Table processing method and device, storage medium, program product and electronic equipment

Abstract

The invention discloses a table processing method and device, a storage medium, a program product and electronic equipment. Relates to the field of artificial intelligence. The method comprises the following steps: acquiring a plurality of newly added business tables in a target storage area; an adjacency matrix is constructed based on the incidence relation between the multiple service tables, elements in the adjacency matrix represent the incidence intensity between the two service tables, and the incidence intensity represents the probability value that the two service tables belong to the same user; the adjacency matrix and the multiple service tables are processed through a target recognition model, the probability value that each service table belongs to different users is obtained, and the target recognition model comprises a graph neural network and a table data representation learning model; and determining users associated with each service table according to the probability value that each service table belongs to different users. Through the method and the device, the problem of low recognition accuracy when the user associated with the service table is recognized in the related technology is solved.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence, and more specifically, to a table processing method, device, storage medium, program product, and electronic device. Background Art

[0002] In financial risk assessment, it is usually necessary to comprehensively consider multiple dimensions of customer information, such as credit records, transaction behaviors, etc. This information is often scattered in different business tables, and records such as transaction behaviors usually involve a large amount of data and need to be recorded in multiple tables. However, not all tables directly record the customer's customer information (for example, only the first page of the table records user information, while other pages of the table omit user information). Therefore, the step of determining the relationship between the customer and the table is particularly important. At present, related technologies rely on predefined rules or patterns to identify the aforementioned relationship, but these rules or patterns are difficult to effectively adapt to the ever-changing financial environment and user behavior, resulting in low recognition accuracy.

[0003] To address the above-mentioned problems, no effective solution has been proposed yet. Summary of the invention

[0004] The main purpose of the present application is to provide a form processing method, device, storage medium, program product and electronic device to solve the problem of low recognition accuracy when identifying users associated with business forms in related technologies.

[0005] In order to achieve the above-mentioned purpose, according to one aspect of the present application, a table processing method is provided. The method includes: obtaining multiple newly added business tables in the target storage area, wherein a single business table is used to record the business information of a single user, some of the multiple business tables include the user information of the user, and the remaining business tables do not include the user information of the user; constructing an adjacency matrix based on the association relationship between the multiple business tables, wherein the elements in the adjacency matrix represent the association strength between the two business tables, and the association strength represents the probability value that the two business tables belong to the same user; using the target recognition model to process the adjacency matrix and the multiple business tables to obtain the probability value of each business table belonging to a different user, wherein the target recognition model includes a graph neural network and a table data representation learning model; according to the probability value of each business table belonging to a different user, determining the user associated with each business table.

[0006] Furthermore, the table processing method also includes: taking the business table as a node, constructing edges between associated nodes based on the association relationship between multiple business tables to obtain a target graph structure; and constructing an adjacency matrix based on the target graph structure.

[0007] Furthermore, the table processing method also includes: processing the adjacency matrix and multiple business tables through a graph neural network to obtain a first vector representation of each business table; using a table data representation learning model to process multiple business tables to obtain a second vector representation of each business table; for each business table, fusing the first vector representation and the second vector representation of the business table to obtain a target vector of the business table; processing the target vector of each business table through the fully connected layer and the normalized index layer in the target recognition model to obtain a probability value that each business table belongs to a different user.

[0008] Furthermore, the table processing method also includes: determining a first hyperparameter corresponding to the graph neural network, and determining a second hyperparameter corresponding to the table data representation learning model; based on the first hyperparameter and the second hyperparameter, weightedly fusing the first vector representation and the second vector representation of the business table to obtain a target vector.

[0009] Furthermore, the table processing method further includes: for each business table, determining a maximum probability value from the probability values ​​corresponding to the business table; and determining a user corresponding to the maximum probability value as a user associated with the business table.

[0010] Furthermore, the table processing method also includes: obtaining a training sample set, wherein the training samples in the training sample set include multiple sample business tables, sample adjacency matrices corresponding to the multiple sample business tables, and the true labels of the training samples represent the users to which each sample business table belongs; training an initial recognition model based on the training sample set to obtain a target recognition model.

[0011] Furthermore, the table processing method also includes: determining multiple sample business tables from historical business tables in the target storage area, wherein the historical business tables are business tables of users to which they have been determined; constructing sample adjacency matrices corresponding to the multiple sample business tables based on associations between the multiple sample business tables; and determining training samples based on the multiple sample business tables and the sample adjacency matrices corresponding to the multiple sample business tables.

[0012] In order to achieve the above-mentioned purpose, according to another aspect of the present application, a table processing device is provided. The device includes: a first acquisition module, which is used to acquire multiple newly added business tables in the target storage area, wherein a single business table is used to record the business information of a single user, some of the multiple business tables include the user information of the user, and the remaining business tables do not include the user information of the user; a first construction module, which is used to construct an adjacency matrix based on the association relationship between the multiple business tables, wherein the elements in the adjacency matrix represent the association strength between the two business tables, and the association strength represents the probability value that the two business tables belong to the same user; a first processing module, which is used to process the adjacency matrix and multiple business tables using a target recognition model to obtain the probability value that each business table belongs to a different user, wherein the target recognition model includes a graph neural network and a table data representation learning model; a first determination module, which is used to determine the user associated with each business table based on the probability value that each business table belongs to a different user.

[0013] Furthermore, the first construction module also includes: a first construction submodule, which is used to use the business table as a node, and to construct edges between associated nodes based on the association relationship between multiple business tables to obtain a target graph structure; and a second construction submodule, which is used to construct an adjacency matrix based on the target graph structure.

