Finance and accounting supervision system based on Internet big data technology

Abstract

The invention discloses a finance and accounting supervision system based on Internet big data technology, and relates to the technical field of risk management, and the system comprises a collection module which is used for collecting financial data from a plurality of different financial management platforms through the Internet; the processing module is used for carrying out standardization, cleaning and duplicate removal operation on the collected financial data; the comparison module is used for performing intelligent comparison of four dimensions of money, transaction time, transaction objects and invoice numbers on the processed financial data by adopting an accurate matching algorithm, generating a comparison report and marking abnormal data; the anomaly detection module is used for constructing an abnormal behavior prediction model based on a machine learning algorithm of big data, training the abnormal behavior prediction model by utilizing historical financial data, inputting abnormal data into the abnormal behavior prediction model, and performing real-time detection and generating early warning information; and the analysis module performs time sequence modeling based on the historical financial data, predicts a future financial trend and potential financial risks, and generates a financial trend report.

Description

Technical Field

[0001] The present invention relates to the technical field of risk management, and in particular to a financial accounting supervision system based on Internet big data technology. Background Art

[0002] With the rapid development of Internet technology and big data, the field of financial management has gradually shifted towards automation, intelligence and data-driven. Traditional financial management systems mainly rely on manual operations and decentralized management methods. When processing large amounts of financial data, they are prone to data lag, information asymmetry and human errors. In recent years, enterprises have gradually adopted ERP (enterprise resource planning) and CRM (customer relationship management) integrated systems to improve the efficiency of financial management. However, with the expansion of enterprise scale and the increase of cross-platform operations, the diversity and complexity of financial data have also increased significantly. In particular, there are still major technical challenges in data synchronization, standardization and consistency between different financial systems. In addition, financial compliance requirements are becoming increasingly stringent. Enterprises need to quickly respond to and adapt to the ever-changing tax and accounting policies and regulations, which further exacerbates the complexity and risks of financial management.

[0003] Although the existing financial management systems have been able to achieve partial automation of financial data processing, there are still obvious deficiencies in data integration and anomaly detection. First, most of the existing systems lack a unified data standardization and cleaning mechanism. Financial data across platforms often have different formats, and data redundancy and duplication are serious, resulting in the inability to effectively compare and analyze data. Secondly, most of the existing anomaly detection methods rely on pre-set rules, lack flexibility and intelligence, and cannot cope with complex and changing financial environments. Especially when processing large-scale, real-time financial transaction data, the existing systems have major defects in the accuracy and timeliness of data anomaly identification, risk prediction, and early warning generation. In addition, the existing technology also lags behind in financial trend analysis and compliance review, and cannot make full use of historical data and the latest policies and regulations for intelligent decision-making support. These deficiencies make it easy for enterprises to have data inconsistencies, compliance risks, and inefficient financial management when facing a complex financial environment. Summary of the invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides a financial and accounting supervision system based on Internet big data technology to solve the problem that most existing systems lack a unified data standardization and cleaning mechanism, cross-platform financial data often have different formats, and data redundancy and duplication problems are serious, resulting in the inability to effectively perform data comparison and analysis.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0007] In a first aspect, an embodiment of the present invention provides a financial and accounting supervision system based on Internet big data technology, which includes:

[0008] Acquisition module, processing module, comparison module, anomaly detection module, analysis module, review module and recommendation module,

[0009] The collection module is used to collect financial data from multiple different financial management platforms via the Internet and transmit the collected financial data to the processing module;

[0010] The processing module is used to standardize, clean and deduplicate the collected financial data, pass the processed standardized data to the comparison module, and pass the duplicate data to the anomaly detection module and the analysis module;

[0011] The comparison module uses an accurate matching algorithm to perform intelligent comparison of the processed financial data in four dimensions: amount, transaction time, transaction object, and invoice number, identifies data inconsistencies between different platforms, generates a comparison report and marks abnormal data, and transmits the compared abnormal data to the anomaly detection module;

[0012] The anomaly detection module builds an abnormal behavior prediction model based on a machine learning algorithm based on big data, uses historical financial data to train the abnormal behavior prediction model, inputs abnormal data into the abnormal behavior prediction model, analyzes transaction data and predicts potential abnormal behaviors, monitors real-time financial transactions, detects and generates early warning information in real time, transmits abnormal data to the analysis module, and transmits early warning information to the audit module for compliance review;

[0013] The analysis module performs time series modeling based on historical financial data, predicts future financial trends and potential financial risks, generates financial trend reports, and transmits the analysis results to the audit module;

[0014] The audit module is used to conduct compliance audit on the financial data of the enterprise according to the latest financial policies and regulations, pass the audit results to the recommendation module, and feedback the compliance audit report to the financial personnel;

[0015] The recommendation module is used to generate rectification suggestions for financial data that does not meet compliance requirements.

[0016] As a preferred solution of the financial and accounting supervision system based on Internet big data technology described in the present invention, wherein: the collection of financial data from multiple different financial management platforms through the Internet refers to,

[0017] Connect with different financial management platforms such as ERP, CRM, tax system and bank statement system through the Internet, and use the API interface provided by each platform to achieve real-time synchronization and collection of data;

[0018] The acquisition module is connected to the processing module, and the processed data is transmitted from the processing module to other modules.

