Automatic process processing method for accounting data

Abstract

The invention provides an accounting data automatic process processing method and system. An internal business system of an enterprise is connected through an adaptive interface to collect data, and the data is cleaned in combination with an OCR technology and a professional vocabulary library. An improved random forest algorithm is applied, a classification model is constructed for attribute weighting according to accounting professional knowledge, and accurate classification and standardization of data are achieved. According to a preset financial process rule, processes such as reimbursement and purchase payment are automatically processed and monitored in real time, and early warning is timely performed when abnormity occurs. And the processed data is encrypted and stored, a backup and authority management mechanism is established, and a financial statement can be automatically generated and visually displayed. The problems that traditional accounting data processing is low in efficiency and prone to errors are solved, the data processing efficiency and accuracy are improved, data safety is guaranteed, and efficient support is provided for enterprise financial management.

Description

Technical Field

[0001] The present invention relates to the field of data processing, and specifically to an automatic process for processing accounting data. Background Art

[0002] In the traditional accounting data processing process, a large amount of accounting data, such as daily expense reimbursement forms, invoice information, bank statement data of enterprises, etc., all need to be manually input into the financial system by accounting personnel. Taking the processing of expense reimbursement forms as an example, accounting personnel need to check one by one the information such as employee name, department, reimbursement item, amount, invoice number, etc. on the reimbursement form, and accurately input this information into the financial software system. This process not only consumes a large amount of time and manpower, but also is prone to data entry errors due to inevitable negligence in manual operations. Once data entry errors occur, a series of subsequent work such as financial statement generation, cost accounting, tax declaration, etc. will be affected, which may lead to inaccurate financial data and bring potential risks to the decision-making and operation of enterprises. In addition, the low efficiency of manual input makes it difficult for the accounting department to quickly complete data processing work when facing a large amount of data, affecting the overall timeliness of financial work. Therefore, there is an urgent need for a process method that can automatically, accurately and efficiently process accounting data. Summary of the Invention

[0003] The purpose of the present invention is to provide an automatic process for processing accounting data to solve the problems raised in the above background art.

[0004] To achieve the above purpose, the present invention provides the following technical solution: An automatic process for processing accounting data, including the following steps:

[0005] Step 1, Data collection: Use an adaptation interface to connect to the enterprise's OA, procurement, and sales internal business systems, regularly capture the electronic expense reimbursement forms submitted by employees from the OA system every day, obtain purchase orders and corresponding invoice data from the procurement system, and then use optical character recognition technology to scan paper invoices and bank statement bills. After scanning, according to the accounting professional vocabulary library and grammar rules, automatically verify and correct the error characters recognized by OCR, and at the same time unify the data format, remove duplicate data, and fill in missing values to complete data cleaning and obtain the collected data;

[0006] Step 2, Data classification and standardization: Use an improved machine learning algorithm to construct a classification model with reference to the enterprise's accounting subject system, and input the collected data cleaned in Step 1 into the classification model to accurately classify it into the corresponding accounting subjects;

[0007] Step 3, Automated Process Handling: For the reimbursed data, procurement data, and sales data classified in Step 2, process them according to the enterprise's financial approval and accounting processing procedures, where the processing procedure is the process rules preset in the system;

[0008] Step 4, Data Storage and Analysis: The processed accounting data in Step 3 is encrypted and stored in the enterprise database, and a regular backup mechanism is established to prevent data loss; according to the responsibilities of personnel, set access permissions for different data to ensure data security; at the same time, use data analysis tools and algorithms to automatically generate financial statement data including but not limited to balance sheets, income statements, and cash flow statements based on the stored data; through data visualization technology, present financial data in intuitive charts;

[0009] In Step 2, an improved machine learning algorithm is used. Specifically, an improved random forest algorithm that assigns different weights to different attributes according to accounting expertise is used. A classification model is constructed through the improved random forest algorithm, and the collected data is automatically and accurately classified through the classification model.

[0010] Preferably, the specific implementation steps of the improved random forest algorithm are as follows:

[0011] Data Preparation Phase: From the enterprise's historical accounting data, construct an accounting subject sample data set by using stratified sampling combined with dynamic ratio adjustment. Let the historical accounting data set be , and the data set contains various accounting data samples with different proportions. Among them, the accounting data samples include but are not limited to expense reimbursement data , procurement data , and sales data corresponding accounting subject samples; when performing stratified sampling, first divide into non-overlapping layers according to business types, that is, . For each layer , its data volume is denoted as and the importance level is denoted as . The importance level is determined by comprehensive evaluation by accounting domain experts according to the influence degree of data of different business types on accounting subject classification, and the value range is , and . Dynamically adjust the extraction ratio according to the data volume and importance level of each layer. After calculating the extraction ratio of each layer according to the above formula, randomly extract data from each layer according to the corresponding ratio to construct a diversified training subset ;

[0012] Decision Tree Training Phase: For each constructed training subset , respectively carry out the training work of the decision tree; during the node splitting process of the decision tree, based on the analytic hierarchy process combined with the experience of accounting experts, automatically assign weights to the decision tree node splitting attributes, and according to the calculated weight distribution method, for each training subset perform recursive partitioning to construct multiple decision trees;

[0013] Random forest integration stage: Combine the multiple trained decision trees into a random forest classification model, that is, the classification model;

[0014] Data classification execution stage: Input the collected data cleaned in step 1 into the constructed classification model. Each decision tree in the classification model independently classifies and judges the input data. Each decision tree analyzes each feature of the input data according to the rules and attribute weights obtained from its own training, and then gives a classification result. After all decision trees complete the classification judgment, the final classification result of the input data is determined through a voting mechanism;

[0015] Standardization operation stage: After completing data classification, strictly standardize the classified data according to accounting standards and the financial system formulated within the enterprise.

