Large-model government affair service intelligent management method, system, equipment and medium

Abstract

The invention discloses a large-model government affair service intelligent management method, system and device and a medium, belongs to the technical field of artificial intelligence, and aims to solve the technical problem of how to improve the automation and intelligence degree of government affair service by using a large-model technology, improve the government affair service efficiency and further improve the user experience. According to the technical scheme, the method comprises the steps of government affair service data collection, wherein specific affair handling items and public common affair handling item original data are collected from multiple channels; government affair service original data treatment: carrying out data cleaning on the collected original data, removing noise and abnormal values in the original data, obtaining cleaned data, and carrying out affair handling item classification on the cleaned data; training a pre-training large model: training the pre-training large model by using the mass processed data, so that the pre-training large model continuously learns and evolves, and obtaining the trained pre-training large model to meet the requirements of government affair services; and processing the government affair problem and giving a processing result.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and specifically to a large-scale model of intelligent management method, system, equipment and medium for government service. Background Art

[0002] With the full arrival of the digital age and the accelerated advancement of informatization, the importance of government service halls as a key interactive platform between government and the public is increasing day by day. However, the limitations of the traditional government service model are gradually emerging. The inefficiency of manual processing, information lag and cumbersome service processes have become bottlenecks restricting the improvement of government service efficiency.

[0003] The mature application of large-scale model technology, especially large-scale pre-trained language models, has opened up a new path for the intelligent transformation of government services. Through deep learning and natural language processing technology, these models can demonstrate unprecedented capabilities in understanding and responding to public needs. They can not only quickly parse complex user inquiries and provide accurate service guidance, but also mine valuable information from massive data, provide data-driven insights for decision makers, and help optimize policies and reengineer processes.

[0004] Therefore, how to use big model technology to improve the automation and intelligence of government services, improve the efficiency of government services, and thus improve user experience is a technical problem that needs to be solved urgently. Summary of the invention

[0005] The technical task of the present invention is to provide a large-model government service intelligent management method, system, equipment and medium to solve the problem of how to use big model technology to improve the automation and intelligence of government services, improve the efficiency of government services, and thus improve user experience.

[0006] The technical task of the present invention is achieved in the following way: a large-scale model of government service intelligent management method, the method is as follows:

[0007] Government service data collection: collect original data on specific services and common services for the public from multiple channels;

[0008] Governance of government service raw data: clean the collected raw data, remove noise and outliers in the raw data, obtain the cleaned data, and then classify the cleaned data into service items;

[0009] Pre-trained large model training: Use massive processed data to train the pre-trained large model, so that the pre-trained large model can continuously learn and evolve, and obtain the trained pre-trained large model to meet the needs of government services;

[0010] Handle government affairs issues and provide results: Handle government affairs issues through the trained pre-trained large model and provide feedback on the results.

[0011] As the preferred option, specific services refer to the collection of all specific services that can be handled in the government service hall; among them, the specific service items such as certificate processing, tax consultation and social security inquiry can be provided by the administrative approval bureau at the specific application location. The centralized provision of this series of services aims to achieve "one-stop" government services, allowing data to run more and the public to run less, which greatly facilitates the public and enterprises and improves the convenience and satisfaction of government services;

[0012] Common matters for the public refer to matters that are most frequently consulted and handled by the public, which are collected and analyzed by government service agencies through various channels, so as to optimize resource allocation and service processes and give priority to high-frequency demands. For example, by designing and distributing questionnaires, first-hand information can be obtained directly from service recipients to understand their specific pain points and high-frequency demands in government services. At the same time, data analysis using social media platforms, such as monitoring public discussions and hot topics on social networks such as Weibo, WeChat, and forums, can provide insights into the public's real-time needs and potential expectations, especially when there is a surge in service demand during special periods such as policy changes, holidays, etc. The collection and analysis of this information will help government service agencies to adjust service priorities in a timely manner, optimize service supply, and ensure that resources can be effectively invested in matters that the public is most concerned about, thereby improving overall service efficiency and public satisfaction. In addition, by establishing a regular feedback mechanism, such as an online evaluation system and a customer service hotline, public opinions can be continuously collected to achieve continuous improvement of services and ensure that government services always meet the actual needs of the public.

[0013] As a preferred method, data cleaning is as follows:

[0014] Remove duplicate data: Purify the original government service data to ensure the uniqueness of each record, thereby eliminating redundancy, avoiding analysis bias, and improving the accuracy of pre-trained large model training and decision-making;

[0015] Filling missing values: By using statistical methods such as mean and median filling or using regression and KNN interpolation prediction algorithms, the gaps in the original government data can be effectively filled to ensure the integrity of the data and the reliability of the analysis;

[0016] Handling outliers: This involves identifying extreme values ​​in the data set that are obviously inconsistent with the overall trend, judging their rationality through statistical analysis or domain knowledge, and then correcting or eliminating outliers to reduce data noise and improve the accuracy of data analysis and the predictive performance of the model.