[0014] Furthermore, the first processing module also includes: a first processing sub-module, which is used to process the adjacency matrix and multiple business tables through a graph neural network to obtain a first vector representation of each business table; a second processing sub-module, which is used to process multiple business tables using a table data representation learning model to obtain a second vector representation of each business table; a third processing sub-module, which is used to fuse the first vector representation and the second vector representation of each business table for each business table to obtain a target vector of the business table; and a fourth processing sub-module, which is used to process the target vector of each business table through a fully connected layer and a normalized index layer in a target recognition model to obtain a probability value that each business table belongs to a different user.

[0015] Furthermore, the third processing sub-module also includes: a determination unit, used to determine a first hyperparameter corresponding to the graph neural network, and to determine a second hyperparameter corresponding to the tabular data representation learning model; a processing unit, used to perform weighted fusion of the first vector representation and the second vector representation of the business table based on the first hyperparameter and the second hyperparameter to obtain a target vector.

[0016] Furthermore, the first determination module further includes: a first determination submodule, for determining, for each business table, a maximum probability value from probability values ​​corresponding to the business table; and a second determination submodule, for determining the user corresponding to the maximum probability value as the user associated with the business table.

[0017] Furthermore, the table processing device also includes: a second acquisition module, used to obtain a training sample set, wherein the training samples in the training sample set include multiple sample business tables, sample adjacency matrices corresponding to the multiple sample business tables, and the true labels of the training samples represent the users to which each sample business table belongs; a second processing module, used to train an initial recognition model based on the training sample set to obtain a target recognition model.

[0018] Furthermore, the table processing device also includes: a second determination module, used to determine multiple sample business tables from the historical business tables in the target storage area, wherein the historical business tables are business tables of users to which they have been determined; a second construction module, used to construct a sample adjacency matrix corresponding to the multiple sample business tables based on the association relationship between the multiple sample business tables; and a third determination module, used to determine training samples based on the multiple sample business tables and the sample adjacency matrices corresponding to the multiple sample business tables.

[0019] In order to achieve the above-mentioned purpose, according to another aspect of the present application, a computer-readable storage medium is provided, the computer-readable storage medium including a stored executable program, wherein when the executable program runs, the device where the computer-readable storage medium is located is controlled to execute the above-mentioned table processing method.

[0020] In order to achieve the above-mentioned purpose, according to another aspect of the present application, an electronic device is provided, the electronic device comprising a memory storing an executable program; and a processor for running the program, wherein the above-mentioned table processing method is executed when the program is running.

[0021] In order to achieve the above-mentioned purpose, according to another aspect of the present application, a computer program product is provided, comprising computer instructions, which implement the steps of the above-mentioned table processing method when executed by a processor.

[0022] In an embodiment of the present application, an adjacency matrix is ​​constructed based on the association relationship between multiple newly added business tables, so that the known complex relationship between the business tables is converted into a quantifiable adjacency matrix. By setting the target recognition model to include a graph neural network and a table data representation learning model, and using the target recognition model to process the adjacency matrix and multiple business tables, the probability value of each business table belonging to a different user is obtained, and the graph neural network is used to effectively obtain the complex relationship between the tables, accurately capture context information and relationship dependencies, and use the table data representation learning model to learn the internal features of the table, so as to determine the probability value of the business table belonging to different users based on the internal features of the table and the cross-table features, thereby improving the accuracy of the determined probability value. By determining the users associated with each business table based on the probability value of each business table belonging to different users, the accuracy of identifying the users associated with the business table can be effectively improved.

[0023] It can be seen that the solution provided in the present application achieves the purpose of determining the users associated with the business table based on the internal characteristics and cross-table characteristics of the table, realizes the technical effect of improving the accuracy of identifying users associated with the business table, and solves the problem of low recognition accuracy when identifying users associated with the business table in the related technology. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] The drawings constituting a part of the present application are used to provide a further understanding of the present application. The illustrative embodiments and descriptions of the present application are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0025] Figure 1 is a hardware structure block diagram of a computer terminal provided according to an embodiment of the present application;

[0026] Figure 2 is a flowchart of a table processing method provided according to an embodiment of the present application;

[0027] Figure 3 is a schematic diagram of a table processing device provided according to an embodiment of the present application;

[0028] Figure 4 It is a structural block diagram of an electronic device according to an embodiment of the present application. DETAILED DESCRIPTION

[0029] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present application.

[0030] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0031] It should be noted that the collected information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data for display, data for analysis, etc.) involved in this application are information and data authorized by the user or fully authorized by all parties, and the collection, storage, use, processing, transmission, provision, disclosure and application of relevant data are in compliance with relevant laws, regulations and standards, necessary confidentiality measures are taken, and public order and good customs are not violated, and corresponding operation entrances are provided for users to choose to authorize or refuse. For example, an interface is set up between this system and relevant users or institutions to provide users with corresponding operation entrances for users to choose to agree or refuse the results of automated decision-making; if the user chooses to refuse, the expert decision-making process will be entered.