[0019] As a preferred solution of the financial and accounting supervision system based on Internet big data technology described in the present invention, the specific steps for standardizing, cleaning and deduplicating the collected financial data are as follows:

[0020] Based on the API interface, the processing module first performs structured processing on the collected financial data, identifies the data source, uses a machine learning-based classification algorithm to identify the invoice number, amount, and transaction time in the financial data, and classifies the financial data into three different data types: invoice, voucher, and transaction record, and passes them to the processing module;

[0021] The processing module converts the data formats of different platforms into .JSON format through data conversion algorithms, and removes erroneous data through cleaning algorithms;

[0022] By comparing invoice numbers, transaction times and amounts, the processing module uses a hash algorithm to remove duplicate financial data;

[0023] Standardize and score the processed financial data and judge the completeness and consistency of the data;

[0024] The standardized financial data is first passed to the comparison module. The system will compare the data from different platforms in four dimensions: amount, transaction time, transaction object and invoice number, and generate a comparison report.

[0025] Then, the standardized financial data is divided into two copies in proportion, and one copy is passed to the anomaly detection module;

[0026] At the same time, another copy is passed to the analysis module.

[0027] As a preferred solution of the financial and accounting supervision system based on Internet big data technology described in the present invention, wherein: the precise matching algorithm is used to perform intelligent comparison of the processed financial data in four dimensions: amount, transaction time, transaction object and invoice number. The specific steps are:

[0028] The processing module passes the standardized data to the comparison module, and represents each piece of financial data as a four-dimensional vector, corresponding to the four dimensions of amount, transaction time, transaction object, and invoice number. The expression is:

[0029] V i=[M i , T i , O i , F i ];

[0030] Among them, i is the index variable, V i represents the i-th financial data vector, M i represents the amount of the ith financial data, T i represents the transaction time of the ith financial data, O i represents the transaction object of the i-th financial data, F i Indicates the invoice number of the i-th financial data.

[0031] As a preferred solution of the financial and accounting supervision system based on Internet big data technology described in the present invention, the identification of data inconsistencies between different platforms, the generation of a comparison report and the marking of abnormal data, and the transmission of the compared abnormal data to the abnormality detection module are specifically as follows:

[0032] The comparison module determines the similarity between two data vectors by calculating the weighted cosine similarity and obtains the similarity Sim. kj ;

[0033] Set the similarity threshold τ;

[0034] When Sim kj When ≥τ, the two data are considered to match;

[0035] When Sim kj When <τ, the two data are considered mismatched and marked as abnormal data;

[0036] A comparison report is generated based on the marked abnormal data, and the abnormal data and the comparison report will be passed to the anomaly detection module.

[0037] As a preferred solution of the financial and accounting supervision system based on Internet big data technology described in the present invention, wherein: the machine learning algorithm based on big data builds an abnormal behavior prediction model, and uses historical financial data to train the abnormal behavior prediction model. The specific steps are:

[0038] Select the multi-layer neural network MLP as the core model of the abnormal behavior prediction model;

[0039] For the amount, the transaction amount is used directly and normalized to the interval [0, 1].

[0040] For transaction time, the time information is converted into timestamps, and the periodic characteristics of transaction time are captured through sine and cosine transforms;

[0041] For transaction objects, a hash function is used to encode the transaction object into a fixed-length numerical vector;

[0042] For invoice numbers, they are also digitized using a hash function, and similar invoice numbers are grouped into the same category through bucketing.

[0043] Input the processed financial data into the abnormal behavior prediction model to obtain the abnormal probability value, which is expressed as:

[0044]

[0045] φy(X y )=log(1+X y )·sin(X y +αy);

[0046] Among them, P ab represents the probability of anomaly between transaction a and transaction b, y is the index variable, e is the natural constant, W y represents the weight of the y-th dimension, φ y (X y ) represents the y-th dimension input feature X y The nonlinear transformation function is performed, φ represents the nonlinear transformation function, sin represents the sine function, α y represents the phase adjustment parameter of the y-th dimension;

[0047] Based on the abnormal probability, a threshold is set to judge the abnormal risk. When the abnormal probability exceeds the set threshold, the early warning mechanism is triggered.

[0048] As a preferred solution of the financial and accounting supervision system based on Internet big data technology described in the present invention, wherein: the abnormal data is input into the abnormal behavior prediction model, the transaction data is analyzed and potential abnormal behaviors are predicted, real-time financial transactions are monitored, and real-time detection and early warning information are generated. The specific steps are:

[0049] Collect real-time financial data, perform data structuring, classification, data cleaning, standardization, feature extraction, encoding and deduplication processing on the real-time financial data, input the processed real-time financial data into the abnormal behavior prediction model, and calculate the probability value P of abnormal behavior based on the input data and historical data training weights. ab ;

[0050] Setting Thresholds§;

[0051] When P ab When the set threshold is exceeded, the system considers the transaction as a high-risk transaction and generates real-time warning information;

[0052] The system monitors every financial transaction in real time and analyzes potential abnormal behavior;

[0053] When the system detects abnormal behavior, it generates an early warning message and passes the information to the audit module for compliance review.

