A government service data processing method and system based on artificial intelligence

Through the government service data processing method based on artificial intelligence, using Gaussian hybrid model, EM algorithm, t-SNE algorithm and other technologies, the problem of inefficient processing of traditional government data is solved, and the efficient processing and intelligence of government data is realized, and the improvement of the efficiency of government data is improved. Data utilization rate provides scientific basis for government decision-making.

CN118656770BActive Publication Date: 2024-11-12SHANDONG DADAOZHILI EDUCATION TECH GRP CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411147377.X
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-08-21
Publication Date
2024-11-12
Estimated Expiration
2044-08-21

AI Technical Summary

Technical Problem

Traditional government data processing relies on manual labor, resulting in low processing efficiency and insufficient data utilization, which cannot meet the needs of modern government services. The complexity and diversity of government data put higher requirements on data processing technology.

Method used

A government service data processing method based on artificial intelligence is designed, including data integration, preprocessing, outlier detection, dimensionality reduction processing and data analysis, using Gaussian hybrid model, EM algorithm, t-SNE algorithm, decision tree model and XGBoost Algorithms and other technologies conduct data prediction and risk warning, and push government analysis results.

Benefits of technology

It improves the efficiency and accuracy of government data processing, makes government services more intelligent and automated, reduces labor costs, and provides a scientific basis for decision-making.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118656770B_ABST
    Figure CN118656770B_ABST
Patent Text Reader

Abstract

The present invention relates to the field of government data management, and discloses a government service data processing method and system based on artificial intelligence. The method comprises the following steps: collecting government service data from various government service departments, integrating the government service data, and obtaining service data to be processed; pre-processing the service data to be processed, obtaining standardized data, inputting the standardized data into a Gaussian mixture model, and iteratively updating the Gaussian mixture model using an EM algorithm to output first sample service data; performing outlier detection on the first sample service data, obtaining second sample service data, and performing dimensionality reduction processing on the second sample service data based on a t-SNE algorithm to obtain target feature data; inputting the target feature data into a data analysis model, performing prediction through the data analysis model, and outputting government analysis results; pushing the government analysis results to corresponding government service departments, and displaying the government analysis results through data visualization technology; the present invention makes government services more intelligent and automated.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of government data management, and in particular to a government service data processing method and system based on artificial intelligence. Background Art

[0002] As the division of labor in public management services becomes increasingly detailed and complex, the amount of government data is growing explosively. Traditional data processing still relies on manual work, which can no longer meet the needs of modern government services. There are problems such as low processing efficiency and insufficient data utilization. At the same time, the complexity and diversity of government data also put forward higher requirements for data processing technology. Therefore, it is of great significance to study a government service data processing method and system based on artificial intelligence to improve the efficiency and accuracy of data processing. Summary of the invention

[0003] The purpose of the present invention is to solve the above problems and to design a government service data processing method and system based on artificial intelligence.

[0004] A first aspect of the present invention provides a government service data processing method based on artificial intelligence, the government service data processing method based on artificial intelligence comprising the following steps:

[0005] Collect government service data from various government service departments, and integrate the government service data to obtain service data to be processed;

[0006] Preprocessing the service data to be processed to obtain standardized data, inputting the standardized data into a Gaussian mixture model, and iteratively updating the Gaussian mixture model using an EM algorithm to output first sample service data;

[0007] Performing outlier detection on the first sample service data to obtain second sample service data, and performing dimensionality reduction processing on the second sample service data based on a t-SNE algorithm to obtain target feature data;

[0008] Inputting the target feature data into a data analysis model to perform prediction through the data analysis model and outputting government affairs analysis results, wherein the government affairs analysis results at least include service demand prediction, policy solution prediction and risk warning;

[0009] The government affairs analysis results are pushed to the corresponding government affairs service departments, and the government affairs analysis results are displayed through data visualization technology.

[0010] Optionally, in a first implementation manner of the first aspect of the present invention, preprocessing the service data to be processed to obtain standardized data, inputting the standardized data into a Gaussian mixture model, and iteratively updating the Gaussian mixture model using an EM algorithm to output sample service data includes:

[0011] Acquire the service data to be processed, and clean the service data to be processed to obtain the cleaned service data to be processed;

[0012] The cleaned service data to be processed is normalized by using Z-score standardization, and the data value range of the cleaned service data to be processed is scaled to [0, 1] or [-1, 1] to obtain standardized data;

[0013] Calculating the posterior probability from each Gaussian distribution for each data point of the standardized data, and obtaining the parameters of each Gaussian distribution based on the posterior probability using a weighted least squares method;

[0014] Repeat the E and M steps of the iterative EM algorithm until the algorithm converges, end the iteration, weight the posterior probability of the data points in each cluster, and then use the weighted data points as the first sample service data.

