Financial statement automatic proofreading method and device based on OCR (Optical Character Recognition)

Abstract

The invention is suitable for the technical field of Internet, and provides an OCR (Optical Character Recognition)-based financial statement automatic proofreading method and device. In the embodiment, after a financial department obtains an initial financial statement file, text information in the initial financial statement file is automatically recognized and extracted through the OCR technology, then financial subject name matching is conducted on the text information through a preset keyword library, and therefore the subject amount of each financial subject in the text information is obtained; and after confidence evaluation is carried out on the financial statement file, a subject amount lower than a preset threshold value is determined to be checked and modified, and finally a checked target financial statement file is obtained. According to the method and the device, the financial department can realize automatic extraction and matching of financial report information through the OCR and the corresponding algorithm rule, and performs final verification through the financial rule, so that the extraction efficiency and the analysis efficiency of financial data are greatly improved, and the manpower expenditure is reduced.

Description

Technical Field

[0001] The present application relates to the field of Internet technology, and in particular to a method and device for automatically proofreading financial statements based on OCR recognition. Background Art

[0002] Due to the high frequency of futures trading, the margin balance of bank accounts is constantly changing. Financial personnel need to make fund transfer operations based on the real-time balance and pre-set thresholds to ensure user withdrawals and improve fund utilization. Some data need to be manually queried from the financial trading platform in real time and added to the bank balance for statistics.

[0003] With the rapid development of information technology, the amount of financial data generated by enterprises has increased dramatically. Due to business needs, it is necessary to collect the three major financial statements of related companies or units for analysis, so as to obtain their operating conditions as the basis for subsequent decision-making. However, since the three major financial statements of related companies or units are not standardized, and the document styles and clarity provided are different, they need to be manually sorted and summarized into designated templates. The three major financial statements exist in pictures or PDFs. Business development needs to collect data for many consecutive years. Manual operations need to check and enter them into the corresponding templates one by one, which consumes a lot of manpower and is prone to errors. Summary of the invention

[0004] In view of this, the present application provides a method and device for automatic proofreading of financial statements based on OCR recognition, so as to realize automatic extraction and matching of financial report information, improve the extraction efficiency and analysis efficiency of financial data, and reduce manpower expenditure.

[0005] The first aspect of the present application provides a method for automatic proofreading of financial statements based on OCR recognition, the method comprising: Obtaining an initial financial statement file, and using OCR technology to automatically identify and extract financial text information from the initial financial statement file; Matching the financial text information with the financial subject names through a preset keyword library to obtain financial information of each financial subject, wherein the financial information at least includes the subject name and the subject amount; Conduct confidence assessment on the financial information of each financial account, determine the target account amount whose confidence is lower than a preset threshold, and mark the target account amount; The target account amount is checked and modified, and the initial financial statement file is updated with the modified target account amount to obtain the initial financial statement file checked and modified by the user, and the debit and credit balance check is performed on the checked and modified initial financial statement file, and after the check passes, the target financial statement file is generated.

[0006] Optionally, the confidence assessment of the financial information of each financial item includes: The confidence of the financial information of each financial subject is determined by comparing the matching degree of the financial text information recognized by OCR with the standard financial subject name library, and the confidence of the financial information of each financial subject is evaluated according to the confidence.

[0007] Optionally, after generating the target financial statement file, the method further includes: An account amount verification record is determined according to the target account amount and the modified target account amount, and a verification report corresponding to the target financial statement file is generated according to the verification result of the debit and credit balance verification and the account amount verification record.

[0008] Optionally, the automatically identifying and extracting the financial text information in the initial financial statement file using OCR technology includes: Determine a target OCR tool according to the initial financial statement file, obtain a financial statement image in the initial financial statement file, and pre-process the financial statement image; The target OCR tool recognizes the text in the financial report image, and after correcting the text, it is determined as the financial text information.

[0009] Optionally, after generating the target financial statement file, the method further includes: Determine the target financial system, connect with the target financial system through the API interface, and enter the target financial report file into the target financial system.