[0032] Example 1

[0033] According to an embodiment of the present application, an embodiment of a table processing method is also provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0034] The method embodiment provided in the first embodiment of the present application can be executed in a mobile terminal, a computer terminal or a similar computing device. Figure 1 FIG. 1 shows a hardware structure block diagram of a computer terminal (or mobile device) for implementing a table processing method. Figure 1 As shown, the computer terminal 10 (or mobile device) may include one or more (102a, 102b, ..., 102n are used to illustrate) processors 102 (the processor 102 may include but is not limited to a processing device such as a microprocessor MCU or a programmable logic device FPGA), a memory 104 for storing data, and a transmission device 106 for communication functions. In addition, it may also include: a display, an input / output interface (I / O interface), a universal serial bus (USB) port (which may be included as one of the ports of the BUS bus), a network interface, a power supply and / or a camera. It can be understood by those skilled in the art that Figure 1 The structure shown is only for illustration and does not limit the structure of the above electronic device. Figure 1 More or fewer components as shown, or with Figure 1 Different configurations are shown.

[0035] It should be noted that the one or more processors 102 and / or other data processing circuits described above may generally be referred to herein as "data processing circuits". The data processing circuits may be embodied in whole or in part as software, hardware, firmware, or any other combination thereof. In addition, the data processing circuit may be a single independent processing module, or may be incorporated in whole or in part into any of the other components in the computer terminal 10 (or mobile device). As described in the embodiments of the present application, the data processing circuit acts as a processor control (e.g., selection of a variable resistor terminal path connected to an interface).

[0036] The memory 104 can be used to store software programs and modules of application software, such as the program instructions / data storage device corresponding to the table processing method in the embodiment of the present application. The processor 102 executes various functional applications and data processing by running the software programs and modules stored in the memory 104, that is, realizing the above-mentioned table processing method. The memory 104 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, the memory 104 may further include a memory remotely arranged relative to the processor 102, and these remote memories may be connected to the computer terminal 10 via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0037] The transmission device 106 is used to receive or send data via a network. The specific example of the above network may include a wireless network provided by a communication provider of the computer terminal 10. In one example, the transmission device 106 includes a network adapter (Network Interface Controller, NIC), which can be connected to other network devices through a base station so as to communicate with the Internet. In one example, the transmission device 106 can be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0038] The display may be, for example, a touch screen liquid crystal display (LCD) that enables a user to interact with a user interface of the computer terminal 10 (or mobile device).

[0039] Under the above operating environment, this application provides Figure 2 The table processing method shown. Figure 2 is a flow chart of a table processing method provided according to an embodiment of the present application, such as Figure 2 As shown, the method comprises the following steps:

[0040] Step S201 : acquiring multiple newly added business tables in the target storage area, wherein a single business table is used to record business information of a single user, some of the multiple business tables include user information of the user, and the remaining business tables do not include user information of the user.

[0041] Optionally, electronic devices, application systems, servers and other devices may be used as the execution subject of the present application. In this embodiment, the target processing system is used as the execution subject to execute the above-mentioned table processing method.

[0042] Optionally, the target storage area may be a database or system for storing financial business data, which may include various business tables, such as a credit record table, a transaction behavior table, etc. The newly added multiple business tables refer to tables newly added in the target storage area, which contain the latest business information about different users.

[0043] Optionally, each business table records a user's specific business data. For example, a business table may record a user's credit history, which is equivalent to the above-mentioned business information. Since the business information of a single user may also involve a large amount of data, a certain type of business information (such as transaction behavior) of a single user may be recorded in multiple business tables. However, not all tables directly record the customer's customer information. For example, only the first page of the transaction behavior table records a user's user information, while the transaction behavior tables on other pages omit the user's user information and only record the business information. Therefore, it is necessary to determine the relationship between these business tables that do not record user information and the user.

[0044] Step S202: construct an adjacency matrix based on the association relationship between the multiple business tables, wherein an element in the adjacency matrix represents the association strength between two business tables, and the association strength represents the probability value that the two business tables belong to the same user.

[0045] In some embodiments, the target processing system may first construct a target graph structure based on the association relationship between multiple business tables, and then construct an adjacency matrix based on the target graph structure. Each node in the target graph structure represents a business table, and the edge represents the association relationship and association strength between the business tables. The association relationship between the business tables is used to represent whether the two business tables belong to the same user, and the association strength between the business tables represents the probability value that the two business tables belong to the same user.

[0046] Optionally, the target processing system may determine the association strength between the business tables based on the user information in the business tables. For example, if two business tables record the user information of the same user, the association strength is determined to be 1, and an edge is constructed between the corresponding nodes; if the two business tables do not record the user information of the same user, the association strength is determined to be 0, and no edge is constructed between the corresponding nodes.

[0047] Optionally, the adjacency matrix is ​​a matrix used to represent the connection relationship between nodes in a graph structure, and the elements in the matrix represent the strength of association between two nodes (i.e., business tables). In this embodiment, the adjacency matrix is ​​used to quantify the possibility that business tables belong to the same user. Optionally, the adjacency matrix can be expressed as G = (V, E), where V represents a node set, and each node v i ∈V is a business table. E is an edge set, each edge e i,j ∈E represents the table v i and the form v j The strength of the association between them.

[0048] Step S203, using the target recognition model to process the adjacency matrix and multiple business tables to obtain the probability value of each business table belonging to a different user, wherein the target recognition model includes a graph neural network and a table data representation learning model.

[0049] Optionally, after the adjacency matrix is ​​determined, the target recognition model can be used to process the adjacency matrix and multiple business tables to obtain the probability value of each business table belonging to a different user. For example, the adjacency matrix and multiple business tables are input into the target recognition network model, and the first vector representation of each business table is determined based on the input data by the graph neural network, and the second vector representation of each business table is determined based on the input data by the table data representation learning model. Thus, the target recognition model determines the probability value of each business table belonging to a different user based on the first vector representation and the second vector representation of each business table.

