CN117575816A

CN117575816A - Financial data detection method and device, storage medium and electronic equipment

Info

Publication number: CN117575816A
Application number: CN202311468580.2A
Authority: CN
Inventors: 黎文江; 彭郢; 郭松鹭; 杨俊�; 谢斌; 汪洋
Original assignee: Chongqing Ant Consumer Finance Co ltd
Current assignee: Chongqing Ant Consumer Finance Co ltd
Priority date: 2023-11-06
Filing date: 2023-11-06
Publication date: 2024-02-20

Abstract

The specification discloses a financial data detection method, a device, a storage medium and electronic equipment, wherein the method comprises the following steps: determining subject structure combination dimension information aiming at a financial management system, controlling the financial management system to carry out financial data summarization processing based on the subject structure dimension information to obtain financial evidence summarization data, inputting the financial evidence summarization data into a financial evidence detection model to output a financial evidence detection result under a target abnormal index dimension, and carrying out abnormal evidence data processing based on the financial evidence detection result.

Description

Financial data detection method and device, storage medium and electronic equipment

Technical Field

The present disclosure relates to the field of computer technologies, and in particular, to a method and apparatus for detecting financial data, a storage medium, and an electronic device.

Background

Since financial data of a subject such as an enterprise and an organization directly and significantly affects both society and the subject themselves, it is necessary to perform supervision and inspection on the financial data and to record and feed back the relevant inspection data on an irregular basis in order to promote orderly development of production and management, and thus it is possible to determine the current financial situation based on the inspection data, thereby facilitating adjustment in management policy.

Disclosure of Invention

The specification provides a financial data detection method, a device, a storage medium and electronic equipment, wherein the technical scheme is as follows:

in a first aspect, the present description provides a method of financial data detection, the method comprising:

determining subject structure combination dimension information aiming at a financial management system, and controlling the financial management system to carry out financial data summarization processing based on the subject structure dimension information to obtain financial evidence summarization data;

inputting the financial document summary data into a financial document detection model, and outputting a financial document detection result under a target abnormal index dimension, wherein the target abnormal index dimension is an abnormal index dimension associated with the subject structure combination dimension information;

and carrying out abnormal credential data processing based on the financial credential detection result.

In a second aspect, the present specification provides a financial credential detection model training method, the method comprising:

creating an initial financial credential detection model for a financial management system;

determining sample subject structure combination dimension information aiming at a financial management system from a dimension combination abnormal index dimension rule base through a financial terminal, controlling the financial management system to carry out financial data summarization processing based on the sample subject structure dimension information to obtain sample financial evidence summarization data, and labeling sample data labels on the sample financial evidence summarization data based on the sample abnormal index dimension;

And performing at least one round of model training on the initial financial voucher detection model based on the sample financial voucher summary data and the sample data label to obtain a model-trained financial voucher detection model.

In a third aspect, the present description provides a financial data detection apparatus, the apparatus comprising:

the information determining module is used for determining subject structure combination dimension information aiming at the financial management system, and controlling the financial management system to carry out financial data summarization processing based on the subject structure dimension information to obtain financial evidence summarization data;

the model processing module is used for inputting the financial evidence summary data into a financial evidence detection model and outputting a financial evidence detection result under a target abnormal index dimension, wherein the target abnormal index dimension is an abnormal index dimension associated with the subject structure combination dimension information;

and the data processing module is used for processing abnormal credential data based on the financial credential detection result.

In a fourth aspect, the present description provides a financial credential detection model training apparatus, the apparatus comprising:

the model creation module is used for creating an initial financial credential detection model for the financial management system;

The data summarizing module is used for determining sample subject structure combination dimension information aiming at the financial management system from a dimension combination abnormal index dimension rule base through a financial end, controlling the financial management system to carry out financial data summarizing processing based on the sample subject structure dimension information to obtain sample financial voucher summarized data, and labeling sample data labels on the sample financial voucher summarized data based on the sample abnormal index dimension;

and the model training module is used for carrying out at least one round of model training on the initial financial evidence detection model based on the sample financial evidence summary data and the sample data label to obtain a model-trained financial evidence detection model.

In a fifth aspect, the present description provides a computer storage medium storing at least one instruction adapted to be loaded by a processor and to perform the method steps of one or more embodiments of the present description.

In a sixth aspect, the present description provides a computer program product storing at least one instruction adapted to be loaded by a processor and to perform the method steps of one or more embodiments of the present description.

In a seventh aspect, the present description provides an electronic device, which may include: a processor and a memory; wherein the memory stores a computer program adapted to be loaded by the processor and to perform the method steps of one or more embodiments of the present description.

The technical scheme provided by some embodiments of the present specification has the following beneficial effects:

in one or more embodiments of the present disclosure, a service platform controls a financial management system to perform financial data summarization processing based on subject structure dimension information by determining subject structure combination dimension information for the financial management system to obtain financial document summarization data, inputs the financial document summarization data into a financial document detection model to output a financial document detection result under a target abnormal index dimension, and the target abnormal index dimension is equivalent to an abnormal index dimension focused based on subject structure combination dimension information in an actual scene.

Drawings

In order to more clearly illustrate the technical solutions of the present specification or the prior art, the following description will briefly introduce the drawings that are required to be used in the embodiments or the prior art descriptions, it is obvious that the drawings in the following description are only some embodiments of the present specification, and other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.

FIG. 1 is a schematic illustration of a scenario of a financial data detection system provided herein;

FIG. 2 is a flow chart of a method for detecting financial data provided in the present specification;

FIG. 3 is a schematic view of a pooled subject architecture provided herein;

FIG. 4 is a schematic illustration of an exemplary financial data structure provided herein;

FIG. 5 is a schematic flow chart of maintenance and update of a dimension rule base of a dimension combination anomaly index provided by the present specification;

FIG. 6 is a schematic flow chart of maintenance and update of a dimension rule base of a dimension combination anomaly index provided by the present specification;

FIG. 7 is an exemplary model architectural diagram of a financial credential detection model provided herein;

FIG. 8 is a schematic diagram of a financial data testing apparatus provided herein;

FIG. 9 is a schematic diagram of a training device for a financial document detection model provided in the present specification;

fig. 10 is a schematic structural diagram of an electronic device provided in the present specification.

Detailed Description

The following description of the embodiments of the present invention will be made apparent from, and elucidated with reference to, the drawings of the present specification, in which embodiments described are only some, but not all, embodiments of the present specification. All other embodiments, which can be made by one of ordinary skill in the art without undue burden from the present disclosure, are intended to be within the scope of the present disclosure.

In the description of the present specification, it should be understood that the terms "first," "second," and the like are used for descriptive purposes only and are not to be construed as indicating or implying relative importance. In the description of the present specification, it should be noted that, unless expressly specified and limited otherwise, "comprise" and "have" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, article, or apparatus that comprises a list of steps or elements is not limited to only those listed steps or elements but may include other steps or elements not listed or inherent to such process, method, article, or apparatus. The specific meaning of the terms in this specification will be understood by those of ordinary skill in the art in the light of the specific circumstances. In addition, in the description of the present specification, unless otherwise indicated, "a plurality" means two or more. "and/or", describes an association relationship of an association object, and indicates that there may be three relationships, for example, a and/or B, and may indicate: a exists alone, A and B exist together, and B exists alone. The character "/" generally indicates that the context-dependent object is an "or" relationship.

