Optimization processing method and device of audit model, equipment and storage medium
By inference processing of the audit problem list and preset prompt words, multiple inference audit reports are generated, and target audit models are determined by comparing and determining the target audit model, the problems of low efficiency and high labor cost are solved, and more efficient and high-quality audit report generation is achieved.
Patent Information
- Application Number
- CN202510335734.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-03-20
- Publication Date
- 2025-06-24
AI Technical Summary
When a large number of audit problems arise, the way auditors write audit reports based on the audit problems can easily lead to low efficiency in generating audit reports and high labor costs.
By obtaining audit report data, entering the audit problem list and preset prompt words into the trained audit model for inference processing, generating multiple inference audit reports, comparing them with the original audit report, and determining the target audit model to generate the final audit report.
It improves the efficiency of generating audit reports, reduces labor costs, and improves the quality of generation.
Smart Images
Figure CN120198237A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of computer technology, and particularly to an optimization processing method, device, equipment and storage medium for an audit model. Background Art
[0002] A large model refers to a "large parameter" model trained using large-scale data and powerful computing capabilities, which can improve expression ability and prediction performance to handle more complex tasks and data. Among them, the large model can also be a specific model such as an audit model.
[0003] In the related art, when an audit report needs to be written, an auditor writes the audit report through a computer device according to pre-determined audit questions to complete the corresponding audit work.
[0004] However, in the related art, when a large number of audit questions are generated, the way that an auditor writes an audit report according to the audit questions easily leads to low efficiency in generating the audit report and high labor costs. Summary of the Invention
[0005] This application provides an optimization processing method, device, equipment and storage medium for an audit model to solve the problem that when a large number of audit questions are generated, the way that an auditor writes an audit report according to the audit questions easily leads to low efficiency in generating the audit report and high labor costs.
[0006] In a first aspect, this application provides an optimization processing method for an audit model, which is applied to a computer device and includes:
[0007] Obtain any audit report data, where the audit report data includes an audit question list and an audit report;
[0008] Input the audit question list and a preset prompt word into each trained audit model for inference processing to obtain multiple first inference audit reports, where each trained audit model is obtained by training according to each preset audit model;
[0009] Compare each first inference audit report with the audit report to obtain multiple comparison values;
[0010] Determine a target audit model from each of the trained audit models according to each comparison value;
[0011] Output the target audit model, where the target audit model is used to generate an audit report according to an input list of audit questions to be processed.
[0012] In a possible design, the training process of each preset audit model includes: obtaining all historical audit report data within a preset time; screening the all historical audit report data to obtain multiple screened historical audit report data; determining whether the quantity corresponding to the screened historical audit report data is less than a first preset threshold; if it is determined that the quantity corresponding to the screened historical audit report data is not less than the first preset threshold, determining part of the multiple historical audit report data as a training set; and training each preset audit model according to the training set and a preset training method to obtain each trained audit model.
[0013] In a possible design, the historical audit report data includes a historical audit issue list and a historical audit report; correspondingly, the screening of the all historical audit report data to obtain multiple screened historical audit report data includes: determining each historical audit issue list and each historical audit report corresponding to the all historical audit report data; screening the each historical audit issue list and the each historical audit report to obtain each screened historical audit issue list and each screened historical audit report; and determining the each screened historical audit issue list and the each screened historical audit report as the corresponding multiple screened historical audit report data.
[0014] In a possible design, the screening of the each historical audit issue list and the each historical audit report to obtain each screened historical audit issue list and each screened historical audit report includes: inputting any historical audit issue in any historical audit issue list and the corresponding historical audit report into a preset general audit model for processing to determine whether the historical audit issue matches the historical audit report; if the historical audit issue matches the historical audit report, determining the historical audit issue as a screened historical audit issue; if the historical audit issue does not match the historical audit report, filtering out the historical audit issue; determining the each screened historical audit issue and the historical audit report as a screened historical audit issue list and a screened historical audit report; and traversing the remaining each historical audit issue list and each historical audit report to obtain each screened historical audit issue list and each screened historical audit report.
