CN118211832A - Financial tax data risk monitoring method, system, electronic equipment and storage medium - Google Patents

Abstract

The invention discloses a financial and tax data risk monitoring method and system, electronic equipment and a computer readable storage medium. According to the method, original financial and tax data are collected and preprocessed, a risk monitoring model is built by means of a machine learning algorithm, financial and tax data requests are responded, financial and tax information to be checked and relevant characteristic information sets are obtained, classification and check processing are conducted, abnormal information is marked, and risk early warning information is generated. The method improves the efficiency and accuracy of financial tax data risk management, is favorable for timely finding and coping with financial tax risks, optimizes financial management flow, and adapts to changes of financial tax policies and environments.

Description

Financial and tax data risk monitoring method, system, electronic device and storage medium

Technical Field

Embodiments of the present invention relate to the field of financial and tax data processing and risk assessment, and in particular, to a financial and tax data risk monitoring method, system, electronic device and storage medium.

Background technique

With the development of economy and the progress of globalization, financial and tax data plays an increasingly important role in enterprise management, policy formulation and risk assessment. The accuracy and completeness of financial and tax data are crucial to the healthy operation of enterprises, the tax collection and management of the government and the economic security of society. However, due to the complexity and diversity of financial and tax data, as well as the possible human errors or system loopholes in the processing process, financial and tax data often face various risks, such as fraud, tax evasion, misreporting, etc. These risks may not only cause enterprises to suffer economic losses or legal penalties, but also affect the economic order and tax fairness of the entire society.

Traditional financial and tax data risk monitoring methods usually rely on manual auditing and simple statistical analysis. These methods are inefficient when processing large-scale, high-dimensional financial and tax data, and it is difficult to accurately identify and assess potential risks.

The above information disclosed in this Background section is only for enhancement of understanding of the background of the inventive concept and therefore it may contain information that does not form the prior art that is already known in this country to a person of ordinary skill in the art.

Summary of the invention

The content of the present invention is used to introduce the concepts in a brief form, which will be described in detail in the detailed implementation section below. The content of the present disclosure is not intended to identify the key features or essential features of the technical solution claimed for protection, nor is it intended to limit the scope of the technical solution claimed for protection.

Some embodiments of the present invention propose a financial and tax data risk monitoring method, system, electronic device and storage medium to solve one or more of the technical problems mentioned in the above background technology section.

In a first aspect, some embodiments of the present invention provide a method for monitoring financial and tax data risks, the method comprising the following steps:

Collecting original financial and tax data, and preprocessing the collected original financial and tax data, wherein the preprocessing includes data standardization, and the original financial and tax data includes corporate financial statements, tax return forms, audit reports, invoice data, and transaction records;

Based on the pre-processed original financial and tax data, a risk monitoring model is constructed using a machine learning algorithm;

Determine the characteristic variables and target variables of the risk monitoring model, wherein the characteristic variables include indicators reflecting the financial and taxation status of the enterprise, and the target variable is the risk level of the financial and taxation data;

In response to receiving the fiscal and taxation data request information sent by the platform, obtaining the fiscal and taxation information to be verified, the first fiscal and taxation feature information set and the second fiscal and taxation feature information set, wherein the first fiscal and taxation feature information set and the second fiscal and taxation feature information set are generated based on historical fiscal and taxation data;

Classify each first fiscal and taxation characteristic information in the first fiscal and taxation characteristic information set to obtain a first classification result, and classify each second fiscal and taxation characteristic information in the second fiscal and taxation characteristic information set to obtain a second classification result;

Inputting the to-be-verified financial and tax information into the pre-built risk monitoring model to obtain the to-be-verified financial and tax feature information, wherein the to-be-verified financial and tax feature information includes feature category information;

Based on the first classification result and the second classification result, verify the financial and tax characteristic information to be verified to obtain a verification result;

If the verification result meets the preset conditions, the financial and tax information to be verified is marked as abnormal information;

In response to the financial and tax information to be verified being marked as abnormal information, corresponding risk warning information is generated based on the characteristic variables and target variables of the risk monitoring model, and the risk warning information includes risk type information, risk level information, risk source information and risk description information.

Optionally, in the above embodiment provided by the present invention, it also includes:

Define a risk warning information structure, wherein each node in the risk warning information structure corresponds to a variable in the risk type, risk level, risk source and risk description, and the connection relationship between each node is the dependency relationship between the variables;

Based on the risk warning information structure, calculating the conditional probability of each of the nodes, the conditional probability being the probability of each of the nodes taking different values;

Based on the characteristic variables and target variables in the risk monitoring model, analyzing the conditional probability to obtain an analysis result;

Based on the analysis results of the conditional probability, risk warning information is generated.

Optionally, in the above embodiment provided by the present invention, it also includes:

Obtain user financial and tax behavior log information from the platform according to the preset data format;

Filter the user's financial and tax behavior log information according to the preset screening principles to identify missing value data;

Correct missing value data according to the preset cleaning algorithm;

The user's financial and tax behavior log information is unified according to preset unified rules, and the unified user's financial and tax behavior log information is input into the pre-trained risk monitoring model to obtain the first financial and tax feature information.

Optionally, in the above embodiment provided by the present invention, it further includes: acquiring multi-source platform financial and tax history information according to a preset data format;

The financial and tax historical information of the multi-source platforms is unified according to preset unified rules, and the unified financial and tax historical information of the multi-source platforms is input into the pre-trained risk monitoring model to obtain the second financial and tax feature information.

Optionally, in the above embodiment provided by the present invention, it also includes:

If the feature category information included in the to-be-verified fiscal and taxation feature information meets the first category condition, matching the to-be-verified fiscal and taxation feature information with each first fiscal and taxation feature information in the first fiscal and taxation feature information set to generate a matching result, thereby obtaining a matching result set;

If the feature category information included in the to-be-verified fiscal and taxation feature information meets the second category condition, matching the to-be-verified fiscal and taxation feature information with each second fiscal and taxation feature information in the second fiscal and taxation feature information set to generate a matching result, thereby obtaining a matching result set;

If there is a matching result that meets the preset matching condition in the matching result set, the first preset verification information is determined as the verification result;

If there is no matching result satisfying the preset matching condition in the matching result set, the second preset verification information is determined as the verification result.

Optionally, in the above embodiment provided by the present invention, it also includes:

Obtaining a training sample set and an initial risk monitoring model, wherein each training sample in the training sample set includes sample financial and tax abnormality information and sample financial and tax abnormality feature information, and the initial risk monitoring model includes a full convolution module, a residual convolution module sequence, and a pooling module;

Selecting training samples from the training sample set and performing training steps specifically include:

Input the sample tax exception information included in the training sample into the full convolution module of the initial risk monitoring model to obtain the full convolution feature information;

Input the full convolution feature information into the residual convolution module sequence of the initial risk monitoring model to obtain residual convolution information;

Input the residual convolution information into the pooling module of the initial risk monitoring model to obtain the initial financial and tax abnormal feature information;

Based on a preset loss function, determine the abnormal difference value between the initial financial and tax abnormal feature information and the sample financial and tax abnormal feature information included in the training sample;

If the abnormal difference value is less than the target value, the initial risk monitoring model is determined as the risk monitoring model;

If the abnormal difference value is greater than or equal to the target value, the relevant parameters in the initial risk monitoring model are adjusted, the adjusted initial risk monitoring model is renamed as the initial risk monitoring model, and the training step is re-executed.