[0016] Preferably, during the node splitting process of the decision tree, based on the analytic hierarchy process combined with the experience of accounting experts, automatically assign weights to the decision tree node splitting attributes. The specific implementation logic is as follows:

[0017] Step A, determine the attribute set and decision objective: First, the relevant attributes of the training subset In the expense reimbursement data, the attributes include "reimbursement item", "reimbursement amount", "reimbursement time", "reimbursing person's department"; for procurement data, the attributes are "procurement item category", "procurement quantity", "supplier", "procurement date", and for sales data, "sales item", "sales amount", "sales time", "salesperson's department"; classify the above accurately classified accounting attribute data into the corresponding accounting subjects as the decision objective;

[0018] Step B, construct the judgment matrix: Combine the experience of accounting experts and the sample attributes of accounting subjects for different business types to construct the judgment matrix. Let the of the training subset attribute sets be , for each attribute data in the accounting subject sample, construct the judgment matrix: , where represents the importance degree of attribute relative to attribute , and satisfies , , when constructing the judgment matrix, the 1-9 scale method is adopted to compare the relative importance between every two types of attributes in each accounting subject sample. Specifically, if attribute is equally important as attribute , then ; if is slightly more important than , ; significantly more important, ; strongly more important, ; extremely more important, ; if it is between the two, take the values 2, 4, 6, and 8 respectively;

[0019] Step C. Calculate the attribute weight vector: Use the eigenvector method to calculate the maximum eigenvalue of the judgment matrix and its corresponding eigenvector . By solving the equation , the eigenvector is obtained;

[0020] Step D. Consistency test: Calculate the consistency index , and the formula is , where is the number of attributes; then look up the average random consistency index , which is obtained from the standard table according to the number of attributes ; calculate the consistency ratio , and the formula is . When , accept the consistency of the judgment matrix, that is, the judgment of the expert is reasonable and the obtained weight vector is valid; if , then readjust the judgment matrix until the consistency test passes;

[0021] Step E. Integrate the expert weight results: For each attribute in the training subset , integrate the weights given by the experts. Using the simple average method, assume the number of experts is , and the weight of attribute given by the rd expert is . Then the final weight of the integrated attribute is calculated by the formula: ;

[0022] Step F. Apply the weights to the decision tree node splitting: Apply the obtained final attribute weights to the decision tree node splitting process.

[0023] Preferably, the eigenvector method is used to calculate the maximum eigenvalue of the judgment matrix and its corresponding eigenvector The specific steps are as follows:

[0024] Normalize the judgment matrix column by column: Assume the constructed judgment matrix is , for each column of the matrix, calculate the sum of the elements in this column , and then divide each element in the matrix by the sum of the elements in this column to obtain the normalized matrix element . Through the above calculations, complete the column-by-column normalization of the entire judgment matrix to obtain the normalized matrix ;

[0025] Sum the elements of the normalized matrix row by row: For the normalized matrix , calculate the sum of the elements in each row ;

[0026] Normalize the row sum vector to obtain the eigenvector : Normalize the row sum vector , that is, calculate to obtain the eigenvector . Each component in the eigenvector corresponds to the relative weight of each attribute in the accounting subject sample;

[0027] Calculate the maximum eigenvalue : Calculate to obtain the vector , and then calculate , where is the -th element of the vector . The maximum eigenvalue is used for subsequent consistency tests to judge the rationality of expert judgments.

[0028] Preferably, the specific content of the automated process in step 3 is as follows:

[0029] Preset process rules: For reimbursement and sales processes, the enterprise presets approvers and approval sequences corresponding to different amount ranges, different reimbursement and sales items; for the procurement payment process, the enterprise determines the approval levels divided according to factors such as the procurement order amount and supplier type; write the above preset rules into code logic as the basis for subsequent automated process execution;

[0030] Reimbursement and sales process handling: According to the classification results of the collected data in invention step 2, automatically screen out expense reimbursement data and sales data; Determine the approver corresponding to each reimbursement and sales order according to the preset rules, and send the reimbursement form to the approver's work interface through the internal message push mechanism or by integrating with the enterprise office software, along with detailed reimbursement and sales information, including but not limited to the names, departments, reimbursement items, amounts, and invoices of the reimbursers and salespersons; After receiving the reimbursement and sales form, the approver conducts the approval operation and selects to approve, reject, or return for modification; If the approver approves, automatically transfer the data to the financial accounting link; If rejected or returned for modification, promptly notify the reimburser or salesperson and explain the reasons; In the financial accounting link, according to the reimbursement and sales form data, automatically generate accounting vouchers in accordance with accounting standards and the enterprise's financial system to complete the accounting process;

[0031] Purchase payment process handling: According to the classification results of the collected data in invention step 2, automatically screen out purchase data, and real-time monitor the purchase order data and the associated invoice data in the purchase data; When the purchase data information matches and meets the preset payment conditions, automatically generate a payment application, including but not limited to the purchase order number, supplier information, payment amount, and payment method; After the payment application is generated, according to the preset approval hierarchy rules, push the application to the corresponding approvers in sequence. After the approver conducts the approval operation and the approval is passed, if all levels of approval are completed, automatically send an instruction to the enterprise's fund payment system to trigger the payment operation to complete the payment;

[0032] Data transfer and processing progress monitoring: Real-time monitor the input data transfer path and processing progress in the reimbursement and sales process handling and the purchase payment process handling; That is, by setting marks and timestamps at each process node, record the time when the data enters and leaves each link, as well as the current location;

[0033] Exception handling and reminder: Continuously check whether there are abnormal situations in the data transfer and processing process; If the approval times out, immediately start the reminder mechanism and send reminder emails or text messages to the approver, reimburser, or relevant business responsible person to urge them to complete the approval operation; If it is found that the data does not meet the preset rules, send a notice to inform the details of the abnormality and suspend the relevant process until the problem is solved.

