Tax data processing method and system
By using large-scale pre-trained models for deep learning analysis in tax data processing, potential tax rules and abnormal patterns are discovered, and combined with data standardization, standardization, automation and intelligent processing, the problem of low efficiency and accuracy of tax data management is solved, and efficient and accurate tax data processing and self-repair functions are achieved.
Patent Information
- Application Number
- CN202411946305.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-12-27
- Publication Date
- 2025-04-25
AI Technical Summary
In the prior art, tax data management is relatively low efficiency and accuracy. The traditional processing method requires the processing of tax data in each province separately. The development cycle is long and the cost is high. Due to the diverse data formats, the error rate is high.
A large-scale pre-trained model is used to conduct deep learning analysis of tax data, discover potential tax rules and abnormal patterns, and improve data consistency and accuracy through data standardization, standardization, automation and intelligent processing.
It significantly improves the intelligence level of tax data processing, reduces the workload of manual review, improves the efficiency and accuracy of tax management, reduces the error rate, and realizes the self-repair function of data.
Smart Images
Figure CN119885011A_ABST
Abstract
Description
Technical Field
[0001] The present application belongs to the field of computer technology, and specifically relates to a method and system for processing tax data. Background Art
[0002] Tax data refers to all kinds of information and materials related to taxation, including various records of individuals and enterprises in the process of taxation. These data not only involve basic information of taxpayers, but also include information on economic activities such as income, expenditure, assets, liabilities, etc. Understanding tax data is of great significance for personal and corporate financial management, compliance operations, and legal tax avoidance. Although corporate tax data can be queried in major provincial tax bureaus, due to the wide variety of information and complex formats, it is easy to cause problems such as difficulty in data statistical analysis and use.
[0003] At present, traditional technical means require the tax data of each province to be processed separately, which has a long development cycle and high development costs. Since each province has multiple data styles for the same tax type, the traditional processing method is cumbersome and has a high error rate.
[0004] Application Contents
[0005] The purpose of the embodiments of the present application is to provide a method and system for processing tax data to address the defects of low efficiency and accuracy of tax management in the prior art.
[0006] In order to solve the above technical problems, this application is implemented as follows:
[0007] In a first aspect, a method for processing tax data is provided, comprising the following steps:
[0008] Process tax data using existing tax rules and check whether the processed tax data meets the preset characteristic standards;
[0009] In the case where the processed tax data does not meet the preset feature standards, a large-scale pre-trained model is used to conduct deep learning analysis on the tax data to discover potential tax rules and abnormal patterns;
[0010] The tax data is reprocessed based on the potential tax rules and the abnormal pattern, and the potential tax rules are stored if the processed tax data meets a preset characteristic standard.
[0011] In a second aspect, a tax data processing system is provided, comprising:
[0012] A checking module is used to process tax data through existing tax rules and check whether the processed tax data meets the preset characteristic standards;
[0013] A learning module, for performing deep learning analysis on the tax data using a large-scale pre-trained model to discover potential tax rules and abnormal patterns when the processed tax data does not meet the preset feature standards;
[0014] A processing module is used to reprocess the tax data based on the potential tax rules and the abnormal pattern, and store the potential tax rules when the processed tax data meets the preset characteristic standard.
[0015] The embodiment of the present application utilizes a large-scale pre-trained model to perform deep learning analysis on tax data to discover potential tax rules and abnormal patterns, which can significantly improve the level of intelligence in tax data processing, reduce the workload of manual review, and improve the efficiency and accuracy of tax management. BRIEF DESCRIPTION OF THE DRAWINGS
[0016] Figure 1 It is a flow chart of a method for processing tax data provided by an embodiment of the present application;
[0017] Figure 2 It is a structural diagram of a tax data processing system provided in an embodiment of the present application. DETAILED DESCRIPTION
[0018] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.
[0019] In the prior art, due to the various formats of tax data, more code needs to be modified when fixing problems, and the mapping program needs to be restarted when identifying compatible logic, which makes updates difficult and has a wide impact. In addition, due to the complexity of parsing PDF files, there may be some subtle errors that are not easy to detect, and these errors cannot be automatically fixed, resulting in limited automatic repair capabilities.
[0020] In order to improve the efficiency of data analysis and use, reduce the error rate, and ensure comprehensive self-healing capabilities, this application embodiment proposes a new solution. This solution will improve the existing processing method and is committed to solving the problem of differences in corporate tax data among provinces. Specifically, this application embodiment will improve the existing technology in the following ways:
[0021] Data standardization: In order to ensure the consistency and integrity of the data, the embodiment of the present application performs data standardization processing to keep the data in a consistent format and structure.