[0015] In a possible design, determining a target audit model from the trained audit models according to the comparison values includes: screening the comparison values according to a second preset threshold to obtain one or more screened comparison values; determining corresponding one or more first inference audit reports from the first inference audit reports according to the one or more screened comparison values; generating a corresponding evaluation report in response to an evaluation operation of an auditor on the one or more first inference audit reports; determining a target inference audit report from the one or more first inference audit reports according to the evaluation report; and determining a corresponding target audit model from the trained audit models according to the target inference audit report.
[0016] In a possible design, the historical audit report data includes a historical audit problem list and a historical audit report; correspondingly, after determining whether the number of all the screened historical audit report data is less than a first preset threshold, it further includes: if it is determined that the number of the screened historical audit report data is less than the first preset threshold, obtaining historical audit problems of each type in each historical audit problem list; generating corresponding audit problem lists according to the historical audit problems of each type; inputting the audit problem lists and preset prompt words into a preset general audit model for inference processing to obtain corresponding second inference audit reports; generating adjusted audit reports in response to an adjustment operation of an auditor on each second inference audit report; generating corresponding combined audit report data according to the audit problem lists and the adjusted audit reports; and integrating the combined audit report data into the multiple screened historical audit report data to complete the supplementation of the multiple screened historical audit report data.
[0017] In a possible design, it further includes: determining the remaining part of the multiple historical audit report data as a test set; and inputting the test set into each trained audit model for testing to complete the testing of each trained audit model.
[0018] In a second aspect, the present application provides an optimization processing device for an audit model, which is applied to a computer device and includes:
[0019] A first acquisition module, configured to acquire any audit report data, where the audit report data includes an audit problem list and an audit report;
[0020] A first processing module, configured to input the audit problem list and preset prompt words into each trained audit model for inference processing to obtain multiple first inference audit reports, where each trained audit model is obtained by training according to each preset audit model;
[0021] A comparison module, configured to compare each first inference audit report with the audit report to obtain a plurality of comparison values;
[0022] A first determination module, configured to determine a target audit model from the trained audit models according to the comparison values;
[0023] An output module, configured to output the target audit model, where the target audit model is used to generate an audit report according to an input list of audit issues to be processed.
[0024] In a third aspect, the present application provides a computer device, including: at least one processor and a memory;
[0025] The memory stores computer-executable instructions;
[0026] The at least one processor executes the computer-executable instructions stored in the memory, so that the at least one processor executes the optimization processing method of the audit model described in the first aspect and various possible designs of the first aspect above.
[0027] In a fourth aspect, the present application provides a computer storage medium, in which computer-executable instructions are stored. When a processor executes the computer-executable instructions, the optimization processing method of the audit model described in the first aspect and various possible designs of the first aspect above is implemented.
[0028] The optimization processing method, device, equipment and storage medium of the audit model provided by the present application obtain any audit report data, where the audit report data includes a list of audit issues and an audit report; input the list of audit issues and a preset prompt word into each trained audit model for inference processing to obtain a plurality of first inference audit reports, where each trained audit model is obtained by training according to each preset audit model; compare each first inference audit report with the audit report to obtain a plurality of comparison values; determine a target audit model from each trained audit model according to the comparison values; output the target audit model, where the target audit model is used to generate an audit report according to an input list of audit issues to be processed, thereby improving the generation efficiency of the audit report and reducing the labor cost. BRIEF DESCRIPTION OF THE DRAWINGS
[0029] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present application. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.
[0030] Figure 1Schematic diagram of the application scenario of the optimization processing method for the audit model provided by the embodiments of this application;
[0031] Figure 2 Schematic flow of the optimization processing method for the audit model provided by the embodiments of this application Figure 1 ;
[0032] Figure 3 Schematic flow of the optimization processing method for the audit model provided by the embodiments of this application Figure 2 ;
[0033] Figure 4 Schematic diagram of the structure of the optimization processing device for the audit model provided by the embodiments of this application;
[0034] Figure 5 Schematic diagram of the hardware structure of the computer device provided by the embodiments of this application. Detailed implementation manners
[0035] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings in the embodiments of this application. Apparently, the described embodiments are some but not all of the embodiments of this application. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in this application without making creative efforts fall within the scope of protection of this application.
[0036] In the technical solutions of this application, the collection, storage, use, processing, transmission, provision, and disclosure of information such as financial data or user data comply with the provisions of relevant laws and regulations and do not violate public order and good customs.