[0015] Optionally, in a second implementation of the first aspect of the present invention, performing outlier detection on the first sample service data to obtain second sample service data includes:

[0016] Determine the number of clusters of the K-means clustering algorithm and select k initial samples as the initial cluster centers;

[0017] Acquire first sample service data, calculate the Euclidean distance between each data point in the first sample service data and the initial cluster center, and assign each data point to a cluster to which the initial cluster center with the smallest distance belongs;

[0018] Re-determine the cluster center for the obtained cluster, recalculate until convergence, and determine whether the distance from the data point in the first sample service data to the cluster center is greater than a threshold. If the data point does not belong to the cluster or the distance is greater than the threshold, determine that the data point is an abnormal data point;

[0019] Abnormal data points in the first sample service data are eliminated to obtain second sample service data.

[0020] Optionally, in a third implementation of the first aspect of the present invention, performing dimensionality reduction processing on the second sample service data based on the t-SNE algorithm to obtain target feature data includes:

[0021] Get the second sample service data, set the t-SNE algorithm loss function, and initialize the t-SNE algorithm iteration number, learning rate, and momentum;

[0022] Calculating the conditional probabilities of the two data points in the second sample service data, calculating the joint probability distribution based on the conditional probabilities of the two data points, and performing normalization processing;

[0023] Randomly generate low-dimensional random numbers and calculate the joint probability distribution of low-dimensional sample space points in the t distribution. Use gradient descent to calculate the optimized gradient and get the current low-dimensional data. Iterate the loop so that the loss function reaches the minimum value and obtain the target feature data.

[0024] Optionally, in a fourth implementation of the first aspect of the present invention, the process of constructing the data analysis model includes the following steps:

[0025] Obtain historical service data of the government service department, split the historical service data into a training sample data set and a test sample data set, use the Bayesian method to fill in the missing sample data in the training sample data set, and discretize the sample data in the training sample data set;

[0026] Construct a decision tree model, input the processed training sample data set into the decision tree model for training, calculate the information gain of each attribute of the root node of the decision tree, select the attribute corresponding to the maximum value as the attribute of the root node, and then recursively calculate the attributes of other child nodes from top to bottom;

[0027] Add a Bayesian node between the parent node and the child node, check the attributes of the parent node, and prune the decision tree. Use a two-layer collaborative correction method to build the baseline layer and the real-time layer to obtain a decision tree model based on the two-layer XGBoost algorithm.

[0028] The test sample data set and the Pelican algorithm were used to optimize the parameters of the decision tree model based on the double-layer XGBoost algorithm and establish a data analysis model.

[0029] Optionally, in a fifth implementation of the first aspect of the present invention, the decision tree model based on the double-layer XGBoost algorithm is optimized by using the test sample data set and the Pelican algorithm to establish a data analysis model, including:

[0030] Set the population size, search space dimension and maximum number of iterations, initialize the pelican population, construct a fitness value function based on the test sample data set, calculate the fitness value of each individual in the initial pelican population, and record the position of the current best individual as the global optimal solution;

[0031] Carry out reverse differential evolution on the pelican population, compare the fitness values ​​of individuals before and after evolution, and select the best individuals to form the updated pelican population;

[0032] Recalculate the fitness values ​​of the updated individuals in the population, select the current best individual, and perform adaptive mutation on the best individual to update the position of the individuals in the pelican population;

[0033] Confirm whether the maximum number of iterations is met. If not, continue to iterate. If it is met, assign the optimal hyperparameters to the decision tree model based on the double-layer XGBoost algorithm, and use the optimal hyperparameters to establish a data analysis model.

[0034] Optionally, in a sixth implementation of the first aspect of the present invention, the target feature data is input into a data analysis model to perform prediction through the data analysis model, and output a government affairs analysis result, including:

[0035] Input the target feature data into the data analysis model, use the target feature data as the reference layer input of the data analysis model, make decisions along the tree structure, and finally reach a leaf node, calculate the reference prediction value, and obtain the output result of the reference layer;

[0036] The output results of the baseline layer are used as the input of the real-time layer, the input baseline prediction value is verified, the error prediction value is iteratively calculated, and the output results of the real-time layer are obtained. The output results of the baseline layer and the real-time layer are summarized to obtain the final prediction result to output the government analysis results.