[0017] As a preferred option, the categories of matters to be handled are as follows:

[0018] Task category decomposition: complex tasks are divided into small units and classified according to specific situations, which is convenient for the pre-trained large model to understand and process;

[0019] To-do item coding: Assign a unique code to each to-do item to facilitate subsequent data management and model training.

[0020] As a preference, the pre-trained large model training is as follows:

[0021] Model and fine-tuning method selection, as follows:

[0022] Pre-trained model selection: Given that government services cover a wide range, from regulatory consultation to business processing, each service may involve a large amount of text understanding and logical reasoning; therefore, a pre-trained large model with strong language understanding and generation capabilities is selected, and the pre-trained large model is pre-trained on large-scale text data, with rich language knowledge and context understanding capabilities, laying a solid foundation for subsequent fine-tuning;

[0023] Fine-tuning strategy determination: In view of the characteristics of government services, such as high precision requirements, dense professional terms, and frequent policy changes, it is necessary to select fine-tuning strategies such as transfer learning, incremental learning, and joint learning to ensure that the pre-trained large model can accurately capture the special needs of the government service field while maintaining generalization capabilities;

[0024] Develop a training strategy as follows:

[0025] Data preprocessing: Before training, the data is carefully preprocessed by text cleaning, word segmentation, labeling, and balancing to ensure the quality and consistency of the input data;

[0026] Training parameter setting: According to the characteristics of the pre-trained large model and data characteristics, set the training parameters of learning rate, batch size, and number of iterations to promote effective learning of the pre-trained large model;

[0027] Selection of evaluation indicators: Define the evaluation indicators of accuracy, recall rate and F1 score to measure the performance of the pre-trained large model in the government service scenario, and ensure the clarity and quantification of the training objectives;

[0028] Training and iteration are as follows:

[0029] Initial training: Use the preprocessed data set to perform initial training on the selected pre-trained large model and observe the performance of the pre-trained large model on the benchmark data;

[0030] Performance monitoring: During the training process, the performance indicators of the pre-trained large model are continuously monitored, and the training strategy is adjusted in time to prevent overfitting or underfitting. The training strategy includes dynamically adjusting the learning rate and increasing regularization.

[0031] Iterative optimization: Based on the performance of the pre-trained large model on the validation set, multiple iterative trainings are performed until the pre-trained large model reaches the expected performance standards in key indicators, ensuring the robustness and accuracy of the pre-trained large model in government service scenarios;

[0032] Verification and testing, as follows:

[0033] Cross-validation: Use methods such as k-fold cross-validation to evaluate the performance of the pre-trained large model on different data subsets to ensure the generalization ability of the pre-trained large model;

[0034] Real-world scenario testing: In a controlled environment, the pre-trained large model is tested in real government service scenarios to verify its stability and effectiveness in actual applications;

[0035] Pre-trained large model deployment and monitoring, as follows:

[0036] Model deployment: Deploy the pre-trained large model that has been trained and verified to the production environment for the government service system to call;

[0037] Continuous monitoring: After deployment, continuously monitor the operating status and performance indicators of the pre-trained large model, and regularly retrain or fine-tune the pre-trained large model to adapt to the latest changes and needs in the government service field.

[0038] Preferably, the problem is handled as follows:

[0039] Pre-trained big model processes questions: When a user raises a service demand, the pre-trained big model first parses the question stem into structured language according to the classification of the service items. When the user raises an ambiguous question, the pre-trained big model provides options for the user to choose from based on the missing parts in the structure until the structure is complete.

[0040] Query the database based on the field: Based on the structured language of the question stem decomposed by the pre-trained large model, the relevant database is automatically queried to obtain the detailed information required for the service;

[0041] Pre-trained large model polishing stage: Use the pre-trained large model to integrate the queried information into easy-to-understand responses, and perform language polishing to ensure the accuracy and friendliness of the information.

[0042] A large-scale government service intelligent management system, the system comprising:

[0043] The data collection module is used to collect the original data of specific services and common services for the public from multiple channels;

[0044] The data governance module is used to clean the collected raw data, remove noise and outliers in the raw data, obtain the cleaned data, and then classify the cleaned data into service items;

[0045] The pre-trained large model training module is used to train the pre-trained large model with massive processed data, so that the pre-trained large model can continuously learn and evolve, and obtain the trained pre-trained large model to meet the needs of government services;

[0046] The government affairs problem processing module is used to process government affairs problems through the trained pre-trained large model and provide feedback on the processing results.