In the related art, the quality anomaly detection of the financial data, namely the financial data detection, usually uses an anomaly rule base created by a financial expert to enumerate all anomaly rules in detail in the anomaly rule base, and the object to be detected-the financial credential data are checked one by one according to the rules in the anomaly rule base.

The present specification is described in detail below with reference to specific examples.

Referring to fig. 1, a schematic view of a scenario of a financial data detection system provided in the present specification is provided. As shown in FIG. 1, the financial data detection system may include at least a client cluster and a service platform 100.

The client cluster may include at least one client, as shown in fig. 1, specifically including a client 1 corresponding to a user 1, a client 2 corresponding to a user 2, …, and a client n corresponding to a user n, where n is an integer greater than 0.

Each client in the client cluster may be a communication-enabled electronic device including, but not limited to: wearable devices, handheld devices, personal computers, tablet computers, vehicle-mounted devices, smart phones, computing devices, or other processing devices connected to a wireless modem, etc. Electronic devices in different networks may be called different names, for example: a user equipment, an access terminal, a subscriber unit, a subscriber station, a mobile station, a remote terminal, a mobile device, a user terminal, a wireless communication device, a user agent or user equipment, a cellular telephone, a cordless telephone, a personal digital assistant (personal digital assistant, PDA), an electronic device in a 5G network or future evolution network, and the like.

The service platform 100 may be a separate server device, such as: rack-mounted, blade, tower-type or cabinet-type server equipment or hardware equipment with stronger computing capacity such as workstations, mainframe computers and the like is adopted; the server cluster may also be a server cluster formed by a plurality of servers, and each server in the server cluster may be formed in a symmetrical manner, wherein each server is functionally equivalent and functionally equivalent in a transaction link, and each server may independently provide services to the outside, and the independent provision of services may be understood as no assistance of another server is needed.

In one or more embodiments of the present description, the service platform 100 may establish a communication connection with at least one client in a client cluster, based on which interactions of data in the financial data detection process, such as online transaction data interactions, are completed.

It should be noted that, the service platform 100 establishes a communication connection with at least one client in the client cluster through a network for interactive communication, where the network may be a wireless network, or may be a wired network, where the wireless network includes, but is not limited to, a cellular network, a wireless local area network, an infrared network, or a bluetooth network, and the wired network includes, but is not limited to, an ethernet network, a universal serial bus (universal serial bus, USB), or a controller area network. In one or more embodiments of the specification, techniques and/or formats including HyperText Mark-up Language (HTML), extensible markup Language (Extensible Markup Language, XML), and the like are used to represent data exchanged over a network (e.g., target compression packages). All or some of the links may also be encrypted using conventional encryption techniques such as secure socket layer (Secure Socket Layer, SSL), transport layer security (Transport Layer Security, TLS), virtual private network (Virtual Private Network, VPN), internet protocol security (Internet Protocol Security, IPsec), and the like. In other embodiments, custom and/or dedicated data communication techniques may also be used in place of or in addition to the data communication techniques described above.

The embodiments of the financial data detection system provided in the present specification are the same concept as the financial data detection method in one or more embodiments, and the execution subject corresponding to the financial data detection method in one or more embodiments of the present specification may be the service platform 100 described above. The implementation process of the embodiment of the financial data detection system may be described in detail in the following method embodiments, which are not described herein.

Based on the schematic view of the scenario shown in fig. 1, a detailed description of a financial data detection method provided in one or more embodiments of the present disclosure is provided below.

Referring to fig. 2, a flow diagram of a method of financial data detection, which may be implemented in dependence on a computer program, may be run on a von neumann system-based financial data detection device, is provided for one or more embodiments of the present description. The computer program may be integrated in the application or may run as a stand-alone tool class application. The financial data detection device may be a service platform.

Specifically, the financial data detection method comprises the following steps:

s102: determining subject structure combination dimension information aiming at a financial management system, and controlling the financial management system to carry out financial data summarization processing based on the subject structure dimension information to obtain financial evidence summarization data;

Background of financial data detection: along with the rapid development of the main bodies such as enterprises, the financial data volume of the main bodies such as enterprises is larger and larger, and the problem of the financial quality is also more and more serious, so that a service platform corresponding to the main bodies such as enterprises is required to adopt an efficient financial quality management method so as to ensure the accuracy, the integrity and the consistency of the financial data, the accuracy of the financial data is the basis of the decision of the main bodies such as enterprises, institutions and the like, and inaccurate financial data can have great influence.

For a service platform, a financial management (software) system for a subject such as an enterprise to perform financial management is generally associated with the service platform, such as a GL financial management (software) system: the system comprises two main functions of accounting and management of financial information, and can be subdivided into sub-functions of accounting subject management, credential input, accounting analysis, cash management, budget management and the like. Accounting data in the GL system is very important to the body of the enterprise, etc., because they contain important information such as financial status, operational benefit, etc. of the enterprise.

In practical application, along with the growth of the scale of the main body and the business of enterprises and the like, the amount of the financial data in the financial management system for performing financial management by the service platform is increased increasingly, and thus, certain challenges are brought to the management and quality of the financial data. In an actual application environment, various problems may exist in accounting data in a financial management (software) system such as a GL system, such as data repetition, data deletion, data error, etc., and phenomena such as accounting subject error, multiple entry, missing recording, error recording, etc. are not spent due to diversity of data sources, human factors of entry operation, flexibility of system setting, etc. These problems not only result in the accuracy and authenticity of the financial data becoming unreliable, thereby affecting the management decisions of the enterprise, but may also result in serious financial risks and losses as a result. Based on the method, the service platform can adopt the financial data detection method related to one or more embodiments of the specification to avoid financial risks and loss to ensure the quality of financial data, and improve or even avoid the limitation of financial data detection.

The subject structure combination dimension information refers to (financial) accounting subject structure (COA) combination dimension information, the accounting subject structure (COA) combination dimension information is a COA dimension combination, a common (financial) accounting subject structure COA can refer to fig. 3, fig. 3 is a schematic diagram of a collected subject structure, COA dimensions such as a company section dimension, a cost section dimension, an area end section dimension and the like are shown in fig. 3, the subject structure combination dimension information can be regarded as a COA dimension combination, and financial information corresponding to a designated COA dimension combination indicated by the subject structure combination dimension information is screened out from a large amount of financial data related to a financial management system by presetting subject structure combination dimension information, so that targeted financial data extraction is realized.

For example, taking the department structure combination dimension information as a department segment-department segment as an example, the financial terminal can select the department structure combination dimension information as a department segment-department segment through a service platform, the service platform determines the department segment-department segment combination dimension information aiming at a financial management system, and controls the financial management system to carry out financial data summarization processing based on the department segment-department segment to obtain financial evidence summarization data indicated by the department segment-department segment;

For another example, the set subject structure combination dimension information may also be "section-subject section-cost section", "section-subject section-product section", "section-subject section-area section", or the like.