[0037] The second aspect of the present invention provides a government service data processing system based on artificial intelligence, the government service data processing system based on artificial intelligence includes a data integration module, a data preprocessing module, a data dimension reduction module, a government analysis module and a result push module, wherein:

[0038] A data integration module is used to collect government service data from various government service departments, integrate the government service data, and obtain service data to be processed;

[0039] A data preprocessing module, used for preprocessing the service data to be processed to obtain standardized data, inputting the standardized data into a Gaussian mixture model, and iteratively updating the Gaussian mixture model using an EM algorithm to output first sample service data;

[0040] A data dimension reduction module, used to perform outlier detection on the first sample service data to obtain second sample service data, and perform dimension reduction processing on the second sample service data based on a t-SNE algorithm to obtain target feature data;

[0041] A government affairs analysis module, used for inputting the target characteristic data into a data analysis model to make predictions through the data analysis model, and outputting government affairs analysis results, wherein the government affairs analysis results at least include service demand predictions, policy solution predictions and risk warnings;

[0042] The result push module is used to push the government affairs analysis results to the corresponding government affairs service department and display the government affairs analysis results through data visualization technology.

[0043] Optionally, in a first implementation of the second aspect of the present invention, the data preprocessing module includes a data cleaning submodule, a normalization processing submodule, a first calculation submodule and an iteration submodule, wherein:

[0044] A data cleaning submodule is used to obtain the service data to be processed, and clean the service data to be processed to obtain the cleaned service data to be processed;

[0045] The normalization processing submodule is used to normalize the cleaned service data to be processed by using Z-score standardization, and scale the data value range of the cleaned service data to be processed to [0, 1] or [-1, 1] to obtain standardized data;

[0046] A first calculation submodule, configured to calculate a posterior probability from each Gaussian distribution for each data point of the standardized data, and obtain a parameter of each Gaussian distribution based on the posterior probability using a weighted least square method;

[0047] The iterative submodule is used to repeatedly iterate the E step and the M step of the EM algorithm until the algorithm converges, terminate the iteration, weight the posterior probability of the data points in each cluster, and then use the obtained weighted data points as the first sample service data.

[0048] Optionally, in a second implementation of the second aspect of the present invention, the data dimension reduction module includes a selection submodule, a second calculation submodule, a judgment submodule and a elimination submodule, wherein:

[0049] The selection submodule is used to determine the number of clusters of the K-means clustering algorithm and select k initial samples as the initial cluster centers;

[0050] A second calculation submodule is used to obtain the first sample service data, calculate the Euclidean distance from each data point in the first sample service data to the initial cluster center, and assign each data point to the cluster to which the initial cluster center with the smallest distance belongs;

[0051] A judgment submodule, used to redetermine the cluster center of the obtained cluster, recalculate until convergence, and judge whether the distance from the data point in the first sample service data to the cluster center is greater than a threshold. If the data point does not belong to the cluster or the distance is greater than the threshold, the data point is determined to be an abnormal data point;

[0052] The elimination submodule is used to eliminate abnormal data points in the first sample service data to obtain second sample service data.

[0053] In the technical solution provided by the present invention, government service data is collected from various government service departments, and data integration is performed on the government service data to obtain service data to be processed; the service data to be processed is preprocessed to obtain standardized data, and the standardized data is input into a Gaussian mixture model, and the Gaussian mixture model is iteratively updated using an EM algorithm to output first sample service data; outlier detection is performed on the first sample service data to obtain second sample service data, and dimension reduction processing is performed on the second sample service data based on the t-SNE algorithm to obtain target feature data; the target feature data is input into a data analysis model to perform prediction through the data analysis model and output a government analysis result; the government analysis result is pushed to the corresponding government service department, and the government analysis result is displayed through data visualization technology; the present invention makes government services more intelligent and automated, reduces labor costs, improves the efficiency and accuracy of government data processing, and provides a scientific basis for decision-making. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Various other advantages and benefits will become apparent to those of ordinary skill in the art by reading the following detailed description of the preferred embodiment.The drawings are only for the purpose of illustrating the preferred embodiments and are not to be construed as limiting the invention.