[0047] Preferably, the data governance module includes:

[0048] Data cleaning submodule, used to remove duplicate data, fill missing values, and handle outliers;

[0049] The task classification submodule is used to break down the tasks into small units, classify them according to specific circumstances, and assign a unique code to each task;

[0050] The pre-trained large model training module includes:

[0051] The model selection and fine-tuning submodule is used to select a pre-trained large model with strong language understanding and generation capabilities and formulate a fine-tuning strategy; the fine-tuning strategy includes transfer learning, incremental learning, or joint learning;

[0052] The training strategy formulation submodule is used to perform pre-processing operations such as text cleaning, word segmentation, labeling, and balancing on the data. Then, according to the characteristics of the pre-trained large model and data features, the training parameters of learning rate, batch size, and number of iterations are set, and the accuracy, recall rate, and F1 score evaluation indicators are defined to measure the performance of the pre-trained large model in the government service scenario, ensuring the clarity and quantification of the training objectives;

[0053] The training and iteration submodule is used to use the preprocessed data to perform initial training on the selected pre-trained large model. During the training process, the performance indicators of the pre-trained large model are continuously monitored. The training strategy is adjusted in time by dynamically adjusting the learning rate and increasing regularization to prevent overfitting or underfitting. Multiple iterative training is then performed until the pre-trained large model reaches the expected performance standard in terms of key indicators.

[0054] The verification and testing submodule is used to evaluate the performance of the pre-trained large model on different data subsets using the k-fold cross-validation method to ensure the generalization ability of the pre-trained large model. In addition, the pre-trained large model is tested in real government service scenarios under a controlled environment to verify its stability and effectiveness in actual applications.

[0055] The pre-trained large model deployment and monitoring submodule is used to deploy the pre-trained large model that has been trained and verified to the production environment for the government service system to call; and after deployment, continuously monitor the operating status and performance indicators of the pre-trained large model, and regularly retrain or fine-tune the pre-trained large model to adapt to the latest changes and needs in the field of government services;

[0056] The problem handling modules include:

[0057] The pre-trained large model processes the question stem submodule. When a user raises a service demand, the pre-trained large model first parses the question stem into structured language according to the classification of the service items. When the user raises an ambiguous question, the pre-trained large model provides options for the user to choose from based on the missing parts in the structure until the structure is completed.

[0058] The database query submodule is used to query the database based on the fields. That is, based on the structured language that is decomposed from the question stem based on the pre-trained large model, the relevant database is automatically queried to obtain the detailed information required for the service.

[0059] The pre-trained large model polishing sub-module is used to use the pre-trained large model to integrate the queried information into easy-to-understand responses and perform language polishing to ensure the accuracy and friendliness of the information.

[0060] An electronic device comprising: a memory and at least one processor;

[0061] Wherein, the memory stores computer-executable instructions;

[0062] The at least one processor executes the computer execution instructions stored in the memory, so that the at least one processor executes the intelligent management method of government services based on the large model as described above.

[0063] A computer-readable storage medium stores computer execution instructions. When a processor executes the computer execution, the intelligent management method of government service based on the large model as mentioned above is implemented.

[0064] The large-scale model of the government service intelligent management method, system, device and medium of the present invention have the following advantages:

[0065] (1) The present invention uses advanced large model technology to intelligently transform the service process of the government service hall. Through data collection, data governance, model training and problem handling, it achieves efficient response to the public's service needs, optimizes the government service experience, improves service efficiency, and effectively relieves the work pressure of staff;

[0066] (ii) The present invention focuses on the integrated application of large-scale pre-trained language models in government services. Through natural language processing, machine learning, knowledge graph construction and advanced human-computer interaction design, it aims to create an efficient, personalized and responsive government service platform that can accurately analyze user needs, optimize service processes, build a government knowledge base, and provide a seamless multimodal interactive experience, comprehensively promote the digital and intelligent transformation of government services, and achieve the service goal of "data runs, people save worry";

[0067] (III) Through continuous learning and self-optimization, the present invention can adapt to changing needs and ensure that government services always keep pace with the times; at the same time, intelligent government services can continuously adjust service strategies by analyzing user behavior and feedback, realize personalized service push, and improve user experience;

[0068] (IV) The present invention uses large-scale pre-trained language models to innovate government services, aiming to automate information retrieval, appointments, and form filling to improve efficiency; enhance user experience through precise natural language interaction; take over standardized tasks to reduce the burden on staff; use continuous learning and multiple verifications to improve the accuracy and reliability of services; and ultimately build a government service system that is both intelligent and efficient and can ensure legality and compliance, comprehensively improving public satisfaction and government service quality;

[0069] (V) The present invention can improve efficiency: through the automated processing capabilities of large models, it can realize functions such as rapid information retrieval, intelligent appointment scheduling, and automatic form filling, significantly shortening the public's waiting time and speeding up business processing;

[0070] (VI) The present invention improves user satisfaction: With the powerful natural language understanding and generation capabilities of the large model, it provides a personalized, high-quality service experience, ensuring that every interaction accurately meets the needs of the public, reducing misunderstandings and repeated communication, thereby improving satisfaction;