[0050] Among them, the graph neural network is a deep learning model used to process structured graph data, which can capture the complex relationship between nodes and transmit information. The graph neural network can encode the adjacency matrix to learn the correlation features between business tables. The tabular data representation learning model can be a hypergraph-enhanced tabular data learning model. The tabular data representation learning model can use a hypergraph structure to model tabular data, such as converting tabular data into a hypergraph structure, where each node represents a cell in the table, and the edge represents the relationship between cells to capture the characteristics of the table such as arrangement invariance and hierarchical structure. The tabular data representation learning model is used to learn and encode the internal features of tabular data, such as cell content, table structure, etc. By combining it with the graph neural network, the information within and across tables can be integrated, thereby avoiding the problem of information loss or one-sided understanding caused by focusing only on a single table or data source.

[0051] Step S204 , determining the user associated with each business table according to the probability value that each business table belongs to a different user.

[0052] Optionally, after obtaining the probability values ​​that each business table belongs to different users, the target processing system may determine the maximum probability value from the probability values ​​corresponding to each business table, thereby determining the user corresponding to the maximum probability value as the user associated with the business table.

[0053] In some embodiments, the target processing system can determine a target business table from multiple business tables, wherein the target business table refers to a business table that does not include user information of the user, thereby determining the user associated with the target business table based on the probability value that the target business table belongs to different users. For a business table that includes user information, the user to whom the user information in the business table belongs is directly determined as the user associated with the business table.

[0054] Optionally, after determining the users associated with each business form, the target processing system may integrate the business information in the business forms belonging to the same user, and then evaluate the transaction risk value of the user based on the integrated information, thereby facilitating improving the accuracy of transaction risk evaluation.

[0055] In an embodiment of the present application, an adjacency matrix is ​​constructed based on the association relationship between multiple newly added business tables, so that the known complex relationship between the business tables is converted into a quantifiable adjacency matrix. By setting the target recognition model to include a graph neural network and a table data representation learning model, and using the target recognition model to process the adjacency matrix and multiple business tables, the probability value of each business table belonging to a different user is obtained, and the graph neural network is used to effectively obtain the complex relationship between the tables, accurately capture context information and relationship dependencies, and use the table data representation learning model to learn the internal features of the table, so as to determine the probability value of the business table belonging to different users based on the internal features of the table and the cross-table features, thereby improving the accuracy of the determined probability value. By determining the users associated with each business table based on the probability value of each business table belonging to different users, the accuracy of identifying the users associated with the business table can be effectively improved.

[0056] It can be seen that the solution provided in the present application achieves the purpose of determining the users associated with the business table based on the internal characteristics and cross-table characteristics of the table, realizes the technical effect of improving the accuracy of identifying users associated with the business table, and solves the problem of low recognition accuracy when identifying users associated with the business table in the related technology.

[0057] Optionally, in the table processing method provided in the embodiment of the present application, an adjacency matrix is ​​constructed based on the association relationship between multiple business tables, including: taking the business table as a node, constructing edges between associated nodes based on the association relationship between the multiple business tables to obtain a target graph structure; and constructing an adjacency matrix based on the target graph structure.

[0058] Optionally, before constructing the adjacency matrix, the business tables are first abstracted as nodes in a graph structure. Each business table is regarded as an independent node, and the edges between the nodes reflect the association relationship and the strength of the association between the business tables.

[0059] Optionally, the target processing system may determine the association strength between the business tables based on the user information in the business tables. For example, if two business tables record the user information of the same user, the association strength is determined to be 1, and an edge is constructed between the corresponding nodes; if the two business tables do not record the user information of the same user, the association strength is determined to be 0, and no edge is constructed between the corresponding nodes.

[0060] Optionally, after determining the target graph structure, the target processing system can construct an adjacency matrix based on the target graph structure, and each element in the adjacency matrix indicates whether there is an edge between two nodes, and the weight of the edge, that is, the strength of association between the business tables. When constructing the adjacency matrix, for two business tables with an associated relationship, their corresponding elements will be assigned a non-zero value, which reflects the degree of association between them, while business tables without an associated relationship will correspond to a zero value.

[0061] It should be noted that through the above method, the complex business table association relationships are effectively converted into mathematical representations that can be processed by graph neural networks, so that the model can learn and utilize these relationships, thereby improving the model processing effect and further improving the accuracy of identifying the association relationship between users and tables.

[0062] Optionally, in the table processing method provided in the embodiment of the present application, the adjacency matrix and multiple business tables are processed using a target recognition model to obtain a probability value that each business table belongs to a different user, including: processing the adjacency matrix and multiple business tables through a graph neural network to obtain a first vector representation of each business table; processing multiple business tables using a table data representation learning model to obtain a second vector representation of each business table; for each business table, fusing the first vector representation and the second vector representation of the business table to obtain a target vector of the business table; processing the target vector of each business table through a fully connected layer and a normalized index layer in the target recognition model to obtain a probability value that each business table belongs to a different user.

[0063] Optionally, after inputting the adjacency matrix and multiple business tables into the target recognition model, we can process the adjacency matrix and multiple business tables through a graph neural network to obtain the first vector representation of each business table. These vectors represent the cross-table relationship features carried by the business tables after they are associated through the graph structure. The purpose is to introduce the information transmission mechanism of the node key, and learn the feature representation of the business table containing the cross-table relationship by aggregating the features of the adjacent nodes.