[0054] As a preferred solution of the financial and accounting supervision system based on Internet big data technology described in the present invention, the time series modeling is performed based on historical financial data to predict future financial trends and potential financial risks, and generate financial trend reports. The analysis results are transmitted to the audit module, specifically,

[0055] The acquisition module collects historical financial data from multiple financial management platforms. The processing module standardizes, cleans and removes duplicates of these data. The processing module passes these data to the analysis module and enters the time series modeling stage.

[0056] In time series modeling, the autoregressive integrated moving average ARIMA model is selected as the basis for prediction, and the expression is:

[0057] X t =&1X t-1 +&2X t-2 +…+& p X t-p +θ1∈ t-1 +θ2∈ t-2 +…+θ u ∈ t-u +∈ t ;

[0058] Among them, X t represents the financial data at time t, & p The table is the u-th natural regression coefficient, θ u represents the uth sliding average coefficient, ∈ t is a white noise term, p and u are index variables;

[0059] X-based t Calculate the predicted value of financial risk, the expression is:

[0060]

[0061] Among them, F t is the predicted value of financial risk at time t, represents the number of historical time points, ψ z (X t-z ) represents the nonlinear transformation function at the zth time point, z represents the index variable, W z represents the weight of the zth time point;

[0062] A financial trend report is generated based on the transaction amount, transaction time, transaction object, invoice number and financial risk forecast value. The analysis module passes the generated financial trend report and risk assessment report to the audit module.

[0063] As a preferred solution of the financial and accounting supervision system based on Internet big data technology described in the present invention, the financial data of the enterprise is audited for compliance according to the latest financial policies and regulations, and the audit results are transmitted to the recommendation module. At the same time, the compliance audit report is fed back to the financial personnel. Specifically,

[0064] Through the API interface connection between the system and the tax bureau and the Ministry of Finance, the system automatically captures and parses the latest policy documents, extracts tax rate changes, invoice management and tax reporting requirements, and generates tax rate inspection rules based on the parsed policies and regulations;

[0065] The system compares each piece of financial data with the compliance rules and calculates the compliance score of each piece of data based on the generated tax rate check rules. The expression is:

[0066]

[0067] Among them, S c represents the compliance score of the enterprise's financial data, with a value range of [0, 1], indicating the degree of compliance, i represents the index variable, N represents the total number of financial data items, and W i The weight of the ith financial data, α i represents the adjustment coefficient of the ith financial data, X i represents the difference between the i-th financial data and regulatory requirements, μ i Indicates the compliance standard value in policies and regulations;

[0068] After the audit is completed, the system will pass the audit results to the recommendation module, and the audit report will also be simultaneously fed back to the financial personnel to help them identify problems in a timely manner and make data corrections.

[0069] As a preferred solution of the financial and accounting supervision system based on Internet big data technology described in the present invention, wherein: the generation of rectification suggestions for financial data that does not meet compliance requirements refers to:

[0070] The system extracts historical compliance records, past corrective actions, and financial operation logs from the company’s financial history database;

[0071] Regarding tax rates, invoice management and tax filing time, the system will combine industry best practices to provide recommendations that comply with the latest policies and regulations;

[0072] Through the preset rule base, the system automatically matches the causes of non-compliance items and generates corresponding rectification steps;

[0073] The system will generate clear operational steps to guide financial personnel on how to perform rectification in the enterprise financial management platform.

[0074] The beneficial effects of the present invention are: through modular design, including an acquisition module, a processing module, a comparison module, an anomaly detection module, an analysis module, an audit module and a recommendation module, it is possible to collect data from multiple financial management platforms, perform standardized processing and intelligent comparison, and rely on machine learning models to achieve anomaly detection and financial risk prediction, effectively solving the data consistency, anomaly detection and compliance audit problems in the prior art, and belonging to the field of financial big data intelligent analysis and risk management technology. BRIEF DESCRIPTION OF THE DRAWINGS

[0075] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0076] Figure 1 This is a system diagram of the financial and accounting supervision system based on Internet big data technology in Example 1. DETAILED DESCRIPTION

[0077] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings.

[0078] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0079] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0080] Example 1, reference Figure 1 , which is the first embodiment of the present invention, and provides a financial and accounting supervision system based on Internet big data technology, including the following steps:

[0081] Acquisition module, processing module, comparison module, anomaly detection module, analysis module, review module and recommendation module,

[0082] A collection module is used to collect financial data from multiple different financial management platforms through the Internet and transmit the collected financial data to the processing module;

[0083] Connect with different financial management platforms such as ERP, CRM, tax system and bank statement system through the Internet, and use the API interface provided by each platform to achieve real-time synchronization and collection of data. The API interface can ensure the real-time transmission of data between systems to avoid data lag or loss. The API interface can ensure the stability and accuracy of data synchronization and avoid inconsistent financial information due to data delay or interruption;

[0084] The acquisition module is connected to the processing module, and the processed data is passed from the processing module to other modules to ensure that standardized data flows uniformly in the system;

[0085] The financial management platform includes but is not limited to the enterprise resource planning ERP system, customer relationship management CRM system, tax management system, and bank statement system.