[0034] Preferably, an accounting data automatic process system includes a data collection module, a data classification and standardization module, an automatic process processing module, and a data storage and analysis module; the data collection module includes an adaptation interface development component and an OCR scanning and data cleaning component, which can realize the data connection with the enterprise internal business system and the data collection and cleaning of paper bills; the data classification and standardization module integrates a machine learning algorithm module and a data standardization processing component, which can complete the accurate classification and standardization operation of data; the automatic process processing module has a process rule preset component and a monitoring and warning component, which can automatically process the data process according to the preset rules and give timely warnings when abnormalities occur; the data storage and analysis module includes a data encrypted storage component, a report generation component, and a data visualization component, which realizes the secure storage, report generation, and visualization display of data.

[0035] Preferably, an electronic device includes: a processor and a memory, wherein a computer program callable by the processor is stored in the memory:

[0036] The processor executes an accounting data automatic process system by calling the computer program stored in the memory.

[0037] Preferably, a computer program product stored on a computer-readable medium includes a computer-readable program, which provides a user input interface to implement an accounting data automatic process system when executed on an electronic device.

[0038] Compared with the prior art, the beneficial effects of the present invention are: improving data processing efficiency: automatically and regularly capturing data from enterprise internal business systems such as OA, procurement, and sales through the adaptation interface, and using optical character recognition technology to scan paper bills, while automatically completing data cleaning, replacing the traditional manual entry method. This greatly reduces the time and labor costs of data collection.

[0039] Improving data processing accuracy: Using an improved machine learning algorithm, especially an improved random forest algorithm to construct a classification model, and combining the experience of accounting experts to assign weights to different attributes, can accurately classify the collected data into the corresponding accounting subjects. Compared with the traditional manual classification or simple machine learning classification methods, the classification error rate is greatly reduced.

[0040] Enhance data security and confidentiality: In terms of data storage, mature and highly secure encryption algorithms in the industry are used to encrypt and store accounting data, such as the AES encryption algorithm, effectively preventing data from being stolen or tampered with during storage. At the same time, a regular backup mechanism is established to prevent data loss due to unexpected situations such as hardware failures and virus attacks, ensuring data persistence and availability. Access permissions are set for different data according to personnel responsibilities, and only authorized personnel can access the corresponding data, further ensuring data security.

[0041] Optimize enterprise financial management decision support: Utilize data analysis tools and algorithms to automatically generate various financial statements such as balance sheets, income statements, and cash flow statements based on the stored data, and present the financial data in intuitive charts through data visualization technology. Brief Description of the Drawings

[0042] Figure 1 It is a schematic flow chart of the method of the present invention;

[0043] Figure 2 It is a schematic implementation flow chart of the improved random forest algorithm of the present invention;

[0044] Figure 3 It is a schematic diagram of the system structure of the present invention;

[0045] Figure 4 It is a schematic diagram of the structure of an electronic device. Detailed Embodiments

[0046] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without making creative efforts shall fall within the protection scope of the present invention.

[0047] Embodiment 1

[0048] Please refer to Figure 1-2 , the present invention provides a technical solution: An automatic process method for accounting data, including the following steps:

[0049] Step 1, data collection: Use an adaptation interface to connect to the enterprise OA, procurement, and sales internal business systems, regularly capture the electronic expense reimbursement forms submitted by employees from the OA system every day, obtain procurement orders and corresponding invoice data from the procurement system, and then use optical character recognition technology to scan paper invoices and bank statement bills. After scanning, according to the accounting professional vocabulary library and grammar rules, automatically verify and correct the error characters recognized by OCR, and at the same time unify the data format, remove duplicate data, and fill in missing values to complete data cleaning and obtain the collected data;

[0050] Step 2: Data Classification and Standardization: Using an improved machine learning algorithm, a classification model is constructed with reference to the enterprise accounting subject system. The collected data cleaned in Step 1 is input into the classification model and accurately classified into the corresponding accounting subjects.

[0051] Step 3: Automated Process Processing: For the classified reimbursement data, procurement data, and sales data in Step 2, they are processed according to the enterprise financial approval and accounting processing procedures, where the processing procedures are the process rules preset in the system. The specific content of the processing procedures is as follows:

[0052] Preset Process Rules: For the reimbursement and sales processes, the enterprise presets the approvers and approval sequences corresponding to different amount ranges, different reimbursement and sales items; for the procurement payment process, the enterprise determines the approval levels divided according to factors such as the procurement order amount and supplier type; the above preset rules are written into code logic as the basis for the subsequent automated process execution.