[0064] Optionally, the convolutional layer in the graph neural network can update the feature representation by the following formula:

[0065]

[0066] Among them, H (l+1) H represents the node embedding matrix of the L+1th convolutional layer; (l) represents the node embedding matrix of the Lth convolutional layer; σ represents the activation function, such as ReLU; is the adjacency matrix of the graph plus the identity matrix, which is used to account for self-loops; express The degree matrix of The elements on the diagonal are The sum of each row; W (l) Represents the weight matrix of layer L. By stacking multiple convolutional layers, we can learn the complex relationship between tables and obtain the embedded representation of each table.

[0067] Optionally, the table data representation learning model is used to process the business table, extract the internal features of the table content and structure, and output a second vector representation, which reflects the business table's own information and does not include cross-table relationships. The purpose of this step is to use the model to capture the details and complexity of the business table and generate a feature vector for each table based on its own content.

[0068] Optionally, after obtaining the first vector representation and the second vector representation, the target processing system may combine the first vector representation (i.e., the cross-table relationship feature) and the second vector representation (i.e., the table internal feature) to obtain a more comprehensive feature representation, i.e., the target vector. For example, the target processing system may directly concatenate the first vector representation and the second vector representation to obtain the target vector, or the target processing system may perform weighted fusion on the first vector representation and the second vector representation to obtain the target vector.

[0069] After obtaining the target vector, the target processing system can transform the target vector through a fully connected layer, map it to a space that can represent user classification, and then map the transformed vector to a probability value through a normalized exponential layer, thereby obtaining the probability value that each business table belongs to a different user.

[0070] It should be noted that through the above process, the internal features of the table and the cross-table relationship features are effectively acquired, and the probability value of each business table belonging to a different user is determined based on the features of these two dimensions, thereby improving the accuracy of the determined probability value.

[0071] Optionally, in the table processing method provided in an embodiment of the present application, the first vector representation and the second vector representation of the business table are fused to obtain a target vector of the business table, including: determining a first hyperparameter corresponding to the graph neural network, and determining a second hyperparameter corresponding to the table data representation learning model; based on the first hyperparameter and the second hyperparameter, weighted fusion of the first vector representation and the second vector representation of the business table is performed to obtain a target vector.

[0072] In some embodiments, the target vector may be determined by:

[0073] F=αQ+βK

[0074] Wherein, F represents the target vector, α represents the first hyperparameter, β represents the second hyperparameter, Q represents the first vector representation, and K represents the second vector representation. The specific values ​​of α and β are learned by the target recognition model during the training process.

[0075] It should be noted that by determining the target vector in a weighted fusion manner, the accuracy of the determined target vector is improved, thereby improving the accuracy of identifying the user associated with the business table.

[0076] Optionally, in the table processing method provided in the embodiment of the present application, the users associated with each business table are determined based on the probability values ​​that each business table belongs to different users, including: for each business table, determining the maximum probability value from the probability values ​​corresponding to the business table; and determining the user corresponding to the maximum probability value as the user associated with the business table.

[0077] For example, for a certain business table, suppose the target recognition model outputs the probability value that the business table belongs to three users A, B, and C, among which the probability value corresponding to user A is 50%, the probability value corresponding to user B is 30%, and the probability value corresponding to user C is 20%, then user A is determined as the user associated with the business table.

[0078] It should be noted that, through the above process, it is possible to determine the user associated with the business table according to the size of the quantitative value output by the model, thereby improving the accuracy of identifying the user associated with the business table.

[0079] Optionally, in the table processing method provided in an embodiment of the present application, the target recognition model is trained in the following manner: obtaining a training sample set, wherein the training samples in the training sample set include multiple sample business tables, sample adjacency matrices corresponding to the multiple sample business tables, and the true labels of the training samples represent the users to which each sample business table belongs; training an initial recognition model based on the training sample set to obtain a target recognition model.

[0080] Optionally, the sample business tables can be understood as historical business tables in the target storage area. These sample business tables are business tables of known users. The sample adjacency matrix is ​​constructed based on multiple sample business tables. The sample adjacency matrix is ​​used to quantify the degree of association between the sample business tables.

[0081] After obtaining the training sample set, the target processing model can train the initial recognition model based on the training sample set to obtain the target recognition model. The loss function used in the training process can be a cross entropy loss function, and the initial recognition model includes an initial graph neural network and an initial table data representation learning model.

[0082] It should be noted that by using known data to train the initial recognition model to obtain the target recognition model, the initial recognition model can effectively learn how to extract key features from the business table and adjacency matrix and associate them with user information during the training process, thereby improving the accuracy of the trained target recognition model.

[0083] Optionally, in the table processing method provided in the embodiment of the present application, training samples are obtained in the following manner: determining multiple sample business tables from historical business tables in the target storage area, wherein the historical business tables are business tables of users to which they have been determined; constructing a sample adjacency matrix corresponding to the multiple sample business tables based on the association relationship between the multiple sample business tables; and determining training samples based on the multiple sample business tables and the sample adjacency matrix corresponding to the multiple sample business tables.

[0084] Optionally, the historical business table may be a table that has been processed and the associated user has been determined. The target processing system may randomly select a historical business table from the target storage area as a sample business table, or the target processing system may select historical business tables corresponding to multiple users from the target storage system as sample business tables, and some of the multiple sample business tables do not include user information.

[0085] After determining multiple sample business tables, the target processing system can construct sample adjacency matrices corresponding to the multiple sample business tables based on the association relationship between the multiple sample business tables. The process of constructing the sample adjacency matrix is ​​the same as the process of constructing the adjacency matrix, so it will not be repeated here.

[0086] After obtaining the sample adjacency matrix, the target processing system can combine multiple sample business tables and the sample adjacency matrices corresponding to the multiple sample business tables into training samples, wherein the multiple sample business tables in different training samples are not completely the same.

