Automatic accounting voucher generation method and device based on large finance and tax model

Abstract

The invention discloses an automatic accounting voucher generation method and device based on a large finance and tax model. The method comprises the steps of performing data preprocessing on collected financial data to obtain an effective financial data set; continuously pre-training the finance and taxation large model based on the effective finance data set to obtain a finance large model with finance and taxation knowledge, and pre-marking the effective finance data set based on prompt word engineering and GPT4 to obtain a sample data set with marking information; according to the sample data set, instruction fine tuning is carried out on the large financial model with finance and taxation knowledge, and the large financial model with a key element extraction function and a voucher generation function is obtained; performing key element extraction on the to-be-generated voucher input data by using a key element extraction function to obtain key elements of the to-be-generated voucher input data; and performing voucher generation on the key elements by using a voucher generation model according to a pre-generated voucher generation instruction, and obtaining an entry voucher corresponding to the to-be-generated voucher input data.

Description

Technical Field

[0001] The present invention relates to the field of digital accounting automation technology, and more specifically, to a method and device for automatically generating accounting vouchers based on a financial and taxation big model. Background Art

[0002] With the continuous development of the digital economy, management experience and business methods in all walks of life have undergone tremendous changes. As the central hub of corporate data, it is necessary for the finance department to strengthen its own capacity building, enhance the practical value of financial information for corporate operational management decisions, and realize the continuous empowerment of finance for business development. In the context of today's digital world, the processing of financial and tax accounting information has become increasingly complex, and the demand for technologies that have both financial and tax knowledge and digital intelligence capabilities is growing rapidly.

[0003] The finance and taxation big model improves various core and professional finance and taxation capabilities by continuing pre-training and fine-tuning on the open source base model. Compared with the general big model, it shows unique professional advantages in the field of digital accounting. Especially in key tasks such as accounting subject classification and invoicing information extraction, the finance and taxation big model can more accurately identify and process financial data, ensuring the efficiency and accuracy in compliance with accounting rules and electronic voucher generation, significantly improving the professional performance of digital accounting, and combining RAG (Retrieval-Augmented Generation) technology to ensure the compliance of generated data, providing strong support for enterprises to operate stably in a complex and changing market environment. However, there is currently no large model-based method for digital accounting automation. Summary of the invention

[0004] In view of the deficiencies in the prior art, the present invention provides a method and device for automatically generating accounting vouchers based on a large financial and taxation model.

[0005] According to one aspect of the present invention, a method for automatically generating accounting vouchers based on a financial and taxation model is provided, comprising:

[0006] Preprocess the collected financial data to obtain valid financial data sets, where the financial data includes accounting vouchers, financial statements, and tax data;

[0007] Continue to pre-train the finance and taxation big model based on the effective financial data set to obtain the finance big model with finance and taxation knowledge, and pre-label the effective financial data set based on the prompt engineering and GPT4 to obtain a sample data set with labeled information, where the labeled information includes the key elements of the feature extraction task and the voucher elements of the voucher generation task;

[0008] Based on the sample data set with annotated information, the financial model with financial and tax knowledge is fine-tuned to obtain a financial model with key element extraction and voucher generation functions;

[0009] Utilize the key element extraction function to extract key elements from the input data of the voucher to be generated according to the pre-generated key element extraction instructions, and obtain the key elements of the input data of the voucher to be generated;

[0010] The voucher generation model is used to generate vouchers for key elements according to the pre-generated voucher generation instructions, and the entry vouchers corresponding to the input data of the vouchers to be generated are obtained.

[0011] Optionally, the collected financial data is preprocessed to obtain a valid financial data set, including:

[0012] Desensitize the financial data according to preset rules to obtain desensitized financial data;

[0013] Regular rules are used to clean the desensitized financial data to obtain cleaned financial data, where the cleaning operations include noise removal, deduplication, text cleaning, and segmentation;

[0014] The cleaned financial data is processed through the statistical method of Z-score, constant filling and regular matching to obtain high-quality financial data;

[0015] Perform data enhancement on high-quality financial data to obtain effective financial data sets.

[0016] Optionally, the method further includes: optimizing the key factor extraction function according to the real-time financial data collected in real time, and the implementation process is:

[0017] Use the key element extraction function to extract key elements from real-time financial data and obtain real-time key element information of real-time financial data;

[0018] When the real-time key element information does not conform to the preset element format, the hyperparameters of the key element extraction function are adjusted until the output real-time key element information conforms to the preset element format, thereby obtaining an optimized key element extraction function.