Further, in the specification, corresponding abnormal index dimensions are associated with the subject structure combination dimension information, the corresponding abnormal index dimensions are used for indicating financial evidence detection under the corresponding abnormal index dimensions of financial evidence summary data extracted from the subject structure combination dimension information, the target abnormal index dimensions associated with the subject structure combination dimension information are recorded and configured in a service platform by a financial end according to expert experience, and in practical application, different abnormal index dimensions (dimensions) are associated with different subject structure combination dimension information;

alternatively, the abnormal index dimension associated with the subject structure combination dimension information may be index data corresponding to subject structure combination data contained as "subject structure combination dimension information", where the index data and the subject structure combination data both belong to subject structure combination dimension information.

In this specification, the subject structure combination dimension information may be one or more groups, and each group of subject structure combination dimension information corresponds to a target abnormality index dimension.

For example, taking subject structure combination dimension information as "department segment-subject segment.+ -." as an example, the associated abnormal index dimension may be "net-ring ratio fluctuation rate", and the "net-ring ratio fluctuation rate" index may be used to indicate a fluctuation condition of measuring subject net in a corresponding accounting subject by a certain department, where the subject net is taken as a basis and an important index of enterprise financial management analysis, and abnormal fluctuation is an index of important attention of accounting.

For example, taking the subject structure combination dimension information as "department segment-subject segment.+ -.)" as an example, the abnormal index dimension associated with the subject structure combination dimension information may be "cash flow fluctuation rate", and the "cash flow fluctuation rate" may be used to indicate the net amount change condition of the relevant subjects such as cash of the operation activity of a certain department and cash of the investment activity;

for example, taking the subject structure combination dimension information as "department segment-subject segment-cost segment.+ -.)" as an example, the abnormal index dimension associated with the subject structure combination dimension information may be "segment value fluctuation", where "segment value fluctuation" may be used to indicate the change condition of the cost corresponding to a certain accounting subject compared with the historical value;

for example, taking the subject structure combination dimension information as a "department segment-subject segment-product segment" as an example, the dimension of the abnormal index associated with the subject structure combination dimension information may be a "balance direction abnormal" index, and the "balance direction abnormal" index may be used to indicate that the balance direction is opposite to the direction that should be used to measure the end of the period of a product corresponding to a certain accounting subject in a certain department;

For example, taking the subject structure combination dimension information as "department segment-subject segment-area segment.+ -.)" as an example, the abnormal index dimension associated with the subject structure combination dimension information may be "balance fluctuation rate", and the "balance fluctuation rate" index may be used to indicate the balance change condition of the area corresponding to a certain accounting subject in a certain department.

In a specific implementation scenario, a service platform pre-creates and maintains a dimension rule base with dimension combination abnormal index, wherein the dimension combination abnormal index dimension rule base comprises a plurality of reference subject structure combination dimension information and rules of the reference subject structure combination dimension information corresponding to the reference abnormal index dimension; specifically, for the relevant abnormal index dimension corresponding to the possible combination situation of the COA segment values (i.e. the reference subject structure combination dimension information), the financial terminal performs input setting to the service platform by means of expert experience, and the service platform stores the reference subject structure combination dimension information and the corresponding reference abnormal index dimension thereof as a detection rule in a dimension combination abnormal index dimension rule base.

Under an implementation scene, a current financial terminal can select subject structure combination dimension information aiming at a financial management system from a dimension combination abnormal index dimension rule base, because the subject structure combination dimension information is associated with a target abnormal index dimension in the dimension combination abnormal index dimension rule base, based on the subject structure combination dimension information selected by the financial terminal and the corresponding target abnormal index dimension, a service platform firstly determines subject structure combination dimension information and then sends the subject structure combination dimension information to the financial management system, financial data managed by the financial management system usually process a large amount of financial detail data, one or more pieces of financial detail data are processed into one or more pieces of financial voucher data, the financial management system can screen and collect a large amount of financial voucher data according to the subject structure combination dimension information, so that financial voucher summary data are generated, and the service platform can acquire the voucher summary data from the financial management system.

For example, as shown in fig. 4, fig. 4 is a schematic diagram of an exemplary financial data structure, where the structure of financial detail data managed by the financial management system may be the structure shown in fig. 4, the structure of financial detail data substantially covers an external primary key field portion, an external source field portion, a COA field (in fig. 4, the COA field is 12 bytes), an amount field portion, the structure of financial credential data substantially covers an external source field portion, a COA field (in fig. 4, the COA field is 12 bytes), a cash flow subject marking field portion, an amount field portion, and a financial credential summary data is a financial credential multidimensional summary data structure, focusing on summary data in different combinations of COA fields, a form such as a "partial field-subject field" corresponding to subject structure combination dimension information, a form such as a "partial field-subject field-cost field" corresponding to subject structure combination dimension information, and so on.

Further, the accounting (or referred to as finance) detail data refers to accounting data with high fine granularity;

the accounting (or financial) voucher data is accounting data corresponding to a transaction voucher, for example, a transaction order may include a plurality of commodities, each commodity is a piece of detail accounting data, and the voucher corresponds to the transaction order.

The financial voucher collecting data is that after accounting (or called financial) voucher data is collected according to different dimensions, the collected voucher meaning can be changed correspondingly, for example, the financial voucher collecting data is collected according to a department section-subject section-cost section', and the focus of the data is the cost generated by the department under the subject.

S104: inputting the financial document summary data into a financial document detection model, and outputting a financial document detection result under a target abnormal index dimension, wherein the target abnormal index dimension is an abnormal index dimension associated with the subject structure combination dimension information;

the financial credential detection result may be an abnormal result or a normal result for the financial credential data;

in one or more embodiments of the present description, a financial credential detection model is obtained in advance based on machine learning model training;

the input to the financial instrument detection model is financial instrument summary data;

the output of the financial document detection model is a financial document detection result after the model performs financial document detection at least from the target abnormal index dimension.

Optionally, after the financial credential summary data is input to the financial credential detection model, the financial credential summary data corresponds to a target abnormal index dimension focused by a user, the control financial credential detection model focuses on performing financial credential detection from the target abnormal index dimension, so as to obtain a first financial credential detection sub-result under the target abnormal index dimension, and the control financial credential detection model can also perform financial credential detection from other non-target abnormal index dimensions to obtain a second financial credential detection sub-result, and the first financial credential detection sub-result and the second financial credential detection sub-result are combined to generate a financial credential detection result, wherein the financial credential detection result contains detection result information under the target abnormal index dimension.

S106: and carrying out abnormal credential data processing based on the financial credential detection result.

Optionally, whether an abnormal result exists is determined based on the financial credential detection result, financial credential data indicated by the abnormal result is obtained, the obtained financial credential data is abnormal credential data, and abnormal credential data processing is performed based on a preset abnormal credential data processing rule.

In one possible implementation, the abnormality pre-warning may be performed based on the result of the detection of the financial document, and the abnormality pre-warning information is sent to the relevant financial terminal after the abnormality is determined to be detected based on the result of the detection of the financial document, for example, the abnormality pre-warning information includes at least the abnormality document data, the abnormality cause, and so on.