[0055] Figure 1 A schematic diagram of a first embodiment of a government service data processing method based on artificial intelligence provided by an embodiment of the present invention;

[0056] Figure 2 A schematic diagram of a second embodiment of the government service data processing method based on artificial intelligence provided by an embodiment of the present invention;

[0057] Figure 3 A schematic diagram of a third embodiment of the government service data processing method based on artificial intelligence provided by an embodiment of the present invention;

[0058] Figure 4 A schematic diagram of the structure of an artificial intelligence-based government service data processing system provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0059] The terms "first", "second", "third", "fourth", etc. (if any) in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "including" or "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, device, product or device that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0060] For ease of understanding, the specific process of the embodiment of the present invention is described below. Figure 1 A schematic diagram of a first embodiment of a method for processing government service data based on artificial intelligence provided by an embodiment of the present invention, the method specifically comprises the following steps:

[0061] Step 101: Collect government service data from various government service departments, integrate the government service data, and obtain service data to be processed;

[0062] Step 102: pre-process the service data to be processed to obtain standardized data, input the standardized data into a Gaussian mixture model, and iteratively update the Gaussian mixture model using an EM algorithm to output first sample service data;

[0063] In this embodiment, the service data to be processed is obtained, and the service data to be processed is cleaned to obtain the cleaned service data to be processed; the cleaned service data to be processed is normalized by using Z-score standardization, and the data value range of the cleaned service data to be processed is scaled to between [0, 1] or [-1, 1] to obtain standardized data; the posterior probability from each Gaussian distribution is calculated for each data point of the standardized data, and the parameters of each Gaussian distribution are obtained based on the posterior probability by using the weighted least squares method; the E step and the M step of the iterative EM algorithm are repeated until the algorithm converges, and the iteration is terminated, the posterior probability of the data points in each cluster is weighted, and the weighted data points obtained are used as the first sample service data.

[0064] Step 103: perform outlier detection on the first sample service data to obtain second sample service data, and perform dimensionality reduction processing on the second sample service data based on the t-SNE algorithm to obtain target feature data;

[0065] In this embodiment, the second sample service data is obtained, the t-SNE algorithm loss function is set, and the number of iterations, learning rate and momentum of the t-SNE algorithm are initialized; the conditional probabilities of the data points in the second sample service data are calculated, and the joint probability distribution is calculated based on the conditional probabilities of the data points and normalized; low-dimensional random numbers are randomly generated, and the joint probability distribution of the low-dimensional sample space points in the t distribution is calculated, and the optimized gradient is calculated by using gradient descent, and the current low-dimensional data is obtained. The loss function is iterated in a loop so as to reach the minimum value and obtain the target feature data.

[0066] Step 104: input the target feature data into the data analysis model to make predictions through the data analysis model, and output the government affairs analysis results, wherein the government affairs analysis results at least include service demand predictions, policy solution predictions, and risk warnings;

[0067] In this embodiment, the target feature data is input into the data analysis model, and the target feature data is used as the baseline layer input of the data analysis model. Decisions are made along the tree structure, and finally a leaf node is reached to calculate the baseline prediction value to obtain the output result of the baseline layer. The output result of the baseline layer is used as the input of the real-time layer, the input baseline prediction value is verified, the error prediction value is iteratively calculated, and the output result of the real-time layer is obtained. The output results of the baseline layer and the real-time layer are summarized to obtain the final prediction result to output the government analysis result.

[0068] In this embodiment, service demand forecasting can predict the changing trend of public demand for government services in the future, which helps to plan service resources in advance, optimize service processes, and improve service efficiency. Policy plan forecasting can predict the social, economic, and environmental impacts that may occur after the implementation of policies, which helps policymakers to have a more comprehensive understanding of the potential impact of policies and thus formulate more scientific and reasonable policy plans. Risk warning can warn and evaluate risks, and can quickly take measures to prevent risks from expanding or turning into actual hazards.

[0069] Step 105: Push the government affairs analysis results to the corresponding government affairs service department, and display the government affairs analysis results through data visualization technology.