[0019] Optionally, the method further includes: optimizing the voucher generation model according to the real-time financial data collected in real time, and the implementation process is:

[0020] Use the updated key element extraction function to extract key elements from real-time financial data and obtain real-time key element information of real-time financial data;

[0021] Use the voucher generation model to generate vouchers for real-time key element information and obtain real-time accounting vouchers for real-time financial data;

[0022] When the real-time accounting voucher does not conform to the preset accounting voucher format, the voucher generation model is hyper-parameter adjusted until the output real-time accounting voucher conforms to the preset accounting voucher format, thereby obtaining an optimized voucher generation model.

[0023] According to another aspect of the present invention, there is provided a device for automatically generating accounting vouchers based on a financial and taxation model, comprising:

[0024] The preprocessing module is used to preprocess the collected financial data to obtain valid financial data sets, where the financial data includes accounting vouchers, financial statements, and tax data;

[0025] The training module is used to continue pre-training the finance and taxation big model based on the effective financial data set, obtain the financial big model with finance and taxation knowledge, and pre-label the effective financial data set based on the prompt engineering and GPT4 to obtain a sample data set with labeled information, where the labeled information includes the key elements of the feature extraction task and the voucher elements of the voucher generation task;

[0026] The instruction fine-tuning module is used to fine-tune the instructions of the financial model with financial and tax knowledge based on the sample data set with annotated information, so as to obtain a financial model with key element extraction function and voucher generation function;

[0027] An extraction module is used to extract key elements from the input data of the voucher to be generated by using a key element extraction function according to a pre-generated key element extraction instruction, and obtain the key elements of the input data of the voucher to be generated;

[0028] The generation module is used to generate vouchers for key elements using the voucher generation model according to the pre-generated voucher generation instructions, and obtain the entry voucher corresponding to the input data of the voucher to be generated.

[0029] Optionally, the preprocessing module includes:

[0030] The desensitization submodule is used to perform desensitization processing on the financial data according to preset rules to obtain desensitized financial data;

[0031] The cleaning submodule is used to clean the desensitized financial data using regular rules to obtain cleaned financial data, where the cleaning operations include noise removal, deduplication, text cleaning, and slicing;

[0032] The processing submodule is used to process the cleaned financial data through the statistical method of Z-score, constant filling and regular matching to obtain high-quality financial data;

[0033] Perform data enhancement on high-quality financial data to obtain effective financial data sets.

[0034] Optionally, the device further includes: a first optimization module, which is used to optimize the key factor extraction function according to the real-time financial data collected in real time, and the implementation process is:

[0035] Use the key element extraction function to extract key elements from real-time financial data and obtain real-time key element information of real-time financial data;

[0036] When the real-time key element information does not conform to the preset element format, the hyperparameters of the key element extraction function are adjusted until the output real-time key element information conforms to the preset element format, thereby obtaining an optimized key element extraction function.

[0037] Optionally, the device further includes: a second optimization module, which is used to optimize the voucher generation model according to the real-time financial data collected in real time, and the implementation process is:

[0038] Use the updated key element extraction function to extract key elements from real-time financial data and obtain real-time key element information of real-time financial data;

[0039] Use the voucher generation model to generate vouchers for real-time key element information and obtain real-time accounting vouchers for real-time financial data;

[0040] When the real-time accounting voucher does not conform to the preset accounting voucher format, the voucher generation model is hyper-parameter adjusted until the output real-time accounting voucher conforms to the preset accounting voucher format, thereby obtaining an optimized voucher generation model.

[0041] According to another aspect of the present invention, a computer-readable storage medium is provided, wherein the storage medium stores a computer program, and the computer program is used to execute the method described in any one of the above aspects of the present invention.

[0042] According to another aspect of the present invention, an electronic device is provided, comprising: a processor; a memory for storing instructions executable by the processor; the processor is configured to read the executable instructions from the memory and execute the instructions to implement the method described in any one of the above aspects of the present invention.

[0043] The present invention proposes a digital accounting automation method based on a large financial and taxation model, and the beneficial effects are as follows:

[0044] 1. The accuracy of factor extraction of the financial and taxation big model has increased by 40%, and the score of voucher generation has increased by 4%. This method can significantly improve the efficiency and accuracy of accounting work and provide strong support for the digital transformation in the financial and taxation field.