[0010] A second aspect of the present application provides an automatic proofreading device for financial statements based on OCR recognition, the device comprising: A text information extraction unit, used to obtain an initial financial statement file, and automatically identify and extract financial text information in the initial financial statement file using OCR technology; A financial information obtaining unit, used to match the financial text information with the financial subject name through a preset keyword library to obtain the financial information of each financial subject, wherein the financial information at least includes the subject name and the subject amount; A target account amount determination unit, used to perform confidence assessment on the financial information of each financial account, determine the target account amount whose confidence is lower than a preset threshold, and mark the target account amount; The verification unit is used to verify and modify the target account amount, and update the initial financial statement file with the modified target account amount to obtain the initial financial statement file verified and modified by the user, perform debit and credit balance verification on the verified and modified initial financial statement file, and generate the target financial statement file after the verification passes.

[0011] Optionally, the confidence evaluation of the financial information of each financial account in the target account amount determination unit includes: The confidence of the financial information of each financial subject is determined by comparing the matching degree of the financial text information recognized by OCR with the standard financial subject name library, and then the confidence of the financial information of each financial subject is evaluated based on the confidence.

[0012] Optionally, after the verification unit generates the target financial statement file, the device further includes: The proofreading report generating unit is used to determine the account amount proofreading record according to the target account amount and the modified target account amount, and to generate the proofreading report corresponding to the target financial statement file according to the verification result of the debit and credit balance verification and the account amount proofreading record.

[0013] Optionally, the text information extraction unit automatically identifies and extracts the financial text information in the initial financial statement file using OCR technology, including: Determine a target OCR tool according to the initial financial statement file, obtain a financial statement image in the initial financial statement file, and pre-process the financial statement image; The target OCR tool recognizes the text in the financial report image, and after correcting the text, it is determined as the financial text information.

[0014] Optionally, after the verification unit generates the target financial statement file, the device further includes: The system input unit is used to determine the target financial system, connect with the target financial system through an API interface, and input the target financial report file into the target financial system.

[0015] In the embodiment provided by the present application, after the financial department obtains the initial financial report file, it automatically identifies and extracts the text information in the initial financial report file through OCR technology, and then matches the text information with the financial account name through the preset keyword library, thereby obtaining the account amount of each financial account in the text information; after performing a confidence assessment, it determines the account amount below the preset threshold for verification and modification, and finally obtains the proofread target financial report file. Through this application, the financial department can realize the automatic extraction and matching of financial report information through OCR and corresponding algorithm rules, and perform the final verification through financial rules, which greatly improves the efficiency of financial data extraction and analysis, and reduces manpower expenditure. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 A flow chart of the method provided in the embodiment of the present application; Figure 2 A structural diagram of a device provided in an embodiment of the present application; Figure 3 A schematic diagram of the internal structure of a computer device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0017] Exemplary embodiments will be described in detail herein, examples of which are shown in the accompanying drawings. When the following description refers to the drawings, the same numbers in different drawings represent the same or similar elements unless otherwise indicated. The implementations described in the following exemplary embodiments do not represent all implementations consistent with the present application. Instead, they are merely examples of devices and methods consistent with some aspects of the present application as detailed in the appended claims.

[0018] The terms used in this application are for the purpose of describing specific embodiments only and are not intended to limit this application. The singular forms of "a", "said" and "the" used in this application and the appended claims are also intended to include plural forms unless the context clearly indicates other meanings. It should also be understood that the term "and / or" used herein refers to and includes any or all possible combinations of one or more associated listed items.

[0019] It should be understood that although the terms first, second, third, etc. may be used in the present application to describe various information, these information should not be limited to these terms. These terms are only used to distinguish the same type of information from each other. For example, without departing from the scope of the present application, the first information may also be referred to as the second information, and similarly, the second information may also be referred to as the first information. Depending on the context, the word "if" as used herein may be interpreted as "at the time of" or "when" or "in response to determining".

[0020] The present application provides a method and device for automatic proofreading of financial statements based on OCR recognition, so as to improve the efficiency of extracting and analyzing financial data and reduce manpower expenditure.

[0021] The technical solution of the present application is described in detail with specific embodiments below. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described in detail in some embodiments.

[0022] like Figure 1 As shown, it is a flow chart of a margin balance reminder method based on automated simulated keyboard and mouse operation provided by the present application, and the process includes: Step S101, obtaining an initial financial report file, and using OCR technology to automatically identify and extract financial text information in the initial financial report file.

[0023] In this embodiment, the initial financial statement file can be a scanned copy in paper form, or a document in electronic format such as PDF, image (such as JPG, PNG), etc. If necessary, the collected financial statement file needs to be converted into a format supported by OCR software to obtain the initial financial statement file.