[0053] Reimbursement and Sales Process Processing: According to the classification results of the collected data in Step 2 of the invention, the expense reimbursement data and sales data are automatically screened out; according to the preset rules, the approver corresponding to each reimbursement and sales order is determined, and the reimbursement form is sent to the approver's work interface through the internal message push mechanism or integrated with the enterprise office software, and detailed reimbursement and sales information is attached, including but not limited to the names, departments, reimbursement items, amounts, and bills of the reimbursers and salespersons; after receiving the reimbursement and sales forms, the approver conducts an approval operation and selects to approve, reject, or return for modification; if the approver approves, the data is automatically transferred to the financial accounting link; if rejected or returned for modification, the reimburser or salesperson is notified in a timely manner and the reason is explained; in the financial accounting link, according to the reimbursement and sales form data, accounting vouchers are automatically generated in accordance with accounting standards and enterprise financial systems to complete the accounting processing.

[0054] Procurement Payment Process Processing: According to the classification results of the collected data in Step 2 of the invention, the procurement data is automatically screened out, and the procurement order data and the associated invoice data in the procurement data are monitored in real time; when the procurement data information matches and meets the preset payment conditions, a payment application is automatically generated, including but not limited to the procurement order number, supplier information, payment amount, and payment method; after the payment application is generated, according to the preset approval level rules, the application is sequentially pushed to the corresponding approvers, and the approvers conduct approval operations. After the approval is passed, if all levels of approval are completed, an instruction is automatically sent to the enterprise's fund payment system to trigger the payment operation to complete the payment.

[0055] Data Flow and Processing Progress Monitoring: Real-time monitoring of the input data flow path and processing progress in the reimbursement and sales process handling and the procurement payment process handling; that is, by setting marks and timestamps at each process node to record the time when data enters and leaves each link, as well as the current location.

[0056] Exception Handling and Reminder: Continuously check whether there are abnormal situations in the data flow and processing process; if the approval times out, immediately start the reminder mechanism and send reminder emails or text messages to the approver, reimburser, or relevant business person in charge to urge them to complete the approval operation; if it is found that the data does not conform to the preset rules, send a notice to inform the details of the abnormality and suspend the relevant process until the problem is solved.

[0057] Step 4: Data Storage and Analysis: The processed accounting data in Step 3 is encrypted and stored in the enterprise database, and a regular backup mechanism is established to prevent data loss; access permissions are set for different data according to personnel responsibilities to ensure data security; at the same time, using data analysis tools and algorithms, automatically generate financial statement data including but not limited to balance sheets, income statements, and cash flow statements based on the stored data; through data visualization technology, present financial data in intuitive charts.

[0058] Among them, for the processed accounting data encrypted and stored in the enterprise database, the RSA asymmetric encryption algorithm is specifically used. This encryption algorithm belongs to existing mature technology and is briefly described here. The RSA asymmetric encryption algorithm uses a pair of keys, namely the public key and the private key. The public key is used to encrypt data, and the private key is used to decrypt data. In the enterprise accounting data storage scenario, the public key can be distributed to relevant personnel who need to upload or access data, such as financial personnel, auditors, etc. When these personnel upload accounting data to the database, the public key is used to encrypt the data. During the transmission and storage of the encrypted data, even if it is obtained by a third party, without the corresponding private key, the data cannot be decrypted. And the enterprise data management department with the private key can decrypt the data when needed to ensure the security and confidentiality of the data.

[0059] In Step 2, an improved machine learning algorithm is used, specifically the improved random forest algorithm that assigns different weights to different attributes for accounting professional knowledge. A classification model is constructed through the improved random forest algorithm, and the collected data is automatically and accurately classified through the classification model.

[0060] The specific implementation steps of the improved random forest algorithm are as follows:

[0061] Data Preparation Phase: From the enterprise's historical accounting data, construct an accounting subject sample data set by using stratified sampling combined with dynamic ratio adjustment. Let the historical accounting data set be , data set Include various accounting data samples with different proportions, where the accounting data samples include but are not limited to expense reimbursement data , procurement data , sales data and the corresponding accounting subject samples; when conducting stratified sampling, first divide into non-overlapping layers according to business types, that is . For each layer , record its data volume as and the importance level as . The importance level is determined by the comprehensive evaluation of accounting field experts on the impact degree of data of different business types on accounting subject classification, and the value range is , and . Dynamically adjust the extraction ratio according to the data volume and importance level of each layer. After calculating the extraction ratio of each layer according to the above formula, randomly extract data from each layer according to the corresponding ratio to construct a diversified training subset ;

[0062] Based on the above description, an example is given here: for example, assume , ; , ; , . Then , , , and each subset contains rich and proportionally different business data samples of various types.

[0063] Advantages in the data preparation stage: Compared with traditional algorithms, the training subsets constructed in this way can cover the diversity of accounting data more comprehensively, enable the model to learn the complex relationships between data features and accounting subjects in different business scenarios, improve the classification ability of the model for complex business data, and reduce classification biases caused by uneven data distribution.

[0064] Decision tree training stage: For each constructed training subset , conduct the training work of decision trees respectively; during the node splitting process of decision trees, based on the method combining the analytic hierarchy process and accounting expert experience, automatically assign weights to the decision tree node splitting attributes, and recursively partition each training subset according to the calculated weight assignment method to construct multiple decision trees;

[0065] This method integrates accounting business logic into the decision tree construction process. Compared with traditional algorithms, decision trees can capture key attributes more accurately during classification, improving classification accuracy. Especially when dealing with accounting data with clear business rules, it can effectively avoid misclassification caused by simply relying on general indicators and ignoring the business essence.

[0066] Random forest integration stage: Combine the multiple trained decision trees mentioned above into a random forest classification model, that is, the classification model.

[0067] Compared with traditional algorithms, the improved random forest can better capture the potential connections between accounting data. When facing data classification involving multiple business processes, it can comprehensively consider various factors, further improving the classification accuracy rate and meeting the needs of enterprises for processing complex financial data.