[0045] 2. Reduce the labor costs required for generating traditional accounting documents and eliminate the difficulties and errors faced by traditional methods when dealing with complex business scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] A more complete understanding of exemplary embodiments of the present invention may be obtained by referring to the following drawings:

[0047] Figure 1 It is a flowchart of a method for automatically generating accounting vouchers based on a financial and taxation model provided by an exemplary embodiment of the present invention;

[0048] Figure 2 It is another flow chart of a method for automatically generating accounting vouchers based on a financial and taxation model provided by an exemplary embodiment of the present invention;

[0049] Figure 3 It is a structural schematic diagram of an automatic generation device for accounting vouchers based on a large financial and taxation model provided by an exemplary embodiment of the present invention;

[0050] Figure 4 This is a structure of an electronic device provided by an exemplary embodiment of the present invention. DETAILED DESCRIPTION

[0051] Below, the exemplary embodiments according to the present invention will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments of the present invention, and it should be understood that the present invention is not limited to the exemplary embodiments described here.

[0052] It should be noted that the relative arrangement of components and steps, the numerical expressions and numerical values ​​set forth in these embodiments do not limit the scope of the present invention unless specifically stated otherwise.

[0053] Those skilled in the art can understand that the terms "first" and "second" in the embodiments of the present invention are only used to distinguish different steps, devices or modules, etc., and neither represent any specific technical meaning nor indicate the necessary logical order between them.

[0054] It should also be understood that, in the embodiments of the present invention, “plurality” may refer to two or more than two, and “at least one” may refer to one, two or more than two.

[0055] It should also be understood that any component, data or structure mentioned in the embodiments of the present invention can generally be understood as one or more, unless explicitly limited or otherwise indicated in the context.

[0056] In addition, the term "and / or" in the present invention is only a description of the association relationship of the associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in the present invention generally indicates that the associated objects before and after are in an "or" relationship.

[0057] It should also be understood that the description of the various embodiments of the present invention focuses on the differences between the various embodiments, and the same or similar aspects thereof can be referenced to each other, and for the sake of brevity, they will not be described one by one.

[0058] At the same time, it should be understood that for the convenience of description, the sizes of the various parts shown in the drawings are not drawn according to the actual proportional relationship.

[0059] The following description of at least one exemplary embodiment is merely illustrative in nature and is in no way intended to limit the invention, its application, or uses.

[0060] Technologies, methods, and equipment known to ordinary technicians in the relevant art may not be discussed in detail, but where appropriate, the technologies, methods, and equipment should be considered part of the specification.

[0061] It should be noted that like reference numerals and letters refer to similar items in the following figures, and therefore, once an item is defined in one figure, it need not be further discussed in subsequent figures.

[0062] Embodiments of the present invention can be applied to electronic devices such as terminal devices, computer systems, servers, etc., which can operate with many other general or special computing system environments or configurations. Examples of well-known terminal devices, computing systems, environments and / or configurations suitable for use with electronic devices such as terminal devices, computer systems, servers, etc. include, but are not limited to: personal computer systems, server computer systems, thin clients, thick clients, handheld or laptop devices, microprocessor-based systems, set-top boxes, programmable consumer electronics, network personal computers, small computer systems, large computer systems, and distributed cloud computing technology environments including any of the above systems, etc.

[0063] Electronic devices such as terminal devices, computer systems, servers, etc. can be described in the general context of computer system executable instructions (such as program modules) executed by computer systems. Generally, program modules can include routines, programs, object programs, components, logic, data structures, etc., which perform specific tasks or implement specific abstract data types. Computer systems / servers can be implemented in a distributed cloud computing environment, where tasks are performed by remote processing devices linked through a communication network. In a distributed cloud computing environment, program modules can be located on local or remote computing system storage media including storage devices.