[0087] It should be noted that, through the above method, the effective construction of training samples is achieved, thereby further improving the model training effect.

[0088] It can be seen that the solution provided in the present application achieves the purpose of determining the users associated with the business table based on the internal characteristics and cross-table characteristics of the table, realizes the technical effect of improving the accuracy of identifying users associated with the business table, and solves the problem of low recognition accuracy when identifying users associated with the business table in the related technology.

[0089] It should be noted that the steps shown in the flowcharts of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and that, although a logical order is shown in the flowcharts, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0090] Example 2

[0091] The embodiment of the present application also provides a table processing device. It should be noted that the table processing device of the embodiment of the present application can be used to execute the table processing method provided in the embodiment of the present application. The table processing device provided in the embodiment of the present application is introduced below.

[0092] According to an embodiment of the present application, a table processing device for implementing the above table processing method is also provided. Figure 3 As shown, the device comprises:

[0093] A first acquisition module 301 is used to acquire a plurality of newly added business tables in a target storage area, wherein a single business table is used to record business information of a single user, some of the plurality of business tables include user information of the user, and the remaining business tables do not include user information of the user;

[0094] A first construction module 302 is used to construct an adjacency matrix based on the association relationship between multiple business tables, wherein an element in the adjacency matrix represents the association strength between two business tables, and the association strength represents the probability value that the two business tables belong to the same user;

[0095] The first processing module 303 is used to process the adjacency matrix and multiple business tables using a target recognition model to obtain a probability value that each business table belongs to a different user, wherein the target recognition model includes a graph neural network and a table data representation learning model;

[0096] The first determining module 304 is used to determine the user associated with each service table according to the probability value that each service table belongs to a different user.

[0097] In an embodiment of the present application, an adjacency matrix is ​​constructed based on the association relationship between multiple newly added business tables, so that the known complex relationship between the business tables is converted into a quantifiable adjacency matrix. By setting the target recognition model to include a graph neural network and a table data representation learning model, and using the target recognition model to process the adjacency matrix and multiple business tables, the probability value of each business table belonging to a different user is obtained, and the graph neural network is used to effectively obtain the complex relationship between the tables, accurately capture context information and relationship dependencies, and use the table data representation learning model to learn the internal features of the table, so as to determine the probability value of the business table belonging to different users based on the internal features of the table and the cross-table features, thereby improving the accuracy of the determined probability value. By determining the users associated with each business table based on the probability value of each business table belonging to different users, the accuracy of identifying the users associated with the business table can be effectively improved.

[0098] It can be seen that the solution provided in the present application achieves the purpose of determining the users associated with the business table based on the internal characteristics and cross-table characteristics of the table, realizes the technical effect of improving the accuracy of identifying users associated with the business table, and solves the problem of low recognition accuracy when identifying users associated with the business table in the related technology.

[0099] Optionally, in the table processing device provided in the embodiment of the present application, the first construction module also includes: a first construction sub-module, used to take the business table as a node, and build edges between associated nodes based on the association relationship between multiple business tables to obtain a target graph structure; a second construction sub-module, used to build an adjacency matrix based on the target graph structure.

[0100] Optionally, in the table processing device provided in the embodiment of the present application, the first processing module also includes: a first processing sub-module, which is used to process the adjacency matrix and multiple business tables through a graph neural network to obtain a first vector representation of each business table; a second processing sub-module, which is used to process multiple business tables using a table data representation learning model to obtain a second vector representation of each business table; a third processing sub-module, which is used to fuse the first vector representation and the second vector representation of each business table for each business table to obtain a target vector of the business table; and a fourth processing sub-module, which is used to process the target vector of each business table through a fully connected layer and a normalized index layer in a target recognition model to obtain a probability value that each business table belongs to a different user.

[0101] Optionally, in the table processing device provided in the embodiment of the present application, the third processing sub-module also includes: a determination unit, used to determine a first hyperparameter corresponding to the graph neural network, and to determine a second hyperparameter corresponding to the table data representation learning model; a processing unit, used to perform weighted fusion of the first vector representation and the second vector representation of the business table based on the first hyperparameter and the second hyperparameter to obtain a target vector.

[0102] Optionally, in the table processing device provided in the embodiment of the present application, the first determination module also includes: a first determination sub-module, used to determine, for each business table, the maximum probability value from the probability values ​​corresponding to the business table; and a second determination sub-module, used to determine the user corresponding to the maximum probability value as the user associated with the business table.

[0103] Optionally, in the table processing device provided in the embodiment of the present application, the table processing device also includes: a second acquisition module, used to acquire a training sample set, wherein the training samples in the training sample set include multiple sample business tables, sample adjacency matrices corresponding to the multiple sample business tables, and the true labels of the training samples represent the users to which each sample business table belongs; a second processing module, used to train an initial recognition model based on the training sample set to obtain a target recognition model.

[0104] Optionally, in the table processing device provided in the embodiment of the present application, the table processing device also includes: a second determination module, used to determine multiple sample business tables from historical business tables in the target storage area, wherein the historical business tables are business tables of users to which they have been determined; a second construction module, used to construct a sample adjacency matrix corresponding to the multiple sample business tables based on the association relationship between the multiple sample business tables; and a third determination module, used to determine training samples based on the multiple sample business tables and the sample adjacency matrices corresponding to the multiple sample business tables.