[0022] Data normalization: Data normalization aims to eliminate redundant information in the data, improve data availability, and facilitate data analysis.
[0023] Data automation: With the help of automation tools, the embodiments of the present application can process large amounts of data quickly and accurately, thereby improving work efficiency.
[0024] Data intelligence: Using artificial intelligence technology, the embodiments of the present application can build an intelligent data management system that automatically detects and corrects errors to improve the accuracy and reliability of data.
[0025] Introducing AI big model technical capabilities: By introducing AI big models, the embodiments of the present application can better realize the self-repair function of data and reduce losses caused by data quality issues.
[0026] By comprehensively improving the above aspects, the embodiments of the present application can significantly improve the quality and effectiveness of corporate tax data and provide better decision support and services for enterprises. In addition, by introducing new technical means, the embodiments of the present application can better realize the self-repair function of data and reduce the losses caused by data quality problems.
[0027] Specifically, the embodiment of the present application provides a method and application of a large model-based rule discovery device in tax data processing. The method uses a large-scale pre-trained model to perform deep learning analysis on tax data to automatically discover potential tax rules and abnormal patterns. The data of each module is processed through existing rules, and the necessary features of the module are checked based on the results. When the necessary features are abnormal, the large model is used to re-extract the rules and process the data until the necessary features meet the requirements, and the new rules are automatically stored.
[0028] The specific steps include:
[0029] Data collection and preprocessing: Collect raw data from multiple tax systems and perform cleaning and standardization.
[0030] Feature extraction: Use the pre-trained large model to extract features from the processed data and generate a high-dimensional feature vector.
[0031] Rule discovery: Automatically identify and extract potential rules and abnormal patterns in tax data through the self-learning ability of the model.
[0032] Rule verification and optimization: Verify and optimize the discovered rules to ensure their accuracy and reliability.
[0033] Application implementation: Apply the optimized rules to actual tax data processing to improve the efficiency and accuracy of tax management.
[0034] In the following, in conjunction with the accompanying drawings, a method for processing tax data provided by an embodiment of the present application is described in detail through specific embodiments and application scenarios.
[0035] like Figure 1 FIG. 1 is a flow chart of a method for processing tax data provided in an embodiment of the present application, the method comprising the following steps:
[0036] Step 101, process the tax data using existing tax rules, and check whether the processed tax data meets the preset characteristic standards.
[0037] Specifically, the tax data can be matched with the rule configuration table and the corresponding tax rules can be loaded; the tax data can be preliminarily processed using the model rule valuer, and the processed data can be analyzed and processed by the central processing unit; the data output by the central processing unit can be value processed by the value processor, and the module feature checker can be used to check whether the processed data meets the preset feature standards.
[0038] In this embodiment, after processing the tax data through the existing tax rules and checking whether the processed tax data meets the preset characteristic standards, it also includes: when the processed tax data does not meet the preset characteristic standards, if the number of non-compliance with the rules reaches a preset threshold, an alarm information is generated through an alarm processor to prompt the user to perform manual intervention; when the processed tax data meets the preset characteristic standards, the mapping results are stored and output.
[0039] Step 102, when the processed tax data does not meet the preset feature standards, use a large-scale pre-trained model to perform deep learning analysis on the tax data to discover potential tax rules and abnormal patterns.
[0040] Specifically, a large-scale pre-trained model can be used to extract features from tax data to generate a high-dimensional feature vector; through the self-learning ability of the large-scale pre-trained model, based on the feature vector, the potential rules and abnormal patterns in the tax data can be identified and extracted.
[0041] In this embodiment, the tax data can be processed into multiple module data using a module classifier, and multiple module extractors can be used to extract rules from each module data respectively; the extracted potential tax rules and abnormal patterns can be summarized through a unified module rule processor.
[0042] Step 103: reprocess the tax data based on the potential tax rules and the abnormal pattern, and store the potential tax rules if the processed tax data meets the preset characteristic standard.
[0043] The embodiment of the present application utilizes a large-scale pre-trained model to perform deep learning analysis on tax data to discover potential tax rules and abnormal patterns, which can significantly improve the level of intelligence in tax data processing, reduce the workload of manual review, and improve the efficiency and accuracy of tax management.