[0064] Exemplary Methods

[0065] Figure 1 1 is a flow chart of a method for automatically generating an accounting voucher based on a tax model according to an exemplary embodiment of the present invention. This embodiment can be applied to electronic devices, such as Figure 1 As shown, the method 100 for automatically generating accounting vouchers based on the financial and tax model includes the following steps:

[0066] Step 101, preprocessing the collected financial data to obtain a valid financial data set, wherein the financial data includes accounting vouchers, financial statements, and tax data;

[0067] Step 102: Continue pre-training the finance and taxation big model based on the valid financial data set to obtain a financial big model with finance and taxation knowledge, and pre-label the valid financial data set based on the prompt engineering and GPT4 to obtain a sample data set with labeled information, where the labeled information includes the key elements of the element extraction task and the voucher elements of the voucher generation task;

[0068] Step 103, fine-tuning the instructions of the financial model with financial and tax knowledge according to the sample data set with annotated information, so as to obtain a financial model with key element extraction function and voucher generation function;

[0069] Step 104, using the key element extraction function to extract key elements from the input data of the voucher to be generated according to the pre-generated key element extraction instructions, to obtain the key elements of the input data of the voucher to be generated;

[0070] Step 105, using the voucher generation model to generate vouchers for key elements according to the pre-generated voucher generation instructions, and obtaining the entry voucher corresponding to the input data of the voucher to be generated.

[0071] Specifically, the present invention proposes a digital accounting automation method based on a large financial and tax model, which can realize the extraction of key information of bill elements and the automatic generation of accounting vouchers, thereby reducing the inefficiency and labor costs of manual operations. Figure 2As shown in the figure, the overall process of digital accounting based on the financial and taxation big model can be divided into three stages: professional ability training of the financial and taxation big model, accurate extraction of key invoicing elements, compliance verification, invoicing, and automatic generation of accounting vouchers.

[0072] The process of professional capacity training for the financial and taxation model is as follows:

[0073] Data is a key element of the big model. To ensure the effectiveness of the big model, high-quality and wide-ranging data is provided. The data set is divided into continued pre-training and instruction fine-tuning data sets. The data set construction process is as follows:

[0074] 1. Data collection: The collected financial data, including accounting vouchers, financial statements, tax data, etc., covers financial and tax scenarios and operations, and combines experts in the field of finance, taxation and accounting to ensure the diversity and accuracy of the data.

[0075] 2. Data preprocessing:

[0076] 1) Desensitize the original data according to preset rules to ensure privacy and data compliance.

[0077] 2) Use regular rules to clean the data, including noise removal, deduplication, text cleaning, and segmentation.

[0078] 3) Through the statistical method of Z-score, constant filling, regular matching, etc., the problems of data anomaly, missing data, and redundancy are reduced, and the quality of the data set is further improved.

[0079] 4) Use synonym replacement, sentence structure adjustment, word insertion or deletion and other methods to enhance data.

[0080] 3. Data labeling: Pre-label the financial and tax data based on advanced large models such as prompt engineering and GPT4, and then review and revise them by business experts to provide supervised learning training samples for the model. For the factor extraction task, the key elements involved are labeled, such as the transaction object, amount, and subject. For the voucher generation task, the voucher elements involved are labeled, such as subject, amount, and entry direction. By labeling the data, the model can learn the structure and characteristics of the accounting voucher, as well as the corresponding labeling standards.

[0081] In the context of the digital transformation of the finance and taxation industry, this article selects 14B's Tongyi Qianwen model as the base model, which has excellent Chinese and English understanding and generation capabilities, providing a solid foundation for applications in the finance and taxation field. The Tongyi Qianwen model has rich general semantic knowledge and various general technical capabilities, such as code writing, summary summarization, multiple rounds of clarification, etc., but it does not yet have the ability to solve specific tasks in the finance and taxation field. Therefore, using a finance and taxation data set consisting of finance and taxation books, accounting standards, finance and taxation knowledge data, accounting vouchers, financial statements, etc. for continued pre-training, the model can learn the language patterns and knowledge of the finance and taxation professional background, which helps the model better understand and process related vocabulary and concepts. Through fine-tuning, the model's ability to understand and extract key information related to finance and taxation and generate accounting entry vouchers has been improved.

[0082] The purpose of fine-tuning the instructions is to enhance the capabilities of the large model and further integrate professional knowledge in the field of finance and taxation into the model, so that it has stronger capabilities in factor extraction, compliance verification and automatic generation of accounting documents. The specific steps include:

[0083] 1. Build a formatting example. Design the instruction format and build an instruction template.

[0084] 2. In terms of hyperparameter setting, the full parameter fine-tuning method helps the model to adjust the model parameters more finely to adapt to new tasks while maintaining the previously learned knowledge; the half-precision floating point format is used to maintain the calculation accuracy while significantly improving the calculation speed and reducing the use of video memory, thereby improving training efficiency; to avoid overfitting while maintaining the generalization ability of the model, this article sets the training cycle to one epoch; a lower learning rate (1e-5).