[0037] It should be noted that in the embodiments of this application, some industry-existing solutions such as certain software, components, and models may be mentioned. They should be considered exemplary, and their purpose is only to illustrate the feasibility in the implementation of the technical solutions of this application, but it does not mean that the applicant has already or necessarily used this solution.
[0038] A large model refers to a "large-parameter" model trained using large-scale data and powerful computing capabilities. These models are usually constructed by deep neural networks and have billions or even hundreds of billions of parameters, which can improve the expressive ability and prediction performance to handle more complex tasks and data. Among them, the large model can also be a specific model such as an audit model. In related technologies, when an audit report needs to be written, the auditor writes the audit report through a computer device according to pre-determined audit questions to complete the corresponding audit work. However, in related technologies, when a large number of audit questions arise, the way the auditor writes the audit report according to the audit questions easily leads to low efficiency in generating the audit report and high labor costs.
[0039] To solve the above technical problems, the embodiments of the present application propose the following technical concept: The inventor considered each trained audit model, based on each trained audit model, inferred the audit question list in any audit report data to obtain multiple inferred audit reports, compared the audit report in the audit report data with the inferred audit reports to obtain multiple comparison values, and determined the target audit model from each trained audit model using each comparison value, and generated the audit report through the target audit model, thereby improving the efficiency of generating the audit report and reducing the labor cost.
[0040] Figure 1 It is a schematic diagram of the application scenario of the optimization processing method of the audit model provided by the embodiments of the present application.
[0041] As Figure 1 shown, this scenario includes: a display terminal 101 and a computer device 102.
[0042] Among them, the display terminal 101 can be a display screen or a terminal such as a personal computer.
[0043] The computer device 102 can be an independent device or a cluster composed of multiple devices.
[0044] The computer device 102 obtains any audit report data, where the audit report data includes an audit question list and an audit report; inputs the audit question list and a preset prompt word into each trained audit model for inference processing to obtain multiple first inferred audit reports; compares each first inferred audit report with the audit report to obtain multiple comparison values; determines the target audit model from each trained audit model according to each comparison value; and outputs the target audit model to the display terminal 101. The following uses detailed embodiments for detailed description.
[0045] Figure 2 It is a schematic flow chart of the optimization processing method of the audit model provided by the embodiments of the present application Figure 1, the execution subject of this embodiment may be Figure 1 the computer device in the embodiment shown, and there is no special limitation here in this embodiment. As Figure 2 shown, the method includes:
[0046] S201: Obtain any audit report data, where the audit report data includes an audit issue list and an audit report.
[0047] In this embodiment, the audit issue list is a summary list of various issues found during the audit process.
[0048] In this embodiment, the audit report is a written document issued after auditing the audited party according to auditing standards and relevant requirements.
[0049] S202: Input the audit issue list and preset prompt words into each trained audit model for inference processing to obtain multiple first inference audit reports, where each trained audit model is obtained by training according to each preset audit model.
[0050] In this embodiment, the preset prompt words are texts or instructions input during the operation of the audit model, used to instruct the audit model to complete the corresponding operation.
[0051] S203: Compare each first inference audit report with the audit report to obtain multiple comparison values.
[0052] In this embodiment, the comparison can be a similarity comparison or other comparisons; the comparison values can be similarity values or other values.
[0053] Exemplarily, each first inference audit report is compared with the audit report in a similar manner through a preset bilingual evaluation substitution method to obtain multiple similarity values.
[0054] S204: Determine the target audit model from each trained audit model according to each comparison value.
[0055] Specifically, step S204 specifically includes:
[0056] S2041: Screen each comparison value according to a second preset threshold to obtain one or more screened comparison values.
[0057] In this embodiment, the second preset threshold can be any value among 50%, 60%, or 70%, or other values.
[0058] Specifically, each comparison value is screened according to the second preset threshold to eliminate the comparison values less than the second preset threshold, and then one or more screened comparison values are obtained.
[0059] S2042: Determine one or more corresponding first inference audit reports from each of the first inference audit reports according to one or more filtered comparison values.
[0060] S2043: Generate a corresponding evaluation report in response to the auditor's evaluation operation on one or more first inference audit reports.
