the highest similarity to the requester of the resource request to be tested, first, it is necessary to obtain the behavior sequence of each requester in the resource request to be tested and the historical resource requests. The behavior sequence usually comes from the interaction data between the requester and the requested party. This data can be obtained through log files, databases, or specific data collection tools, and then useful features are extracted from them to form a behavior sequence. These features can characterize the behavior patterns, timestamps, behavior types, etc. of the requester and the requested party. The method of obtaining the behavior sequence is not limited here.

[0120] Next, the similarity between the requester behavior sequence of the resource request to be tested and each requester behavior sequence in the historical resource request is calculated. When calculating the similarity of the behavior sequence, a variety of measurement methods can be used, including but not limited to cosine similarity (measuring the similarity by calculating the cosine value of the angle between two vectors) and dynamic time warping (measuring the similarity between two time series, allowing the time series to be stretched and moved on the time axis).

[0121] When using cosine similarity calculation, the similarity calculation formula can be used as follows:

[0122]

[0123] Among them, s represents the similarity, and the value range is [-1,1]; X i represents the i-th element in the requester behavior sequence of the resource request to be tested; Y i It represents the i-th element in the requester behavior sequence in the historical resource request of this calculation; n is the number of elements in the requester behavior sequence of the resource request to be tested or in the requester behavior sequence in the historical resource request of this calculation.

[0124] Finally, the requester of the historical resource request whose behavior sequence is most similar to the requester of the resource request to be tested is determined as the replacement requester. This is equivalent to combining the idea of ​​collaborative filtering and using the similarity between requesters to predict the user request results, which can provide more accurate prediction results.

[0125] Step 3033: Obtain corresponding values ​​of the replacement requester and the requested party of the resource request to be tested in the filling matrix as predicted request values.

[0126] The replacement requester, as the requester with the most similar behavior sequence to the requester of the resource request to be tested, can better represent its resource request situation. By taking the corresponding values ​​of the replacement requester and the requested party of the resource request to be tested in the filling matrix as the predicted request value, a more accurate predicted request result can be obtained.

[0127] In summary, through steps 2031 to 3033, even if at least one of the requester and the requested party has not appeared in historical resource requests, its request result can be predicted, making the application scenarios of this embodiment more diverse and more flexible.

[0128] Step 304: Determine the request result of the resource request to be tested according to the predicted request value and the preset threshold.

[0129] The preset threshold can be set based on experience and is not limited here. Optionally, two preset thresholds can be set, namely a first preset threshold and a second preset threshold. If the predicted request value is greater than the first preset threshold, the request result of the resource request to be tested is determined to be providing resources; if the predicted request value is less than the second preset threshold, the request result of the resource request to be tested is determined to be forwarding resource request.

[0130] It should be noted that, in order to ensure the accuracy of the prediction, the first preset threshold and the second preset threshold may be equal or different. When the first preset threshold and the second preset threshold are different, when the predicted request value is between the two, it may be impossible to predict the request result based on the predicted request value and the preset threshold. At this time, an indication that the request result of the resource request to be tested cannot be determined can be output. In this way, it can be ensured that the accuracy of the obtained request result is significantly improved, thereby improving the reliability of the request result.

[0131] For example, x is the predicted request value. If x<0.7, it is determined that the request result of the resource request to be tested is a forwarding resource request. If 0.7≤x≤0.8, since the prediction accuracy cannot be guaranteed, an indication that the request result of the resource request to be tested cannot be determined is output. If x>0.8, it is determined that the request result of the resource request to be tested is to provide resources.

[0132] The resource request result prediction method provided by the above-mentioned embodiment of the present disclosure can automatically identify the request result of the resource request to be tested, thereby improving the recognition efficiency and accuracy.

[0133] In the financial field, compared with manually analyzing data on loans provided by banks or other financial institutions to customers to determine whether an unknown institution is a real lending institution, the resource request result prediction method provided in the above embodiment can efficiently identify traffic diversion behavior, realize machine automation operation, reduce labor costs, and greatly improve the accuracy of traffic diversion identification.

[0134] In the field of data sharing, compared with manually analyzing the service provider's data provision records to determine whether the service provider directly provides data, the resource request result prediction method provided in the above embodiment can efficiently determine the service provider's data provision capability and achieve automatic identification with high accuracy.

[0135] Further references Figure 4 As an implementation of the methods shown in the above figures, the present disclosure provides an embodiment of a resource request result prediction device, and the device embodiment is Figure 3A Corresponding to the method embodiment shown, the device can be specifically applied to various electronic devices.