[0085] 3. Since the SFT data generated for bill element extraction and voucher entry is a question-and-answer instruction data, in order to avoid damaging the instruction effect of the base model, this paper adopts a general data ratio (1:5).

[0086] 4. For the task of extracting bill elements, the effect of the large model is evaluated by using business experts to score. The evaluation criteria are shown in Table 1 below.

[0087] Table 1. Scoring criteria for factor extraction

[0088]

[0089] For voucher generation tasks, business experts are asked to score based on the quality of task completion, with the score range from 1 to 5 points. The scoring criteria are detailed in Table 2 below.

[0090] Table 2 Credential generation scoring criteria

[0091]

[0092] The process of extracting key accounting elements is as follows:

[0093] By designing targeted prompt words, the tax model is guided to focus on key information in the document, such as basic customer information, invoicing information, contract settlement, amount, etc. Combined with RAG technology, key information is audited to avoid risks. The following are examples of relevant prompt word design and output.

[0094] 1. Explicitly construct the prompt to extract the invoice details.

[0095] 2. Identify key information such as invoice type, product amount, and buyer’s name.

[0096] 3. Design a JSON structure containing key values ​​such as invoice type, invoice amount, and buyer name.

[0097] 4. Directly use background information and user input to fill in the values ​​of each JSON field.

[0098] 5. Review output to ensure data accuracy and completeness.

[0099]

[0100] Automatic generation of accounting vouchers

[0101] When generating voucher entries, it is necessary to combine the above-mentioned extracted and approved invoicing elements so that the big model can analyze key information, such as commodity classification abbreviations, commodity details, buyer and seller information, etc. At the same time, considering the specific industry background of the enterprise, the financial and tax big model can be used to generate voucher entries that meet the accounting scenarios and accounting rules. The relevant design of prompt words and voucher generation examples are as follows.

[0102] 1. Clearly construct the prompt to analyze the key information in the input, such as invoice type, purchaser name, invoice amount, etc.

[0103] 2. Determine business attributes, select accounting subjects, determine debit amounts, credit amounts and calculate taxes.

[0104] 3. Design a JSON structure containing business attributes and accounting subjects.

[0105] 4. Use user input directly to fill in the values ​​of each JSON field.

[0106] 5. Review output to ensure data accuracy and completeness.

[0107]

[0108] The key technical points of the present invention are:

[0109] 1. The present invention uses advanced large-scale language models, such as GPT-4, combined with prompt engineering technology to automatically pre-label financial and tax data.

[0110] 2. The financial and taxation big model trained based on the financial and taxation data set can accurately identify and extract key information, generate entries that comply with accounting standards, and realize the automated processing of digital accounting.

[0111] The present invention proposes a digital accounting automation method based on a large financial and taxation model, and the beneficial effects are as follows:

[0112] 1. The accuracy of factor extraction of the financial and taxation big model has increased by 40%, and the score of voucher generation has increased by 4%. This method can significantly improve the efficiency and accuracy of accounting work and provide strong support for the digital transformation in the financial and taxation field.

[0113] 2. Reduce the labor costs required for generating traditional accounting documents and eliminate the difficulties and errors faced by traditional methods when dealing with complex business scenarios.

[0114] Exemplary Devices

[0115] Figure 3 1 is a schematic diagram of a structure of an automatic generating device for accounting vouchers based on a large financial and taxation model provided by an exemplary embodiment of the present invention. Figure 3 As shown, the device 300 includes:

[0116] The preprocessing module 310 is used to perform data preprocessing on the collected financial data to obtain a valid financial data set, wherein the financial data includes accounting vouchers, financial statements, and tax data;

[0117] The training module 320 is used to continue pre-training the finance and taxation big model based on the effective financial data set, obtain the finance big model with finance and taxation knowledge, and pre-label the effective financial data set based on the prompt engineering and GPT4 to obtain a sample data set with labeling information, wherein the labeling information includes the key elements of the element extraction task and the voucher elements of the voucher generation task;

[0118] The instruction fine-tuning module 330 is used to fine-tune the instruction of the financial model with financial and tax knowledge according to the sample data set with annotation information, so as to obtain the financial model with key element extraction function and voucher generation function;

[0119] An extraction module 340 is used to extract key elements from the input data of the voucher to be generated by using a key element extraction function according to a pre-generated key element extraction instruction, and obtain key elements of the input data of the voucher to be generated;

[0120] The generation module 350 is used to generate vouchers for key elements using the voucher generation model according to the pre-generated voucher generation instructions, and obtain the entry voucher corresponding to the input data of the voucher to be generated.