[0068] Data classification execution stage: Input the collected data cleaned in step 1 into the constructed classification model. Each decision tree in the classification model independently classifies and judges the input data. Each decision tree analyzes each feature of the input data according to the rules and attribute weights obtained from its own training, and then gives a classification result. After all decision trees complete the classification and judgment, the final classification result of the input data is determined through a voting mechanism.

[0069] An example is given here for determining the final classification result of the input data through the voting mechanism:

[0070] Initialize vote counting: For all possible accounting subject categories, create corresponding vote count records, and set the initial votes to 0. Suppose the accounting subject categories are "Administrative Expenses - Travel Expenses", "Administrative Expenses - Office Expenses", "Selling Expenses - Business Promotion Expenses", etc. Then prepare a "counter" for each of these categories to record the votes, and initially record them as 0 votes.

[0071] Count the voting results of decision trees: After each decision tree completes the classification and judgment of the input data, add 1 to the votes of the category corresponding to the classification result it gives. For example, if the first decision tree determines that a certain expense reimbursement data belongs to "Administrative Expenses - Travel Expenses", then add 1 to the votes of "Administrative Expenses - Travel Expenses"; if the second decision tree determines that it belongs to "Selling Expenses - Business Promotion Expenses", then add 1 to the votes of "Selling Expenses - Business Promotion Expenses", and so on until the votes of all decision trees are counted.

[0072] Find the category with the most votes: After counting the votes of all decision trees, compare the votes of each accounting subject category. Find the category with the most votes among them. If the votes of "Administrative Expenses - Travel Expenses" are the most among all categories, then it is the "leader" of this classification.

[0073] Handling the situation of equal vote counts: If there are multiple categories with the same number of votes, further processing is required. A random selection method can be adopted to randomly pick one from the categories with the same number of votes as the final classification result; or a second-round voting can be conducted to let the decision trees related to these categories with the same number of votes vote again to decide; it is also possible to combine the actual situation of accounting operations and, based on prior knowledge, preferentially select the categories that are more common and more in line with business logic in enterprise accounting operations.

[0074] Determining the final classification result: After the above steps, the determined category is the final classification result of the input data. This result will be used for subsequent processing of accounting data, such as bookkeeping, generating financial statements, etc.

[0075] Compared with the traditional simple majority voting, this weighted voting method can more reasonably synthesize the classification results of decision trees, avoid the influence of the misclassification of individual low-quality decision trees on the final result, improve the reliability and stability of data classification, and especially can effectively improve the classification quality when dealing with large-scale accounting data.

[0076] Standardization operation stage: After completing data classification, strictly in accordance with accounting standards and the financial system formulated within the enterprise, perform standardization processing on the classified data; for example, in terms of unifying the amount precision and converting measurement units, traditional methods may lack in-depth integration with accounting standards and the personalized financial system of the enterprise. In the standardization process of the present invention, not only general accounting standards are followed, but also specific business requirements and financial accounting habits of the enterprise are combined. For example, for enterprises in some special industries, there may be higher requirements for the amount precision. On the basis of retaining two decimal places, special settings for the rounding rule of the third decimal place are also required; in the conversion of measurement units, full consideration is given to the customary usage in the internal business processes of the enterprise. For example, in a manufacturing enterprise, for raw material procurement data, some special measurement units are accurately converted according to the conversion table used by the enterprise for a long time.

[0077] Here, taking the unification of amount precision as an example, let the original amount data be , and the data after unifying the precision be . For enterprises in special industries, if there are special rounding rules for the amount precision, assuming the rule is that when the third decimal place is greater than or equal to it is carried forward, and less than it is discarded ( set according to the enterprise requirements), then the calculation formula is:

[0078]

[0079] In the conversion of measurement units, assume that there is a specific conversion table within the enterprise. For raw material procurement data, the original measurement unit data is , and the unified measurement unit data after conversion is , the conversion relationship is , is the conversion function determined according to the conversion table.

[0080] Compared with the traditional standardization method, the improved method can better meet the actual financial management needs of enterprises, ensure seamless connection of the processed data in the internal financial processes of enterprises, and improve the availability and accuracy of data in subsequent financial analysis, report generation and other links.

[0081] For the node splitting process of the decision tree, a method based on the analytic hierarchy process combined with the experience of accounting experts is used to automatically assign weights to the decision tree node splitting attributes. The specific implementation logic is as follows:

[0082] Step A, determine the attribute set and decision goal: First, train the relevant attributes of the subset . Specifically, in the expense reimbursement data, the attributes include "reimbursement item", "reimbursement amount", "reimbursement time", "reimburser's department"; for procurement data, the attributes are "procurement item category", "procurement quantity", "supplier", "procurement date", and for sales data, "sales item", "sales amount", "sales time", "seller's department"; classify the above accurate accounting attribute data into the corresponding accounting subjects as the decision goal;

[0083] Step B, construct the judgment matrix: Combine the experience of accounting experts and the sample attributes of accounting subjects for different business types to construct the judgment matrix. Let the attribute sets of the training subset be . For each attribute data in the accounting subject sample, construct the judgment matrix: , where represents the importance degree of attribute relative to attribute , and satisfies , . When constructing the judgment matrix, the 1-9 scale method is used to compare the relative importance of various attributes in each accounting subject sample pairwise. Specifically, if attribute is equally important as attribute , then ; if is slightly more important than , ; significantly more important, ; strongly more important, ; extremely more important, ; if it is between the two, take the values 2, 4, 6, 8 respectively;

[0084] For example, experts generally believe that the "reimbursement item" directly determines the accounting subject to which the expense belongs, and is extremely important for determining the accounting subject compared to the "reimbursement time". Then, in the judgment matrix, the position of the "reimbursement item" row and the "reimbursement time" column is assigned 9; according to , the position of the "reimbursement time" row and the "reimbursement item" column is assigned . If experts judge that the "reimbursement amount" and the "department of the reimburser" are equally important for determining the accounting subject, both the "reimbursement amount" row and the "department of the reimburser" column, and the "department of the reimburser" row and the "reimbursement amount" column in the matrix are assigned 1. In this way, the assignment for pairwise comparison of all attributes is completed, and a matrix applicable to the importance judgment of expense reimbursement data attributes is constructed.