[0105] It should be noted that the first acquisition module 301, the first construction module 302, the first processing module 303 and the first determination module 304 correspond to steps S201 to S204 in Example 1, and the four modules and the corresponding steps implement the same examples and application scenarios, but are not limited to the contents disclosed in the above-mentioned Example 1. It should be noted that the above-mentioned modules or units may be hardware components or software components stored in a memory (e.g., memory 104) and processed by one or more processors (e.g., processors 102a, 102b, ..., 102n), and the above-mentioned modules may also be part of the device and may be run in the computer terminal 10 provided in Example 1.

[0106] Example 3

[0107] An embodiment of the present application may provide an electronic device, Figure 4is a structural block diagram of an electronic device according to an embodiment of the present application. Figure 4 As shown, the electronic device may include: one or more ( Figure 4 (only one is shown) processor 1002, memory 1004, storage controller, and peripheral interface, wherein the peripheral interface is connected to the radio frequency module, audio module and display.

[0108] Among them, the memory can be used to store software programs and modules, such as program instructions / modules corresponding to the methods and devices in the embodiments of the present application, and the processor executes various functional applications and data processing by running the software programs and modules stored in the memory, that is, realizing the above-mentioned method. The memory may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory may further include a memory remotely arranged relative to the processor, and these remote memories may be connected to the terminal via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0109] The processor can call the information and application stored in the memory through the transmission device to perform the following steps: obtain multiple business tables newly added in the target storage area, wherein a single business table is used to record the business information of a single user, some of the multiple business tables include the user information of the user, and the remaining business tables do not include the user information of the user; construct an adjacency matrix based on the association relationship between the multiple business tables, wherein the elements in the adjacency matrix represent the association strength between the two business tables, and the association strength represents the probability value that the two business tables belong to the same user; use the target recognition model to process the adjacency matrix and multiple business tables to obtain the probability value of each business table belonging to a different user, wherein the target recognition model includes a graph neural network and a table data representation learning model; determine the user associated with each business table based on the probability value that each business table belongs to a different user.

[0110] The processor can also call the information and application stored in the memory through the transmission device to perform the following steps: taking the business table as a node, building edges between associated nodes based on the association relationship between multiple business tables to obtain a target graph structure; and building an adjacency matrix based on the target graph structure.

[0111] The processor can also call the information and applications stored in the memory through the transmission device to perform the following steps: process the adjacency matrix and multiple business tables through a graph neural network to obtain a first vector representation of each business table; use a table data representation learning model to process multiple business tables to obtain a second vector representation of each business table; for each business table, fuse the first vector representation and the second vector representation of the business table to obtain a target vector of the business table; process the target vector of each business table through the fully connected layer and the normalized index layer in the target recognition model to obtain a probability value that each business table belongs to a different user.

[0112] The processor can call the information and application programs stored in the memory through the transmission device to perform the following steps: determine the first hyperparameter corresponding to the graph neural network, and determine the second hyperparameter corresponding to the table data representation learning model; based on the first hyperparameter and the second hyperparameter, perform weighted fusion on the first vector representation and the second vector representation of the business table to obtain the target vector.

[0113] The processor can call the information and application stored in the memory through the transmission device to perform the following steps: for each business table, determine the maximum probability value from the probability values ​​corresponding to the business table; determine the user corresponding to the maximum probability value as the user associated with the business table.

[0114] The processor can call the information and application stored in the memory through the transmission device to perform the following steps: obtain a training sample set, wherein the training samples in the training sample set include multiple sample business tables, sample adjacency matrices corresponding to the multiple sample business tables, and the true labels of the training samples represent the users to which each sample business table belongs; train an initial recognition model based on the training sample set to obtain a target recognition model.

[0115] The processor can call the information and application stored in the memory through the transmission device to perform the following steps: determine multiple sample business tables from the historical business tables in the target storage area, wherein the historical business tables are business tables of users to which they have been determined; construct a sample adjacency matrix corresponding to the multiple sample business tables based on the association relationship between the multiple sample business tables; determine the training samples based on the multiple sample business tables and the sample adjacency matrix corresponding to the multiple sample business tables.

[0116] In an embodiment of the present application, an adjacency matrix is ​​constructed based on the association relationship between multiple newly added business tables, so that the known complex relationship between the business tables is converted into a quantifiable adjacency matrix. By setting the target recognition model to include a graph neural network and a table data representation learning model, and using the target recognition model to process the adjacency matrix and multiple business tables, the probability value of each business table belonging to a different user is obtained, and the graph neural network is used to effectively obtain the complex relationship between the tables, accurately capture context information and relationship dependencies, and use the table data representation learning model to learn the internal features of the table, so as to determine the probability value of the business table belonging to different users based on the internal features of the table and the cross-table features, thereby improving the accuracy of the determined probability value. By determining the users associated with each business table based on the probability value of each business table belonging to different users, the accuracy of identifying the users associated with the business table can be effectively improved.

[0117] It can be seen that the solution provided in the present application achieves the purpose of determining the users associated with the business table based on the internal characteristics and cross-table characteristics of the table, realizes the technical effect of improving the accuracy of identifying users associated with the business table, and solves the problem of low recognition accuracy when identifying users associated with the business table in the related technology.

[0118] It can be understood by those skilled in the art that Figure 4 The structure shown is for illustration only, and the electronic device may also be a smart phone (such as an Android phone, an iOS phone, etc.), a tablet computer, a PDA, a mobile Internet device (Mobile Internet Devices, MID), a PAD, or other terminal devices. Figure 4 The structure of the electronic device is not limited. Figure 4 More or fewer components (such as network interfaces, display devices, etc.) shown in, or having Figure 4 Different configurations are shown.