[0024] Then, depending on the preset OCR tool or service, which can be commercial software such as Adobe Acrobat, open source tools such as Tesseract, or cloud services such as Google Cloud Vision API, Amazon Textract, Baidu AI OCR, etc., the initial financial report file is loaded into the OCR software, and the OCR tool will automatically process the image file, recognize and extract text information. And output the recognized text into an editable format such as TXT, CSV, Word document, etc., so as to obtain the above financial text information.

[0025] In another embodiment, the method of automatically identifying and extracting financial text information from the initial financial statement file using OCR technology includes: Determine a target OCR tool according to the initial financial statement file, obtain a financial statement image in the initial financial statement file, and pre-process the financial statement image; The target OCR tool recognizes the text in the financial report image, and after correcting the text, it is determined as the financial text information.

[0026] In this embodiment, when selecting an OCR tool to automatically identify and extract financial text information from an initial financial report file, multiple factors need to be considered, including recognition accuracy, format support, compatibility, ease of use, and cost. For example, when the format of the initial financial report file is PDF, an OCR tool that supports PDF is selected. Or when the initial financial report file contains multiple languages, an OCR tool that supports multiple languages, such as ABBYY FineReader, should be selected. In order to improve the recognition accuracy, the scanned or photographed financial report image needs to be preprocessed, including denoising, image enhancement, edge detection, rotation correction, and other operations.

[0027] In this embodiment, the method for correcting the text is as follows: 1. Use the built-in proofreading function of the OCR software to identify common errors, such as misrecognition of characters, layout errors, omissions and repetitions, and confusion of language and symbols. Then use natural language processing (NLP) technology to analyze the text context to identify and correct recognition results that do not match the context.

[0028] 2. Improve the accuracy of character recognition through machine learning model optimization, such as deep learning models, and then use rule engines and dictionary matching to correct errors based on language rules and common vocabulary libraries.

[0029] 3. Develop a user-friendly interactive interface that allows users to manually adjust the recognition results. This semi-automatic method can significantly improve the recognition accuracy of specific documents.

[0030] Step S102, matching the financial text information with the financial account names through a preset keyword library to obtain the financial information of each financial account, wherein the financial information at least includes the account name and the account amount.

[0031] In this embodiment, a database containing account codes, account names, account attributes and other information can be established based on the company's own financial accounts and report formats. This can be done by collecting historical financial reports and vouchers, extracting the account names and codes therein, and then collating them into a standardized database. In addition, national or international accounting standards can also be referenced, which usually provide a set of standard accounting account lists that can serve as the basis for the keyword library.

[0032] Step S103, performing confidence evaluation on the financial information of each financial account, determining the target account amount whose confidence is lower than a preset threshold, and marking the target account amount.

[0033] In this embodiment, the confidence of the financial information of each financial account is determined by comparing the matching degree of the financial text information recognized by OCR with the standard financial account name library. For example, the nature and frequency of OCR recognition errors are first analyzed, including common error types such as character misrecognition, missing characters, extra characters, etc. Then the accuracy of the OCR recognition results is modeled. The model learns the relationship between the recognition results and the actual values ​​based on historical data. The confidence is calculated by comparing the difference between the model prediction value and the actual value. The calculated confidence is compared with the preset threshold to determine whether the confidence of each account amount is lower than the threshold. When the target account amount is determined to be lower than the threshold, it is marked in red.

[0034] Step S104, verify and modify the target account amount, and update the initial financial statement file with the modified target account amount to obtain the initial financial statement file verified and modified by the user, perform a debit and credit balance check on the verified and modified initial financial statement file, and generate the target financial statement file after the check passes.

[0035] In this embodiment, performing a debit / credit balance check is an important step in accounting to ensure the accuracy and completeness of accounting information. The following are the steps to implement a debit / credit balance check: Trial balance: Trial balance is a method of checking the correctness of account records by preparing a trial balance. It includes two methods: trial balance of occurrence amount and trial balance of balance: 1) Trial balance of transactions: Compare the total debit transactions of all accounts in this period with the total credit transactions of this period. The two should be equal. This is based on the accounting rule of "every debit must have a credit, and the debits and credits must be equal".

[0036] 2) Trial balance: Check whether the total debit balance of all accounts is equal to the total credit balance. This reflects the constant relationship between assets and equity (liabilities and owner's equity).

[0037] The principle of balance between credits and debits: The principle of balance between credits and debits is the basic principle of accounting, which requires that every accounting transaction must involve at least two accounting accounts at the same time, and the debit amount is equal to the credit amount, that is, assets = liabilities + owner's equity.