[0071] (VII) The present invention reduces the pressure on staff: the pre-trained large model can undertake a large number of standardized and highly repetitive tasks, such as answering common questions and preliminary data review, so that government service staff can focus on more complex and challenging tasks, effectively reducing work intensity and improving work efficiency;

[0072] (VIII) The present invention enhances accuracy and reliability: It is particularly emphasized that the present invention is committed to improving the accuracy and reliability of the service process. The pre-trained large model ensures the accuracy of information processing and reduces human errors through continuous learning and self-optimization; multiple verification mechanisms are set up in key links, such as legal clause analysis and policy applicability judgment, to ensure that the decision-making basis is correct;

[0073] (IX) The present invention realizes intelligent and efficient services: By integrating the above technologies, the present invention strives to build a highly intelligent and responsive government service, which can not only efficiently handle daily affairs, but also provide accurate guidance in complex situations, helping the public to smoothly complete various government affairs, while ensuring that each step of the operation is carried out within the framework of laws and regulations, thus ensuring the legality and standardization of the service;

[0074] (10) The data collection module of the present invention includes the collection of specific service items and the collection of common service items of the public to comprehensively cover the needs of government services; the data governance module ensures data quality and structuring through data cleaning and service item classification, which is convenient for big model understanding and learning; the pre-trained big model training module selects appropriate big models and fine-tuning strategies, and conducts personalized training based on the characteristics of government services; the problem handling module analyzes user needs through big models, automatically queries the database and integrates information, and provides accurate and friendly service feedback;

[0075] (XI) The present invention is used to realize the intelligent upgrade of government services. It can automatically process government service matters according to user needs, improve efficiency, reduce labor costs, and enhance user experience. It has self-learning and optimization capabilities, and can continuously improve service quality as the usage time increases. It can also collect and analyze user feedback to provide data support for the continuous improvement of government services. It has an integrated security mechanism to ensure the security and privacy protection of user data. BRIEF DESCRIPTION OF THE DRAWINGS

[0076] The present invention is further described below in conjunction with the accompanying drawings.

[0077] Attached Figure 1 The flowchart of the intelligent management method of government services for a large model. DETAILED DESCRIPTION

[0078] The following detailed description is made of the large-scale model of the government service intelligent management method, system, device and medium of the present invention with reference to the drawings and specific embodiments of the specification.

[0079] Embodiment 1:

[0080] As attached Figure 1 As shown, this embodiment provides a large-scale model of government service intelligent management method, the method is as follows:

[0081] S1. Government service data collection: collect original data on specific services and common services for the public from multiple channels;

[0082] S2. Governance of original data of government services: clean the collected original data, remove noise and outliers in the original data, obtain the cleaned data, and then classify the service items based on the cleaned data;

[0083] S3. Pre-trained large model training: Use massive processed data to train the pre-trained large model, so that the pre-trained large model can continuously learn and evolve, and obtain a trained pre-trained large model to meet the needs of government services;

[0084] S4. Process government affairs issues and give processing results: Process government affairs issues through the trained pre-trained large model and provide feedback on the processing results.

[0085] The specific services in step S1 of this embodiment refer to the collection of all specific services that can be handled in the government service hall; among them, the specific service items such as certificate processing, tax consultation and social security inquiry can be provided by the administrative approval bureau of the specific application location. The centralized provision of this series of services aims to achieve "one-stop" government services, allowing data to travel more and the public to run less, which greatly facilitates the public and enterprises and improves the convenience and satisfaction of government services.

[0086] The common matters for the public in step S1 of this embodiment refer to the matters that the public most frequently consults and handles, which are collected and analyzed by government service agencies through various channels, so as to optimize resource allocation and service processes and give priority to high-frequency demands. For example, by designing and distributing questionnaires, first-hand information can be directly obtained from service recipients to understand their specific pain points and high-frequency demands in government services. At the same time, data analysis can be performed using social media platforms, such as monitoring public discussions and hot topics on social networks such as Weibo, WeChat, and forums, so as to gain insight into the public's real-time demands and potential expectations, especially in the case of a surge in service demands during special periods such as policy changes and holidays. The collection and analysis of this information can help government service agencies to timely adjust service priorities, optimize service supply, and ensure that resources can be effectively invested in matters that the public is most concerned about, thereby improving overall service efficiency and public satisfaction. In addition, by establishing a normalized feedback mechanism, such as an online evaluation system and a customer service hotline, public opinions can be continuously collected to achieve continuous improvement of services and ensure that government services always meet the actual needs of the public.