[0136] like Figure 4 As shown, the resource request result prediction device 400 of this embodiment includes: a sparse matrix establishment module 401, a prediction filling module 402, a prediction value acquisition module 403 and a prediction module 404.

[0137] The sparse matrix establishment module 401 is used to establish a sparse matrix according to at least two historical resource requests, wherein the first dimension and the second dimension of the sparse matrix respectively correspond to the requesting party and the requested party included in each of the historical resource requests, and the elements in the sparse matrix are used to represent the request results in the historical resource requests;

[0138] A prediction and filling module 402 is used to predict and fill in the empty values ​​of the elements in the sparse matrix to obtain a filling matrix;

[0139] The predicted value acquisition module 403 is used to obtain the element value corresponding to the resource request to be tested in the filling matrix as the predicted request value;

[0140] The prediction module 404 is used to determine the request result of the resource request to be tested according to the predicted request value and a preset threshold.

[0141] In this embodiment, the specific processing of the resource request result prediction device 400 and the technical effects thereof can be referred to in Figure 3A The relevant descriptions of step 301, step 302, step 303 and step 304 in the corresponding embodiment are not repeated here.

[0142] In some optional implementations, the sparse matrix establishment module 401 includes:

[0143] The acquisition unit 4011 is used to acquire at least two historical resource requests, wherein the historical resource requests include a requester, a requested party, and a request result, and the request result is used to represent the result of the requester requesting the resource from the requested party;

[0144] An initial establishment unit 4012 is configured to generate an initial matrix based on the at least two historical resource requests, wherein a first dimension and a second dimension of the initial matrix correspond to a requester and a requested party included in each of the historical resource requests, respectively, and elements of the initial matrix are all null values;

[0145] The assignment unit 4013 is used to perform the following operations for each historical resource request:

[0146] For the target elements of the requester and the requested party corresponding to the historical resource request in the initial matrix, if the request result of the historical resource request indicates providing resources, the target element is set to the first parameter, and if the request result of the historical resource request indicates forwarding the resource request, the target element is set to the second parameter.

[0147] In some optional implementations, the prediction filling module 402 includes:

[0148] A decomposition unit 4021, used for decomposing the sparse matrix into two decomposition matrices;

[0149] A dense unit 4022, configured to obtain a dense matrix based on the two decomposed matrices, wherein the dense matrix and the sparse matrix are of the same type;

[0150] The replacement unit 4023 is used to replace the empty values ​​of elements in the sparse matrix with the values ​​of elements at corresponding positions in the dense matrix to obtain a filled matrix.

[0151] In some optional embodiments, the decomposition unit 4021 includes:

[0152] a decomposition component (not shown in the figure), used for decomposing the sparse matrix into a first decomposition matrix and a second decomposition matrix, wherein the first dimension of the sparse matrix is ​​the first dimension of the first decomposition matrix, the second dimension of the sparse matrix is ​​the second dimension of the second decomposition matrix, and the second dimension of the first decomposition matrix and the first dimension of the second decomposition matrix are preset hidden layer dimensions;

[0153] A loss calculation component (not shown in the figure), used to calculate the loss function between the dot product result of the first decomposition matrix and the second decomposition matrix and the sparse matrix;

[0154] An iterative component (not shown in the figure) is used to update the first decomposition matrix and / or the second decomposition matrix according to the loss function.

[0155] In some optional implementations, the above-mentioned prediction value acquisition module 403 includes:

[0156] The first situation unit 4031 is used for obtaining the corresponding values ​​of the requester and the requested party of the resource request to be tested in the filling matrix as the predicted request value if both the requester and the requested party of the resource request to be tested have appeared in the historical resource request;

[0157] The second situation unit 4032 is used to obtain the requestor with the highest similarity with the requestor of the resource request to be tested in the historical resource requests as a replacement requestor if at least one of the requestor and the requested party of the resource request to be tested has not appeared in the historical resource requests; and obtain the corresponding values ​​of the replacement requestor and the requested party of the resource request to be tested in the filling matrix as predicted request values.

[0158] In some optional implementations, the second situation unit 4032 includes:

[0159] A behavior sequence acquisition component (not shown in the figure), used to acquire the behavior sequence of each requester in the resource request to be tested and the historical resource request;

[0160] A similarity calculation component (not shown in the figure), used to calculate the similarity between the behavior sequence of the requester of the resource request to be tested and the behavior sequence of the requester in each historical resource request;

[0161] The replacement determination component (not shown in the figure) is used to determine the requester of the historical resource request with the greatest similarity in behavior sequence with the requester of the resource request to be tested as the replacement requester.