[0121] Optionally, the preprocessing module 310 includes:

[0122] The desensitization submodule is used to perform desensitization processing on the financial data according to preset rules to obtain desensitized financial data;

[0123] The cleaning submodule is used to clean the desensitized financial data using regular rules to obtain cleaned financial data, where the cleaning operations include noise removal, deduplication, text cleaning, and slicing;

[0124] The processing submodule is used to process the cleaned financial data through the statistical method of Z-score, constant filling and regular matching to obtain high-quality financial data;

[0125] Perform data enhancement on high-quality financial data to obtain effective financial data sets.

[0126] Optionally, the device 300 further includes: a first optimization module, configured to optimize the key factor extraction function according to the real-time financial data collected in real time, and the implementation process is:

[0127] Use the key element extraction function to extract key elements from real-time financial data and obtain real-time key element information of real-time financial data;

[0128] When the real-time key element information does not conform to the preset element format, the hyperparameters of the key element extraction function are adjusted until the output real-time key element information conforms to the preset element format, thereby obtaining an optimized key element extraction function.

[0129] Optionally, the device 300 further includes: a second optimization module, configured to optimize the voucher generation model according to the real-time financial data collected in real time, and the implementation process is as follows:

[0130] Use the updated key element extraction function to extract key elements from real-time financial data and obtain real-time key element information of real-time financial data;

[0131] Use the voucher generation model to generate vouchers for real-time key element information and obtain real-time accounting vouchers for real-time financial data;

[0132] When the real-time accounting voucher does not conform to the preset accounting voucher format, the voucher generation model is hyper-parameter adjusted until the output real-time accounting voucher conforms to the preset accounting voucher format, thereby obtaining an optimized voucher generation model.

[0133] Exemplary Electronic Devices

[0134] Figure 4 This is a structure of an electronic device provided by an exemplary embodiment of the present invention. Figure 4 As shown, the electronic device 40 includes one or more processors 41 and a memory 42 .

[0135] The processor 41 may be a central processing unit (CPU) or other forms of processing units having data processing capabilities and / or instruction execution capabilities, and may control other components in the electronic device to perform desired functions.

[0136] The memory 42 may include one or more computer program products, which may include various forms of computer-readable storage media, such as volatile memory and / or non-volatile memory. The volatile memory may include, for example, random access memory (RAM) and / or cache memory (cache), etc. The non-volatile memory may include, for example, read-only memory (ROM), hard disk, flash memory, etc. One or more computer program instructions may be stored on the computer-readable storage medium, and the processor 41 may run the program instructions to implement the methods of the software programs of the various embodiments of the present invention described above and / or other desired functions. In one example, the electronic device may also include: an input device 43 and an output device 44, which are interconnected via a bus system and / or other forms of connection mechanisms (not shown).

[0137] In addition, the input device 43 may also include, for example, a keyboard, a mouse, etc.

[0138] The output device 44 can output various information to the outside, and can include, for example, a display, a speaker, a printer, a communication network and a remote output device connected thereto.

[0139] Of course, to simplify, Figure 4 Only some of the components related to the present invention in the electronic device are shown, and components such as a bus, an input / output interface, etc. are omitted. In addition, the electronic device may further include any other appropriate components according to specific application conditions.

[0140] Exemplary computer program products and computer-readable storage media

[0141] In addition to the above-mentioned methods and devices, an embodiment of the present invention may also be a computer program product, which includes computer program instructions, which, when executed by a processor, enable the processor to execute the steps of the method according to various embodiments of the present invention described in the above-mentioned "Exemplary Method" section of this specification.

[0142] The computer program product may be written in any combination of one or more programming languages ​​to write program code for performing the operations of the embodiments of the present invention, including object-oriented programming languages ​​such as Java, C++, etc., and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user computing device, partially on the user device, as a separate software package, partially on the user computing device and partially on a remote computing device, or entirely on a remote computing device or server.

[0143] In addition, an embodiment of the present invention may also be a computer-readable storage medium having computer program instructions stored thereon, which, when executed by a processor, enable the processor to execute the steps of the method according to various embodiments of the present invention described in the above “Exemplary Method” section of this specification.