Optionally, in the above embodiment provided by the present invention, it also includes:

Collect public opinion information and business environment change information related to the finance and taxation fields;

Based on the public opinion information and business environment change information, the training sample set required by the risk monitoring model is updated;

Retraining the risk monitoring model using the updated training sample set;

Evaluate the performance of the retrained risk monitoring model;

Based on the performance evaluation results, determine whether to replace the original model with the updated risk monitoring model.

In a second aspect, some embodiments of the present invention provide a financial and tax data risk monitoring system, the system comprising:

Data collection and processing module, where users collect original financial and tax data and pre-process the collected original financial and tax data, where the pre-processing includes data standardization. The original financial and tax data includes corporate financial statements, tax return forms, audit reports, invoice data, and transaction records;

A model building module, used to build a risk monitoring model using a machine learning algorithm based on the pre-processed original financial and tax data;

A variable determination module module is used to determine the characteristic variables and target variables of the risk monitoring model, wherein the characteristic variables include indicators reflecting the financial and taxation status of the enterprise, and the target variable is the risk level of the financial and taxation data;

A feature generation module, configured to obtain, in response to receiving the request information for fiscal and tax data sent by the platform, the fiscal and tax information to be verified, the first fiscal and tax feature information set and the second fiscal and tax feature information set, wherein the first fiscal and tax feature information set and the second fiscal and tax feature information set are generated based on historical fiscal and tax data;

a classification processing module, configured to classify each first fiscal and taxation characteristic information in the first fiscal and taxation characteristic information set to obtain a first classification result, and classify each second fiscal and taxation characteristic information in the second fiscal and taxation characteristic information set to obtain a second classification result;

A feature classification module, used for inputting the to-be-verified financial and tax information into the pre-built risk monitoring model to obtain the to-be-verified financial and tax feature information, wherein the to-be-verified financial and tax feature information includes feature classification information;

A verification module, configured to verify the financial and tax characteristic information to be verified based on the first classification result and the second classification result to obtain a verification result;

An abnormal information marking module is used to mark the financial and tax information to be verified as abnormal information if the verification result meets the preset conditions;

The early warning information generation module is used to generate corresponding risk early warning information in response to the financial and tax information to be verified being marked as abnormal information, based on the characteristic variables and target variables of the risk monitoring model, wherein the risk early warning information includes risk type information, risk level information, risk source information and risk description information.

In a third aspect, some embodiments of the present invention provide an electronic device comprising: one or more processors; a storage device on which one or more programs are stored, and when the one or more programs are executed by the one or more processors, the one or more processors implement the method described in any implementation method in the above-mentioned first aspect.

In a fourth aspect, some embodiments of the present invention provide a computer-readable medium having a computer program stored thereon, wherein when the program is executed by a processor, the method described in any one of the implementations in the first aspect above is implemented.

The above-mentioned embodiments of the present disclosure have the following beneficial effects: by comprehensively using machine learning algorithms, data preprocessing technology and risk model construction, efficient and accurate risk monitoring of fiscal and tax data is achieved. This method not only significantly improves the accuracy and efficiency of risk identification, but also can respond to dynamic risk warning needs in real time, and promptly discover and deal with potential risks. At the same time, by collecting multi-dimensional fiscal and tax data and related information, continuously updating and optimizing the risk monitoring model, ensuring that it always keeps pace with the latest risk model, it brings significant technical advantages and practical application value to the field of fiscal and tax data risk monitoring.

BRIEF DESCRIPTION OF THE DRAWINGS

The above and other features, advantages and aspects of the embodiments of the present invention will become more apparent with reference to the following detailed description in conjunction with the accompanying drawings. Throughout the accompanying drawings, the same or similar reference numerals represent the same or similar elements. It should be understood that the drawings are schematic and that components and elements are not necessarily drawn to scale.

FIG1 is a flow chart of a method for monitoring financial and tax data risks according to an embodiment of the present invention;

FIG2 is a functional block diagram of a financial and taxation data risk monitoring system according to an embodiment of the present invention;

FIG. 3 is a schematic diagram of the structure of an electronic device suitable for implementing some embodiments of the present invention.

Detailed ways

Embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although certain embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be construed as being limited to the embodiments set forth herein. On the contrary, these embodiments are provided to provide a more thorough and complete understanding of the present disclosure. It should be understood that the drawings and embodiments of the present disclosure are only for exemplary purposes and are not intended to limit the scope of protection of the present disclosure.

It should also be noted that, for ease of description, only the parts related to the invention are shown in the drawings. In the absence of conflict, the embodiments and features in the embodiments of the present disclosure may be combined with each other.

It should be noted that the concepts such as "first" and "second" mentioned in the present disclosure are only used to distinguish different devices, modules or units, and are not used to limit the order or interdependence of the functions performed by these devices, modules or units.

It should be noted that the modifications of "one" and "plurality" mentioned in the present disclosure are illustrative rather than restrictive, and those skilled in the art should understand that unless otherwise clearly indicated in the context, it should be understood as "one or more".

The names of the messages or information exchanged between multiple devices in the embodiments of the present disclosure are only used for illustrative purposes and are not used to limit the scope of these messages or information.

The present disclosure will be described in detail below with reference to the accompanying drawings and in conjunction with embodiments.

FIG1 shows a process 100 of some embodiments of the financial and tax data risk monitoring method according to the present disclosure. The financial and tax data risk monitoring method comprises the following steps:

101. Collecting original financial and tax data, and preprocessing the collected original financial and tax data, wherein the preprocessing includes data standardization, and the original financial and tax data includes corporate financial statements, tax return forms, audit reports, invoice data, and transaction records;

It should be noted that original financial and tax data refers to the original data related to finance and taxation generated by an enterprise in its daily business activities. These data are unprocessed and come directly from the enterprise's financial statements, tax return forms, audit reports, invoices, transaction records, etc.

In some embodiments, first, original financial and tax data are collected from multiple sources (such as corporate financial statements, tax returns, audit reports, invoice data, transaction records, etc.) These data usually contain a large amount of corporate financial and tax information and are the basis for assessing financial and tax risks.

102. Based on the pre-processed original financial and tax data, a risk monitoring model is constructed using a machine learning algorithm;

In some embodiments, preprocessing can be a series of operations performed on the original data before data analysis or machine learning, such as cleaning, conversion, standardization, etc., to make it meet the requirements of subsequent analysis or model training. Based on the preprocessed original fiscal and tax data, a risk monitoring model is constructed using a machine learning algorithm. The machine learning algorithms that can be selected include decision trees, random forests, support vector machines, neural networks, etc. These algorithms can automatically learn the inherent laws and risk characteristics of fiscal and tax data, thereby realizing the identification and assessment of potential risks. When building a model, it is necessary to determine characteristic variables and target variables. Characteristic variables are indicators that reflect the fiscal and tax status of an enterprise, such as income, cost, profit, tax burden, etc. The target variable is the risk level of fiscal and tax data, which can be high, medium, and low, or a specific risk score.