Further, the financial terminal can acquire the abnormal credential data based on the abnormal early warning information to carry out expert-terminal financial audit so as to confirm whether the abnormal credential data is abnormal, correct the abnormal financial (accounting) data of the financial management system, and correct the accounting detail data corresponding to the abnormal credential data, thereby improving the quality and accuracy of the accounting data.

Optionally, the service platform performs the determining the subject structure combination dimension information for the financial management system, which may be:

the service platform determines subject structure combination dimension information for the financial management system from a dimension combination anomaly index dimension rule base through the financial end, in some embodiments, the subject structure combination dimension information is associated with a target anomaly index dimension in the dimension combination anomaly index dimension rule base.

Referring to fig. 5, fig. 5 is a schematic flow chart of maintenance and update of a dimension rule base of a dimension combination anomaly index according to one or more embodiments of the present disclosure. Specific:

s202: acquiring reference subject structure combination dimension information, wherein the reference subject structure combination dimension information consists of a plurality of subject structure sections;

the reference subject structure combination dimension information is used for updating and maintaining a dimension combination abnormal index dimension rule base, and the dimension combination abnormal index dimension rule base is a database formed by rules corresponding to various reference subject structure combination dimension information.

For each piece of reference subject structure combination dimension information, the reference subject structure combination dimension information is composed of a plurality of subject structure segments, wherein the subject structure segments are field results in an accounting subject structure (COA for short), for example, the subject structure segments comprise COA dimensions including but not limited to company segment dimension, department segment dimension, cost segment dimension, regional end segment dimension and the like;

S204: determining a reference abnormal index dimension corresponding to the reference subject structure combination dimension information, and establishing a dimension combination abnormal index dimension mapping relation between the reference subject structure combination dimension information and the reference abnormal index dimension;

s206: updating the dimension mapping relation of at least one dimension combination abnormal index into a dimension rule base of the dimension combination abnormal index.

Specifically, for the "combination of COA segment values" of interest for performing abnormality detection of financial data, as one type of reference subject structure combination dimension information, the reference subject structure combination dimension information is composed of a plurality of subject structure segments. The reference subject structure combination dimension information is set by the finance end through the expert experience and is input to the service platform, for example, the reference subject structure combination dimension information is 'department section-subject section-region section' and 'department section-subject section-product section' and 'department section-subject section' and the like.

After the service platform acquires the reference subject structure combination dimension information input by the financial terminal, determining a reference abnormal index dimension corresponding to the reference subject structure combination dimension information, wherein the reference abnormal index dimension can also be a financial data abnormal detection dimension selected by the financial terminal aiming at the reference subject structure combination dimension information, such as whether a net-ring ratio fluctuation rate dimension is an abnormal dimension, such as whether a cash flow fluctuation rate is an abnormal dimension, and the like, the service platform takes the reference subject structure combination dimension information and the reference abnormal index dimension corresponding to the reference subject structure combination dimension information as a detection rule, and the detection rule exists in the form of a dimension combination abnormal index dimension mapping relation of the reference subject structure combination dimension information and the reference abnormal index dimension, so that the dimension combination abnormal index dimension mapping relation is stored in a dimension combination abnormal index dimension rule library.

In the specification, corresponding abnormal index dimensions are associated with the subject structure combination dimension information, the corresponding abnormal index dimensions are used for indicating financial evidence summary data extracted from the subject structure combination dimension information to carry out financial evidence detection under the corresponding abnormal index dimensions, the target abnormal index dimensions associated with the subject structure combination dimension information are configured by a financial end according to expert experience recorded in a service platform, and in practical application, different abnormal index dimensions (dimensions) are associated with different subject structure combination dimension information;

s208: determining sample subject structure combination dimension information aiming at a financial management system from the dimension combination abnormal index dimension rule base through a financial terminal, wherein the sample subject structure combination dimension information is associated with sample abnormal index dimensions in the sample dimension combination abnormal index dimension rule base;

s210: controlling the financial management system to carry out financial data summarization processing based on the sample subject structure dimension information to obtain sample financial voucher summarization data;

under the actual implementation scene, a financial evidence detection model needs to be trained, and sample data acquisition and labeling are related to before the financial evidence detection model.

Specifically, the financial terminal may select sample subject structure combination dimension information for the financial management system from a dimension combination abnormal index dimension rule base, the sample subject structure combination dimension information is used to obtain sample financial credential summary data serving as a model training sample, because the sample subject structure combination dimension information is associated with sample abnormal index dimensions in the dimension combination abnormal index dimension rule base, based on this, the service platform determines the sample subject structure combination dimension information selected by the financial terminal and the corresponding sample abnormal index dimensions, then the service platform sends the sample subject structure combination dimension information to the financial management system, the financial data managed by the financial management system usually processes a large amount of financial detail data, processes one or more pieces of financial credential data into one or more pieces of financial credential data, and the financial management system performs screening and summarizing processing on a large amount of financial data fields of specified COA combination dimensions based on the sample subject structure combination dimension information, so as to generate sample financial credential summary data, and the service platform may obtain the sample credential summary data from the financial credential management system.

Typically, a large amount of sample financial credential summary data will be collected from the perspective of structural dimension information from different orders.

S212: and labeling the sample data labels on the sample financial voucher collecting data based on the sample abnormal index dimension, taking the sample data labels and the sample financial voucher collecting data as model training data of an initial financial voucher detection model, and generating the financial voucher detection model after model training by adopting the sample data labels and the sample financial voucher collecting data.

In one possible implementation, the sample financial document summary data may be marked in a semi-supervised data marking manner, that is, a sample data tag is marked on a portion of the sample financial document summary data. The service platform can be marked by adopting expert service, sample data marking is carried out by adopting expert service according to corresponding sample abnormal index dimensions set by a financial staff according to a dimension combination abnormal index dimension rule base, a financial evidence detection model is trained by adopting semi-supervised learning, the financial staff does not need to mark all sample financial evidence summary data with sample data labels, only part of sample financial evidence summary data in the sample financial evidence summary data need to be marked, and the sample data labels can be data abnormal probability or data detection results (abnormal results or normal results) marked from the sample abnormal index dimensions.

In a feasible implementation mode, in order to ensure the training quality, the system sets different labeling quantity ratios for sample financial voucher summarized data of different kinds of the dimension information of the subject structure, at least the dimension summary according to different sample subject structure dimension information is considered, more data after partial combination dimension summary is generated, less data after partial combination dimension summary is generated, the labeling by adopting the same ratio can be avoided, and the model training effect is ensured.

Specifically, labeling the sample financial credential summary data with a sample data tag based on the sample anomaly index dimension may be:

a2: the service platform determines first financial evidence summary data and first abnormal index dimensions corresponding to the first financial evidence summary data from the sample financial evidence summary data, and determines target abnormal index dimension results corresponding to the first abnormal index dimensions;

the first financial document summary data is data that the sample financial document summary data needs to be marked with a data tag, and the data in the sample financial document summary data except the first financial document summary data does not need to be marked.