[0144] The computer readable storage medium can adopt any combination of one or more readable media. The readable medium can be a readable signal medium or a readable storage medium. The readable storage medium can include, for example, but is not limited to, a system, system or device of electricity, magnetism, light, electromagnetic, infrared, or semiconductor, or any combination of the above. More specific examples (non-exhaustive list) of readable storage media include: an electrical connection with one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above.

[0145] The basic principle of the present invention is described above in conjunction with specific embodiments. However, it should be pointed out that the advantages, strengths, effects, etc. mentioned in the present invention are only examples and not limitations, and it cannot be considered that these advantages, strengths, effects, etc. must be possessed by each embodiment of the present invention. In addition, the specific details disclosed above are only for the purpose of illustration and facilitation of understanding, rather than limitation, and the above details do not limit the present invention to being implemented by adopting the above specific details.

[0146] Each embodiment in this specification is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the system embodiment, since it basically corresponds to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0147] The block diagrams of the devices, systems, equipment, and systems involved in the present invention are only illustrative examples and are not intended to require or imply that they must be connected, arranged, and configured in the manner shown in the block diagram. As will be appreciated by those skilled in the art, these devices, systems, equipment, and systems can be connected, arranged, and configured in any manner. Words such as "including", "comprising", "having", etc. are open words, referring to "including but not limited to", and can be used interchangeably with them. The words "or" and "and" used here refer to the words "and / or" and can be used interchangeably with them, unless the context clearly indicates otherwise. The word "such as" used here refers to the phrase "such as but not limited to", and can be used interchangeably with it.

[0148] The method and system of the present invention may be implemented in many ways. For example, the method and system of the present invention may be implemented by software, hardware, firmware or any combination of software, hardware, firmware. The above order of steps for the method is only for illustration, and the steps of the method of the present invention are not limited to the order specifically described above, unless otherwise specifically stated. In addition, in some embodiments, the present invention may also be implemented as a program recorded in a recording medium, which includes machine-readable instructions for implementing the method according to the present invention. Thus, the present invention also covers a recording medium storing a program for executing the method according to the present invention.

[0149] It should also be noted that in the system, device and method of the present invention, each component or each step can be decomposed and / or recombined. These decompositions and / or recombinations should be regarded as equivalent schemes of the present invention. The above description of the disclosed aspects is provided to enable any technician in the field to make or use the present invention. Various modifications to these aspects are very obvious to those skilled in the art, and the general principles defined here can be applied to other aspects without departing from the scope of the present invention. Therefore, the present invention is not intended to be limited to the aspects shown here, but in accordance with the widest range consistent with the principles and novel features disclosed here.

[0150] The above description has been given for the purpose of illustration and description. In addition, this description is not intended to limit the embodiments of the present invention to the forms disclosed herein. Although a number of example aspects and embodiments have been discussed above, those skilled in the art will recognize certain variations, modifications, changes, additions and sub-combinations thereof.

Claims

1. A method for automatically generating accounting vouchers based on a large financial and taxation model, characterized in that: include: Performing data preprocessing on the collected financial data to obtain a valid financial data set, wherein the financial data includes accounting vouchers, financial statements, and tax data; Based on the valid financial data set, the financial and taxation big model is further pre-trained to obtain a financial big model with financial and taxation knowledge, and the valid financial data set is pre-labeled based on the prompt engineering and GPT4 to obtain a sample data set with labeling information, wherein the labeling information includes the key elements of the element extraction task and the voucher elements of the voucher generation task; Fine-tune the instructions of the financial model with financial and tax knowledge according to the sample data set with annotated information, so as to obtain the financial model with key element extraction function and voucher generation function; Utilize the key element extraction function to extract key elements from the input data of the voucher to be generated according to the pre-generated key element extraction instruction, and obtain the key elements of the input data of the voucher to be generated; The voucher generation model is used to generate vouchers for the key elements according to the pre-generated voucher generation instructions, and the entry voucher corresponding to the input data of the voucher to be generated is obtained.

2. The method according to claim 1, characterized in that Preprocess the collected financial data to obtain valid financial data sets, including: Desensitizing the financial data according to preset rules to obtain desensitized financial data; Using regular rules to perform a cleaning operation on the desensitized financial data to obtain cleaned financial data, wherein the cleaning operation includes removing noise, removing duplicates, text cleaning, and slicing; Process the cleaned financial data by using the statistical method of Z-score, constant filling and regular matching to obtain high-quality financial data; The high-quality financial data is enhanced to obtain the valid financial data set.