103. Determine the characteristic variables and target variables of the risk monitoring model, wherein the characteristic variables include indicators reflecting the financial and taxation status of the enterprise, and the target variable is the risk level of the financial and taxation data;

In some embodiments, in machine learning, feature variables can be used to describe variables that can measure properties or characteristics of a system, object, or event. In risk monitoring of financial and tax data, feature variables can be various indicators that reflect the financial and tax status of an enterprise. The target variable can be the risk level of financial and tax data.

104. In response to receiving the fiscal and taxation data request information sent by the platform, obtaining the fiscal and taxation information to be verified, the first fiscal and taxation characteristic information set, and the second fiscal and taxation characteristic information set, wherein the first fiscal and taxation characteristic information set and the second fiscal and taxation characteristic information set are generated based on historical fiscal and taxation data;

In some embodiments, first, it is necessary to establish a stable communication connection with the platform requesting financial and tax data to ensure that the request information sent by the platform can be received in real time and accurately. Once the financial and tax data request information sent by the platform is received, it is necessary to parse it and extract the specific content of the request, such as the requested financial and tax data type, time range, corporate identification and other key information. According to the parsed request information, the data source storing the financial and tax information to be verified is located, which may be a database, file system or other data storage service. In the located data source, the corresponding financial and tax information is retrieved according to the specific content of the request. The retrieval process may need to consider issues such as data indexing and query optimization to ensure that the required data can be obtained quickly and accurately. The retrieved financial and tax information is integrated to form a set of financial and tax information to be verified for subsequent verification.

In some embodiments, a data source storing historical fiscal and taxation data is accessed. These data have usually been preprocessed and feature extracted, and can be directly used to generate a feature information set. The first fiscal and taxation feature information set may be a first set of features related to the current request extracted from the historical fiscal and taxation data. These features may include statistical information such as total income, total expenditure, tax burden rate, and the trend of changes in the same period of history. The generated feature information set should be able to reflect the main features and laws of historical fiscal and taxation data. The second fiscal and taxation feature information set may be similar to the first fiscal and taxation feature information set, but extracts features from different angles or dimensions, which may be more fine-grained data features, statistical information of different time windows, or other features related to fiscal and taxation risks. The second fiscal and taxation feature information set is intended to provide additional information to enhance the accuracy and comprehensiveness of the verification process. The generated first fiscal and taxation feature information set and the second fiscal and taxation feature information set are stored in a suitable data structure for subsequent use. At the same time, these feature information sets need to be updated regularly or in real time to reflect the latest historical fiscal and taxation data features and laws.

105. Classify each first fiscal and taxation characteristic information in the first fiscal and taxation characteristic information set to obtain a first classification result, and classify each second fiscal and taxation characteristic information in the second fiscal and taxation characteristic information set to obtain a second classification result;

In some embodiments, features for classification processing are selected from the first fiscal and taxation feature information set and the second fiscal and taxation feature information set, respectively. These features should be able to effectively reflect the different aspects and dimensions of fiscal and taxation data to facilitate subsequent classification processing. Select a suitable classification algorithm based on the characteristics and classification requirements of fiscal and taxation data. Common classification algorithms include decision trees, random forests, support vector machines (SVM), logistic regression, neural networks, etc. Taking into account the complexity and nonlinear relationship of fiscal and taxation data, algorithms with strong expressive power, such as random forests or neural networks, can be selected. Collect and organize label information in historical fiscal and taxation data, which usually indicates the category or risk level to which the fiscal and taxation data belongs. Combine the label information with the corresponding feature information to form a training data set for subsequent classification model training.

106. Input the to-be-verified financial and tax information into the pre-built risk monitoring model to obtain the to-be-verified financial and tax feature information, wherein the to-be-verified financial and tax feature information includes feature category information;

It should be noted that feature category information refers to the information contained in the financial and tax feature information to be verified, which is used to describe the category to which the financial and tax data belongs. In the risk monitoring model, financial and tax data are usually divided into different categories, such as normal class, risk class, etc., in order to better identify and manage risks. Feature category information can be a category label directly output by the model, or it can be category information inferred from the feature value. For example, if a feature value exceeds the preset threshold, it can be classified as a risk class; if the feature value is within the normal range, it can be classified as a normal class. By analyzing the feature category information, a preliminary judgment can be made on the risk situation of the financial and tax information to be verified. For example, if the feature category information of a financial and tax information to be verified contains multiple risk class labels, then it can be considered that the financial and tax information has a higher risk; conversely, if the feature category information is mainly normal class labels, then it can be considered that the risk of the financial and tax information is low.

In some instances, first, ensure that the financial and tax information to be verified has been pre-processed in accordance with the format and requirements required by the risk monitoring model. This may include steps such as data cleaning, conversion, and normalization to ensure the accuracy and consistency of the input data. Input the financial and tax information to be verified into the risk monitoring model. This is usually achieved by calling the prediction function or interface of the model, passing the financial and tax information to be verified as an input parameter to the model.

107. Based on the first classification result and the second classification result, verify the fiscal and taxation characteristic information to be verified to obtain a verification result;

In some embodiments, it is confirmed that the data format of the first classification result and the second classification result is consistent with the data format of the fiscal and taxation feature information to be verified, so as to perform subsequent comparison and verification, and set appropriate verification rules according to the characteristics of fiscal and taxation data and business requirements. These rules may include feature value range verification, classification result consistency verification, logic rule verification, etc. Perform range verification on each feature value in the fiscal and taxation feature information to be verified. By comparing the feature value with the preset normal range or threshold, it is determined whether it exceeds the reasonable range. Feature values that exceed the range may indicate data anomalies or risks. Compare the feature category information in the fiscal and taxation feature information to be verified with the first classification result and the second classification result. If the category information of a feature is inconsistent in the first classification result and the second classification result, it is necessary to further analyze the cause and determine whether it is a data error, a model misjudgment or other situations. Apply the preset logic rules to verify the fiscal and taxation feature information to be verified. These logic rules may include associations between different features, business logic constraints, etc. For example, there is a fixed proportional relationship or dependency relationship between certain feature values. By verifying these logic rules, inconsistencies or errors in the data can be found.