[0086] The processing module is used to standardize, clean and deduplicate the collected financial data, pass the processed standardized data to the comparison module, and pass the duplicate data to the anomaly detection module and analysis module;

[0087] Based on the API interface, the processing module first performs structured processing on the collected financial data, identifies the data source, uses a machine learning-based classification algorithm to identify the invoice number, amount, and transaction time in the financial data, and classifies the financial data into three different data types: invoice, voucher, and transaction record, and passes them to the processing module;

[0088] The processing module converts the data formats of different platforms into JSON format through the data conversion algorithm, and removes the erroneous data through the cleaning algorithm. In order to ensure the accuracy and consistency of the data, the system will perform the following operations on the collected data:

[0089] By comparing invoice numbers, transaction times and amounts, the processing module uses a hash algorithm to remove duplicate financial data, ensuring that all data stored in the system is unique and valid;

[0090] The system will verify the integrity and format of the data, including the format of the invoice number and the legitimacy of the amount field, and correct or eliminate erroneous data;

[0091] The processed financial data is standardized and scored, and the integrity and consistency of the data are judged. The expression is:

[0092]

[0093] Where S represents the standardized score, N represents the total number of financial data items, i is the index variable, and W i represents the weight coefficient of the ith financial data, which depends on the importance of the data. The core fields of invoice number and amount have higher weights. i represents the timestamp of the ith financial data, log represents the natural logarithm function, A i Indicates the transaction amount of the i-th financial data;

[0094] The formula calculates the weight, timestamp and amount of each piece of financial data and outputs a standardized score S, which is used to measure the completeness and consistency of the current data set. When S is less than the preset threshold (such as 0.8), the system will trigger the data cleaning program to reorganize the batch of data.

[0095] The standardized score S ranges from 0 to 1;

[0096] Among them, when S=1, it means that the data is completely complete and consistent;

[0097] When S = 0, it means that the data is seriously missing or inconsistent;

[0098] When S = a preset threshold, the error correction mechanism is triggered when it is lower than the preset threshold;

[0099] The standardized financial data is first passed to the comparison module. The system will compare the data from different platforms in four dimensions: amount, transaction time, transaction object and invoice number, and generate a comparison report.

[0100] Then, the standardized financial data is divided into two copies in proportion, and one copy is passed to the anomaly detection module for further abnormal behavior prediction;

[0101] At the same time, another copy is passed to the analysis module for trend analysis and risk prediction.

[0102] The comparison module uses an accurate matching algorithm to intelligently compare the processed financial data in terms of amount, transaction time, transaction object, and invoice number, identifies data inconsistencies between different platforms, generates a comparison report, marks abnormal data, and passes the compared abnormal data to the anomaly detection module;

[0103] The processing module passes the standardized data to the comparison module, and represents each piece of financial data as a four-dimensional vector, corresponding to the four dimensions of amount, transaction time, transaction object, and invoice number. The expression is:

[0104] V i =[M i , T i , Oi , F i ];

[0105] Among them, i is the index variable, V i represents the i-th financial data vector, the financial data vector V i It consists of four elements, corresponding to the four key dimensions of financial data: amount, transaction time, transaction object, and invoice number. i represents the amount of the ith financial data, T i Indicates the transaction time of the ith financial data, which has been converted into a timestamp. i The transaction object representing the i-th financial data is converted into a numerical value through a hash algorithm. i The invoice number representing the i-th financial data is also converted into a numerical value through the hash algorithm;

[0106] The comparison module determines the similarity between two data vectors by calculating the weighted cosine similarity and obtains the similarity Sim. kj , where Sim kj Indicates the similarity between the kth data and the jth data;

[0107] SimilaritySim kj The value range is between 0 and 1. kj =1, it means the two data are completely matched. kj =0, it means the two data do not match at all;

[0108] Set a similarity threshold τ. The threshold τ can be adjusted according to actual business needs. For example, in scenarios where the amount and invoice number have high requirements, the threshold τ can be set higher (such as 0.9) to ensure data consistency.

[0109] When Sim kj When ≥τ, the two data are considered to match;

[0110] When Sim kj When <τ, the two data are considered mismatched and marked as abnormal data;

[0111] Generate a comparison report based on the marked abnormal data, and the abnormal data and the comparison report will be passed to the anomaly detection module;

[0112] Comparison report, including the index of unmatched data items, specific unmatched dimensions (such as amount, transaction time, transaction object and invoice number), and the similarity of each data item. kj .