3. The method according to claim 1, characterized in that Also includes: The key factor extraction function is optimized according to the real-time financial data collected in real time, and the implementation process is as follows: Extracting key elements from the real-time financial data using the key element extraction function to obtain real-time key element information of the real-time financial data; In the case that the real-time key element information does not conform to the preset element format, the key element extraction function is subjected to hyperparameter adjustment until the output real-time key element information conforms to the preset element format, thereby obtaining an optimized key element extraction function.

4. The method according to claim 3, characterized in that Also includes: The voucher generation model is optimized according to the real-time financial data collected in real time, and the implementation process is as follows: Extracting key elements from the real-time financial data using the updated key element extraction function to obtain real-time key element information of the real-time financial data; Using the voucher generation model to generate vouchers for the real-time key element information, and obtaining real-time entry vouchers for the real-time financial data; In the case that the real-time accounting voucher does not conform to the preset accounting voucher format, the voucher generation model is hyper-parameter adjusted until the output real-time accounting voucher conforms to the preset accounting voucher format, thereby obtaining an optimized voucher generation model.

5. A device for automatically generating accounting vouchers based on a large financial and taxation model, characterized in that: include: A preprocessing module is used to perform data preprocessing on the collected financial data to obtain a valid financial data set, wherein the financial data includes accounting vouchers, financial statements, and tax data; A training module is used to continue pre-training the finance and taxation big model based on the valid financial data set to obtain a financial big model with finance and taxation knowledge, and to pre-label the valid financial data set based on the prompt engineering and GPT4 to obtain a sample data set with labeled information, wherein the labeled information includes key elements of the element extraction task and voucher elements of the voucher generation task; An instruction fine-tuning module is used to fine-tune the instructions of the financial model with financial and tax knowledge according to the sample data set with annotation information, so as to obtain the financial model with key element extraction function and voucher generation function; An extraction module, used to extract key elements from the input data of the voucher to be generated by using the key element extraction function according to a pre-generated key element extraction instruction, and obtain the key elements of the input data of the voucher to be generated; A generation module is used to use the voucher generation model to generate vouchers for the key elements according to pre-generated voucher generation instructions, and obtain the entry voucher corresponding to the input data of the voucher to be generated.

6. The device according to claim 5, characterized in that Preprocessing modules include: A desensitization submodule, used to perform desensitization processing on the financial data according to preset rules to obtain desensitized financial data; A cleaning submodule, used for performing a cleaning operation on the desensitized financial data using regular rules to obtain cleaned financial data, wherein the cleaning operation includes noise removal, deduplication, text cleaning, and slicing; A processing submodule, used for processing the cleaned financial data by using a Z-score statistical method, constant filling, and regular matching to obtain high-quality financial data; The high-quality financial data is enhanced to obtain the valid financial data set.

7. The device according to claim 5, characterized in that Also includes: The first optimization module is used to optimize the key factor extraction function according to the real-time financial data collected in real time, and the implementation process is as follows: Extracting key elements from the real-time financial data using the key element extraction function to obtain real-time key element information of the real-time financial data; In the case that the real-time key element information does not conform to the preset element format, the key element extraction function is subjected to hyperparameter adjustment until the output real-time key element information conforms to the preset element format, thereby obtaining an optimized key element extraction function.

8. The method according to claim 7, characterized in that Also includes: The second optimization module is used to optimize the voucher generation model according to the real-time financial data collected in real time, and the implementation process is as follows: Extracting key elements from the real-time financial data using the updated key element extraction function to obtain real-time key element information of the real-time financial data; Using the voucher generation model to generate vouchers for the real-time key element information, and obtaining real-time entry vouchers for the real-time financial data; In the case that the real-time accounting voucher does not conform to the preset accounting voucher format, the voucher generation model is hyper-parameter adjusted until the output real-time accounting voucher conforms to the preset accounting voucher format, thereby obtaining an optimized voucher generation model.

9. A computer-readable storage medium, characterized in that: The storage medium stores a computer program, and the computer program is used to execute the method according to any one of claims 1 to 4.

10. An electronic device, characterized in that: The electronic device comprises: processor; a memory for storing instructions executable by the processor; The processor is used to read the executable instructions from the memory and execute the instructions to implement the method described in any one of claims 1 to 4.

Images