[0087] The data cleaning in step S2 of this embodiment is specifically as follows:

[0088] S2-101. Remove duplicate data: Purify the original government service data to ensure the uniqueness of each record, thereby eliminating redundancy, avoiding analysis bias, and improving the accuracy of pre-trained large model training and decision-making;

[0089] S2-102, Filling missing values: By using statistical methods such as mean and median filling or using regression and KNN interpolation prediction algorithms, the gaps in the original government data can be effectively filled to ensure the integrity of the data and the reliability of the analysis;

[0090] S2-103. Handling outliers: This involves identifying extreme values ​​in the data set that are obviously inconsistent with the overall trend, judging their rationality through statistical analysis or domain knowledge, and then correcting or eliminating outliers to reduce data noise and improve the accuracy of data analysis and the predictive performance of the model.

[0091] The specific classification of the tasks in step S2 of this embodiment is as follows:

[0092] S2-201, Task Category Decomposition: Divide complex tasks into small units and classify them according to specific situations to facilitate understanding and processing by the pre-trained large model;

[0093] S2-202, Service Item Code: Assign a unique code to each service item to facilitate subsequent data management and model training.

[0094] The pre-training large model training in step S3 of this embodiment is specifically as follows:

[0095] S301, model and fine-tuning method selection, as follows:

[0096] S30101. Pre-training model selection: Given that government services cover a wide range, from regulatory consultation to business processing, each service may involve a large amount of text understanding and logical reasoning; therefore, a pre-training large model with strong language understanding and generation capabilities is selected, and the pre-training large model is pre-trained on large-scale text data, with rich language knowledge and context understanding capabilities, laying a solid foundation for subsequent fine-tuning;

[0097] S30102. Determine fine-tuning strategy: In view of the characteristics of government services, such as high precision requirements, dense professional terms, frequent policy changes, etc., it is necessary to select fine-tuning strategies such as transfer learning, incremental learning, and joint learning to ensure that the pre-trained large model can accurately capture the special needs of the government service field while maintaining generalization capabilities;

[0098] S302: Formulate training strategies, as follows:

[0099] S30201. Data preprocessing: Before training, the data is carefully preprocessed by text cleaning, word segmentation, labeling, and balancing to ensure the quality and consistency of the input data;

[0100] S30202. Training parameter setting: according to the characteristics of the pre-trained large model and data characteristics, set the training parameters of learning rate, batch size, and number of iterations to promote effective learning of the pre-trained large model;

[0101] S30203. Evaluation indicator selection: Define the evaluation indicators of accuracy, recall and F1 score to measure the performance of the pre-trained large model in the government service scenario and ensure the clarity and quantifiability of the training objectives;

[0102] S303, training and iteration, as follows:

[0103] S30301. Initial training: Use the preprocessed data set to perform initial training on the selected pre-trained large model and observe the performance of the pre-trained large model on the benchmark data;

[0104] S30302. Performance monitoring: During the training process, continuously monitor the performance indicators of the pre-trained large model and adjust the training strategy in a timely manner to prevent overfitting or underfitting. The training strategy includes dynamically adjusting the learning rate and increasing regularization.

[0105] S30303, Iterative Optimization: Based on the performance of the pre-trained large model on the validation set, multiple iterative trainings are performed until the pre-trained large model reaches the expected performance standards in key indicators, ensuring the robustness and accuracy of the pre-trained large model in the government service scenario;

[0106] S304, verification and testing, as follows:

[0107] S30401, Cross-validation: Use methods such as k-fold cross-validation to evaluate the performance of the pre-trained large model on different data subsets to ensure the generalization ability of the pre-trained large model;

[0108] S30402. Real-world scenario testing: In a controlled environment, the pre-trained large model is tested in real government service scenarios to verify its stability and effectiveness in actual applications;

[0109] S305, deployment and monitoring of pre-trained large models, as follows:

[0110] S30501. Model deployment: Deploy the pre-trained large model that has been trained and verified to the production environment for the government service system to call;

[0111] S30502. Continuous monitoring: After deployment, continuously monitor the operating status and performance indicators of the pre-trained large model, and regularly retrain or fine-tune the pre-trained large model to adapt to the latest changes and needs in the government service field.

[0112] The problem handling in step S4 of this embodiment is specifically as follows:

[0113] S401. Processing the question stem with the pre-trained large model: When a user raises a service demand, the pre-trained large model first parses the question stem into structured language according to the classification of the service items. When the user raises an ambiguous question, the pre-trained large model provides options for the user to choose according to the missing parts in the structure until the structure is completed.

[0114] S402, query the database according to the field: based on the structured language disassembled from the question stem by the pre-trained large model, automatically query the relevant database to obtain the detailed information required for the service;

[0115] S403, pre-trained large model polishing stage: use the pre-trained large model to integrate the queried information into easy-to-understand responses, and perform language polishing to ensure the accuracy and friendliness of the information.