[0113] The anomaly detection module builds an abnormal behavior prediction model based on the machine learning algorithm of big data, uses historical financial data to train the abnormal behavior prediction model, inputs abnormal data into the abnormal behavior prediction model, analyzes transaction data and predicts potential abnormal behaviors, monitors real-time financial transactions, detects and generates early warning information in real time, transmits abnormal data to the analysis module, and transmits early warning information to the audit module for compliance review;

[0114] Multilayer neural network (MLP) is selected as the core model of the abnormal behavior prediction model. MLP can capture complex nonlinear relationships in financial data and is suitable for processing high-dimensional transaction data, such as amount, transaction time, transaction object, and invoice number.

[0115] Extract features such as amount, transaction time, transaction object, and invoice number;

[0116] For the amount, the transaction amount is used directly and normalized to the range of [0, 1] to avoid instability in model training caused by excessively large values.

[0117] For transaction time, the time information is converted into timestamps, and the periodic characteristics of transaction time are captured through sine and cosine transforms;

[0118] For transaction objects, a hash function is used to encode the transaction objects into a fixed-length numerical vector to ensure that different transaction objects can be distinguished by the model;

[0119] For invoice numbers, they are also digitized using a hash function, and similar invoice numbers are grouped into the same category through bucketing.

[0120] Input the processed financial data into the abnormal behavior prediction model to obtain the abnormal probability value, which is expressed as:

[0121]

[0122] φ y (X y )=log(1+X y )·sin(X y +α y );

[0123] Among them, P ab represents the probability of anomaly between transaction a and transaction b, y is the index variable, e is the natural constant, W y represents the weight of the y-th dimension, φ y (X y ) represents the y-th dimension input feature X y The nonlinear transformation function is performed, φ represents the nonlinear transformation function, sin represents the sine function, αy represents the phase adjustment parameter of the y-th dimension;

[0124] Based on the abnormal probability, set a threshold to judge the abnormal risk, and trigger the early warning mechanism when the abnormal probability exceeds the set threshold;

[0125] P ab The value range is [0, 1]. ab =1, it means that the transaction is very likely to have abnormal behavior, and the system will generate an early warning;

[0126] When P ab =0, it means that the transaction is considered normal and there is no abnormality;

[0127] When 0 <P ab When <1, it means that the transaction has certain abnormal behavior. The system determines whether to generate an early warning based on the threshold. For example, when P ab When ≥0.8, the system considers the transaction to have a high abnormal risk and triggers the early warning mechanism;

[0128] Collect real-time financial data, perform data structuring, classification, data cleaning, standardization, feature extraction, encoding and deduplication processing on the real-time financial data, and calculate the probability value P of abnormal behavior based on the input data and historical data training weights. ab ;

[0129] Setting Thresholds§;

[0130] When P ab When the set threshold is exceeded, the system considers the transaction as a high-risk transaction and generates real-time warning information;

[0131] The system monitors every financial transaction in real time and analyzes potential abnormal behavior;

[0132] Abnormal behaviors monitored in real time include but are not limited to:

[0133] Duplicate reimbursements: Capture duplicate reimbursement behaviors by comparing the similarity of invoice numbers and amounts;

[0134] Abnormal fund flow: By analyzing the patterns of transaction objects and transaction time, abnormal fund flow behavior can be captured;

[0135] High-frequency payment: Identify high-frequency payment behaviors by detecting the intervals between transaction times;

[0136] Abnormal transaction amount: By analyzing the difference between the amount and the historical transaction amount, predict whether there is an abnormal large transaction;

[0137] When the system detects abnormal behavior, it generates warning information and passes the information to the audit module for compliance review;

[0138] Warning information, including transaction number;

[0139] Transaction amount, transaction time, transaction object, invoice number;

[0140] The predicted probability of abnormal behavior P ab ;

[0141] Specific types of abnormal behavior (duplicate reimbursements, abnormal fund flows, high-frequency payments, and abnormal transaction amounts).

[0142] The analysis module conducts time series modeling based on historical financial data, predicts future financial trends and potential financial risks, generates financial trend reports, and transmits the analysis results to the audit module;

[0143] The acquisition module collects historical financial data from multiple financial management platforms. The processing module standardizes, cleans and removes duplicates of these data. The processing module passes these data to the analysis module and enters the time series modeling stage.

[0144] In time series modeling, the autoregressive integrated moving average ARIMA model is selected as the basis for prediction. The ARIMA model is suitable for processing financial data with trends and seasonality. It can effectively capture the historical trends of financial data and predict future financial trends. The expression is:

[0145] X t =&1X t-1 +&2X t-2 +…+& p X t-p +θ1∈ t-1 +θ2∈ t-2 +…+θ u ∈ t-u +∈ t ;

[0146] Among them, X t represents the financial data at time t, & p The table is the u-th natural regression coefficient, θ u represents the uth sliding average coefficient, ∈ t is a white noise term, which indicates the random error that cannot be explained by the model. Both p and u are index variables.