[0061] Specifically, generate a corresponding evaluation report in response to the auditor's evaluation operation on one or more first inference audit reports according to the scoring rules.
[0062] Exemplarily, the scoring rules are as follows: 1 point, the quality of the first inference audit report is poor and it is unusable; 2 points, the first inference audit report can be used after modification and has reference significance; 3 points, the first inference audit report can be slightly modified to improve efficiency; 4 points, the first inference audit report hardly needs to be modified; 5 points, the first inference audit report reaches the professional level.
[0063] S2044: Determine the target inference audit report from one or more first inference audit reports according to the evaluation report.
[0064] Specifically, determine the inference audit report with the highest evaluation from one or more first inference audit reports according to the evaluation report; use the inference audit report with the highest evaluation as the target inference audit report.
[0065] In addition, optimize the preset prompt words according to the evaluation report to obtain the optimized prompt words.
[0066] S2045: Determine the corresponding target audit model from each of the trained audit models according to the target inference audit report.
[0067] S205: Output the target audit model, where the target audit model is used to generate an audit report according to the input list of audit problems to be processed.
[0068] In summary, the optimization processing method of the audit model provided in this embodiment obtains any audit report data, where the audit report data includes a list of audit problems and an audit report; inputs the list of audit problems and the preset prompt words into each of the trained audit models for inference processing to obtain multiple first inference audit reports, where each of the trained audit models is trained according to each preset audit model; compares each of the first inference audit reports with the audit report to obtain multiple comparison values; determines the target audit model from each of the trained audit models according to each comparison value; outputs the target audit model, where the target audit model is used to generate an audit report according to the input list of audit problems to be processed, thereby improving the generation efficiency of the audit report and reducing the labor cost.
[0069] In addition, the optimization processing method of the audit model provided in this embodiment determines the target audit model from each trained audit model through each comparison value as the preferred audit model, thereby improving the generation quality of the audit report.
[0070] Figure 3 Schematic flow of the optimization processing method of the audit model provided in the embodiment of the present application Figure 2 . In the embodiment of the present application, based on the provided embodiment, a detailed description is given of the specific implementation method of the training process of each preset audit model in step S202. As Figure 2 shown, the method includes: Figure 3 shown, the method includes:
[0071] S301: Obtain all historical audit report data within a preset time.
[0072] In this embodiment, the preset time can be any time period among the previous 1 year, the previous 3 years, or the previous 5 years, or it can be other time periods.
[0073] S302: Screen all historical audit report data to obtain multiple screened historical audit report data.
[0074] In this embodiment, the historical audit report data includes a historical audit problem list and a historical audit report; correspondingly, step S302 specifically includes:
[0075] S3021: Determine each historical audit problem list and each historical audit report corresponding to all historical audit report data.
[0076] S3022: Screen each historical audit problem list and each historical audit report to obtain each screened historical audit problem list and each screened historical audit report.
[0077] Specifically, step S3022 specifically includes:
[0078] S30221: Input any historical audit problem in any historical audit problem list and the corresponding historical audit report into a preset general audit model for processing to determine whether the historical audit problem matches the historical audit report.
[0079] S30222: If the historical audit problem matches the historical audit report, determine the historical audit problem as the screened historical audit problem; if the historical audit problem does not match the historical audit report, filter out the historical audit problem.
[0080] S30223: Determine each screened historical audit problem and the historical audit report as the screened historical audit problem list and the screened historical audit report.
[0081] In addition, the historical audit problem lists and corresponding historical audit reports can be input into a preset general audit model for processing to determine whether the historical audit reports include content not in the historical audit problem lists, so as to determine the filtered historical audit problem lists and the filtered historical audit reports.
[0082] S30224: Traverse the remaining historical audit problem lists and historical audit reports to obtain the filtered historical audit problem lists and historical audit reports.
[0083] S3023: Determine the filtered historical audit problem lists and the filtered historical audit reports as corresponding multiple filtered historical audit report data.
[0084] Specifically, after the filtered historical audit problem lists and the filtered historical audit reports are in one-to-one correspondence, they are determined as multiple filtered historical audit report data.
[0085] In addition, required general data can be added to the multiple filtered historical audit report data.