The first anomaly index dimension is an anomaly index dimension associated with the first financial credential summary data.

The objective abnormal index dimension result can be marked by adopting (financial) expert service to carry out abnormal index dimension result in advance;

illustratively, determining the first financial credential summary data from the sample financial credential summary data may be as follows:

1. determining sample abnormal index dimensions corresponding to the sample financial evidence summary data, and obtaining preset labeling proportion information corresponding to different sample abnormal index dimensions;

2. first financial credential summary data is determined from the sample financial credential summary data based on the annotation proportion information.

It can be understood that the "sample financial credential summary data" of different dimensions and different kinds of subject structure dimension information are set in advance with different proportions of marking quantity ratios, the marking quantity ratios corresponding to the different kinds of subject structure dimension information form marking ratio information, the marking ratio information can be manually set by a service platform based on expert terminal services, the marking ratio information can be a ratio parameter, so that sample financial credential summary data are obtained by summarizing according to different dimensions of the "sample subject structure dimension information", then first financial credential summary data are determined from the sample financial credential summary data according to the marking ratio information, then marking is performed on the first financial credential summary data, for example, if the marking ratio of the sample anomaly index dimension a is 20%, sample financial credential summary data corresponding to the sample anomaly index dimension a is obtained as the first financial credential summary data, and if the marking ratio of the sample anomaly index dimension B is 30%, sample financial credential summary data corresponding to the sample anomaly index dimension B is obtained as the first.

A4: and labeling the target abnormal index dimension result as a sample data tag of the first financial evidence summary data.

The service platform can call expert service through the financial terminal in advance, manual experience can be adopted to label sample data labels for the first financial voucher summarized data from the target abnormal index dimension, the first financial voucher summarized data only need to be labeled, and the sample data labels can be data abnormal probability or data detection results (abnormal results or normal results) labeled from the sample abnormal index dimension.

In the specification, the sample data tag of the first financial document summary data is marked by a small amount of marking data, so that generalization capability of the financial document detection model in the subsequent model training stage can be enhanced, and the accuracy of the financial document detection model is improved. In the scene of detecting abnormal quality of the financial data, due to the importance of the financial data, when the abnormality occurs, the financial terminal usually calls expert service to clearly judge the abnormality details, the process marks part of data so as to generate a target abnormality index dimension result, a small amount of mark data, namely a sample data label of the target abnormality index dimension result, marks the first financial evidence summary data, and then semi-supervises and learns the financial evidence detection model, so that the accuracy of abnormality detection can be improved;

In one or more embodiments of the present disclosure, a semi-supervised learning method is adopted to perform data labeling, and the semi-supervised learning is used to train and detect the model by combining a small amount of known abnormal data and a large amount of unlabeled data, so that the abnormal data labeling determined by the financial abnormality can be effectively used, and the accuracy of the model can be improved by using the small amount of data in the part; and, establishing an abnormal rule base (namely, a dimension rule base of the dimension combination abnormal index) of the financial data aiming at the financial management system. The financial management system uses the subjects section and the dimension of auxiliary accounting to form the COA section, the COA section is summarized according to different dimensions in the subjects structure dimension information to obtain summarization certificates of different semantics, and corresponding abnormal indexes are defined, so that a set of clear, visual and high-interpretation abnormal rule base is established.

Referring to fig. 6, fig. 6 is a schematic flow chart of maintenance and update of a dimension rule base of a dimension combination anomaly index according to one or more embodiments of the present disclosure. Specific:

s302: creating an initial financial credential detection model for a financial management system;

model creation: creating an initial financial credential detection model based on the machine learning model;

It should be noted that the machine learning model according to one or more embodiments of the present disclosure includes, but is not limited to, fitting of one or more of an artificial intelligence content generation model, a convolutional neural network (Convolutional Neural Network, CNN) model, a deep neural network (Deep Neural Network, DNN) model, a recurrent neural network (Recurrent Neural Networks, RNN), a model, an embedding (embedding) model, a gradient boosting decision tree (Gradient Boosting Decision Tree, GBDT) model, a logistic regression (Logistic Regression, LR) model, and the like.

S302: determining sample subject structure combination dimension information aiming at a financial management system from a dimension combination abnormal index dimension rule base through a financial terminal, controlling the financial management system to carry out financial data summarization processing based on the sample subject structure dimension information to obtain sample financial evidence summarization data, and labeling sample data labels on the sample financial evidence summarization data based on the sample abnormal index dimension;

Specifically, the financial terminal may select sample subject structure combination dimension information for the financial management system from a dimension combination abnormal index dimension rule base, the sample subject structure combination dimension information is used to obtain sample financial credential summary data serving as a model training sample, because the sample subject structure combination dimension information is associated with sample abnormal index dimensions in the dimension combination abnormal index dimension rule base, based on this, the service platform determines the sample subject structure combination dimension information selected by the financial terminal and the corresponding sample abnormal index dimensions, then the service platform sends the sample subject structure combination dimension information to the financial management system, the financial data managed by the financial management system usually processes a large amount of financial detail data, processes one or more pieces of financial credential data into one or more pieces of financial credential data, and the financial management system performs screening and summarizing processing on a large amount of financial data fields of specified COA combination dimensions based on the sample subject structure combination dimension information, so as to generate sample financial credential summary data, and the service platform may obtain the sample credential summary data from the financial credential management system. Typically, a large amount of sample financial credential summary data will be collected from the perspective of structural dimension information from different orders.

S306: and performing at least one round of model training on the initial financial voucher detection model based on the sample financial voucher summary data and the sample data label to obtain a model-trained financial voucher detection model.

In one possible implementation, the model training process is as follows:

model forward propagation training: inputting sample financial document summary data into a financial document detection model for at least one round of model training, and determining a predicted financial document detection result by each round of model training;

Model back propagation fine tuning: in each round of model training process, calculating model loss based on a predicted financial evidence detection result and a sample data label, and adjusting model coefficients of the financial evidence detection model based on the model loss until the financial evidence detection model meets a model training ending condition to obtain a trained domain classification model;

alternatively, the model loss may be calculated using a model loss function set to predict financial credential detection results and sample data tags as inputs, e.g., the model loss function may be a Euclidean distance loss function, a cross entropy loss function, a hinge loss function, and so forth.

In a feasible implementation mode, considering that semi-supervised learning lacks labels due to a large amount of data, the generalization capability of a model is improved, model training can be faster, a pseudo label with weak enhancement data can be used for training in the training process, and in order to improve the accuracy of the pseudo label, the method is introduced into a model structure of a financial evidence detection model by integrating a plurality of different anomaly detection parts, and the plurality of different anomaly detection parts are integrated to serve as an integrated fusion network.