[0070] In an embodiment of the present invention, government service data is collected from various government service departments, and data integration is performed on the government service data to obtain service data to be processed; the service data to be processed is preprocessed to obtain standardized data, and the standardized data is input into a Gaussian mixture model, and the Gaussian mixture model is iteratively updated using an EM algorithm to output first sample service data; outlier detection is performed on the first sample service data to obtain second sample service data, and dimension reduction processing is performed on the second sample service data based on the t-SNE algorithm to obtain target feature data; the target feature data is input into a data analysis model to perform prediction through the data analysis model, and output government analysis results; the government analysis results are pushed to the corresponding government service department, and the government analysis results are displayed through data visualization technology; the present invention makes government services more intelligent and automated, reduces labor costs, improves the efficiency and accuracy of government data processing, and provides a scientific basis for decision-making.

[0071] See also Figure 2 , a schematic diagram of a second embodiment of a government service data processing method based on artificial intelligence provided by an embodiment of the present invention, the method comprising:

[0072] Step 201, determine the number of clusters of the K-means clustering algorithm, and select initial k samples as initial cluster centers;

[0073] Step 202: Obtain the first sample service data, calculate the Euclidean distance from each data point in the first sample service data to the initial cluster center, and assign each data point to the cluster to which the initial cluster center with the smallest distance belongs;

[0074] Step 203: re-determine the cluster center for the obtained cluster, recalculate until convergence, and determine whether the distance from the data point in the first sample service data to the cluster center is greater than a threshold. If the data point does not belong to the cluster or the distance is greater than the threshold, determine that the data point is an abnormal data point;

[0075] Step 204: remove abnormal data points in the first sample service data to obtain second sample service data.

[0076] See also Figure 3 , a schematic diagram of a third embodiment of a government service data processing method based on artificial intelligence provided by an embodiment of the present invention, the method comprising:

[0077] Step 301: Obtain historical service data of the government service department, split the historical service data into a training sample data set and a test sample data set, use the Bayesian method to fill in the missing sample data in the training sample data set, and discretize the sample data in the training sample data set;

[0078] Step 302: construct a decision tree model, input the processed training sample data set into the decision tree model for training, calculate the information gain of each attribute of the root node of the decision tree, select the attribute corresponding to the maximum value as the attribute of the root node, and then recursively calculate the attributes of other child nodes from top to bottom;

[0079] Step 303: Add a Bayesian node between the parent node and the child node, check the attributes of the parent node, prune the decision tree, and use a two-layer collaborative correction method to build a baseline layer and a real-time layer to obtain a decision tree model based on a two-layer XGBoost algorithm.

[0080] In this embodiment, the population size, search space dimension and maximum number of iterations are set, the pelican population is initialized, a fitness value function is constructed based on a test sample data set, the fitness value of each individual in the initial pelican population is calculated, and the position of the current best individual is recorded as the global optimal solution; reverse differential evolution is performed on the pelican population, the fitness values ​​of the individuals before and after evolution are compared, and the best individuals are retained to form an updated pelican population; the fitness values ​​of the individuals in the updated population are recalculated, the current best individual is selected, and the best individual is adaptively mutated to update the position of the individuals in the pelican population; it is confirmed whether the maximum number of iterations is met, if not, the iteration is continued, if it is met, the optimal hyperparameters are assigned to a decision tree model based on the double-layer XGBoost algorithm, and a data analysis model is established using the optimal hyperparameters.

[0081] Step 304: Use the test sample data set and the Pelican algorithm to optimize the parameters of the decision tree model based on the double-layer XGBoost algorithm and establish a data analysis model.

[0082] See also Figure 4 , a schematic diagram of the structure of a government service data processing system based on artificial intelligence provided by an embodiment of the present invention, the system includes a data integration module, a data preprocessing module, a data dimension reduction module, a government service analysis module and a result push module, wherein,

[0083] The data integration module 401 is used to collect government service data from various government service departments, integrate the government service data, and obtain service data to be processed;

[0084] A data preprocessing module 402 is used to preprocess the service data to be processed to obtain standardized data, input the standardized data into a Gaussian mixture model, and iteratively update the Gaussian mixture model using an EM algorithm to output first sample service data;

[0085] The data dimension reduction module 403 is used to perform outlier detection on the first sample service data to obtain second sample service data, and perform dimension reduction processing on the second sample service data based on the t-SNE algorithm to obtain target feature data;

[0086] The government affairs analysis module 404 is used to input the target feature data into the data analysis model to make predictions through the data analysis model and output the government affairs analysis results, wherein the government affairs analysis results at least include service demand predictions, policy solution predictions and risk warnings;

[0087] The result push module 405 is used to push the government affairs analysis results to the corresponding government affairs service departments and display the government affairs analysis results through data visualization technology.