[0147] X-based t Calculate the financial risk prediction value to enhance sensitivity to abnormal financial data while capturing trends and volatility. The expression is:

[0148]

[0149] Among them, Ft is the predicted value of financial risk at time t, represents the number of historical time points, ψ z (X t-z ) represents the nonlinear transformation function at the zth time point, z represents the index variable, W z represents the weight of the zth time point;

[0150] Financial risk prediction value F t The value range is [0,1], indicating the risk level of future financial data;

[0151] When F t =1, it means that the financial data at the current time point may have extremely high risks, and the system will generate a risk warning;

[0152] When F t =0, it means that the financial data has no risk and is a normal fluctuation;

[0153] When 0<F t When it is less than 1, it means that there is a certain risk in the financial data. The specific risk depends on F t The threshold value of the present invention is set to 0.8. If the threshold value is exceeded, it is considered that there is a high financial risk.

[0154] A financial trend report is generated based on the transaction amount, transaction time, transaction object, invoice number and financial risk forecast value. The analysis module passes the generated financial trend report and risk assessment report to the audit module.

[0155] The audit module is used to conduct compliance audits on the company's financial data according to the latest financial policies and regulations, pass the audit results to the recommendation module, and feedback the compliance audit report to the financial staff;

[0156] Through the API interface connection between the system and the tax bureau and the Ministry of Finance, the system automatically captures and parses the latest policy documents, extracts tax rate changes, invoice management and tax reporting requirements, and generates tax rate inspection rules based on the parsed policies and regulations. These rules include tax rules, invoice management rules, and matching of tax reporting time and amount. They will form constraints on corporate financial data in a formulaic way, as expressed in the following expression:

[0157]

[0158] Where R(t) represents the tax rate rule used at time t, b new policy Indicates the effective date of the policy, B new Indicates the tax rate stipulated by the latest financial policy, B old It indicates the old tax rate used before the new policy came into effect;

[0159] The system compares each piece of financial data with the compliance rules and calculates the compliance score of each piece of data based on the generated tax rate check rules. The expression is:

[0160]

[0161] Among them, S c It represents the compliance score of the enterprise's financial data, with a value range of [0, 1], indicating the degree of compliance. c =1, indicating that all financial data are fully compliant; if S c = 0, indicating that the data is seriously non-compliant, i represents the index variable, N represents the total number of financial data items, W i The weight of the ith financial data, α i represents the adjustment coefficient of the ith financial data, X i represents the difference between the i-th financial data and regulatory requirements, μ i Represents a compliance standard value in a policy or regulation, such as a prescribed standard tax rate or invoice number format;

[0162] Set a threshold. When the compliance score of a piece of data is S c When the data is lower than the preset threshold, the system will mark it as abnormal data. The threshold can be flexibly adjusted according to industry standards and enterprise needs. The system generates an overall compliance audit report for the enterprise based on the scores of all data. The audit report includes the compliance score, data items that do not meet the compliance standards, and the reasons for non-compliance (such as incorrect tax rates and inconsistent amounts);

[0163] After the audit is completed, the system will pass the audit results to the recommendation module, and the audit report will also be simultaneously fed back to the financial personnel to help them identify problems in a timely manner and make data corrections.

[0164] The recommendation module is used to generate rectification suggestions for financial data that does not meet compliance requirements;

[0165] The system extracts historical compliance records, past corrective actions, and financial operation logs from the company’s financial history database;

[0166] Regarding tax rates, invoice management and tax filing time, the system will combine industry best practices to provide recommendations that comply with the latest policies and regulations;

[0167] Through the preset rule base, the system automatically matches the reasons for non-compliance and generates corresponding rectification steps. For example, if the invoice number is non-compliant, the system will suggest reissuing a compliant invoice or modifying the invoice information to meet the policy requirements.

[0168] The system will generate clear operational steps to guide financial personnel on how to perform rectification in the enterprise financial management platform.

[0169] This embodiment also provides a computer device, which is suitable for the financial accounting inspection system based on Internet big data technology, including: a memory and a processor; the memory is used to store computer executable instructions, and the processor is used to execute computer executable instructions to realize the financial accounting inspection system based on Internet big data technology as proposed in the above embodiment.

[0170] The computer device may be a terminal, and the computer device includes a processor, a memory, a communication interface, a display screen and an input device connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device may be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device may be a touch layer covering the display screen, or a key, trackball or touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse, etc.

[0171] The present embodiment also provides a storage medium on which a computer program is stored. When the program is executed by a processor, a financial accounting inspection system based on Internet big data technology as proposed in the above embodiment is implemented; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (Static Random Access Memory, referred to as SRAM), electrically erasable programmable read-only memory (Electrically Erasable Programmable Read-Only Memory, referred to as EEPROM), erasable programmable read-only memory (Erasable Programmable Read Only Memory, referred to as EPROM), programmable read-only memory (Programmable Red-Only Memory, referred to as PROM), read-only memory (Read-Only Memory, referred to as ROM), magnetic storage, flash memory, disk or optical disk.

[0172] In summary, the present invention adopts a modular design, including an acquisition module, a processing module, a comparison module, an anomaly detection module, an analysis module, an audit module and a recommendation module. It can collect data from multiple financial management platforms, perform standardized processing and intelligent comparison, and rely on machine learning models to achieve anomaly detection and financial risk prediction, effectively solving the problems of data consistency, anomaly detection and compliance audit in the prior art, and belongs to the field of financial big data intelligent analysis and risk management technology.