[0162] In some optional implementations, the prediction module 404 includes:

[0163] A first prediction unit 4041 is configured to determine that the request result of the resource request to be tested is to provide resources if the predicted request value is greater than a first preset threshold;

[0164] The second prediction unit 4042 is configured to determine that the request result of the resource request to be tested is a forwarding resource request if the predicted request value is less than a second preset threshold.

[0165] It should be noted that the implementation details and technical effects of each module and unit in the resource request result prediction device provided in the embodiment of the present disclosure can be referred to the description of other embodiments in the present disclosure, and will not be repeated here.

[0166] Reference below Figure 5 , which shows a schematic diagram of the structure of a computer system 500 suitable for implementing the electronic device of the present disclosure. Figure 5 The computer system 500 shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present disclosure.

[0167] like Figure 5 As shown, the computer system 500 may include a processing device (e.g., a central processing unit, a graphics processing unit, etc.) 501, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 502 or a program loaded from a storage device 508 to a random access memory (RAM) 503. Various programs and data required for the operation of the computer system 500 are also stored in the RAM 503. The processing device 501, the ROM 502, and the RAM 503 are connected to each other via a bus 504. An input / output (I / O) interface 505 is also connected to the bus 504.

[0168] Typically, the following devices may be connected to the I / O interface 505: input devices 506 including, for example, a touch screen, a touch pad, a keyboard, a mouse, a camera, a microphone, etc.; output devices 507 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; storage devices 508 including, for example, a magnetic tape, a hard disk, etc.; and communication devices 509. The communication devices 509 may allow the computer system 500 to communicate with other devices wirelessly or by wire to exchange data. Although Figure 5 The computer system 500 of an electronic device having various devices is shown, but it should be understood that it is not required to implement or possess all the devices shown. More or fewer devices may be implemented or possessed instead.

[0169] In particular, according to an embodiment of the present disclosure, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present disclosure includes a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network through a communication device 509, or installed from a storage device 508, or installed from a ROM 502. When the computer program is executed by the processing device 501, the above-mentioned functions defined in the method of the embodiment of the present disclosure are executed.

[0170] It should be noted that the computer-readable medium disclosed above may be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present disclosure, a computer-readable storage medium may be any tangible medium containing or storing a program that may be used by or in combination with an instruction execution system, device or device. In the present disclosure, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, in which a computer-readable program code is carried. This propagated data signal may take a variety of forms, including but not limited to an electromagnetic signal, an optical signal, or any suitable combination of the above. The computer readable signal medium may also be any computer readable medium other than a computer readable storage medium, which may send, propagate or transmit a program for use by or in conjunction with an instruction execution system, apparatus or device. The program code contained on the computer readable medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (radio frequency), etc., or any suitable combination of the above.

[0171] The computer-readable medium may be included in the electronic device, or may exist independently without being incorporated into the electronic device.

[0172] The computer readable medium carries one or more programs. When the one or more programs are executed by the electronic device, the electronic device implements the following Figure 3AThe illustrated embodiment and its optional implementations illustrate a method for predicting resource request results.

[0173] Computer program code for performing the operations of the present disclosure may be written in one or more programming languages, or a combination thereof, including object-oriented programming languages, such as Java, Smalltalk, C++, and conventional procedural programming languages, such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., through the Internet using an Internet service provider).

[0174] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present disclosure. In this regard, each square box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some implementations as replacements, the functions marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two square boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0175] The units or modules involved in the embodiments described in the present disclosure may be implemented by software or hardware. The name of a unit or module does not limit the unit itself in some cases. For example, a prediction module may also be described as "used to predict and fill in the null values ​​of the elements in the sparse matrix to obtain a filled matrix".

[0176] The above description is only a preferred embodiment of the present disclosure and an explanation of the technical principles used. Those skilled in the art should understand that the scope of disclosure involved in the present disclosure is not limited to the technical solutions formed by a specific combination of the above technical features, but should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the above disclosed concept. For example, the above features are replaced with the technical features with similar functions disclosed in the present disclosure (but not limited to) by each other to form a technical solution.

Claims

1. A resource request result prediction method, characterized in that: include: Establishing a sparse matrix according to at least two historical resource requests, wherein a first dimension and a second dimension of the sparse matrix respectively correspond to a requester and a requested party included in each of the historical resource requests, and elements in the sparse matrix are used to represent request results in the historical resource requests; Predicting and filling empty values ​​of elements in the sparse matrix to obtain a filled matrix; Obtaining the value of the element corresponding to the resource request to be tested in the filling matrix as the predicted request value; The request result of the resource request to be tested is determined according to the predicted request value and a preset threshold.