[0088] In this embodiment, the data preprocessing module includes a data cleaning submodule, a normalization processing submodule, a first calculation submodule and an iteration submodule, wherein:

[0089] A data cleaning submodule is used to obtain the service data to be processed, clean the service data to be processed, and obtain the cleaned service data to be processed;

[0090] The normalization processing submodule is used to normalize the cleaned service data to be processed by using Z-score standardization, and scale the data value range of the cleaned service data to be processed to [0, 1] or [-1, 1] to obtain standardized data;

[0091] A first calculation submodule is used to calculate the posterior probability from each Gaussian distribution for each data point of the standardized data, and obtain the parameters of each Gaussian distribution based on the posterior probability using a weighted least square method;

[0092] The iterative submodule is used to repeatedly iterate the E step and the M step of the EM algorithm until the algorithm converges, terminate the iteration, weight the posterior probability of the data points in each cluster, and then use the obtained weighted data points as the first sample service data.

[0093] In this embodiment, the data dimension reduction module includes a selection submodule, a second calculation submodule, a judgment submodule and a elimination submodule, wherein:

[0094] The selection submodule is used to determine the number of clusters of the K-means clustering algorithm and select k initial samples as the initial cluster centers;

[0095] A second calculation submodule is used to obtain the first sample service data, calculate the Euclidean distance between each data point in the first sample service data and the initial cluster center, and assign each data point to the cluster to which the initial cluster center with the smallest distance belongs;

[0096] A judgment submodule is used to redetermine the cluster center of the obtained cluster, recalculate until convergence, and judge whether the distance from the data point in the first sample service data to the cluster center is greater than a threshold. If the data point does not belong to the cluster or the distance is greater than the threshold, the data point is determined to be an abnormal data point;

[0097] The elimination submodule is used to eliminate abnormal data points in the first sample service data to obtain the second sample service data.

[0098] Through the implementation of the above scheme, the system includes a data integration module, a data preprocessing module, a data dimension reduction module, a government affairs analysis module and a result push module; the present invention makes government affairs services more intelligent and automated, reduces labor costs, improves the efficiency and accuracy of government affairs data processing, and provides a scientific basis for decision-making.