[0085] Step C: Calculate the attribute weight vector: Use the eigenvector method to calculate the maximum eigenvalue of the judgment matrix and its corresponding eigenvector . By solving the equation , the eigenvector is obtained;

[0086] Step D: Consistency test: Calculate the consistency index , and the formula is , where is the number of attributes; then look up the average random consistency index , which is obtained from the standard table according to the number of attributes ; calculate the consistency ratio , and the formula is . When , accept the consistency of the judgment matrix, that is, the experts' judgment is reasonable and the obtained weight vector is valid; if , then readjust the judgment matrix until the consistency test passes;

[0087] Step E: Integrate the experts' weight results: For each attribute in the training subset , synthesize the weights given by the experts. Using the simple average method, assume the number of experts is , and the weight of the th expert for the attribute is . Then the final weight of the integrated attribute is calculated by the formula: ;

[0088] Step F: Apply the weights to the decision tree node splitting: Apply the obtained final attribute weights to the decision tree node splitting process;

[0089] Among them, use the eigenvector method to calculate the maximum eigenvalue and its corresponding eigenvector The specific steps are as follows:

[0090] Normalize the judgment matrix column by column: Suppose the constructed judgment matrix is , for each column of the matrix , calculate the sum of the elements in this column , and then divide each element in the matrix by the sum of the elements in this column to obtain the normalized matrix element . Through the above calculations, complete the column-by-column normalization of the entire judgment matrix to obtain the normalized matrix ;

[0091] Sum the normalized matrix row by row: For the normalized matrix , calculate the sum of the elements in each row ;

[0092] Normalize the row sum vector to obtain the eigenvector : Normalize the row sum vector , that is, calculate to obtain the eigenvector . Each component in the eigenvector corresponds to the relative weight of each attribute in the accounting subject sample;

[0093] Calculate the maximum eigenvalue : Calculate to obtain the vector , and then calculate , where is the th element of the vector . The maximum eigenvalue is used for subsequent consistency testing to judge the rationality of the expert's judgment.

[0094] Example 2

[0095] Please refer to Figure 3, An accounting data automatic process system, including a data collection module, a data classification and standardization module, an automated process processing module, and a data storage and analysis module; the data collection module includes an adaptation interface development component and an OCR scanning and data cleaning component, which can realize the data connection with the enterprise internal business system and the data collection and cleaning of paper bills; the data classification and standardization module integrates a machine learning algorithm module and a data standardization processing component, which can complete the accurate classification and standardization operation of data; the automated process processing module has a process rule preset component and a monitoring and warning component, which can automatically process the data process according to the preset rules and give a timely warning when an abnormality occurs; the data storage and analysis module includes a data encryption storage component, a report generation component, and a data visualization component, which realizes the secure storage, report generation, and visualization display of data.

[0096] Embodiment 3

[0097] An electronic device according to an exemplary embodiment includes: a processor and a memory, wherein a computer program that can be called by the processor is stored in the memory;

[0098] The processor executes the above-mentioned accounting data automatic process system by calling the computer program stored in the memory.

[0099] Figure 4 It is a schematic structural diagram of an electronic device provided by an embodiment of the present application. This electronic device may have relatively large differences due to different configurations or performances, and can include one or more processors (Central Processing Units, CPUs) and one or more memories. Among them, at least one computer program is stored in the memory, and the at least one computer program is loaded and executed by the processor to implement an accounting data automatic process system provided by each of the above method embodiments.

[0100] This electronic device can also include other components for realizing the functions of the device. For example, this electronic device can also have components such as wired or wireless network interfaces and input / output interfaces for input / output. This is not elaborated in the embodiments of the present application.

[0101] This embodiment also provides a computer program product stored on a computer-readable medium, including a computer-readable program, which provides a user input interface to implement the above-mentioned accounting data automatic process system when executed on an electronic device.

[0102] It should be understood that in various embodiments of the present application, the sequence numbers of the above processes do not imply the order of execution. The order of execution of each process should be determined according to its function and internal logic, and should not constitute any limitation to the implementation process of the embodiments of the present application.

[0103] It should be understood that determining B according to A does not mean determining B solely based on A, but also B can be determined based on A and / or other information.

[0104] The present invention discloses an automatic process method and system for accounting data. The method first collects data from internal business systems such as enterprise OA, procurement, and sales using an adaptation interface, processes paper bills with the help of optical character recognition technology, and cleans the data according to the accounting professional vocabulary library and grammar rules. Then, an improved random forest algorithm is used to construct a classification model, and the data is accurately classified and standardized according to the enterprise accounting subject system. Next, according to the preset enterprise financial approval and accounting processing flow rules, business processes such as expense reimbursement and purchase payment are automatically processed, and the data flow and processing progress are monitored in real time, and anomalies are timely warned. Finally, the processed data is encrypted and stored in the enterprise database, a backup mechanism is established, access permissions are set, and financial statements are generated using data analysis tools and visually displayed. The invention realizes the automation and intelligence of accounting data processing, effectively improves the data processing efficiency and accuracy, enhances data security, provides strong decision-making support for enterprise financial management, and helps enterprises improve their competitiveness.