108. If the verification result meets the preset conditions, the financial and tax information to be verified is marked as abnormal information;

In some embodiments, problems, anomalies or errors found during the verification process are recorded, including specific feature information, classification results, verification rule triggering conditions, etc. A detailed verification report is generated based on the recorded verification results. The report should include information such as the overall situation of the verification, a list of problems found, a specific description of the problems, and recommended treatment methods. According to the list of problems in the verification report, the problems found are handled. It may be necessary to correct data errors, adjust model parameters, or retrain the model. At the same time, the processing results are fed back to relevant personnel or systems to ensure continuous improvement of data quality and model performance. The verification results of the fiscal and tax feature information to be verified are obtained from the verification processing step. These results should include a comprehensive evaluation of the fiscal and tax feature information to be verified, including information on data consistency, logical correctness, classification accuracy, etc. The verification rules and their corresponding preset conditions applied in the verification process are clarified. These preset conditions are set in advance according to business needs and data characteristics, and are used to determine whether the fiscal and tax information is normal. The obtained verification results are compared with the preset conditions one by one. The purpose of this step is to check whether the fiscal and tax feature information to be verified meets all preset verification rules.

109. In response to the financial and tax information to be verified being marked as abnormal information, corresponding risk warning information is generated based on the characteristic variables and target variables of the risk monitoring model, and the risk warning information includes risk type information, risk level information, risk source information and risk description information.

In some embodiments, feature variables related to the financial and tax information to be verified are extracted from the risk monitoring model. These feature variables are determined in the model building phase and are used to describe key information on different risk aspects of financial and tax data. The target variable of the risk monitoring model is clarified, that is, the risk result that the model aims to predict. This can be determined based on historical risk events and business needs. The risk type is determined based on the abnormality of the feature variable and the predicted result of the target variable. The risk type may include tax evasion risk, false reporting risk, illegal operation risk, etc., depending on the training purpose and business scenario of the model. When generating risk type information, the abnormality of the feature variable can be mapped to the corresponding risk category with reference to the preset risk classification rules. Assess the severity of the abnormality and the possible impact on the business, and assign a risk level to each abnormal feature variable or risk type. The risk level can be represented by numbers, color codes or other indicators for quick identification and priority sorting. Taking into account the abnormalities of multiple feature variables, the overall risk level is determined using methods such as weighted average and maximum impact principle.

In some embodiments, the generating of corresponding risk warning information based on the characteristic variables and target variables of the risk monitoring model further includes:

The risk warning information structure is defined by a Bayesian network model, wherein each node in the risk warning information structure corresponds to a variable in the risk type, risk level, risk source and risk description, and the connection relationship between each node is the dependency relationship between the variables;

Based on the risk warning information structure, calculating the conditional probability of each of the nodes, the conditional probability being the probability of each of the nodes taking different values;

Based on the characteristic variables and target variables in the risk monitoring model, analyzing the conditional probability to obtain an analysis result;

Based on the analysis results of the conditional probability, risk warning information is generated.

It should be noted that the Bayesian Network is a graphical model for modeling uncertainty and probabilistic reasoning. It is a directed acyclic graph (DAG) consisting of nodes representing variables and directed edges connecting these nodes. Each node represents a random variable, and the directed edges represent the dependencies between nodes. In addition, the Bayesian network also contains conditional probability distributions, which describe the conditional probability of each node given the value of its parent node. These conditional probabilities constitute the conditional probability table in the network. The basic principle of the Bayesian network is based on Bayes' theorem, which represents probability distribution by describing the conditional dependencies between different variables. This network can capture direct and indirect relationships between variables, so that probabilistic reasoning and prediction can be performed. This makes the Bayesian network a powerful tool for analyzing uncertainty and probabilistic relationships in complex systems. In a Bayesian network, directed edges represent dependencies or causal relationships between variables, and the arrows of the directed edges represent the directionality of the causal relationship (from the parent node to the child node). If there is no connecting edge between nodes, it means that the variables corresponding to the nodes are conditionally independent. This dependency can be reflected through a network model.

In some embodiments, defining the risk warning information structure first requires building a Bayesian network model to define the risk warning information structure based on business requirements and data characteristics. In this structure, each node corresponds to a variable in the risk type, risk level, risk source, and risk description. In the Bayesian network model, the connection relationship between nodes represents the dependency between variables. These dependencies can be determined by historical data, expert knowledge, or business logic. For example, the risk type may directly affect the risk level and risk description, while the risk source may be associated with both the risk type and the risk level.

In some embodiments, historical risk data and related information are collected to calculate the conditional probability of each node in the Bayesian network model. These data should cover different risk situations in order to accurately reflect the dependencies between variables. Based on the collected data, statistical methods are used to calculate the conditional probability of each node taking different values. These conditional probabilities describe the probability of the current node taking each possible value under different parent node states.

Map the characteristic variables and target variables in the risk monitoring model to the corresponding nodes of the Bayesian network model. In this way, the abnormal conditions of the characteristic variables and the predicted results of the target variables can be converted into the state changes of the nodes in the Bayesian network model. According to the mapped node states, analyze the conditional probabilities of each node in the Bayesian network model. Focus on those nodes that are directly related to the abnormal characteristic variables and target variables, as well as the dependencies between them.

Based on the analysis results of conditional probability, the most likely state of each node is determined. These states represent the actual values of risk type, risk level, risk source and risk description. The determined node states are integrated into complete risk warning information. This includes the identification of risk type, assessment of risk level, location of risk source and generation of risk description. Finally, the risk warning information is presented to relevant personnel through appropriate output methods (such as user interface, report documents, etc.).

Through the above steps, risk warning information can be effectively generated based on the characteristic variables and target variables of the Bayesian network model and the risk monitoring model. This method not only considers the dependencies between variables, but also uses historical data and statistical methods to improve the accuracy and reliability of warnings.

In some embodiments, each first fiscal and taxation characteristic information in the first fiscal and taxation characteristic information set is generated by the following steps:

Obtain user financial and tax behavior log information from the platform according to the preset data format;

Filter the user's financial and tax behavior log information according to the preset screening principles to identify missing value data;

Correct missing value data according to the preset cleaning algorithm;

The user's financial and tax behavior log information is unified according to preset unified rules, and the unified user's financial and tax behavior log information is input into the pre-trained risk monitoring model to obtain the first financial and tax feature information.

In some embodiments, first, it is necessary to connect with the platform that stores the user's financial and tax behavior log information to ensure that the required data can be obtained stably and securely. According to the requirements of the risk monitoring model and subsequent processing steps, a standard data format is preset. This format should contain all necessary fields, such as user ID, operation type, operation time, operation amount, etc. According to the preset data format, the user's financial and tax behavior log information is extracted from the platform. This step needs to ensure that the extracted data is complete and accurate, and can truly reflect the user's financial and tax behavior.

Furthermore, according to business needs and data analysis experience, a series of screening principles need to be set. These principles may include requirements for data integrity, rationality, timeliness, etc. The user's financial and tax behavior log information is screened according to the preset screening principles, with the focus on identifying missing value data, which may be caused by the user's failure to fill in, system errors or other reasons. For the identified missing value data, data correction is performed according to the preset cleaning algorithm. The cleaning algorithm may include filling in missing values (such as using the mean, median, mode, etc.), deleting records containing missing values, or using other appropriate data interpolation methods.