[0099] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited by the above embodiments. The above embodiments and descriptions are only preferred examples of the present invention and are not intended to limit the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. A government service data processing method based on artificial intelligence, characterized in that: The government service data processing method based on artificial intelligence includes the following steps: Collect government service data from various government service departments, and integrate the government service data to obtain service data to be processed; Preprocessing the service data to be processed to obtain standardized data, inputting the standardized data into a Gaussian mixture model, and iteratively updating the Gaussian mixture model using an EM algorithm to output first sample service data; Performing outlier detection on the first sample service data to obtain second sample service data, and performing dimensionality reduction processing on the second sample service data based on a t-SNE algorithm to obtain target feature data; Inputting the target feature data into a data analysis model to perform prediction through the data analysis model and outputting government affairs analysis results, wherein the government affairs analysis results at least include service demand prediction, policy solution prediction and risk warning; Pushing the government affairs analysis results to the corresponding government affairs service department, and displaying the government affairs analysis results through data visualization technology; Preprocessing the service data to be processed to obtain standardized data, inputting the standardized data into a Gaussian mixture model, and iteratively updating the Gaussian mixture model using an EM algorithm to output sample service data, including: Acquire the service data to be processed, and clean the service data to be processed to obtain the cleaned service data to be processed; The cleaned service data to be processed is normalized by using Z-score standardization, and the data value range of the cleaned service data to be processed is scaled to [0, 1] or [-1, 1] to obtain standardized data; Calculating the posterior probability from each Gaussian distribution for each data point of the standardized data, and obtaining the parameters of each Gaussian distribution based on the posterior probability using a weighted least squares method; Repeat the E and M steps of the EM algorithm until the algorithm converges, end the iteration, weight the posterior probability of the data points in each cluster, and then use the weighted data points as the first sample service data; Performing outlier detection on the first sample service data to obtain second sample service data includes: Determine the number of clusters of the K-means clustering algorithm and select k initial samples as the initial cluster centers; Acquire first sample service data, calculate the Euclidean distance between each data point in the first sample service data and the initial cluster center, and assign each data point to a cluster to which the initial cluster center with the smallest distance belongs; Re-determine the cluster center for the obtained cluster, recalculate until convergence, and determine whether the distance from the data point in the first sample service data to the cluster center is greater than a threshold. If the data point does not belong to the cluster or the distance is greater than the threshold, determine that the data point is an abnormal data point; Eliminate abnormal data points in the first sample service data to obtain second sample service data; The performing dimensionality reduction processing on the second sample service data based on the t-SNE algorithm to obtain target feature data includes: Get the second sample service data, set the t-SNE algorithm loss function, and initialize the t-SNE algorithm iteration number, learning rate, and momentum; Calculating the conditional probabilities of the two data points in the second sample service data, calculating the joint probability distribution based on the conditional probabilities of the two data points, and performing normalization processing; Randomly generate low-dimensional random numbers and calculate the joint probability distribution of low-dimensional sample space points in the t distribution. Use gradient descent to calculate the optimized gradient and obtain the current low-dimensional data. Iterate the loop so that the loss function reaches the minimum value and obtain the target feature data. The process of constructing the data analysis model includes the following steps: Obtain historical service data of the government service department, split the historical service data into a training sample data set and a test sample data set, use the Bayesian method to fill in the missing sample data in the training sample data set, and discretize the sample data in the training sample data set; Construct a decision tree model, input the processed training sample data set into the decision tree model for training, calculate the information gain of each attribute of the root node of the decision tree, select the attribute corresponding to the maximum value as the attribute of the root node, and then recursively calculate the attributes of other child nodes from top to bottom; Add a Bayesian node between the parent node and the child node, check the attributes of the parent node, and prune the decision tree. Use a two-layer collaborative correction method to build the baseline layer and the real-time layer to obtain a decision tree model based on the two-layer XGBoost algorithm. The test sample data set and the Pelican algorithm were used to optimize the parameters of the decision tree model based on the double-layer XGBoost algorithm and establish a data analysis model; The test sample data set and the Pelican algorithm are used to optimize the parameters of the decision tree model based on the double-layer XGBoost algorithm to establish a data analysis model, including: Set the population size, search space dimension and maximum number of iterations, initialize the pelican population, construct a fitness value function based on the test sample data set, calculate the fitness value of each individual in the initial pelican population, and record the position of the current best individual as the global optimal solution; Carry out reverse differential evolution on the pelican population, compare the fitness values ​​of individuals before and after evolution, and select the best individuals to form the updated pelican population; Recalculate the fitness values ​​of the updated individuals in the population, select the current best individual, and perform adaptive mutation on the best individual to update the position of the individuals in the pelican population; Confirm whether the maximum number of iterations is met. If not, continue to iterate. If it is met, assign the optimal hyperparameters to the decision tree model based on the double-layer XGBoost algorithm, and use the optimal hyperparameters to establish a data analysis model; The target feature data is input into a data analysis model to make predictions through the data analysis model, and output government affairs analysis results, including: Input the target feature data into the data analysis model, use the target feature data as the reference layer input of the data analysis model, make decisions along the tree structure, and finally reach a leaf node, calculate the reference prediction value, and obtain the output result of the reference layer; The output results of the baseline layer are used as the input of the real-time layer, the input baseline prediction value is verified, the error prediction value is iteratively calculated, and the output results of the real-time layer are obtained. The output results of the baseline layer and the real-time layer are summarized to obtain the final prediction result to output the government analysis results.