2. The method according to claim 1, characterized in that The step of establishing a sparse matrix according to at least two historical resource requests includes: Acquire at least two historical resource requests, wherein the historical resource requests include a requester, a requested party, and a request result, and the request result is used to represent the result of the requester requesting the resource from the requested party; Generate an initial matrix based on the at least two historical resource requests, wherein the first dimension and the second dimension of the initial matrix correspond to the requesting party and the requested party included in each of the historical resource requests, respectively, and the element values ​​of the initial matrix are all null values; For each historical resource request, do the following: For the target elements of the requester and the requested party corresponding to the historical resource request in the initial matrix, if the request result of the historical resource request indicates providing resources, the target element is set to the first parameter, and if the request result of the historical resource request indicates forwarding the resource request, the target element is set to the second parameter.

3. The method according to claim 1, characterized in that: The predicting and filling the empty values ​​of the elements in the sparse matrix to obtain a filled matrix includes: Decomposing the sparse matrix into two decomposition matrices; Based on the two decomposed matrices, a dense matrix is ​​obtained, and the dense matrix and the sparse matrix are the same type of matrices; The empty values ​​of elements in the sparse matrix are replaced with the values ​​of elements at corresponding positions in the dense matrix to obtain a filled matrix.

4. The method according to claim 3, characterized in that Decomposing the sparse matrix into two decomposition matrices comprises: Decomposing the sparse matrix into a first decomposition matrix and a second decomposition matrix, wherein the first dimension of the sparse matrix is ​​the first dimension of the first decomposition matrix, the second dimension of the sparse matrix is ​​the second dimension of the second decomposition matrix, and the second dimension of the first decomposition matrix and the first dimension of the second decomposition matrix are preset hidden layer dimensions; Calculate a loss function between a dot product result of the first decomposition matrix and the second decomposition matrix and the sparse matrix; The first decomposition matrix and / or the second decomposition matrix are updated according to the loss function.

5. The method according to claim 1, characterized in that The obtaining a value corresponding to the resource request to be tested in the filling matrix as a predicted request value includes: If both the requester and the requested party of the resource request to be tested have appeared in the historical resource requests, then obtaining corresponding values ​​of the requester and the requested party of the resource request to be tested in the filling matrix as predicted request values; If at least one of the requester and the requested party of the resource request to be tested has not appeared in the historical resource requests, obtain the requester in the historical resource requests with the highest similarity to the requester of the resource request to be tested as the replacement requester; obtain the corresponding values ​​of the replacement requester and the requested party of the resource request to be tested in the filling matrix as the predicted request values.

6. The method according to claim 5, characterized in that The obtaining, as a replacement requester, a requester having the highest similarity with the requester of the resource request to be tested in the historical resource requests, includes: Obtaining a behavior sequence of each requester in the resource request to be tested and the historical resource request; Calculating the similarity between the behavior sequence of the requester of the resource request to be tested and the behavior sequence of the requester in each historical resource request; The requester of the historical resource request whose behavior sequence is most similar to that of the requester of the resource request to be tested is determined as the replacement requester.

7. The method according to claim 1, characterized in that The step of determining the request result of the resource request to be tested according to the predicted request value and the preset threshold value includes: If the predicted request value is greater than a first preset threshold, determining that the request result of the resource request to be tested is to provide resources; If the predicted request value is less than a second preset threshold, it is determined that the request result of the resource request to be tested is a forwarding resource request.

8. A resource request result prediction device, characterized in that: include: A sparse matrix establishment module, used to establish a sparse matrix according to at least two historical resource requests, wherein the first dimension and the second dimension of the sparse matrix respectively correspond to the requesting party and the requested party included in each of the historical resource requests, and the elements in the sparse matrix are used to represent the request results in the historical resource requests; A prediction and filling module, used for predicting and filling the empty values ​​of elements in the sparse matrix to obtain a filling matrix; A predicted value acquisition module, used to obtain the element value corresponding to the resource request to be tested in the filling matrix as the predicted request value; The prediction module is used to determine the request result of the resource request to be tested according to the predicted request value and a preset threshold.

9. An electronic device, characterized in that: include: one or more processors; a storage device having one or more programs stored thereon, When the one or more programs are executed by the one or more processors, the one or more processors are enabled to implement the method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: A computer program is stored thereon, wherein when the computer program is executed by one or more processors, the method according to any one of claims 1 to 7 is implemented.

11. A computer program product, characterized in that The method comprises a computer program / instruction, which implements the method according to any one of claims 1 to 7 when executed by a processor.

Images