[0086] S303: Determine whether the quantity corresponding to the filtered historical audit report data is less than a first preset threshold.
[0087] In this embodiment, the first preset threshold can be any value among 10, 50, or 100, or can be other values.
[0088] In this embodiment, the historical audit report data includes a historical audit problem list and a historical audit report; correspondingly, after step S303, steps a to f are further included:
[0089] Step a: If it is determined that the quantity corresponding to the filtered historical audit report data is less than the first preset threshold, obtain the historical audit problems of each type in each historical audit problem list.
[0090] In this embodiment, the historical audit problems of each type are historical audit problems of the same type.
[0091] Step b: Generate corresponding audit problem lists according to the historical audit problems of each type.
[0092] Step c: Input each audit problem list and a preset prompt word into a preset general audit model for inference processing to obtain a corresponding second inference audit report.
[0093] In this embodiment, the second inference audit report generated by the preset general audit model can be used for subsequent training of each preset audit model.
[0094] Step d: Generate adjusted audit reports for each in response to the auditor's adjustment operations on the second inference audit reports.
[0095] In this embodiment, the adjustment operation may be marking, modifying, supplementing, or other operations.
[0096] In this embodiment, the adjusted audit reports are audit reports of professional level.
[0097] Step e: Generate corresponding combined audit report data according to each audit problem list and each adjusted audit report.
[0098] Specifically, after corresponding each audit problem list with each adjusted audit report one by one, generate multiple corresponding combined audit report data.
[0099] Step f: Integrate the combined audit report data into multiple filtered historical audit report data to complete the supplementation of the multiple filtered historical audit report data.
[0100] S304: If it is determined that the quantity corresponding to the filtered historical audit report data is not less than the first preset threshold, determine part of the multiple historical audit report data as the training set.
[0101] S305: Train each preset audit model according to the training set and the preset training method to obtain each trained audit model.
[0102] In this embodiment, the preset training method is a supervised fine-tuning training method, a reward feedback training method, or other training methods.
[0103] In addition, after step S305, steps A to B are further included:
[0104] Step A: Determine the remaining part of the multiple historical audit report data as the test set.
[0105] Step B: Input the test set into each trained audit model for testing to complete the testing of each trained audit model.
[0106] Specifically, obtain any historical audit problem list in the test set and the corresponding historical audit report; input the historical audit problem list into each trained audit model for testing to obtain each test audit report; if it is determined that each test audit report and the historical audit report meet the preset error, complete the testing of each trained audit model.
[0107] In summary, the optimization method for the audit model provided in this embodiment obtains all historical audit report data within a preset time; screens all historical audit report data to obtain multiple screened historical audit report data; determines whether the quantity corresponding to the screened historical audit report data is less than a first preset threshold; if it is determined that the quantity corresponding to the screened historical audit report data is not less than the first preset threshold, then determines part of the data among the multiple historical audit report data as the training set; trains each preset audit model according to the training set and the preset training method to obtain each trained audit model, laying a foundation for the determination of the target audit model, thereby improving the generation efficiency of subsequent audit reports and reducing subsequent labor costs.
[0108] Figure 4 This is a schematic structural diagram of an optimization processing device for an audit model provided by an embodiment of the present application. As Figure 4 shown, the optimization processing device for the audit model includes: a first acquisition module 401, a first processing module 402, a comparison module 403, a first determination module 404, and an output module 405.
[0109] The first acquisition module 401 is configured to acquire any audit report data, where the audit report data includes an audit problem list and an audit report;
[0110] The first processing module 402 is configured to input the audit problem list and a preset prompt word into each trained audit model for inference processing to obtain multiple first inference audit reports, where each trained audit model is obtained by training each preset audit model;
[0111] The comparison module 403 is configured to compare each first inference audit report with the audit report to obtain multiple comparison values;
[0112] The first determination module 404 is configured to determine a target audit model from each trained audit model according to each comparison value;
[0113] The output module 405 is configured to output the target audit model, where the target audit model is used to generate an audit report according to an input list of audit problems to be processed.