[0044] In the embodiments of the present application, the key innovation is to clean and process a large amount of data to obtain valid tax data, so that users can more conveniently obtain the required valid tax data.
[0045] The key points are:
[0046] 1) Data collection technology
[0047] 2) The module idea of separating business and mapping relationship;
[0048] 3) Use self-developed edit distance to automatically correct project names
[0049] 4) Plug-in real-time alarm method
[0050] The following is a detailed description of the embodiments of the present application with respect to the specific process:
[0051] The tax data mapping method provided in the embodiment of the present application includes the following steps:
[0052] 1. Tax collector: First, collect various tax information of the enterprise through the tax collector.
[0053] 2. Rule configuration table: Match the collected data with the rule configuration table and load the corresponding rules.
[0054] 3. Model rule valuer: Use the model rule valuer to perform preliminary processing on the data.
[0055] 4. Central Processing Unit: The central processing unit receives the processed data and performs further analysis and processing.
[0056] 5. Value processor: The value processor processes the data output by the central processing unit to ensure the accuracy and completeness of the data.
[0057] 6. Module feature verifier: The module feature verifier checks whether the processed data meets the preset feature standards.
[0058] 7. Alarm processor: If the data does not meet the preset feature standards, the alarm processor will generate an alarm message to prompt the user to perform manual intervention.
[0059] 8. Store mapping results: For data that meets the feature criteria, store the mapping results for subsequent use.
[0060] 9. Output mapping results: Finally, the system outputs the mapping results for users to view and analyze.
[0061] 10. Manual processing: For situations that cannot be handled automatically, a manual processing function is provided for manual intervention and processing.
[0062] 11. Re-extract new rules through a large language model: Re-extract new rules through a large language model to optimize the intelligence level of the system.
[0063] 12. Generate new rules: Generate new rules based on actual conditions to adapt to the ever-changing tax environment.
[0064] Through the above process, the embodiment of the present application can effectively integrate various tax information of the enterprise, improve tax management efficiency, reduce tax risks, and achieve more accurate tax reporting.
[0065] like Figure 2 FIG. 1 is a schematic diagram of a tax data processing system provided in an embodiment of the present application, including:
[0066] The checking module 210 is used to process the tax data according to the existing tax rules and check whether the processed tax data meets the preset characteristic standards.
[0067] Specifically, the inspection module 210 is specifically used to match the tax data with the rule configuration table and load the corresponding tax rules; use the model rule valuer to preliminarily process the tax data, and analyze and process the processed data through the central processing unit; perform value processing on the data output by the central processing unit through the value processor, and check whether the processed data meets the preset feature standards through the module feature checker.
[0068] The learning module 220 is used to perform deep learning analysis on the tax data using a large-scale pre-trained model to discover potential tax rules and abnormal patterns when the processed tax data does not meet the preset feature standards.
[0069] Specifically, the learning module 220 is used to extract features from the tax data using a large-scale pre-trained model to generate a high-dimensional feature vector; through the self-learning ability of the large-scale pre-trained model, based on the feature vector, the potential rules and abnormal patterns in the tax data are identified and extracted.
[0070] In this embodiment, the learning module 220 is specifically used to match the tax data with the module feature configuration and load the module features; use the module classifier to process the tax data into multiple module data, and use multiple module extractors to extract rules from each module data respectively; and use the module rule unified processor to summarize the extracted potential tax rules and abnormal patterns.
[0071] The processing module 230 is used to reprocess the tax data based on the potential tax rules and the abnormal pattern, and store the potential tax rules when the processed tax data meets the preset characteristic standard.
[0072] Furthermore, the above system further includes:
[0073] The alarm module is used to generate an alarm message through an alarm processor to prompt a user to perform manual intervention when the processed tax data does not meet the preset characteristic standards and the number of times the rules are not met reaches a preset threshold.
[0074] The output module is used to store and output the mapping result when the processed tax data meets the preset characteristic standards.
[0075] The embodiment of the present application utilizes a large-scale pre-trained model to perform deep learning analysis on tax data to discover potential tax rules and abnormal patterns, which can significantly improve the level of intelligence in tax data processing, reduce the workload of manual review, and improve the efficiency and accuracy of tax management.
[0076] The embodiment of the present application also provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, each process of the above-mentioned tax data processing method embodiment is implemented, and the same technical effect can be achieved. To avoid repetition, it is not repeated here. The computer-readable storage medium is, for example, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.