[0038] Prepare adjusting entries: If you find that the debits and credits are unbalanced, you need to prepare adjusting entries based on the specific circumstances until the debits and credits are balanced.

[0039] Through the above steps, the debits and credits in the financial statements can be balanced, providing a basis for preparing accurate financial statements.

[0040] In another embodiment, after generating the target financial statement file, the method further includes: An account amount verification record is determined according to the target account amount and the modified target account amount, and a verification report corresponding to the target financial statement file is generated according to the verification result of the debit and credit balance verification and the account amount verification record.

[0041] In this embodiment, the proofreading report provides a clear record showing all modifications and proofreading activities, increasing the transparency of financial reports. The proofreading report also facilitates audits, allowing auditors to quickly check and confirm the accuracy and completeness of financial data. If problems are found in the future, the proofreading report can help quickly locate the link where the error occurred, making it easier to find and correct it.

[0042] In another embodiment, after generating the target financial statement file, the method further includes: Determine the target financial system, connect with the target financial system through the API interface, and enter the target financial report file into the target financial system.

[0043] In this embodiment, the data in the financial report file is automatically imported into the target financial system through the API interface, which can reduce the workload of manual input and improve the speed and efficiency of data processing. In addition, data transmission through the API usually includes encryption and security measures to protect the security and privacy of financial data.

[0044] So far, completed Figure 1 The process shown.

[0045] In the embodiment of the present application, after the financial department obtains the initial financial report file, it automatically identifies and extracts the text information in the initial financial report file through OCR technology, and then matches the text information with the financial account name through the preset keyword library, thereby obtaining the account amount of each financial account in the text information; after performing a confidence assessment, it determines the account amount below the preset threshold for verification and modification, and finally obtains the proofread target financial report file. Through this application, the financial department can realize the automatic extraction and matching of financial report information through OCR and corresponding algorithm rules, and perform the final verification through financial rules, which greatly improves the extraction efficiency and analysis efficiency of financial data and reduces manpower expenditure.

[0046] The present application also provides an automatic proofreading device for financial statements based on OCR recognition, such as Figure 2 As shown, the device comprises: The text information extraction unit 201 is used to obtain an initial financial report file, and automatically identify and extract financial text information in the initial financial report file using OCR technology; The financial information obtaining unit 202 is used to match the financial text information with the financial subject name through a preset keyword library to obtain the financial information of each financial subject, wherein the financial information at least includes the subject name and the subject amount; A target account amount determination unit 203 is used to perform confidence evaluation on the financial information of each financial account, determine the target account amount whose confidence is lower than a preset threshold, and mark the target account amount; The verification unit 204 is used to verify and modify the target account amount, and update the initial financial statement file with the modified target account amount to obtain the initial financial statement file verified and modified by the user, perform debit and credit balance verification on the verified and modified initial financial statement file, and generate the target financial statement file after the verification passes.

[0047] In another embodiment, the step of performing confidence evaluation on the financial information of each financial account in the target account amount determination unit includes: The confidence of the financial information of each financial subject is determined by comparing the matching degree of the financial text information recognized by OCR with the standard financial subject name library, and then the confidence of the financial information of each financial subject is evaluated based on the confidence.

[0048] In another embodiment, after the target financial statement file is generated in the verification unit, the device further includes: The proofreading report generating unit 205 is used to determine the account amount proofreading record according to the target account amount and the modified target account amount, and to generate the proofreading report corresponding to the target financial statement file according to the verification result of the debit / credit balance verification and the account amount proofreading record.

[0049] In another embodiment, the text information extraction unit automatically identifies and extracts the financial text information in the initial financial statement file using OCR technology, including: Determine a target OCR tool according to the initial financial statement file, obtain a financial statement image in the initial financial statement file, and pre-process the financial statement image; The target OCR tool recognizes the text in the financial report image, and after correcting the text, it is determined as the financial text information.

[0050] In another embodiment, after the target financial statement file is generated in the verification unit, the device further includes: The system input unit 206 is used to determine the target financial system, connect with the target financial system through an API interface, and input the target financial report file into the target financial system.

[0051] The above embodiment of the present invention provides an automatic proofreading method for financial statements based on OCR recognition, and based on this method, provides an automatic proofreading device for financial statements based on OCR recognition. Through the above method and device, the financial department can realize automatic extraction and matching of financial report information through OCR and corresponding algorithm rules, and perform final verification through financial rules, which greatly improves the extraction efficiency and analysis efficiency of financial data and reduces manpower expenditure.