[0105] Although the embodiments of the present invention have been shown and described, for those of ordinary skill in the art, it can be understood that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. An automatic process method for accounting data, characterized in that, The following steps are involved: Step 1, data collection: Use the adapter interface to connect the company's OA, procurement, and sales internal business systems, grab the electronic expense reimbursement forms submitted by employees from the OA system on a daily basis, obtain the purchase order and corresponding invoice data from the procurement system, and then use optical character recognition technology to scan paper invoices and bank statements. After scanning, automatically check and correct the wrong characters recognized by OCR based on the accounting professional vocabulary library and grammar rules. At the same time, unify the data format, remove duplicate data, fill in missing values, complete data cleaning, and obtain collected data; Step 2: Data classification and standardization: Use the improved machine learning algorithm to build a classification model based on the enterprise accounting subject system, input the cleaned collected data in step 1 into the classification model, and accurately classify it into the corresponding accounting subject; Step 3: Automated process processing: The reimbursement data, purchase data, and sales data classified in step 2 are processed according to the enterprise's financial approval and accounting processing process, where the processing process is a pre-set process rule; Step 4, Data Storage and Analysis: The accounting data processed in step 3 is encrypted and stored in the enterprise database, and a regular backup mechanism is established to prevent data loss; access permissions are set for different data based on personnel responsibilities to ensure data security; at the same time, data analysis tools and algorithms are used to automatically generate financial statement data including but not limited to balance sheet, income statement, and cash flow statement based on the stored data; financial data is presented in intuitive charts through data visualization technology; In step 2, an improved machine learning algorithm is used, specifically an improved random forest algorithm that assigns different weights to different attributes based on accounting expertise. A classification model is constructed through the improved random forest algorithm, and the collected data is automatically and accurately classified through the classification model.

2. The automatic process method for accounting data according to claim 1, characterized in that: The specific implementation steps of the improved random forest algorithm are as follows: Data Preparation Phase: From the enterprise's historical accounting data, construct a sample data set of accounting subjects by using stratified sampling combined with dynamic ratio adjustment. Let the historical accounting data set be , and the data set contains various accounting data samples with different proportions. Among them, the accounting data samples include but are not limited to expense reimbursement data , procurement data , sales data and their corresponding accounting subject samples; during stratified sampling, first divide into non-overlapping layers according to business types, that is . For each layer , its data volume is denoted as and its importance level is denoted as . The importance level is comprehensively evaluated and determined by accounting domain experts according to the influence degree of data of different business types on the classification of accounting subjects, and its value range is , and . Dynamically adjust the extraction ratio according to the data volume and importance level of each layer. After calculating the extraction ratio of each layer according to the above formula, randomly extract data from each layer according to the corresponding ratio to construct a diversified training subset ; Decision tree training phase: For each constructed training subset , the training work of the decision tree is carried out separately; during the node splitting process of the decision tree, based on the analytic hierarchy process combined with the experience of accounting experts, weights are automatically assigned to the decision tree node splitting attributes, and according to the calculated weight assignment method, each training subset is recursively partitioned to construct multiple decision trees; Random forest integration stage: combine the above-mentioned trained multiple decision trees into a random forest classification model, i.e., a classification model; Data classification execution phase: Input the collected data cleaned in step 1 into the constructed classification model. Each decision tree in the classification model independently classifies the input data. Each decision tree analyzes the various features of the input data according to its own training rules and attribute weights, and then gives a classification result. After all decision trees complete the classification judgment, the final classification result of the input data is determined through a voting mechanism. Standardized operation stage: After completing the data classification, the classified data is standardized in strict accordance with accounting standards and the financial system formulated within the enterprise.

3. The automatic process method for accounting data according to claim 2, characterized in that: In the process of node splitting of decision tree, the method based on analytic hierarchy process combined with accounting expert experience automatically assigns weights to the node splitting attributes of decision tree. The specific implementation logic is as follows: Step A, Determine the attribute set and decision objective: First, train the relevant attributes of the subset Specifically, in the expense reimbursement data, the attributes include "reimbursement item", "reimbursement amount", "reimbursement time", and "reimburser's department"; for procurement data, the attributes are "procurement item category", "procurement quantity", "supplier", and "procurement date"; for sales data, "sales item", "sales amount", "sales time", and "seller's department"; classify the above accurately classified accounting attribute data into the corresponding accounting subjects as the decision objective. Step B: Construct a judgment matrix: Combine the experience of accounting experts and the sample attributes of accounting subjects for different business types to construct a judgment matrix. Set the training subset of The attribute set is , for each attribute data in the accounting subject sample, the judgment matrix constructed is: ,in Representation attributes Relative to the attribute The importance of , When constructing the judgment matrix, the 1-9 scale method is used to compare the relative importance of each attribute in each accounting subject sample. Specifically, if the attribute Relative to the attribute Equally important, ;like Compare Slightly more important, ; obviously important, ; Strongly important, ; Extremely important, If it is between the two, The values ​​are 2, 4, 6, and 8 respectively; Step C, calculate the attribute weight vector: Use the eigenvector method to calculate the maximum eigenvalue of the judgment matrix and its corresponding eigenvector , by solving the equation , obtain the eigenvector ; Step D, Consistency Check: Calculate the consistency index , and the formula is , where is the number of attributes; then look up the average random consistency index , which is obtained from the standard table according to the number of attributes ; calculate the consistency ratio , and the formula is . When , accept the consistency of the judgment matrix, that is, the judgment of the expert is reasonable and the obtained weight vector is valid; if , then readjust the judgment matrix until the consistency check passes; Step E, integrating the expert weight results: For the training subset For each attribute, the weights given by the experts are integrated. Using the simple average method, assume the number of experts is , the -th expert gives the weight of attribute as . Then the final weight of the integrated attribute is calculated as follows: ; ; Step F: Apply weights to decision tree node splitting: Apply the final attribute weights obtained to the decision tree node splitting process.