Further, according to some embodiments, the sample financial credential summary data is comprised of first financial credential summary data and second financial credential summary data, the first financial credential summary data being sample financial credential summary data including a sample data tag, the second financial credential summary data being sample financial credential summary data not including a sample data tag;

Further, executing the model training for the initial financial document detection model based on the sample financial document summary data and the sample data tag to obtain a model-trained financial document detection model may be performed by the following manner:

b2: inputting the sample financial document summary data into an initial financial document detection model;

b4: in the model forward propagation training process, generating a sample data pseudo tag for the second financial credential summary data through the initial financial credential detection model, taking the sample data pseudo tag as the sample data tag of the second financial credential summary data, and determining a sample financial credential detection result under a sample abnormal index dimension;

illustratively, as shown in FIG. 7, FIG. 7 is an exemplary model architectural diagram of a financial credential detection model, the initial financial credential detection model including at least a data enhancement network including a data Encoder Encoder and a data Decoder Decoder, and an integrated converged network; the integrated fusion network is a multi-model fusion part using Bagging integration technology, and is composed of a plurality of models (which can be called detection classifiers) for anomaly detection.

In the present specification, based on the data enhancement network of the data Encoder and the data Decoder, considering that the summarized certificates are a string of financial coding text with a small number of characters, the conventional text data enhancement scheme is difficult to use in the text data enhancement network, and the data enhancement network adopting the self Encoder can use the strong feature extraction capability of deep learning to mine deep semantic features, thereby amplifying abnormal features and achieving the effect of data enhancement.

The following is a definition of network processing in the forward propagation of the data enhancement network part:

assuming that the (sample) financial voucher summarized according to the multi-dimensional combination is input as X and y is the corresponding sample data label, an input data set is formed, a small part of first financial voucher summarized data in the data set is provided with a real label through financial labeling, and in addition, a large part of data is not provided with a corresponding real label, and the training is carried out by producing a pseudo label through a data enhancement network and an integration fusion network of a model.

Illustratively, after the (sample) financial credential summary data X is input to the data Encoder, the data Encoder encodes it, and the encoded sample encoding characteristics may be expressed as:

Z＝δ(WX+b)，

Where δ () is the nonlinear activation function of the data Encoder and w and b are the weight and bias characterizations of the data Encoder linear transformation. The sample coding feature may also be referred to as an implicit layer feature Z, and the data Decoder decodes the implicit layer feature Z to obtain reconstructed output model reconstructed data X':

X'＝δ'(W'Z+b')

where δ () ' is the nonlinear activation function of the data Decoder, and W ' and b ' are the weight and bias characterizations of the data Decoder linear transformation.

The model reconstruction data X' corresponds to (sample) financial credential summary data X;

further, executing the generating the sample data pseudo tag for the second financial credential summary data by the initial financial credential detection model may be performed as follows:

1. controlling the data Encoder Encoder to perform feature encoding on the second financial evidence summary data to obtain sample encoding features Z, and controlling the data Decoder Decode to perform data reconstruction on the sample encoding features Z to obtain model reconstruction data X';

2. controlling a plurality of detection classifiers of the integrated fusion network to determine a first financial credential detection result based on the sample coding features and the model reconstruction data;

3. And carrying out fusion processing on the basis of each first financial document detection result to obtain a second financial document detection result, and taking the second financial document detection result as a sample data pseudo tag aiming at the second financial document summary data.

The input (sample) financial document summary data X is encoded by a data Encoder Encoder to obtain a sample encoding characteristic Z with reduced dimension, the degree of abnormality in the financial document summary data is increased, and the input is enhanced strongly; and then the data is decoded by a data Decoder and reconstructed to obtain data X ' "reconstructed by a data model which is the same as the input financial evidence summary data X '" in the same dimension, and then the data X ' "and the data model are similar under the constraint of a loss function, and the input is weakly enhanced. The coded hidden layer feature Z (i.e. the sample coding feature Z) and the decoded output financial evidence summary data X are respectively sent into an integrated fusion network, wherein the integrated fusion network can comprise a plurality of detection classifiers (the number is usually more than 2), and each detection classifier can be of a different model structure;

performing anomaly detection processing on the basis of the sample coding features and the model reconstruction data through each detection classifier to obtain a predicted financial evidence detection sub-result, and then synthesizing the predicted financial evidence detection sub-results to obtain a first financial evidence detection result, wherein the predicted financial evidence detection sub-result is a probability parameter for representing the anomaly degree;

For example, as shown in fig. 7, the integrated fusion network includes at least three models, i.e., an iforst (isolated forest) model, a LOF (local outlier factor) model, and a K-Means (K-Means) model, where in fig. 7, the iforst model mainly focuses on global outliers, the LOF model focuses on local outliers, and the K-Means model is sensitive to outliers and can be used as a supplementary model, and the predicted financial credential detection sub-results of the three classifiers are weighted and summed to obtain a first financial credential detection result;

further, the sample data tag does not exist in the second financial document summary data in the model input data, and here, the first financial document detection result corresponding to the second financial document summary data can be used as the sample data pseudo tag y'.

Based on this, the first financial credential summary data corresponds to a real sample data tag and the second financial credential summary data corresponds to a sample data pseudo tag y'.

B6: in the model back propagation training process, determining model reconstruction data aiming at sample financial voucher collecting data based on the initial financial voucher detection model, and carrying out parameter adjustment on the initial financial voucher detection model based on the model reconstruction data, the sample financial voucher collecting data, the sample financial voucher detection result and the sample data label until the initial financial voucher detection model meets a model finishing training condition to obtain a financial voucher detection model.

1. Firstly, determining data consistency loss based on model reconstruction data and sample financial voucher summarized data, and determining model detection loss based on a sample financial voucher detection result and the sample data label;

2. then, adopting data consistency loss to adjust first network parameters of a data encoder and a data decoder in the data enhancement network, and adopting detection loss to adjust second network parameters of the integrated fusion network;

the data consistency loss may be a set model loss function, which may be a Euclidean distance loss function, a cross entropy loss function, a hinge loss function, or the like.

Illustratively, according to the definition above, after the (sample) financial credential summary data X is input to the data Encoder Encoder, the data Encoder Encoder encodes it, and the encoded sample encoding characteristics can be expressed as:

Z＝δ(WX+b)，

X'＝δ'(W'Z+b')

A data consistency penalty may employ a model penalty function as follows:

LOSS ₁ ＝min _T ||X-[δ'(W'(δ(wx+b)+b')]|| ²

wherein LOSS ₁ Data consistency loss, T is model parameters of a data Decoder and a data Decoder, such as a nonlinear activation function characterization delta (), a linear variation in a data Encoder EncoderThe weight representation W and the bias representation b of the conversion are also represented by nonlinear activation function representation delta () ' in a data encoder Decoder, the weight representation of linear transformation, W ' and the bias representation b ';

specifically, a model detection loss is determined based on the sample financial credential detection result and the sample data tag. And a set model loss function can be adopted, a sample financial certificate detection result and the sample data label are input into the model loss function to output model detection loss, and the model loss function can be a Euclidean distance loss function, a cross entropy loss function, a hinge loss function and the like.

The model detection loss is a set loss function for the integrated fusion network, and is used for fine-tuning model parameters of the integrated fusion network based on the model detection loss.

Optionally, the integrated fusion network is generally composed of a plurality of different detection classifiers, the different detection classifiers can obtain respective prediction financial credential detection sub-results, the prediction financial credential detection sub-results can be regarded as sample financial credential detection results of one detection classifier, and model detection losses of the different detection classifiers can be calculated.