In some embodiments, in order to ensure the consistency and comparability of the data input into the risk monitoring model, a set of unified rules needs to be formulated. These rules may involve the unification of data units, the conversion of time formats, the standardization of text information, etc. According to the formulated unified rules, the user's financial and tax behavior log information is processed. The purpose of this step is to eliminate the differences between the data so that it meets the input requirements of the risk monitoring model.

In some embodiments, the input risk monitoring model generates the first financial and tax feature information to ensure that the risk monitoring model has been properly trained and is ready to receive input data. This includes loading the model, configuring parameters, and necessary preprocessing steps. The uniformly processed user financial and tax behavior log information is input into the pre-trained risk monitoring model. This step needs to ensure the correct input of the data and the normal operation of the model. The risk monitoring model will generate the corresponding first financial and tax feature information based on the input user financial and tax behavior log information. These feature information are the refinement and transformation of the original data, which can more directly reflect the user's financial and tax risk status. Through the above steps, the first financial and tax feature information that meets the requirements can be generated to provide strong support for subsequent risk monitoring and analysis.

In some embodiments, the above method further comprises:

Each second fiscal and taxation characteristic information in the second fiscal and taxation characteristic information set is generated by the following steps:

Obtain historical financial and tax information from multiple source platforms according to the preset data format;

The financial and tax historical information of the multi-source platforms is unified according to preset unified rules, and the unified financial and tax historical information of the multi-source platforms is input into the pre-trained risk monitoring model to obtain the second financial and tax feature information.

In some embodiments, it is first necessary to determine which platforms to obtain financial and tax historical information from. These platforms may include tax systems, financial systems, payment platforms, enterprise information platforms, etc. Data docking is performed with these platforms to ensure that data can be obtained stably. In order to ensure that the data obtained from multiple source platforms are consistent and comparable, a standard data format needs to be set in advance. This format should include key financial and tax information fields, such as taxpayer identification number, tax type, tax period, tax payable, tax paid, etc. According to the preset data format, financial and tax historical information is obtained from each platform. This step requires ensuring that the acquired data is complete and accurate, and the time span meets the analysis requirements. Since data from different platforms may be different and redundant, the acquired data needs to be cleaned. The cleaning steps may include removing duplicate data, correcting erroneous data, filling missing values, etc. In order to eliminate differences between data, a set of unified rules are formulated. These rules may involve conversion of data units (such as unified currency units), adjustment of time formats (such as unifying dates in different formats into a standard format), standardization of field names, etc. Apply unified rules to process financial and tax historical information from different platforms to ensure consistency in format and content of data.

In some embodiments, it is necessary to ensure that the risk monitoring model is loaded and ready to receive processed financial and tax historical information. This may include setting the input parameters of the model, adjusting the operating environment of the model, etc. The unified processed financial and tax historical information of the multi-source platform is input into the pre-trained risk monitoring model. This step requires ensuring the correct input of the data and performing the corresponding format conversion according to the requirements of the model. The risk monitoring model will generate a second financial and tax feature information based on the input financial and tax historical information. These feature information may include the user's tax behavior characteristics, financial status characteristics, historical violation characteristics, etc., which are highly refined and abstract representations of the original data. Through the above steps, a second set of financial and tax feature information can be generated, which will be used for subsequent financial and tax risk monitoring and analysis to help identify potential financial and tax risks and take corresponding prevention and control measures.

In some embodiments, the above method further comprises:

Based on the first classification result and the second classification result, the verification process is performed on the to-be-verified financial and taxation characteristic information, and the verification result obtained includes:

If the feature category information included in the to-be-verified fiscal and taxation feature information meets the first category condition, matching the to-be-verified fiscal and taxation feature information with each first fiscal and taxation feature information in the first fiscal and taxation feature information set to generate a matching result, thereby obtaining a matching result set;

If the feature category information included in the to-be-verified fiscal and taxation feature information meets the second category condition, matching the to-be-verified fiscal and taxation feature information with each second fiscal and taxation feature information in the second fiscal and taxation feature information set to generate a matching result, thereby obtaining a matching result set;

If there is a matching result that meets the preset matching condition in the matching result set, the first preset verification information is determined as the verification result;

If there is no matching result satisfying the preset matching condition in the matching result set, the second preset verification information is determined as the verification result.

In some embodiments, the classification processing of the fiscal and taxation feature information to be verified first requires parsing the fiscal and taxation feature information to be verified to identify the feature category information it contains. The feature category information may include key classification identifiers such as tax type, taxpayer type, and financial statement type. According to the identified feature category information, it is determined whether it meets the first category condition or the second category condition. These conditions may be based on business rules, data types, or judgment logic of specific field values. When the first category condition is met, if the fiscal and taxation feature information to be verified meets the first category condition, the corresponding fiscal and taxation feature information is selected from the first fiscal and taxation feature information set as the matching object. The fiscal and taxation feature information to be verified is matched one by one with each first fiscal and taxation feature information in the first fiscal and taxation feature information set. The matching process may involve comparing field values, calculating similarity, or applying a specific matching algorithm. The results of each match are recorded to form a matching result set. Each matching result may include an identifier of a successful or failed match, as well as relevant matching scores or difference information.

In some embodiments, if the fiscal and taxation characteristic information to be verified meets the second category conditions, the corresponding fiscal and taxation characteristic information is selected from the second fiscal and taxation characteristic information set as the matching object. The matching processing and result set generation need to perform matching processing steps similar to the first category conditions to generate a corresponding matching result set. Traverse the matching result set to check whether there are matching results that meet the preset matching conditions. The preset matching conditions may be based on a threshold judgment of a matching score, a field matching rate, or other business rules. If there is at least one matching result that meets the preset matching conditions in the matching result set, the first preset verification information (such as "verification passed" or a corresponding status code) is determined as the verification result. If there is no matching result that meets the preset matching conditions in the matching result set, the second preset verification information (such as "verification failed" or a corresponding error code) is determined as the verification result.

Feedback the confirmed verification results to the caller or display them to the user. The feedback can be returned through the API, displayed in the user interface, or recorded in the log file. Based on the verification results, the system can take further actions, such as allowing the financial and tax feature information that has passed the verification to enter the next processing link, handling errors for information that has not passed the verification, or requesting the user to re-enter the information. Through the above steps, detailed verification processing of the financial and tax feature information to be verified can be achieved to ensure the accuracy and compliance of the financial and tax data.

In some embodiments, the risk monitoring model is trained by the following steps:

A training sample set and an initial risk monitoring model are obtained, wherein each training sample in the training sample set includes sample financial and tax abnormality information and sample financial and tax abnormality feature information, and the initial risk monitoring model includes a full convolution module, a residual convolution module sequence and a pooling module; a training sample is selected from the training sample set, and a training step is performed, specifically including: inputting the sample financial and tax abnormality information included in the training sample into the full convolution module of the initial risk monitoring model to obtain full convolution feature information; inputting the full convolution feature information into the residual convolution module sequence of the initial risk monitoring model to obtain residual convolution information; inputting the residual convolution information into the pooling module of the initial risk monitoring model to obtain initial financial and tax abnormality feature information; based on a preset loss function, determining the abnormal difference value between the initial financial and tax abnormality feature information and the sample financial and tax abnormality feature information included in the training sample; if the abnormal difference value is less than the target value, determining the initial risk monitoring model as the risk monitoring model; if the abnormal difference value is greater than or equal to the target value, adjusting the relevant parameters in the initial risk monitoring model, renaming the adjusted initial risk monitoring model as the initial risk monitoring model, and re-performing the training step.