4. A method for automatically processing accounting data in a workflow, as claimed in claim 3, wherein: Among them, the eigenvalue method is used to calculate the maximum eigenvalue of the judgment matrix and its corresponding eigenvector The specific steps are as follows: Normalize the judgment matrix column by column: Assume the constructed judgment matrix is , for each column of the matrix , calculate the sum of the elements in this column , and then divide each element in the matrix by the sum of the elements in this column to obtain the normalized matrix element . Through the above calculations, complete the column-by-column normalization of the entire judgment matrix to obtain the normalized matrix ; Sum the normalized matrix row by row: For the normalized matrix , calculate the sum of the elements in each row ; Normalize the row sum vector to obtain the eigenvector : Normalize the row sum vector by calculating to obtain the eigenvector . Each component in the eigenvector corresponds to the relative weight of each attribute in the accounting subject sample; ​ Calculate the maximum eigenvalue : Calculate to obtain the vector , and then calculate , where is the -th element of the vector . The maximum eigenvalue is used for subsequent consistency tests to judge the reasonableness of expert judgments.

5. The automatic process method for accounting data according to claim 1, characterized in that: The specific contents of the automated process processing in step 3 are as follows: Preset process rules: For the reimbursement and sales processes, the enterprise presets the approvers and approval sequences corresponding to different amount ranges, different reimbursement and sales items; for the procurement payment process, the enterprise determines the approval levels divided according to factors such as the procurement order amount and supplier type; write the above preset rules into code logic as the basis for subsequent automated process execution; Reimbursement and sales process handling: According to the classification results of the collected data in step 2 of the invention, automatically screen out expense reimbursement data and sales data; according to the preset rules, determine the approver corresponding to each reimbursement and sales order, and send the reimbursement form to the approver's work interface through the internal message push mechanism or integration with the enterprise office software, and attach detailed reimbursement and sales information, including but not limited to the names, departments, reimbursement items, amounts and bills of the reimbursers and salespersons; after receiving the reimbursement and sales form, the approver performs the approval operation and selects approval, rejection or return for modification; if the approver approves, automatically transfer the data to the financial accounting link; if rejected or returned for modification, notify the reimburser or salesperson in time and explain the reason; in the financial accounting link, according to the reimbursement and sales form data, automatically generate accounting vouchers in accordance with accounting standards and enterprise financial systems to complete the accounting process; Procurement payment process handling: According to the classification results of the collected data in step 2 of the invention, automatically screen out procurement data, and real-time monitor the procurement order data and the associated invoice data in the procurement data; When the procurement data information matches and meets the preset payment conditions, automatically generate a payment application, including but not limited to the procurement order number, supplier information, payment amount and payment method; after the payment application is generated, according to the preset approval level rules, push the application to the corresponding approvers in sequence, and the approvers perform the approval operation. After the approval is passed, if all levels of approvals are completed, automatically send an instruction to the enterprise's fund payment system to trigger the payment operation and complete the payment; Data flow and processing progress monitoring: Real-time monitor the input data flow path and processing progress in the reimbursement and sales process handling and procurement payment process handling; that is, by setting marks and timestamps at each process node, record the time when the data enters and leaves each link, as well as the current location; Exception handling and reminder: Continuously check whether there are abnormal situations in the data flow and processing process; if the approval times out, immediately start the reminder mechanism and send reminder emails or text messages to the approver, reimburser or relevant business person in charge to urge them to complete the approval operation; if it is found that the data does not meet the preset rules, send a notice to inform the details of the abnormality and suspend the relevant process until the problem is solved.

6. An automatic process system for accounting data, characterized in that: An accounting data automatic process method according to any one of claims 1-5, comprising a data acquisition module, a data classification and standardization module, an automated process processing module, and a data storage and analysis module; the data acquisition module includes an adaptation interface development component and an OCR scanning and data cleaning component to achieve data connection with the enterprise internal business system and data acquisition and cleaning of paper bills; the data classification and standardization module integrates a machine learning algorithm module and a data standardization processing component to complete accurate data classification and standardization operations; the automated process processing module has a process rule preset component and a monitoring and warning component to automatically process the data flow according to preset rules and give timely warnings in case of abnormalities; The data storage and analysis module includes a data encryption storage component, a report generation component, and a data visualization component to achieve secure data storage, report generation, and visual display.

7. An electronic device, characterized in that, Comprising: A processor and a memory, wherein the memory stores a computer program that can be called by the processor: The processor executes to implement an accounting data automatic process system according to claim 6 by calling the computer program stored in the memory.

8. A computer program product stored on a computer-readable medium, characterized in that: Including a computer-readable program, when executed on an electronic device, provides a user input interface to implement an accounting data automatic process system according to claim 6.

Images