[0114] In a possible implementation manner, the device further includes:
[0115] A second acquisition module, configured to acquire all historical audit report data within a preset time;
[0116] A screening module, configured to screen all historical audit report data to obtain multiple screened historical audit report data;
[0117] A judgment module, configured to judge whether the quantity corresponding to the filtered historical audit report data is less than a first preset threshold;
[0118] A second determination module, configured to, if it is determined that the quantity corresponding to the filtered historical audit report data is not less than the first preset threshold, determine some of the multiple historical audit report data as a training set;
[0119] A training module, configured to train each preset audit model according to the training set and a preset training method to obtain each trained audit model.
[0120] In a possible implementation manner, the historical audit report data includes a historical audit problem list and a historical audit report; correspondingly, the screening module specifically includes:
[0121] A determination unit, configured to determine each historical audit problem list and each historical audit report corresponding to all the historical audit report data;
[0122] A screening unit, configured to screen each historical audit problem list and each historical audit report to obtain each screened historical audit problem list and each screened historical audit report;
[0123] A determination unit, configured to determine each screened historical audit problem list and each screened historical audit report as corresponding multiple screened historical audit report data.
[0124] In a possible implementation manner, the screening unit specifically includes:
[0125] A processing unit, configured to input any historical audit problem in any historical audit problem list and the corresponding historical audit report into a preset general audit model for processing to judge whether the historical audit problem matches the historical audit report;
[0126] A filtering unit, configured to, if the historical audit problem matches the historical audit report, determine the historical audit problem as a screened historical audit problem; if the historical audit problem does not match the historical audit report, filter out the historical audit problem;
[0127] A determination unit, configured to determine each screened historical audit problem and the historical audit report as a screened historical audit problem list and a screened historical audit report;
[0128] A traversing unit, configured to traverse the remaining historical audit problem lists and historical audit reports to obtain each screened historical audit problem list and each screened historical audit report.
[0129] In a possible implementation manner, the first determination module 404 specifically includes:
[0130] A screening unit, configured to screen each comparison value according to a second preset threshold to obtain one or more screened comparison values;
[0131] A first determination unit, configured to determine one or more corresponding first inference audit reports from the first inference audit reports according to the one or more screened comparison values;
[0132] A generation unit, configured to generate a corresponding evaluation report in response to an evaluation operation of an auditor on the one or more first inference audit reports;
[0133] A second determination unit, configured to determine a target inference audit report from the one or more first inference audit reports according to the evaluation report;
[0134] A third determination unit, configured to determine a corresponding target audit model from the trained audit models according to the target inference audit report.
[0135] In a possible implementation manner, the historical audit report data includes a historical audit problem list and a historical audit report; correspondingly, the apparatus further includes:
[0136] A third acquisition module, configured to, if it is determined that the quantity corresponding to the screened historical audit report data is less than a first preset threshold, acquire historical audit problems of each type in each historical audit problem list;
[0137] A first generation module, configured to generate corresponding audit problem lists according to the historical audit problems of each type;
[0138] A second processing module, configured to input the audit problem lists and a preset prompt word into a preset general audit model for inference processing to obtain corresponding second inference audit reports;
[0139] A second generation module, configured to generate adjusted audit reports in response to an adjustment operation of an auditor on each second inference audit report;
[0140] A third generation module, configured to generate corresponding combined audit report data according to the audit problem lists and the adjusted audit reports;
[0141] An integration module, configured to integrate the combined audit report data into the multiple screened historical audit report data to complete the supplementation of the multiple screened historical audit report data.
[0142] In a possible implementation manner, the apparatus further includes:
[0143] A third determination module, configured to determine the remaining part of the multiple historical audit report data as a test set;
[0144] A third processing module, configured to input the test set into each trained audit model for test processing to complete the testing of each trained audit model.
[0145] The device provided in this embodiment can be used to execute the technical solutions of the above method embodiments. The implementation principles and technical effects are similar, and will not be elaborated here.
[0146] Figure 5 It is a schematic hardware structure diagram of a computer device provided in an embodiment of the present application. As Figure 5 shown, the computer device in this embodiment includes: a processor 501 and a memory 502; the memory stores computer-executable instructions; at least one processor executes the computer-executable instructions stored in the memory, so that at least one processor executes the optimization processing method of the audit model as described above.
[0147] Optionally, the memory 502 can be either independent or integrated with the processor 501.