[0116] Embodiment 2:

[0117] This embodiment provides a large-scale government service intelligent management system, which includes:

[0118] The data collection module is used to collect the original data of specific services and common services for the public from multiple channels;

[0119] The data governance module is used to clean the collected raw data, remove noise and outliers in the raw data, obtain the cleaned data, and then classify the cleaned data into service items;

[0120] The pre-trained large model training module is used to train the pre-trained large model with massive processed data, so that the pre-trained large model can continuously learn and evolve, and obtain the trained pre-trained large model to meet the needs of government services;

[0121] The government affairs problem processing module is used to process government affairs problems through the trained pre-trained large model and provide feedback on the processing results.

[0122] The data governance module in this embodiment includes:

[0123] Data cleaning submodule, used to remove duplicate data, fill missing values, and handle outliers;

[0124] The task classification submodule is used to break down the tasks into small units, classify them according to specific circumstances, and assign a unique code to each task.

[0125] The pre-trained large model training module in this embodiment includes:

[0126] The model selection and fine-tuning submodule is used to select a pre-trained large model with strong language understanding and generation capabilities and formulate a fine-tuning strategy; the fine-tuning strategy includes transfer learning, incremental learning, or joint learning;

[0127] The training strategy formulation submodule is used to perform pre-processing operations such as text cleaning, word segmentation, labeling, and balancing on the data. Then, according to the characteristics of the pre-trained large model and data features, the training parameters of learning rate, batch size, and number of iterations are set, and the accuracy, recall rate, and F1 score evaluation indicators are defined to measure the performance of the pre-trained large model in the government service scenario, ensuring the clarity and quantification of the training objectives;

[0128] The training and iteration submodule is used to use the preprocessed data to perform initial training on the selected pre-trained large model. During the training process, the performance indicators of the pre-trained large model are continuously monitored. The training strategy is adjusted in time by dynamically adjusting the learning rate and increasing regularization to prevent overfitting or underfitting. Multiple iterative training is then performed until the pre-trained large model reaches the expected performance standard in terms of key indicators.

[0129] The verification and testing submodule is used to evaluate the performance of the pre-trained large model on different data subsets using the k-fold cross-validation method to ensure the generalization ability of the pre-trained large model. In addition, the pre-trained large model is tested in real government service scenarios under a controlled environment to verify its stability and effectiveness in actual applications.

[0130] The pre-trained large model deployment and monitoring sub-module is used to deploy the pre-trained and verified large models to the production environment for the government service system to call; and after deployment, continuously monitor the operating status and performance indicators of the pre-trained large models, and regularly retrain or fine-tune the pre-trained large models to adapt to the latest changes and needs in the government service field.

[0131] The problem processing module in this embodiment includes:

[0132] The pre-trained large model processes the question stem submodule. When a user raises a service demand, the pre-trained large model first parses the question stem into structured language according to the classification of the service items. When the user raises an ambiguous question, the pre-trained large model provides options for the user to choose from based on the missing parts in the structure until the structure is completed.

[0133] The database query submodule is used to query the database based on the fields. That is, based on the structured language that is decomposed from the question stem based on the pre-trained large model, the relevant database is automatically queried to obtain the detailed information required for the service.

[0134] The pre-trained large model polishing sub-module is used to use the pre-trained large model to integrate the queried information into easy-to-understand responses and perform language polishing to ensure the accuracy and friendliness of the information.

[0135] Embodiment 3:

[0136] This embodiment also provides an electronic device, including: a memory and at least one processor;

[0137] Wherein, the memory stores computer-executable instructions;

[0138] The at least one processor executes the computer execution instructions stored in the memory, so that the at least one processor executes the large-model intelligent management method for government services described in any one of the present inventions.

[0139] The processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The processor may be a microprocessor or any conventional processor, etc.

[0140] The memory can be used to store computer programs and / or modules. The processor realizes various functions of the electronic device by running or executing the computer programs and / or modules stored in the memory, and calling the data stored in the memory. The memory can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, at least one application required for a function, etc.; the data storage area can store data created according to the use of the terminal, etc. In addition, the memory can also include a high-speed random access memory, and can also include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart memory card (SMC), a secure digital (SD) card, a flash memory card, at least one disk storage period, a flash memory device, or other volatile solid-state storage devices.

[0141] Embodiment 4:

[0142] This embodiment also provides a computer-readable storage medium, in which a plurality of instructions are stored, and the instructions are loaded by a processor, so that the processor executes the large-scale model of the government service intelligent management method in any embodiment of the present invention. Specifically, a system or device equipped with a storage medium can be provided, on which the software program code that implements the functions of any of the above embodiments is stored, and the computer (or CPU or MPU) of the system or device reads and executes the program code stored in the storage medium.

[0143] In this case, the program code itself read from the storage medium can realize the function of any one of the above-mentioned embodiments, and thus the program code and the storage medium storing the program code constitute a part of the present invention.