[0173] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention can be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should be included in the scope of the claims of the present invention.

Claims

1. A financial and accounting supervision system based on Internet big data technology, characterized by: It includes collection module, processing module, comparison module, anomaly detection module, analysis module, review module and recommendation module. The collection module is used to collect financial data from multiple different financial management platforms via the Internet and transmit the collected financial data to the processing module; The processing module is used to standardize, clean and deduplicate the collected financial data, pass the processed standardized data to the comparison module, and pass the duplicate data to the anomaly detection module and the analysis module; The comparison module uses an accurate matching algorithm to perform intelligent comparison of the processed financial data in four dimensions: amount, transaction time, transaction object, and invoice number, identifies data inconsistencies between different platforms, generates a comparison report and marks abnormal data, and transmits the compared abnormal data to the anomaly detection module; The anomaly detection module builds an abnormal behavior prediction model based on a machine learning algorithm based on big data, uses historical financial data to train the abnormal behavior prediction model, inputs abnormal data into the abnormal behavior prediction model, analyzes transaction data and predicts potential abnormal behaviors, monitors real-time financial transactions, detects and generates early warning information in real time, transmits abnormal data to the analysis module, and transmits early warning information to the audit module for compliance review; The analysis module performs time series modeling based on historical financial data, predicts future financial trends and potential financial risks, generates financial trend reports, and transmits the analysis results to the audit module; The audit module is used to conduct compliance audit on the financial data of the enterprise according to the latest financial policies and regulations, pass the audit results to the recommendation module, and feedback the compliance audit report to the financial personnel; The recommendation module is used to generate rectification suggestions for financial data that does not meet compliance requirements.

2. The financial and accounting supervision system based on Internet big data technology as claimed in claim 1, characterized in that: The collection of financial data from multiple different financial management platforms via the Internet refers to: Connect with different financial management platforms such as ERP, CRM, tax system and bank statement system through the Internet, and use the API interface provided by each platform to achieve real-time synchronization and collection of data; The acquisition module is connected to the processing module, and the processed data is transmitted from the processing module to other modules.

3. The financial and accounting supervision system based on Internet big data technology as claimed in claim 2 is characterized by: The method is used to standardize, clean and remove duplicates of the collected financial data, and the specific steps are as follows: Based on the API interface, the processing module first performs structured processing on the collected financial data, identifies the data source, uses a machine learning-based classification algorithm to identify the invoice number, amount, and transaction time in the financial data, and classifies the financial data into three different data types: invoice, voucher, and transaction record, and passes them to the processing module; The processing module converts the data formats of different platforms into JSON format through data conversion algorithms, and removes erroneous data through cleaning algorithms; By comparing invoice numbers, transaction times and amounts, the processing module uses a hash algorithm to remove duplicate financial data; Standardize and score the processed financial data and judge the completeness and consistency of the data; The standardized financial data is first passed to the comparison module. The system will compare the data from different platforms in four dimensions: amount, transaction time, transaction object and invoice number, and generate a comparison report. Then, the standardized financial data is divided into two copies in proportion, and one copy is passed to the anomaly detection module; At the same time, another copy is passed to the analysis module.

4. The financial and accounting supervision system based on Internet big data technology as claimed in claim 3 is characterized by: The precise matching algorithm is used to perform intelligent comparison of the processed financial data in four dimensions: amount, transaction time, transaction object and invoice number. The specific steps are: The processing module passes the standardized data to the comparison module, and represents each piece of financial data as a four-dimensional vector, corresponding to the four dimensions of amount, transaction time, transaction object, and invoice number. The expression is: V i =[M i ,T i ,O i ,F i ]; Among them, i is the index variable, V i represents the i-th financial data vector, M i represents the amount of the ith financial data, T i represents the transaction time of the ith financial data, O i represents the transaction object of the i-th financial data, F i Indicates the invoice number of the i-th financial data.

5. The financial and accounting supervision system based on Internet big data technology as claimed in claim 4 is characterized by: The identification of data inconsistencies between different platforms, the generation of a comparison report and the marking of abnormal data, and the transmission of the compared abnormal data to the abnormality detection module are specifically as follows: The comparison module determines the similarity between two data vectors by calculating the weighted cosine similarity and obtains the similarity Sim. kj ; Set the similarity threshold τ; When Sim kj When ≥τ, the two data are considered to match; When Sim kj <τ, the two data are considered mismatched and marked as abnormal data; A comparison report is generated based on the marked abnormal data, and the abnormal data and the comparison report will be passed to the anomaly detection module.