[0148] When the memory 502 is independently provided, the computer device further includes a bus 503 for connecting the memory 502 and the processor 501.
[0149] An embodiment of the present application further provides a computer storage medium, in which computer-executable instructions are stored. When the processor executes the computer-executable instructions, the optimization processing method of the audit model as described above is implemented.
[0150] An embodiment of the present application further provides a computer program product, including a computer program. When the computer program is executed by a processor, the optimization processing method of the audit model as described above is implemented.
[0151] In several embodiments provided by the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the modules is only a logical function division. In actual implementation, there may be other division methods. For example, multiple modules can be combined or integrated into another system, or some features can be ignored or not executed. Another point, the displayed or discussed couplings or direct couplings or communication connections to each other can be through some interfaces. The indirect couplings or communication connections of the devices or modules can be in electrical, mechanical or other forms.
[0152] The module described as a separation component may or may not be physically separated. The component shown as a module may or may not be a physical unit, that is, it may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to implement the solution of this embodiment.
[0153] In addition, each functional module in various embodiments of the present application may be integrated in a processing unit, or each module may exist physically alone, or two or more modules may be integrated in one unit. The unit formed by the above modules can be implemented in the form of hardware or in the form of a hardware plus software functional unit.
[0154] The integrated module implemented in the form of a software functional module can be stored in a computer-readable storage medium. The above software functional module is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) or a processor to execute some steps of the methods described in various embodiments of the present application.
[0155] It should be understood that the above processor may be a central processing unit (Central Processing Unit, abbreviated as CPU), and may also be other general-purpose processors, digital signal processors (Digital Signal Processor, abbreviated as DSP), application specific integrated circuits (Application Specific Integrated Circuit, abbreviated as ASIC), etc. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc. The steps of the method disclosed in combination with the invention can be directly embodied as being executed by a hardware processor, or executed by a combination of hardware and software modules in the processor.
[0156] The memory may include high-speed RAM memory, and may also include non-volatile storage NVM, such as at least one disk memory, and may also be a USB flash drive, a mobile hard disk, a read-only memory, a magnetic disk or an optical disc, etc.
[0157] The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, an Extended Industry Standard Architecture (EISA) bus, or the like. The bus can be divided into an address bus, a data bus, a control bus, etc. For the sake of convenience in representation, the buses in the attached drawings of this application are not limited to only one bus or one type of bus.
[0158] The above storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random-Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read-only Memory (EPROM), Programmable Read-Only Memory (PROM), Read-Only Memory (ROM), magnetic memory, flash memory, a magnetic disk, or an optical disk. The storage medium can be any available medium that can be accessed by a general-purpose or special-purpose computer.
[0159] An exemplary storage medium is coupled to the processor, enabling the processor to read information from the storage medium and write information to the storage medium. Of course, the storage medium can also be a component of the processor. The processor and the storage medium can be located in an Application Specific Integrated Circuits (ASIC). Of course, the processor and the storage medium can also exist as discrete components in an electronic device or a master control device.
[0160] Those of ordinary skill in the art can understand that all or part of the steps of implementing the above method embodiments can be completed by hardware related to program instructions. The foregoing program can be stored in a computer-readable storage medium. When the program is executed, it performs the steps including the above method embodiments; and the foregoing storage medium includes various media such as ROM, RAM, magnetic disks, or optical disks that can store program codes.
[0161] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than limiting them; although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements on some or all of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present application.
Claims
1. An optimization processing method for an audit model, characterized in that: Applicable to computer equipment, including: Obtaining any audit report data, wherein the audit report data includes an audit question list and an audit report; Inputting the audit question list and preset prompt words into each trained audit model for inference processing to obtain a plurality of first inference audit reports, wherein each trained audit model is obtained by training according to each preset audit model; Comparing each first reasoning audit report with the audit report to obtain a plurality of comparison values; Determining a target audit model from the trained audit models according to the comparison values; The target audit model is output, wherein the target audit model is used to generate an audit report based on the input list of audit issues to be processed.