It can be appreciated that by equating the output of the data enhancement network to the input during the back propagation process in combination with the data consistency loss, to minimize the loss function training the data enhancement network, the better (model) characterization parameters of the data Encoder and the data Encoder Decoder are obtained by the solution;

in one or more embodiments of the present disclosure, on the one hand, a semi-supervised learning method is adopted, and in combination with a small amount of known abnormal data and a large amount of unlabeled data, the semi-supervised learning is used to perform model training on a financial document detection model, so that not only can abnormal data label distinguished by financial distinguishing abnormality be effectively utilized, but also the accuracy of the model can be improved by utilizing the small amount of data; on the other hand, by combining a data enhancement network of an integrated learning technology, semi-supervised learning is performed by creatively introducing a pseudo tag of weak enhancement data in a training process due to the fact that a large amount of data lacks tags, and in order to improve accuracy of the pseudo tag, the scheme improves generalization capability of the model by integrating a plurality of data enhancement networks of different anomaly detection models, so that model training can be faster, and a financial evidence detection model obtained by training can perform financial data detection better.

The financial data detecting apparatus provided in the present specification will be described in detail with reference to fig. 8. Note that, the financial data detecting apparatus shown in fig. 8 is used to perform the method of the embodiment shown in fig. 1 to 6 of the present specification, and for convenience of explanation, only the portion relevant to the present specification is shown, and specific technical details are not disclosed, please refer to the embodiment shown in fig. 1 to 7 of the present specification.

Referring to fig. 8, a schematic diagram of the structure of the financial data detecting apparatus of the present specification is shown. The financial data detection apparatus 1 may be implemented as all or part of a device by software, hardware or a combination of both. According to some embodiments, the financial data detection apparatus 1 comprises an information determination module 11, a financial model processing module 12 and a data processing module 13, in particular for:

the information determining module 11 is configured to determine subject structure combination dimension information for a financial management system, and control the financial management system to perform financial data summarization processing based on the subject structure dimension information to obtain financial credential summarization data;

the model processing module 12 is configured to input the financial document summary data into a financial document detection model, and output a financial document detection result under a target abnormal index dimension, where the target abnormal index dimension is an abnormal index dimension associated with the subject structure combination dimension information;

And the data processing module 13 is used for carrying out abnormal credential data processing based on the financial credential detection result.

Optionally, the information determining module 11 is configured to:

and determining subject structure combination dimension information aiming at the financial management system from a dimension combination abnormal index dimension rule base through a financial terminal, wherein the subject structure combination dimension information is associated with a target abnormal index dimension in the dimension combination abnormal index dimension rule base.

Optionally, the information determining module 11 is configured to:

acquiring reference subject structure combination dimension information, wherein the reference subject structure combination dimension information consists of a plurality of subject structure sections;

determining a reference abnormal index dimension corresponding to the reference subject structure combination dimension information, and establishing a dimension combination abnormal index dimension mapping relation between the reference subject structure combination dimension information and the reference abnormal index dimension;

updating the dimension mapping relation of at least one dimension combination abnormal index into a dimension rule base of the dimension combination abnormal index.

Optionally, the information determining module 11 is configured to:

determining sample subject structure combination dimension information aiming at a financial management system from the dimension combination abnormal index dimension rule base through a financial terminal, wherein the sample subject structure combination dimension information is associated with sample abnormal index dimensions in the sample dimension combination abnormal index dimension rule base;

Controlling the financial management system to carry out financial data summarization processing based on the sample subject structure dimension information to obtain sample financial voucher summarization data;

and labeling the sample data labels on the sample financial voucher collecting data based on the sample abnormal index dimension, taking the sample data labels and the sample financial voucher collecting data as model training data of an initial financial voucher detection model, and generating the financial voucher detection model after model training by adopting the sample data labels and the sample financial voucher collecting data.

Optionally, the information determining module 11 is configured to:

determining first financial evidence summary data and first abnormal index dimensions corresponding to the first financial evidence summary data from the sample financial evidence summary data, and determining target abnormal index dimension results corresponding to the first abnormal index dimensions;

and labeling the target abnormal index dimension result as a sample data tag of the first financial evidence summary data.

Optionally, the information determining module 11 is configured to:

determining sample abnormal index dimensions corresponding to the sample financial evidence summary data, and obtaining preset labeling proportion information corresponding to different sample abnormal index dimensions;

First financial credential summary data is determined from the sample financial credential summary data based on the annotation proportion information.

It should be noted that, when executing the financial data detection method, the financial data detection apparatus provided in the foregoing embodiment is only exemplified by the division of the foregoing functional modules, and in practical application, the foregoing functional allocation may be performed by different functional modules according to needs, that is, the internal structure of the device is divided into different functional modules, so as to complete all or part of the functions described above. In addition, the financial data detection device and the financial data detection method provided in the foregoing embodiments belong to the same concept, which embody the detailed implementation process in the method embodiment, and are not described herein again.

The foregoing description is provided for the purpose of illustration only and does not represent the advantages or disadvantages of the embodiments.

Referring to fig. 9, a schematic diagram of a financial document detection model training apparatus according to the present disclosure is shown. The financial credential detection model training device 2 may be implemented as all or part of a device by software, hardware or a combination of both. According to some embodiments, the financial credential detection model training device 2 comprises a model creation module 21, a data summarization module 22 and a model training module 23, in particular for:

A model creation module 21 for creating an initial financial credential detection model for the financial management system;

the data summarizing module 22 is configured to determine, through a financial end, sample subject structure combination dimension information for a financial management system from a dimension combination abnormal index dimension rule base, control the financial management system to perform financial data summarizing processing based on the sample subject structure dimension information to obtain sample financial credential summarizing data, and label the sample financial credential summarizing data with a sample data tag based on the sample abnormal index dimension;

the model training module 23 is configured to perform at least one round of model training on the initial financial credential detection model based on the sample financial credential summary data and the sample data tag, to obtain a model-trained financial credential detection model.

Optionally, the sample financial document summary data is composed of first financial document summary data and second financial document summary data, the first financial document summary data is sample financial document summary data containing sample data tags, the second financial document summary data is sample financial document summary data not containing sample data tags,

The model training module 23 is configured to:

inputting the sample financial document summary data into an initial financial document detection model;

in the model forward propagation training process, generating a sample data pseudo tag for the second financial credential summary data through the initial financial credential detection model, taking the sample data pseudo tag as the sample data tag of the second financial credential summary data, and determining a sample financial credential detection result under a sample abnormal index dimension;

in the model back propagation training process, determining model reconstruction data aiming at sample financial voucher collecting data based on the initial financial voucher detection model, and carrying out parameter adjustment on the initial financial voucher detection model based on the model reconstruction data, the sample financial voucher collecting data, the sample financial voucher detection result and the sample data label until the initial financial voucher detection model meets a model finishing training condition to obtain a financial voucher detection model.