2. A government service data processing system based on artificial intelligence, characterized in that: The government service data processing system based on artificial intelligence includes a data integration module, a data preprocessing module, a data dimension reduction module, a government service analysis module and a result push module, wherein: A data integration module is used to collect government service data from various government service departments, integrate the government service data, and obtain service data to be processed; A data preprocessing module, used for preprocessing the service data to be processed to obtain standardized data, inputting the standardized data into a Gaussian mixture model, and iteratively updating the Gaussian mixture model using an EM algorithm to output first sample service data; A data dimension reduction module is used to perform outlier detection on the first sample service data to obtain second sample service data, and perform dimension reduction processing on the second sample service data based on the t-SNE algorithm to obtain target feature data: obtain the second sample service data, set the t-SNE algorithm loss function, initialize the number of iterations, learning rate and momentum of the t-SNE algorithm, calculate the conditional probability of each pair of data points in the second sample service data, calculate the joint probability distribution based on the conditional probability of each pair of data points and perform normalization processing, randomly generate low-dimensional random numbers, and calculate the joint probability distribution of low-dimensional sample space points in the t distribution, use gradient descent to calculate the optimized gradient, and obtain the current low-dimensional data, iterate cyclically so that the loss function reaches the minimum value, and obtain the target feature data; The government affairs analysis module is used to input the target feature data into the data analysis model to make predictions through the data analysis model and output the government affairs analysis results, wherein the government affairs analysis results at least include service demand predictions, policy solution predictions and risk warnings: obtain historical service data of the government affairs service department, split the historical service data into a training sample data set and a test sample data set, use the Bayesian method to fill in the missing sample data in the training sample data set, and discretize the sample data in the training sample data set to build a decision tree model, input the processed training sample data set into the decision tree model for training, calculate the information gain of each attribute of the root node of the decision tree, select the attribute corresponding to the maximum value as the attribute of the root node, and then recursively calculate the attributes of other child nodes from top to bottom, add a Bayesian node between the parent node and the child node, check the attributes of the parent node, and prune the decision tree, use a two-layer collaborative correction method to build a baseline layer and a real-time layer, and obtain a decision tree model based on a two-layer XGBoost algorithm, use the test sample data set and Pelican The algorithm optimizes the parameters of the obtained decision tree model based on the double-layer XGBoost algorithm and establishes a data analysis model, wherein the test sample data set and the pelican algorithm are used to optimize the parameters of the obtained decision tree model based on the double-layer XGBoost algorithm and establish a data analysis model, including: setting the population size, search space dimension and maximum number of iterations, initializing the pelican population, constructing a fitness value function based on the test sample data set, calculating the fitness value of each individual in the initial pelican population, and recording the position of the current best individual as the global optimal solution, performing reverse differential evolution on the pelican population, comparing the fitness values ​​of individuals before and after evolution, retaining the best individuals to form an updated pelican population, recalculating the fitness values ​​of the individuals in the updated population, selecting the current best individual, and performing adaptive mutation of the best individual, updating the position of the individual in the pelican population, confirming whether the maximum number of iterations is met, and continuing to iterate if not, and assigning the optimal hyperparameters to the decision tree model based on the double-layer XGBoost algorithm if satisfied, and using the optimal hyperparameters to establish a data analysis model; A result push module is used to push the government affairs analysis results to the corresponding government affairs service department, and display the government affairs analysis results through data visualization technology: input the target feature data into the data analysis model, use the target feature data as the reference layer input of the data analysis model, make decisions along the tree structure, finally reach a leaf node, calculate the reference prediction value, obtain the output result of the reference layer, use the output result of the reference layer as the input of the real-time layer, verify the input reference prediction value, iteratively calculate the error prediction value, obtain the output result of the real-time layer, summarize the output results of the reference layer and the real-time layer, obtain the final prediction result, and output the government affairs analysis results; The data preprocessing module includes a data cleaning submodule, a normalization processing submodule, a first calculation submodule and an iteration submodule, wherein: A data cleaning submodule is used to obtain the service data to be processed, and clean the service data to be processed to obtain the cleaned service data to be processed; The normalization processing submodule is used to normalize the cleaned service data to be processed by using Z-score standardization, and scale the data value range of the cleaned service data to be processed to [0, 1] or [-1, 1] to obtain standardized data; A first calculation submodule, configured to calculate a posterior probability from each Gaussian distribution for each data point of the standardized data, and obtain a parameter of each Gaussian distribution based on the posterior probability using a weighted least square method; The iterative submodule is used to repeatedly iterate the E step and the M step of the EM algorithm until the algorithm converges, terminate the iteration, weight the posterior probability of the data points in each cluster, and then use the obtained weighted data points as the first sample service data; The data dimension reduction module includes a selection submodule, a second calculation submodule, a judgment submodule and a elimination submodule, wherein: The selection submodule is used to determine the number of clusters of the K-means clustering algorithm and select k initial samples as the initial cluster centers; A second calculation submodule is used to obtain the first sample service data, calculate the Euclidean distance from each data point in the first sample service data to the initial cluster center, and assign each data point to the cluster to which the initial cluster center with the smallest distance belongs; A judgment submodule, used to redetermine the cluster center of the obtained cluster, recalculate until convergence, and judge whether the distance from the data point in the first sample service data to the cluster center is greater than a threshold. If the data point does not belong to the cluster or the distance is greater than the threshold, the data point is determined to be an abnormal data point; The elimination submodule is used to eliminate abnormal data points in the first sample service data to obtain second sample service data.