2. The method according to claim 1, characterized in that The training process of each preset audit model includes: Obtain all historical audit report data within a preset time; Screening all the historical audit report data to obtain a plurality of screened historical audit report data; Determine whether the number of historical audit report data after the screening is less than a first preset threshold; If it is determined that the number of the filtered historical audit report data is not less than a first preset threshold, a portion of the plurality of historical audit report data is determined as a training set; Each preset audit model is trained according to the training set and the preset training method to obtain each trained audit model.
3. The method according to claim 2, characterized in that The historical audit report data includes historical audit question lists and historical audit reports; Accordingly, the filtering of all the historical audit report data to obtain a plurality of filtered historical audit report data includes: Determine the historical audit question lists and historical audit reports corresponding to all the historical audit report data; Screening each of the historical audit question lists and each of the historical audit reports to obtain each screened historical audit question list and each screened historical audit report; The screened historical audit question lists and the screened historical audit reports are determined as corresponding multiple screened historical audit report data.
4. The method according to claim 3, characterized in that: The screening of each historical audit question list and each historical audit report to obtain each screened historical audit question list and each screened historical audit report includes: Input any historical audit question in any historical audit question list and the corresponding historical audit report into a preset general audit model for processing to determine whether the historical audit question matches the historical audit report; If the historical audit question matches the historical audit report, the historical audit question is determined as the screened historical audit question; if the historical audit question does not match the historical audit report, the historical audit question is filtered out; Determine each screened historical audit question and the historical audit report as a screened historical audit question list and a screened historical audit report; The remaining historical audit question lists and historical audit reports are traversed to obtain the screened historical audit question lists and historical audit reports.
5. The method according to claim 1, characterized in that Determining a target audit model from the trained audit models according to the comparison values includes: Screening each comparison value according to a second preset threshold value to obtain one or more screened comparison values; Determining one or more corresponding first reasoning audit reports from the first reasoning audit reports according to the one or more filtered comparison values; In response to the auditor's evaluation operation on the one or more first reasoning audit reports, generating a corresponding evaluation report; Determining a target reasoning audit report from among the one or more first reasoning audit reports according to the evaluation report; According to the target inference audit report, a corresponding target audit model is determined from the trained audit models.
6. The method according to claim 2, characterized in that The historical audit report data includes historical audit question lists and historical audit reports; Accordingly, after determining whether the amount of all filtered historical audit report data is less than a first preset threshold, the method further includes: If it is determined that the number corresponding to the filtered historical audit report data is less than the first preset threshold, then obtaining each type of historical audit question in each historical audit question list; Generate a list of corresponding audit questions according to the various types of historical audit questions; Inputting the audit question lists and preset prompt words into a preset universal audit model for inference processing to obtain a corresponding second inference audit report; In response to the auditor's adjustment operation on each second reasoning audit report, generating each adjusted audit report; Generate corresponding combined audit report data according to each audit question list and each adjusted audit report; The combined audit report data is integrated into the plurality of filtered historical audit report data to complete the supplementation of the plurality of filtered historical audit report data.
7. The method according to any one of claims 2 to 4, characterized in that: Also includes: Determine the remaining part of the plurality of historical audit report data as a test set; The test set is input into each trained audit model for test processing to complete the test of each trained audit model.
8. An optimization processing device for an audit model, characterized in that: Applicable to computer equipment, including: A first acquisition module is used to acquire any audit report data, wherein the audit report data includes an audit question list and an audit report; A first processing module, used for inputting the audit question list and preset prompt words into each trained audit model for inference processing to obtain a plurality of first inference audit reports, wherein each trained audit model is obtained by training according to each preset audit model; A comparison module, used for comparing each first reasoning audit report with the audit report to obtain a plurality of comparison values; A first determination module is used to determine a target audit model from the trained audit models according to the comparison values; An output module is used to output the target audit model, wherein the target audit model is used to generate an audit report based on the input list of audit issues to be processed.
9. A computer device, characterized in that: include: at least one processor and memory; The memory stores computer-executable instructions; The at least one processor executes the computer-executable instructions stored in the memory, so that the at least one processor performs the optimization processing method of the audit model according to any one of claims 1 to 7.
10. A computer storage medium, characterized in that: The computer storage medium stores computer execution instructions. When the processor executes the computer execution instructions, the optimization processing method of the audit model as described in any one of claims 1 to 7 is implemented.