In some embodiments, obtaining a training sample set is accomplished by the following steps: Collect samples containing abnormal financial and tax information from historical financial and tax data to construct a training sample set, where each training sample includes sample abnormal financial and tax information and corresponding sample abnormal financial and tax feature information; the sample abnormal financial and tax information may be original data such as financial statements, tax returns, and transaction records; the sample abnormal financial and tax feature information is information processed by feature engineering, such as key features extracted by statistical analysis, pattern recognition, and other methods.

In the step of initializing the risk monitoring model, an initial risk monitoring model is constructed, which includes a full convolution module, a residual convolution module sequence and a pooling module. The full convolution module is used to perform preliminary feature extraction on the input data. The residual convolution module sequence is used to further extract deep feature information and solve the gradient vanishing problem in deep neural networks. The pooling module is used to reduce the dimension of feature information and improve the generalization ability of the model.

The model training specifically includes the following sub-steps: full convolution processing: inputting the sample financial and tax abnormality information in the selected training samples into the full convolution module of the initial risk monitoring model, performing convolution operation, and extracting the full convolution feature information; residual convolution processing: inputting the full convolution feature information into the residual convolution module sequence, and gradually extracting deeper feature information through the stacking of multiple residual blocks to obtain residual convolution information; pooling processing: inputting the residual convolution information into the pooling module, performing dimensionality reduction processing, and obtaining the initial financial and tax abnormality feature information; calculating the abnormal difference value: based on the preset loss function (such as mean square error loss function, cross entropy loss function, etc.), calculating the abnormal difference value between the initial financial and tax abnormality feature information and the sample financial and tax abnormality feature information in the training sample; judgment and adjustment: judging whether the abnormal difference value is less than the preset target value; if so, determining the initial risk monitoring model as the risk monitoring model; if not, adjusting the relevant parameters in the initial risk monitoring model (such as convolution kernel size, step size, learning rate, etc.), and renaming the adjusted model as the initial risk monitoring model, and then re-executing the training step.

In some embodiments, the training steps are repeated until a stop condition is met (such as reaching a preset maximum number of iterations, the abnormal difference value converges, etc.). After iterative training, a risk monitoring model that meets the requirements is obtained. The trained risk monitoring model is applied to the actual financial and tax data risk monitoring task to perform anomaly detection and analysis on new financial and tax data.

Through the above specific embodiments, an effective risk monitoring model can be trained to be used for risk monitoring and anomaly identification of financial and tax data.

In some embodiments, the above method further comprises:

Collect public opinion information and business environment change information related to the finance and taxation fields;

Based on the public opinion information and business environment change information, the training sample set required by the risk monitoring model is updated;

Retraining the risk monitoring model using the updated training sample set;

Evaluate the performance of the retrained risk monitoring model;

Based on the performance evaluation results, determine whether to replace the original model with the updated risk monitoring model.

In some embodiments, first, public opinion information and business environment change information related to the finance and taxation field are continuously collected through a variety of information sources (including news media, social media, government announcements, etc.). This information is strictly screened and sorted to ensure its relevance and accuracy to the finance and taxation field. Next, we deeply analyze the potential risk points or abnormal patterns that this information may generate for the finance and taxation data, and generate or mark new training samples accordingly. These new samples will undergo necessary preprocessing before being added to the original training sample set to ensure data quality. Then, the risk monitoring model is retrained using the updated training sample set. During the training process, we will continuously adjust the parameters and hyperparameters of the model, and verify the model using the validation set to ensure that it is not overfitting and the performance is optimized.

Once the model is retrained, it will be fully evaluated for performance using an independent test set. Evaluation metrics include accuracy, recall, F1 score, etc., and corresponding thresholds and weights are set according to business needs. These evaluation results are compared and analyzed with the performance of the original model to evaluate the effectiveness of the improvement. Based on the performance evaluation results and business needs, we will make a clear decision basis to determine whether to replace the original model with the updated model. In the decision-making process, we will also comprehensively consider other factors such as model complexity and computing resource consumption. If the updated model shows a significant improvement in performance and meets business needs, we will replace the original model with it, deploy and monitor it in the actual environment to ensure its stable operation and continuous optimization.

The above-mentioned embodiments of the present disclosure have the following beneficial effects: through automated data collection, preprocessing, model building and verification processes, manual participation and intervention can be greatly reduced, and the efficiency and accuracy of financial and tax data risk management can be improved. By continuously monitoring and verifying financial and tax data, potential financial and tax risks, such as tax evasion, false tax reporting, etc., can be discovered in a timely manner, so that timely countermeasures can be taken to avoid or reduce losses. Through continuous training and optimization of risk monitoring models, financial management processes can be continuously improved, and the level of intelligence and refinement of financial management can be improved. Through effective financial and tax data risk management, the image and reputation of the company can be enhanced, and the confidence and willingness to cooperate of investors and partners can be enhanced. By continuously collecting and analyzing public opinion information and business environment change information related to the financial and tax fields, it can be ensured that the risk monitoring model adapts to new financial and tax policies and environmental changes in a timely manner, and maintains the advancement and effectiveness of the model.

Further referring to Figure 2, as an implementation of the methods shown in the above figures, the present disclosure provides some embodiments of a financial and taxation data risk monitoring system. These device embodiments correspond to the method embodiments shown in Figure 1. The financial and taxation data risk monitoring system can be specifically applied to various electronic devices.

As shown in FIG. 2 , a financial and tax data risk monitoring system 200 according to some embodiments includes:

The data collection and processing module 201 is used to collect original financial and tax data and pre-process the collected original financial and tax data, wherein the pre-processing includes data standardization. The original financial and tax data includes corporate financial statements, tax return forms, audit reports, invoice data, and transaction records.

A model building module 202, for building a risk monitoring model using a machine learning algorithm based on the pre-processed original financial and tax data;

The variable determination module 203 is used to determine the characteristic variables and target variables of the risk monitoring model, wherein the characteristic variables include indicators reflecting the financial and taxation status of the enterprise, and the target variable is the risk level of the financial and taxation data;

The feature generation module 204 is used to obtain the to-be-verified financial and tax information, the first financial and tax feature information set and the second financial and tax feature information set in response to receiving the financial and tax data request information sent by the platform, wherein the first financial and tax feature information set and the second financial and tax feature information set are generated based on the historical financial and tax data;

A classification processing module 205 is used to classify each first fiscal and taxation characteristic information in the first fiscal and taxation characteristic information set to obtain a first classification result, and to classify each second fiscal and taxation characteristic information in the second fiscal and taxation characteristic information set to obtain a second classification result;

The feature classification module 206 is used to input the to-be-verified financial and tax information into the pre-built risk monitoring model to obtain the to-be-verified financial and tax feature information, wherein the to-be-verified financial and tax feature information includes feature classification information;

Verification module 207, used for verifying the fiscal and taxation characteristic information to be verified based on the first classification result and the second classification result to obtain a verification result;

The abnormal information marking module 208 is used to mark the financial and tax information to be verified as abnormal information if the verification result meets the preset conditions;

The warning information generation module 209 is used to generate corresponding risk warning information in response to the financial and tax information to be verified being marked as abnormal information, based on the characteristic variables and target variables of the risk monitoring model, and the risk warning information includes risk type information, risk level information, risk source information and risk description information.