[0052] This embodiment also discloses a computer device, such as Figure 3 As shown, the computer device includes a processor and a memory, wherein the memory stores at least one instruction, and the at least one instruction is loaded and executed by the processor to implement any of the above-mentioned automatic proofreading methods for financial statements based on OCR recognition.

[0053] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application shall be included in the scope of protection of the present application.

Claims

1. A method for automatic proofreading of financial statements based on OCR recognition, characterized in that: The method comprises: Obtaining an initial financial statement file, and using OCR technology to automatically identify and extract financial text information from the initial financial statement file; Matching the financial text information with the financial subject names through a preset keyword library to obtain financial information of each financial subject, wherein the financial information at least includes the subject name and the subject amount; Conduct confidence assessment on the financial information of each financial account, determine the target account amount whose confidence is lower than a preset threshold, and mark the target account amount; The target account amount is checked and modified, and the initial financial statement file is updated with the modified target account amount to obtain the initial financial statement file checked and modified by the user, and the debit and credit balance check is performed on the checked and modified initial financial statement file, and after the check passes, the target financial statement file is generated.

2. The method according to claim 1, characterized in that The confidence evaluation of the financial information of each financial item includes: The confidence of the financial information of each financial subject is determined by comparing the matching degree of the financial text information recognized by OCR with the standard financial subject name library, and then the confidence of the financial information of each financial subject is evaluated based on the confidence.

3. The method according to claim 1, characterized in that After generating the target financial statement file, the method further includes: An account amount verification record is determined according to the target account amount and the modified target account amount, and a verification report corresponding to the target financial statement file is generated according to the verification result of the debit and credit balance verification and the account amount verification record.

4. The method according to claim 1, characterized in that The automatic identification and extraction of financial text information in the initial financial statement file using OCR technology includes: Determine a target OCR tool according to the initial financial statement file, obtain a financial statement image in the initial financial statement file, and pre-process the financial statement image; The target OCR tool recognizes the text in the financial report image, and after correcting the text, it is determined as the financial text information.

5. The method according to claim 1, characterized in that After generating the target financial statement file, the method further includes: Determine the target financial system, connect with the target financial system through the API interface, and enter the target financial report file into the target financial system.

6. An automatic proofreading device for financial statements based on OCR recognition, characterized in that: The device comprises: A text information extraction unit, used to obtain an initial financial statement file, and automatically identify and extract financial text information in the initial financial statement file using OCR technology; A financial information obtaining unit, used to match the financial text information with the financial subject name through a preset keyword library to obtain the financial information of each financial subject, wherein the financial information at least includes the subject name and the subject amount; A target account amount determination unit, used to perform confidence assessment on the financial information of each financial account, determine the target account amount whose confidence is lower than a preset threshold, and mark the target account amount; The verification unit is used to verify and modify the target account amount, and update the initial financial statement file with the modified target account amount to obtain the initial financial statement file verified and modified by the user, perform debit and credit balance verification on the verified and modified initial financial statement file, and generate the target financial statement file after the verification passes.

7. The device according to claim 6, characterized in that The confidence evaluation of the financial information of each financial account in the target account amount determination unit includes: The confidence of the financial information of each financial subject is determined by comparing the matching degree of the financial text information recognized by OCR with the standard financial subject name library, and then the confidence of the financial information of each financial subject is evaluated based on the confidence.

8. The device according to claim 6, characterized in that After the target financial statement file is generated in the verification unit, the device further comprises: The proofreading report generating unit is used to determine the account amount proofreading record according to the target account amount and the modified target account amount, and to generate the proofreading report corresponding to the target financial statement file according to the verification result of the debit and credit balance verification and the account amount proofreading record.

9. The device according to claim 6, characterized in that The text information extraction unit automatically identifies and extracts the financial text information in the initial financial statement file using OCR technology, including: Determine a target OCR tool according to the initial financial statement file, obtain a financial statement image in the initial financial statement file, and pre-process the financial statement image; The target OCR tool recognizes the text in the financial report image, and after correcting the text, it is determined as the financial text information.

10. The device according to claim 6, characterized in that After the target financial statement file is generated in the verification unit, the device further comprises: The system input unit is used to determine the target financial system, connect with the target financial system through an API interface, and input the target financial report file into the target financial system.

Images