Optionally, the initial financial credential detection model includes at least a data enhancement network and an integrated fusion network, the data enhancement network including a data encoder and a data decoder,

The model training module 23 is configured to:

controlling the data encoder to perform feature encoding on second financial evidence summary data to obtain sample encoding features, and controlling the data decoder to perform data reconstruction on the sample encoding features to obtain model reconstruction data;

controlling a plurality of detection classifiers of the integrated fusion network to determine a first financial credential detection result based on the sample coding features and the model reconstruction data;

and carrying out fusion processing on the basis of each first financial document detection result to obtain a second financial document detection result, and taking the second financial document detection result as a sample data pseudo tag aiming at the second financial document summary data.

Optionally, the model training module 23 is configured to:

determining a data consistency loss based on the model reconstruction data and the sample financial credential summary data, determining a model detection loss based on the sample financial credential detection result and the sample data tag;

performing a first network parameter adjustment on the data encoder and the data decoder in the data enhancement network using the data consistency loss;

and adopting the model to detect loss to adjust second network parameters of the integrated fusion network.

The present disclosure further provides a computer storage medium, where a plurality of instructions may be stored, where the instructions are adapted to be loaded by a processor and executed by the processor, where the specific execution process may refer to the specific description of the embodiment shown in fig. 1 to 7, and the details are not repeated herein.

The present disclosure further provides a computer program product, where at least one instruction is stored, where the at least one instruction is loaded by the processor and executed by the processor to perform the financial data detection method according to the embodiment shown in fig. 1 to 7, and the specific execution process may refer to the specific description of the embodiment shown in fig. 1 to 7, which is not repeated herein.

Referring to fig. 10, a block diagram of an electronic device according to an embodiment of the present disclosure is provided. The electronic device in this specification may include one or more of the following: processor 110, memory 120, input device 130, output device 140, and bus 150. The processor 110, the memory 120, the input device 130, and the output device 140 may be connected by a bus 150.

Processor 110 may include one or more processing cores. The processor 110 connects various parts within the overall terminal using various interfaces and lines, performs various functions of the terminal 100 and processes data by executing or executing instructions, programs, code sets, or instruction sets stored in the memory 120, and invoking data stored in the memory 120. Alternatively, the processor 110 may be implemented in at least one hardware form of digital signal processing (digital signal processing, DSP), field-programmable gate array (field-programmable gate array, FPGA), programmable logic array (programmable logic Array, PLA). The processor 110 may integrate one or a combination of several of a central processor (central processing unit, CPU), an image processor (graphics processing unit, GPU), and a modem, etc. The CPU mainly processes an operating system, a user interface, an application program and the like; the GPU is used for being responsible for rendering and drawing of display content; the modem is used to handle wireless communications. It will be appreciated that the modem may not be integrated into the processor 110 and may be implemented solely by a single communication chip.

The memory 120 may include a random access memory (random Access Memory, RAM) or a read-only memory (ROM). Optionally, the memory 120 includes a non-transitory computer readable medium (non-transitory computer-readable storage medium). Memory 120 may be used to store instructions, programs, code, sets of codes, or sets of instructions.

The input device 130 is configured to receive input instructions or data, and the input device 130 includes, but is not limited to, a keyboard, a mouse, a camera, a microphone, or a touch device. The output device 140 is used to output instructions or data, and the output device 140 includes, but is not limited to, a display device, a speaker, and the like. In the embodiment of the present disclosure, the input device 130 may be a temperature sensor for acquiring an operation temperature of the terminal. The output device 140 may be a speaker for outputting audio signals.

In addition, those skilled in the art will appreciate that the configuration of the terminal illustrated in the above-described figures does not constitute a limitation of the terminal, and the terminal may include more or less components than illustrated, or may combine certain components, or may have a different arrangement of components. For example, the terminal further includes components such as a radio frequency circuit, an input unit, a sensor, an audio circuit, a wireless fidelity (wireless fidelity, WIFI) module, a power supply, a bluetooth module, and the like, which are not described herein again.

In the embodiment of the present specification, the execution subject of each step may be the terminal described above. Optionally, the execution subject of each step is an operating system of the terminal. The operating system may be an android system, an IOS system, or other operating systems, which embodiments of the present specification are not limited to.

In the electronic device of fig. 10, processor 110 may be configured to invoke programs stored in memory 120 and execute to implement the financial data detection method and/or the financial credential detection model training method as described in various method embodiments of the present description.

Those skilled in the art will appreciate that implementing all or part of the above-described methods in accordance with the embodiments may be accomplished by way of a computer program stored on a computer readable storage medium, which when executed may comprise the steps of the embodiments of the methods described above. The storage medium may be a magnetic disk, an optical disk, a read-only memory, a random access memory, or the like.

It should be noted that, information (including but not limited to user equipment information, user personal information, etc.), data (including but not limited to data for analysis, stored data, presented data, etc.), and signals according to the embodiments of the present disclosure are all authorized by the user or are fully authorized by the parties, and the collection, use, and processing of relevant data is required to comply with relevant laws and regulations and standards of relevant countries and regions. For example, objects, data, information, and the like referred to in this specification are all acquired with sufficient authorization.

The foregoing disclosure is only illustrative of the preferred embodiments of the present invention and is not to be construed as limiting the scope of the claims, which follow the meaning of the claims of the present invention.

Claims

1. A method of financial data detection, the method comprising:

2. The method of claim 1, the determining subject structure combination dimension information for a financial management system, comprising:

3. The method of claim 2, the method further comprising:

4. The method of claim 3, further comprising, after updating at least one of the dimension combination anomaly index dimension mappings into a dimension combination anomaly index dimension rule base:

5. The method of claim 4, the tagging the sample financial credential summary data with a sample data tag based on the sample anomaly metrics dimension, comprising:

6. The method of claim 5, the determining first financial credential summary data from the sample financial credential summary data comprising:

7. A financial credential detection model training method, the method comprising:

8. The method of claim 7, the sample financial credential summary data consisting of first financial credential summary data that is sample financial credential summary data that includes a sample data tag and second financial credential summary data that is sample financial credential summary data that does not include a sample data tag,

The model training is performed on the initial financial voucher detection model for at least one round based on the sample financial voucher summary data and the sample data label to obtain a model-trained financial voucher detection model, and the model training method comprises the following steps:

9. The method of claim 8, wherein the initial financial credential detection model comprises at least a data enhancement network and an integrated fusion network, the data enhancement network comprising a data encoder and a data decoder,

the generating, by the initial financial credential detection model, a sample data pseudo tag for the second financial credential summary data, comprising:

10. The method of claim 9, the parameter adjusting the initial financial credential detection model based on the model reconstruction data, the sample financial credential summary data, the sample financial credential detection results, and the sample data tag, comprising:

11. A financial data detection apparatus, the apparatus comprising:

12. A financial instrument detection model training apparatus, the apparatus comprising:

13. A computer storage medium storing a plurality of instructions adapted to be loaded by a processor and to perform the method steps of any one of claims 1 to 6 or 7 to 10.

14. A computer program product storing at least one instruction for loading by a processor and performing the method steps of any of claims 1-6 or 7-10.

15. An electronic device, comprising: a processor and a memory; wherein the memory stores a computer program adapted to be loaded by the processor and to perform the method steps of any of claims 1-6 or 7-10.