It can be understood that the modules recorded in the financial and tax data risk monitoring system 200 correspond to the steps in the financial and tax data risk monitoring method described with reference to FIG1. Therefore, the operations, features and beneficial effects described above for the financial and tax data risk monitoring method are also applicable to the financial and tax data risk monitoring system 200 and the modules contained therein, and will not be repeated here.

Referring to FIG3 below, it shows a schematic diagram of the structure of an electronic device 300 suitable for implementing some embodiments of the present disclosure. The electronic devices in some embodiments of the present disclosure may include but are not limited to mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (personal digital assistants), PADs (tablet computers), PMPs (portable multimedia players), vehicle-mounted terminals (such as vehicle-mounted navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. The terminal device shown in FIG3 is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present disclosure.

As shown in FIG3 , the electronic device 300 may include a processing device (e.g., a central processing unit, a graphics processing unit, etc.) 301, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 302 or a program loaded from a storage device 308 into a random access memory (RAM) 303. In the RAM 303, various programs and data required for the operation of the electronic device 300 are also stored. The processing device 301, the ROM 302, and the RAM 303 are connected to each other via a bus 304. An input/output (I/O) interface 305 is also connected to the bus 304.

Typically, the following devices may be connected to the I/O interface 305: input devices 306 including, for example, a touch screen, a touchpad, a keyboard, a mouse, a camera, a microphone, an accelerometer, a gyroscope, etc.; output devices 307 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; storage devices 308 including, for example, a magnetic tape, a hard disk, etc.; and communication devices 309. The communication device 309 may allow the electronic device 300 to communicate with other devices wirelessly or by wire to exchange data. Although FIG. 3 shows an electronic device 300 with various devices, it should be understood that it is not required to implement or have all the devices shown. More or fewer devices may be implemented or have alternatively. Each box shown in FIG. 3 may represent one device, or may represent multiple devices as needed.

In particular, according to some embodiments of the present disclosure, the process described above with reference to the flowchart can be implemented as a computer software program. For example, some embodiments of the present disclosure include a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a program code for executing the method shown in the flowchart. In some such embodiments, the computer program can be downloaded and installed from the network through the communication device 309, or installed from the storage device 308, or installed from the ROM 302. When the computer program is executed by the processing device 301, the above-mentioned functions defined in the method of some embodiments of the present disclosure are executed.

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

In some embodiments, the client and the server may communicate using any currently known or future developed network protocol such as HTTP (HyperText Transfer Protocol), and may be interconnected with any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network ("LAN"), a wide area network ("WAN"), an internet (e.g., the Internet), and a peer-to-peer network (e.g., an ad hoc peer-to-peer network), as well as any currently known or future developed network.

The above-mentioned computer-readable medium may be included in the above-mentioned electronic device; or it may exist independently without being assembled into the electronic device. The above-mentioned computer-readable medium carries one or more programs. When the above-mentioned one or more programs are executed by the electronic device, the electronic device: collects original financial and tax data, and pre-processes the collected original financial and tax data, wherein the pre-processing includes data standardization, and the original financial and tax data includes corporate financial statements, tax returns, audit reports, invoice data, and transaction records; based on the pre-processed original financial and tax data, a risk monitoring model is constructed using a machine learning algorithm; characteristic variables and target variables of the risk monitoring model are determined, wherein the characteristic variables include indicators reflecting the financial and tax status of the enterprise, and the target variable is the risk level of the financial and tax data; in response to receiving the financial and tax data request information sent by the platform, the financial and tax information to be verified, the first financial and tax characteristic information set and the second financial and tax characteristic information set are obtained, wherein the first financial and tax characteristic information set and the second financial and tax characteristic information set are generated based on historical financial and tax data; Classify each first fiscal and taxation feature information in the first fiscal and taxation feature information set to obtain a first classification result, and classify each second fiscal and taxation feature information in the second fiscal and taxation feature information set to obtain a second classification result; input the fiscal and taxation information to be verified into the pre-constructed risk monitoring model to obtain the fiscal and taxation feature information to be verified, and the fiscal and taxation feature information to be verified includes feature category information; based on the first classification result and the second classification result, verify the fiscal and taxation feature information to be verified to obtain a verification result; if the verification result meets the preset conditions, mark the fiscal and taxation information to be verified as abnormal information; in response to the fiscal and taxation information to be verified being marked as abnormal information, generate corresponding risk warning information based on the feature variables and target variables of the risk monitoring model, and the risk warning information includes risk type information, risk level information, risk source information and risk description information.

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

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

The units described in some embodiments of the present disclosure may be implemented in software or in hardware. The units described may also be provided in a processor, and the functions described above herein may be performed at least in part by one or more hardware logic components. For example, and without limitation, exemplary types of hardware logic components that may be used include: field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), complex programmable logic devices (CPLDs), and the like.

The above descriptions are only some preferred embodiments of the present disclosure and an explanation of the technical principles used. Those skilled in the art should understand that the scope of the invention involved in the embodiments of the present disclosure is not limited to the technical solutions formed by a specific combination of the above-mentioned technical features, but should also cover other technical solutions formed by any combination of the above-mentioned technical features or their equivalent features without departing from the above-mentioned inventive concept. For example, the above-mentioned features are replaced with the technical features with similar functions disclosed in the embodiments of the present disclosure (but not limited to) and the technical solutions formed.

Claims (10)