[0144] The storage medium embodiments for providing the program code include a floppy disk, a hard disk, a magneto-optical disk, an optical disk (such as CD-ROM, CD-R, CD-RW, DVD-ROM, DVD-RAM, DVD-RW, DVD+RW), a magnetic tape, a non-volatile memory card, and a ROM. Alternatively, the program code can be downloaded from a server computer by a communication network.

[0145] In addition, it should be clear that the functions of any of the above embodiments can be implemented not only by executing the program code read by the computer, but also by enabling an operating system operating on the computer to complete part or all of the actual operations based on instructions from the program code.

[0146] In addition, it can be understood that the program code read from the storage medium is written to a memory provided in an expansion board inserted into the computer or written to a memory provided in an expansion unit connected to the computer, and then based on the instructions of the program code, a CPU installed on the expansion board or the expansion unit is enabled to perform part or all of the actual operations, thereby realizing the functions of any of the above-mentioned embodiments.

[0147] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, 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 invention.

Claims

1. A large-scale model of government service intelligent management method, characterized in that: The method is as follows: Government service data collection: collect original data on specific services and common services for the public from multiple channels; Governance of government service raw data: clean the collected raw data, remove noise and outliers in the raw data, obtain the cleaned data, and then classify the cleaned data into service items; Pre-trained large model training: Use massive processed data to train the pre-trained large model, so that the pre-trained large model can continuously learn and evolve, and obtain the trained pre-trained large model to meet the needs of government services; Handle government affairs issues and provide results: Handle government affairs issues through the trained pre-trained large model and provide feedback on the results.

2. The large-scale model-based intelligent management method for government services according to claim 1 is characterized in that: Specific matters refer to the collection of all specific matters that can be handled in the government service hall; among them, specific matters include certificate processing, tax consultation and social security inquiry; Common matters for the public refer to matters that are most frequently consulted and handled by the public, which are collected and analyzed by government service agencies through various channels, with the goal of optimizing resource allocation and service processes and giving priority to handling high-frequency demands.

3. The large-scale model-based intelligent management method for government services according to claim 1 is characterized in that: The data cleaning is as follows: Remove duplicate data: Purify the original government service data to ensure the uniqueness of each record, thereby eliminating redundancy, avoiding analysis bias, and improving the accuracy of pre-trained large model training and decision-making; Filling missing values: By using statistical methods such as mean and median filling or using regression and KNN interpolation prediction algorithms, the gaps in the original government data can be effectively filled to ensure the integrity of the data and the reliability of the analysis; Handling outliers: This involves identifying extreme values ​​in the data set that are obviously inconsistent with the overall trend, judging their rationality through statistical analysis or domain knowledge, and then correcting or eliminating outliers to reduce data noise and improve the accuracy of data analysis and the predictive performance of the model.

4. The large-scale model-based intelligent management method for government services according to claim 1 is characterized in that: The specific categories of matters are as follows: Task category decomposition: complex tasks are divided into small units and classified according to specific situations, which is convenient for the pre-trained large model to understand and process; To-do item coding: Assign a unique code to each to-do item to facilitate subsequent data management and model training.

5. The large-scale model-based intelligent management method for government services according to claim 1 is characterized in that: The specific training of the pre-trained large model is as follows: Model and fine-tuning method selection, as follows: Pre-trained model selection: Select a pre-trained large model with strong language understanding and generation capabilities, and pre-train the pre-trained large model on large-scale text data to have rich language knowledge and context understanding capabilities, laying a solid foundation for subsequent fine-tuning; Fine-tuning strategy determination: Select fine-tuning strategies such as transfer learning, incremental learning, and joint learning to ensure that the pre-trained large model can accurately capture the special needs of the government service field while maintaining generalization capabilities; Develop a training strategy as follows: Data preprocessing: Before training, the data is carefully preprocessed by text cleaning, word segmentation, labeling, and balancing to ensure the quality and consistency of the input data; Training parameter setting: According to the characteristics of the pre-trained large model and data characteristics, set the training parameters of learning rate, batch size, and number of iterations to promote effective learning of the pre-trained large model; Selection of evaluation indicators: Define the evaluation indicators of accuracy, recall and F1 score to measure the performance of the pre-trained large model in the government service scenario, ensuring the clarity and quantification of the training objectives; Training and iteration are as follows: Initial training: Use the preprocessed data set to perform initial training on the selected pre-trained large model and observe the performance of the pre-trained large model on the benchmark data; Performance monitoring: During the training process, the performance indicators of the pre-trained large model are continuously monitored, and the training strategy is adjusted in time to prevent overfitting or underfitting. The training strategy includes dynamically adjusting the learning rate and increasing regularization. Iterative optimization: Based on the performance of the pre-trained large model on the validation set, multiple iterative trainings are performed until the pre-trained large model reaches the expected performance standards in key indicators, ensuring the robustness and accuracy of the pre-trained large model in government service scenarios; Verification and testing, as follows: Cross-validation: Use methods such as k-fold cross-validation to evaluate the performance of the pre-trained large model on different data subsets to ensure the generalization ability of the pre-trained large model; Real-world scenario testing: In a controlled environment, the pre-trained large model is tested in real government service scenarios to verify its stability and effectiveness in actual applications; Pre-trained large model deployment and monitoring, as follows: Model deployment: Deploy the pre-trained large model that has been trained and verified to the production environment for the government service system to call; Continuous monitoring: After deployment, continuously monitor the operating status and performance indicators of the pre-trained large model, and regularly retrain or fine-tune the pre-trained large model to adapt to the latest changes and needs in the government service field.