[0077] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the existence of other identical elements in the process, method, article or device including the element.
[0078] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, a magnetic disk, or an optical disk), and includes a number of instructions for a terminal (which can be a mobile phone, a computer, a server, an air conditioner, or a network device, etc.) to execute the methods described in each embodiment of the present application.
[0079] The embodiments of the present application are described above in conjunction with the accompanying drawings, but the present application is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the guidance of the present application, ordinary technicians in this field can also make many forms without departing from the purpose of the present application and the scope of protection of the claims, all of which are within the protection of the present application.
Claims
1. A method for processing tax data, characterized in that: The following steps are involved: Process tax data using existing tax rules and check whether the processed tax data meets the preset characteristic standards; In the case where the processed tax data does not meet the preset feature standards, a large-scale pre-trained model is used to conduct deep learning analysis on the tax data to discover potential tax rules and abnormal patterns; The tax data is reprocessed based on the potential tax rules and the abnormal pattern, and the potential tax rules are stored if the processed tax data meets a preset characteristic standard.
2. The method according to claim 1, characterized in that The tax data is processed by the existing tax rules, and the processed tax data is checked to see whether it meets the preset characteristic standards, specifically including: Matching the tax data with the rule configuration table and loading the corresponding tax rules; Performing preliminary processing on the tax data using a model rule valuer, and analyzing and processing the processed data using a central processing unit; The data output by the central processor is value processed by the value processor, and the module feature checker is used to check whether the processed data meets the preset feature standard.
3. The method according to claim 2, characterized in that After processing the tax data using the existing tax rules and checking whether the processed tax data meets the preset characteristic standards, the method further includes: In the case where the processed tax data does not meet the preset characteristic standard, if the number of times it does not meet the rule reaches a preset threshold, an alarm processor generates an alarm message to prompt the user to perform manual intervention; When the processed tax data meets the preset characteristic standard, the mapping result is stored and output.
4. The method according to claim 1, characterized in that: The use of large-scale pre-trained models to conduct deep learning analysis on tax data to discover potential tax rules and abnormal patterns, including: Use large-scale pre-trained models to extract features from tax data and generate high-dimensional feature vectors; Through the self-learning ability of the large-scale pre-trained model, potential rules and abnormal patterns in the tax data are identified and extracted based on the feature vector.
5. The method according to claim 1, characterized in that The use of large-scale pre-trained models to conduct deep learning analysis on tax data to discover potential tax rules and abnormal patterns, including: Matching the tax data with the module feature configuration and loading the module features; Using a module classifier to process the tax data into a plurality of module data, and using a plurality of module extractors to extract rules from each module data respectively; The extracted potential tax rules and exception patterns are summarized through the module rule unified processor.
6. A tax data processing system, characterized in that: include: A checking module is used to process tax data through existing tax rules and check whether the processed tax data meets the preset characteristic standards; A learning module, for performing deep learning analysis on the tax data using a large-scale pre-trained model to discover potential tax rules and abnormal patterns when the processed tax data does not meet the preset feature standards; A processing module is used to reprocess the tax data based on the potential tax rules and the abnormal pattern, and store the potential tax rules when the processed tax data meets the preset characteristic standard.
7. The system according to claim 6, characterized in that The checking module is specifically used to match the tax data with the rule configuration table and load the corresponding tax rules; perform preliminary processing on the tax data using the model rule valuer, and analyze and process the processed data through the central processing unit; The data output by the central processor is value processed by the value processor, and the module feature checker is used to check whether the processed data meets the preset feature standard.
8. The system according to claim 7, characterized in that Also includes: An alarm module is used to generate an alarm message through an alarm processor to prompt a user to perform manual intervention when the processed tax data does not meet the preset characteristic standard and the number of times the rule is not met reaches a preset threshold; The output module is used to store and output the mapping result when the processed tax data meets the preset characteristic standards.
9. The system according to claim 6, characterized in that The learning module is specifically used to extract features from tax data using a large-scale pre-trained model to generate a high-dimensional feature vector; through the self-learning ability of the large-scale pre-trained model, based on the feature vector, the potential rules and abnormal patterns in the tax data are identified and extracted.
10. The system according to claim 6, characterized in that The learning module is specifically used to match the tax data with the module feature configuration and load the module features; use a module classifier to process the tax data into multiple module data, and use multiple module extractors to extract rules from each module data respectively; and use a module rule unified processor to summarize the extracted potential tax rules and abnormal patterns.