6. The financial and accounting supervision system based on Internet big data technology as claimed in claim 5, characterized in that: The machine learning algorithm based on big data builds an abnormal behavior prediction model, and uses historical financial data to train the abnormal behavior prediction model. The specific steps are: Select the multi-layer neural network MLP as the core model of the abnormal behavior prediction model; For the amount, the transaction amount is used directly and normalized to the interval [0, 1]. For transaction time, the time information is converted into timestamps, and the periodic characteristics of transaction time are captured through sine and cosine transforms; For transaction objects, a hash function is used to encode the transaction object into a fixed-length numerical vector; For invoice numbers, they are also digitized using a hash function, and similar invoice numbers are grouped into the same category through bucketing. Input the processed financial data into the abnormal behavior prediction model to obtain the abnormal probability value, which is expressed as: φy(X y )=log(1+X y )·sin(X y +αy); Among them, P ab represents the probability of anomaly between transaction a and transaction b, y is the index variable, e is the natural constant, W y represents the weight of the y-th dimension, φ y (X y ) represents the y-th dimension input feature X y The nonlinear transformation function is performed, φ represents the nonlinear transformation function, sin represents the sine function, and αy represents the phase adjustment parameter of the y-th dimension; Based on the abnormal probability, a threshold is set to judge the abnormal risk. When the abnormal probability exceeds the set threshold, the early warning mechanism is triggered.

7. The financial and accounting supervision system based on Internet big data technology as claimed in claim 6, characterized in that: The specific steps of inputting abnormal data into the abnormal behavior prediction model, analyzing transaction data and predicting potential abnormal behavior, monitoring real-time financial transactions, and detecting and generating early warning information in real time are as follows: Collect real-time financial data, perform data structuring, classification, data cleaning, standardization, feature extraction, encoding and deduplication processing on the real-time financial data, input the processed real-time financial data into the abnormal behavior prediction model, and calculate the probability value P of abnormal behavior based on the input data and historical data training weights. ab ; Setting Thresholds§; When P ab When the set threshold is exceeded, the system considers the transaction as a high-risk transaction and generates real-time warning information; The system monitors every financial transaction in real time and analyzes potential abnormal behavior; When the system detects abnormal behavior, it generates an early warning message and passes the information to the audit module for compliance review.

8. The financial and accounting supervision system based on Internet big data technology as claimed in claim 7, characterized in that: The time series modeling is based on historical financial data to predict future financial trends and potential financial risks, and generate financial trend reports. The analysis results are passed to the audit module, specifically, The acquisition module collects historical financial data from multiple financial management platforms. The processing module standardizes, cleans and removes duplicates of these data. The processing module passes these data to the analysis module and enters the time series modeling stage. In time series modeling, the autoregressive integrated moving average ARIMA model is selected as the basis for prediction, and the expression is: X t =&1X t-1 +&2X t-2 +…+& p X t-p +θ1∈ t-1 +θ2∈ t-2 +…+θ u ∈ t-u +∈ t ; Among them, X t represents the financial data at time t, & p The table is the u-th natural regression coefficient, θ u represents the u-th sliding average coefficient, ∈t is the white noise term, and p and u are both index variables; X-based t Calculate the predicted value of financial risk, the expression is: Among them, F t is the predicted value of financial risk at time t, represents the number of historical time points, ψ z (X t-z ) represents the nonlinear transformation function at the zth time point, z represents the index variable, W z represents the weight of the zth time point; A financial trend report is generated based on the transaction amount, transaction time, transaction object, invoice number and financial risk forecast value. The analysis module passes the generated financial trend report and risk assessment report to the audit module.

9. The financial and accounting supervision system based on Internet big data technology as claimed in claim 8, characterized in that: The company's financial data is audited for compliance according to the latest financial policies and regulations, and the audit results are passed to the recommendation module. At the same time, the compliance audit report is fed back to the financial personnel. Specifically, Through the API interface connection between the system and the tax bureau and the Ministry of Finance, the system automatically captures and parses the latest policy documents, extracts tax rate changes, invoice management and tax reporting requirements, and generates tax rate inspection rules based on the parsed policies and regulations; The system compares each piece of financial data with the compliance rules and calculates the compliance score of each piece of data based on the generated tax rate check rules. The expression is: Among them, S c represents the compliance score of the enterprise's financial data, with a value range of [0,1], indicating the degree of compliance, i represents the index variable, N represents the total number of financial data items, and W i The weight of the ith financial data, α i represents the adjustment coefficient of the ith financial data, X i represents the difference between the i-th financial data and regulatory requirements, μ i Indicates the compliance standard value in policies and regulations; Set a threshold. When the compliance score of a piece of data is S c When the data is lower than the preset threshold, the system will mark it as abnormal data. The system will generate an overall compliance audit report for the enterprise based on the scores of all data. After the audit is completed, the system will pass the audit results to the recommendation module, and the audit report will also be simultaneously fed back to the financial personnel to help them identify problems in a timely manner and make data corrections.

10. The financial and accounting supervision system based on Internet big data technology as claimed in claim 9, characterized in that: Generating rectification suggestions for financial data that does not meet compliance requirements means: The system extracts historical compliance records, past corrective actions, and financial operation logs from the company’s financial history database; Regarding tax rates, invoice management and tax filing time, the system will combine industry best practices to provide recommendations that comply with the latest policies and regulations; Through the preset rule base, the system automatically matches the causes of non-compliance items and generates corresponding rectification steps; The system will generate clear operational steps to guide financial personnel on how to perform rectification in the enterprise financial management platform.

Images