6. The method for intelligent management of government services based on a large model according to any one of claims 1 to 5, characterized in that: The problem is solved as follows: Pre-trained big model processes questions: When a user raises a service demand, the pre-trained big model first parses the question stem into structured language according to the classification of the service items. When the user raises an ambiguous question, the pre-trained big model provides options for the user to choose from based on the missing parts in the structure until the structure is complete. Query the database based on the field: Based on the structured language of the question stem decomposed by the pre-trained large model, the relevant database is automatically queried to obtain the detailed information required for the service; Pre-trained large model polishing stage: Use the pre-trained large model to integrate the queried information into easy-to-understand responses, and perform language polishing to ensure the accuracy and friendliness of the information.

7. A large-scale government service intelligent management system, characterized in that: The system includes: The data collection module is used to collect the original data of specific services and common services for the public from multiple channels; The data governance module is used to clean the collected raw data, remove noise and outliers in the raw data, obtain the cleaned data, and then classify the cleaned data into service items; The pre-trained large model training module is used to train the pre-trained large model with massive processed data, so that the pre-trained large model can continuously learn and evolve, and obtain the trained pre-trained large model to meet the needs of government services; The government affairs problem processing module is used to process government affairs problems through the trained pre-trained large model and provide feedback on the processing results.

8. The large-scale model-based intelligent government service management system according to claim 7 is characterized in that: The data governance modules include: Data cleaning submodule, used to remove duplicate data, fill missing values, and handle outliers; The task classification submodule is used to break down the tasks into small units, classify them according to specific circumstances, and assign a unique code to each task; The pre-trained large model training module includes: The model selection and fine-tuning submodule is used to select a pre-trained large model with strong language understanding and generation capabilities and formulate a fine-tuning strategy; the fine-tuning strategy includes transfer learning, incremental learning, or joint learning; The training strategy formulation submodule is used to perform pre-processing operations such as text cleaning, word segmentation, labeling, and balancing on the data. Then, according to the characteristics of the pre-trained large model and data features, the training parameters of learning rate, batch size, and number of iterations are set, and the accuracy, recall rate, and F1 score evaluation indicators are defined to measure the performance of the pre-trained large model in the government service scenario, ensuring the clarity and quantification of the training objectives; The training and iteration submodule is used to use the preprocessed data to perform initial training on the selected pre-trained large model. During the training process, the performance indicators of the pre-trained large model are continuously monitored. The training strategy is adjusted in time by dynamically adjusting the learning rate and increasing regularization to prevent overfitting or underfitting. Multiple iterative training is then performed until the pre-trained large model reaches the expected performance standard in terms of key indicators. The verification and testing submodule is used to evaluate the performance of the pre-trained large model on different data subsets using the k-fold cross-validation method to ensure the generalization ability of the pre-trained large model. In addition, the pre-trained large model is tested in real government service scenarios under a controlled environment to verify its stability and effectiveness in actual applications. The pre-trained large model deployment and monitoring submodule is used to deploy the pre-trained large model that has been trained and verified to the production environment for the government service system to call; and after deployment, continuously monitor the operating status and performance indicators of the pre-trained large model, and regularly retrain or fine-tune the pre-trained large model to adapt to the latest changes and needs in the field of government services; The problem handling modules include: The pre-trained large model processes the question stem submodule. When a user raises a service demand, the pre-trained large model first parses the question stem into structured language according to the classification of the service items. When the user raises an ambiguous question, the pre-trained large model provides options for the user to choose from based on the missing parts in the structure until the structure is completed. The database query submodule is used to query the database based on the fields. That is, based on the structured language that is decomposed from the question stem based on the pre-trained large model, the relevant database is automatically queried to obtain the detailed information required for the service. The pre-trained large model polishing sub-module is used to use the pre-trained large model to integrate the queried information into easy-to-understand responses and perform language polishing to ensure the accuracy and friendliness of the information.

9. An electronic device, characterized in that: include: memory and at least one processor; Wherein, the memory stores computer-executable instructions; The at least one processor executes the computer execution instructions stored in the memory, so that the at least one processor executes the large model government service intelligent management method as described in any one of claims 1 to 6.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer execution instructions, and when the processor executes the computer execution, the large-model intelligent management method for government services as described in any one of claims 1 to 6 is implemented.

Images