1. A method for monitoring financial and tax data risks, comprising: Collecting original financial and tax data, and preprocessing the collected original financial and tax data, wherein the preprocessing includes data standardization, and the original financial and tax data includes corporate financial statements, tax return forms, audit reports, invoice data, and transaction records; Based on the pre-processed original financial and tax data, a risk monitoring model is constructed using a machine learning algorithm; Determine the characteristic variables and target variables of the risk monitoring model, wherein the characteristic variables include indicators reflecting the financial and taxation status of the enterprise, and the target variable is the risk level of the financial and taxation data; In response to receiving the fiscal and tax data request information sent by the platform, obtaining the fiscal and tax information to be verified, the first fiscal and tax feature information set and the second fiscal and tax feature information set, wherein the first fiscal and tax feature information set and the second fiscal and tax feature information set are generated based on historical fiscal and tax data; Classify each first fiscal and taxation characteristic information in the first fiscal and taxation characteristic information set to obtain a first classification result, and classify each second fiscal and taxation characteristic information in the second fiscal and taxation characteristic information set to obtain a second classification result; Inputting the to-be-verified financial and tax information into the pre-built risk monitoring model to obtain the to-be-verified financial and tax feature information, wherein the to-be-verified financial and tax feature information includes feature category information; Based on the first classification result and the second classification result, verify the financial and tax characteristic information to be verified to obtain a verification result; If the verification result meets the preset conditions, the financial and tax information to be verified is marked as abnormal information; In response to the financial and tax information to be verified being marked as abnormal information, corresponding risk warning information is generated based on the characteristic variables and target variables of the risk monitoring model, and the risk warning information includes risk type information, risk level information, risk source information and risk description information. 2. The method for monitoring financial and tax data risks according to claim 1, wherein generating corresponding risk warning information based on the characteristic variables and target variables of the risk monitoring model comprises: The risk warning information structure is defined by a Bayesian network model, wherein each node in the risk warning information structure corresponds to a variable in the risk type, risk level, risk source and risk description, and the connection relationship between each node is the dependency relationship between the variables; Based on the risk warning information structure, calculating the conditional probability of each of the nodes, the conditional probability being the probability of each of the nodes taking different values; Based on the characteristic variables and target variables in the risk monitoring model, analyzing the conditional probability to obtain an analysis result; Based on the analysis results of the conditional probability, risk warning information is generated. 3. The method for monitoring financial and tax data risks according to claim 1, wherein each first financial and tax characteristic information in the first financial and tax characteristic information set is generated by the following steps: Obtain user financial and tax behavior log information from the platform according to the preset data format; Filter the user's financial and tax behavior log information according to the preset screening principles to identify missing value data; Correct missing value data according to the preset cleaning algorithm; The user's financial and tax behavior log information is unified according to preset unified rules, and the unified user's financial and tax behavior log information is input into the pre-trained risk monitoring model to obtain the first financial and tax feature information. 4. The method for monitoring financial and tax data risks according to claim 1, wherein each second financial and tax characteristic information in the second financial and tax characteristic information set is generated by the following steps: Obtain historical financial and tax information from multiple source platforms according to the preset data format; The financial and tax historical information of the multi-source platforms is unified according to preset unified rules, and the unified financial and tax historical information of the multi-source platforms is input into the pre-trained risk monitoring model to obtain the second financial and tax feature information. 5. The method for monitoring financial and tax data risks according to claim 1, characterized in that the verification process is performed on the financial and tax feature information to be verified based on the first classification result and the second classification result, and the verification result obtained includes: If the feature category information included in the to-be-verified fiscal and taxation feature information meets the first category condition, matching the to-be-verified fiscal and taxation feature information with each first fiscal and taxation feature information in the first fiscal and taxation feature information set to generate a matching result, thereby obtaining a matching result set; If the feature category information included in the to-be-verified fiscal and taxation feature information meets the second category condition, matching the to-be-verified fiscal and taxation feature information with each second fiscal and taxation feature information in the second fiscal and taxation feature information set to generate a matching result, thereby obtaining a matching result set; If there is a matching result that meets the preset matching condition in the matching result set, the first preset verification information is determined as the verification result; If there is no matching result satisfying the preset matching condition in the matching result set, the second preset verification information is determined as the verification result. 6. The method for monitoring financial and tax data risks according to claim 1, wherein the risk monitoring model is trained by the following steps: Obtaining a training sample set and an initial risk monitoring model, wherein each training sample in the training sample set includes sample financial and tax abnormality information and sample financial and tax abnormality feature information, and the initial risk monitoring model includes a full convolution module, a residual convolution module sequence, and a pooling module; Selecting training samples from the training sample set and performing training steps specifically include: Input the sample tax exception information included in the training sample into the full convolution module of the initial risk monitoring model to obtain the full convolution feature information; Input the full convolution feature information into the residual convolution module sequence of the initial risk monitoring model to obtain residual convolution information; Input the residual convolution information into the pooling module of the initial risk monitoring model to obtain the initial financial and tax abnormal feature information; Based on a preset loss function, determine the abnormal difference value between the initial financial and tax abnormal feature information and the sample financial and tax abnormal feature information included in the training sample; If the abnormal difference value is less than the target value, the initial risk monitoring model is determined as the risk monitoring model; If the abnormal difference value is greater than or equal to the target value, the relevant parameters in the initial risk monitoring model are adjusted, the adjusted initial risk monitoring model is renamed as the initial risk monitoring model, and the training step is re-executed. 7. The method for monitoring financial and tax data risks according to claim 6, characterized in that the method further comprises: Collect public opinion information and business environment change information related to the finance and taxation fields; Based on the public opinion information and business environment change information, the training sample set required by the risk monitoring model is updated; Retraining the risk monitoring model using the updated training sample set; Evaluate the performance of the retrained risk monitoring model; Based on the performance evaluation results, determine whether to replace the original model with the updated risk monitoring model. 8. A financial and tax data risk monitoring system, characterized in that the system comprises: Data collection and processing module, where users collect original financial and tax data and pre-process the collected original financial and tax data, where the pre-processing includes data standardization. The original financial and tax data includes corporate financial statements, tax return forms, audit reports, invoice data, and transaction records; A model building module, used to build a risk monitoring model based on the pre-processed original financial and tax data using a machine learning algorithm; A variable determination module module is used to determine the characteristic variables and target variables of the risk monitoring model, wherein the characteristic variables include indicators reflecting the financial and taxation status of the enterprise, and the target variable is the risk level of the financial and taxation data; A feature generation module, configured to obtain, in response to receiving the request information for fiscal and tax data sent by the platform, the fiscal and tax information to be verified, the first fiscal and tax feature information set and the second fiscal and tax feature information set, wherein the first fiscal and tax feature information set and the second fiscal and tax feature information set are generated based on historical fiscal and tax data; a classification processing module, configured to classify each first fiscal and taxation characteristic information in the first fiscal and taxation characteristic information set to obtain a first classification result, and classify each second fiscal and taxation characteristic information in the second fiscal and taxation characteristic information set to obtain a second classification result; A feature category module, used for inputting the to-be-verified financial and tax information into the pre-built risk monitoring model to obtain the to-be-verified financial and tax feature information, wherein the to-be-verified financial and tax feature information includes feature category information; A verification module, configured to verify the financial and tax characteristic information to be verified based on the first classification result and the second classification result to obtain a verification result; An abnormal information marking module is used to mark the financial and tax information to be verified as abnormal information if the verification result meets the preset conditions; The early warning information generation module is used to generate corresponding risk early warning information in response to the financial and tax information to be verified being marked as abnormal information, based on the characteristic variables and target variables of the risk monitoring model, wherein the risk early warning information includes risk type information, risk level information, risk source information and risk description information. 9. An electronic device comprising: one or more processors; a storage device having one or more programs stored thereon; When the one or more programs are executed by the one or more processors, the one or more processors implement the method according to any one of claims 1 to 7. 10. A computer-readable storage medium having executable instructions stored thereon, which, when executed by a processor, enables the processor to implement the method according to any one of claims 1 to 7.