[0119] A person of ordinary skill in the art can understand that all or part of the steps in the various methods of the above embodiments can be completed by instructing the hardware related to the terminal device through a program, and the program can be stored in a computer-readable storage medium, and the storage medium may include: a flash drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, etc.

[0120] Example 4

[0121] The embodiment of the present application further provides a storage medium. Optionally, in this embodiment, the storage medium can be used to store the program code executed by the table processing method provided in the first embodiment.

[0122] Optionally, in this embodiment, the above storage medium may be located in any computer terminal in a computer terminal group in a computer network, or in any mobile terminal in a mobile terminal group.

[0123] The present application also provides a computer program product, which, when executed on a data processing device, is suitable for executing the program steps of the table processing method.

[0124] The serial numbers of the above-mentioned embodiments of the present application are for description only and do not represent the advantages or disadvantages of the embodiments.

[0125] In the above embodiments of the present application, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, please refer to the relevant description of other embodiments.

[0126] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of units or modules, which can be electrical or other forms.

[0127] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0128] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0129] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, in essence, or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions to enable a computer device (which can be a personal computer, a server or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), mobile hard disk, disk or optical disk and other media that can store program codes.

[0130] The above is only a preferred implementation of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present application. These improvements and modifications should also be regarded as the scope of protection of the present application.

Claims

1. A table processing method, characterized in that: include: Acquire multiple newly added business tables in the target storage area, wherein a single business table is used to record business information of a single user, some of the multiple business tables include user information of the user, and the remaining business tables do not include user information of the user; Constructing an adjacency matrix based on the association relationship between the multiple business tables, wherein the elements in the adjacency matrix represent the association strength between two business tables, and the association strength represents the probability value that the two business tables belong to the same user; Processing the adjacency matrix and the multiple business tables using a target recognition model to obtain a probability value that each business table belongs to a different user, wherein the target recognition model includes a graph neural network and a table data representation learning model; The user associated with each business table is determined according to the probability value that each business table belongs to a different user.

2. The method according to claim 1, characterized in that Constructing an adjacency matrix based on the association relationship between the multiple business tables includes: Taking the business table as a node, building edges between associated nodes based on the association relationship between the multiple business tables to obtain a target graph structure; The adjacency matrix is ​​constructed based on the target graph structure.

3. The method according to claim 1, characterized in that The adjacency matrix and the plurality of business tables are processed by using a target recognition model to obtain a probability value that each business table belongs to a different user, including: Processing the adjacency matrix and the plurality of business tables by means of the graph neural network to obtain a first vector representation of each business table; Processing the plurality of business tables using the table data representation learning model to obtain a second vector representation of each business table; For each business table, fusing the first vector representation and the second vector representation of the business table to obtain a target vector of the business table; The target vectors of each business table are processed by the fully connected layer and the normalized index layer in the target recognition model to obtain the probability value that each business table belongs to a different user.

4. The method according to claim 3, characterized in that The first vector representation and the second vector representation of the business table are fused to obtain a target vector of the business table, including: Determine a first hyperparameter corresponding to the graph neural network, and determine a second hyperparameter corresponding to the tabular data representation learning model; The first vector representation and the second vector representation of the business table are weightedly fused according to the first hyperparameter and the second hyperparameter to obtain the target vector.

5. The method according to claim 1, characterized in that According to the probability value that each business table belongs to different users, the user associated with each business table is determined, including: For each business table, determining a maximum probability value from the probability values ​​corresponding to the business table; The user corresponding to the maximum probability value is determined as the user associated with the business table.

6. The method according to claim 1, characterized in that The target recognition model is trained in the following way: Acquire a training sample set, wherein the training samples in the training sample set include a plurality of sample business tables and sample adjacency matrices corresponding to the plurality of sample business tables, and the true labels of the training samples represent the users to which each sample business table belongs; An initial recognition model is trained based on the training sample set to obtain the target recognition model.

7. The method according to claim 6, characterized in that The training samples are obtained in the following way: Determine a plurality of sample business tables from the historical business tables in the target storage area, wherein the historical business tables are business tables of users to which the users have been determined to belong; Constructing a sample adjacency matrix corresponding to the plurality of sample business tables based on the association relationship between the plurality of sample business tables; The training samples are determined based on the multiple sample business tables and sample adjacency matrices corresponding to the multiple sample business tables.

8. A table processing device, characterized in that: include: A first acquisition module is used to acquire a plurality of newly added business tables in the target storage area, wherein a single business table is used to record business information of a single user, some of the plurality of business tables include user information of the user, and the remaining business tables do not include user information of the user; A first construction module is used to construct an adjacency matrix based on the association relationship between the multiple business tables, wherein the elements in the adjacency matrix represent the association strength between two business tables, and the association strength represents the probability value that the two business tables belong to the same user; A first processing module is used to process the adjacency matrix and the multiple business tables using a target recognition model to obtain a probability value that each business table belongs to a different user, wherein the target recognition model includes a graph neural network and a table data representation learning model; The first determination module is used to determine the user associated with each business table according to the probability value that each business table belongs to different users.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored executable program, wherein when the executable program is executed, the device where the computer-readable storage medium is located is controlled to execute the table processing method according to any one of claims 1 to 7.

10. An electronic device, characterized in that: include: A memory storing an executable program; A processor is used to run the program, wherein the program executes the table processing method described in any one of claims 1 to 7 when running.

11. A computer program product comprising computer instructions, characterized in that: When the computer instructions are executed by a processor, the steps of the table processing method described in any one of claims 1 to 7 are implemented.

Images