Government affair data analysis method and system
By conducting word meaning analysis and multi-dimensional modeling of government affairs data and establishing a demand mining model, the problems of low efficiency and poor accuracy of traditional government affairs data analysis methods are solved, and timely resolution of government affairs needs is achieved.
Patent Information
- Application Number
- CN202411177271.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-26
- Publication Date
- 2024-12-06
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
Traditional government data analysis methods are inefficient and have poor accuracy, making it difficult to cope with complex data analysis needs, resulting in government data not being solved in a timely manner.
By sending data pull applications to multiple government affairs platforms, collecting government affairs resource data for word meaning analysis, constructing word meaning feature sets and multi-dimensional modeling feature data, establishing demand mining models, performing data mining to obtain government affairs feedback demand information, and matching the corresponding work order information.
It improves the accuracy and efficiency of government data analysis and solves government needs problems in a timely manner.
Smart Images

Figure CN119090437A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of data analysis, and specifically to a government data analysis method and system. Background Art
[0002] With the continuous development of information technology, government data processing capabilities have become an important indicator of the government's modern governance level. Traditional government data analysis methods have problems such as low efficiency, poor accuracy, and difficulty in responding to complex data analysis needs, resulting in the inability to promptly resolve objective needs.
[0003] In order to improve the efficiency and accuracy of government data analysis and to promptly resolve existing government needs, how to comprehensively consider government data in multiple dimensions and achieve optimal allocation and efficient use of data resources is an issue that urgently needs to be addressed in this field. Summary of the invention
[0004] In order to solve the above problems, the purpose of the present invention is to provide a government data analysis method and system.
[0005] The purpose of the present invention can be achieved through the following technical solution: A government data analysis method comprises the following steps:
[0006] Step S1: Send a data pull application to several government platforms, and collect government resource data of the corresponding government platforms after the data pull application is approved;
[0007] Step S2: performing word meaning analysis on the government resource data of each government platform, and then constructing a word meaning feature set of each government platform;
[0008] Step S3: performing index analysis on the semantic feature sets of several government platforms, and generating multi-dimensional modeling feature data of the corresponding government platforms;
[0009] Step S4: construct a corresponding demand mining model based on the multi-dimensional modeling feature data of each government affairs platform;
[0010] Step S5: The demand mining model performs data mining on the government resource data of the corresponding government platforms, thereby obtaining government feedback demand information;
[0011] Step S6: Match the corresponding government work order information according to the government feedback demand information of each government platform.
[0012] Furthermore, a data extraction application is sent to several government platforms. After the data extraction application is approved, the process of collecting government resource data of the corresponding government platforms includes:
[0013] Several government platforms are numbered and recorded as i, i=1, 2, 3, ..., n, where n is a natural number greater than 0. Several data pulling applications are generated in a set application period and sent to several government platforms. When several government platforms receive any data pulling application, it is determined whether the IP address of the data pulling application is in a preset legal address form. If so, a pulling file is constructed according to the data pulling application, and the pulling file collects the government resource data corresponding to the corresponding government platform. If not, the corresponding data pulling application is included in a preset blacklist, and a new data pulling application is generated to replace the data pulling application included in the blacklist. A new pulling file is generated by the replaced data pulling application to collect the government resource data corresponding to the government platform; until the government resource data corresponding to several government platforms are collected, the government resource data includes administrative management data, public security data, public opinion monitoring data and people's livelihood service data.
[0014] Furthermore, the process of performing word meaning analysis on the government resource data of each government platform and then constructing a word meaning feature set of each government platform includes:
[0015] The word meaning analysis includes word meaning information extraction, word bag model construction and word meaning feature matching;
[0016] The content of word meaning information extraction is: setting extraction words corresponding to administrative management data, public safety data, public opinion monitoring data and people's livelihood service data, processing administrative management data, public safety data, public opinion monitoring data and people's livelihood service data into several word segments respectively, marking the parts of each word segment that are consistent with the extracted words as their corresponding word meaning information, and removing the parts of the word segment that are inconsistent with the extracted words;
[0017] The contents of constructing the bag-of-words model are as follows: inputting the semantic information of each government platform into the pre-constructed initial neural network model, classifying the semantic information of the same frequency in each government platform into a group of terms, summarizing the terms of different frequencies into a preset blank statistical table, and then forming a term statistical table corresponding to each government platform, constructing a corresponding term sparse matrix according to the frequencies corresponding to several groups of terms in each term statistical table, vectorizing the text data of each term sparse matrix, and then obtaining several term feature vectors corresponding to each term sparse matrix, summarizing all the term feature vectors of each government platform, and then constructing a bag-of-words model of the government resource data corresponding to each government platform;
[0018] The bag-of-words model consists of an administrative management sub-model, a public safety sub-model, a public opinion monitoring sub-model and a people's livelihood service sub-model. The bag-of-words model constructed corresponding to each government affairs platform is used to match the semantic features of the government affairs resource data corresponding to each government affairs platform. The semantic features include administrative management semantic features, public safety semantic features, public opinion monitoring semantic features and people's livelihood service semantic features. The administrative management semantic features, public safety semantic features, public opinion monitoring semantic features and people's livelihood service semantic features corresponding to each government affairs platform are summarized and constructed into their respective corresponding semantic feature sets.
[0019] Furthermore, the process of performing index analysis on the semantic feature sets of several government platforms and generating multi-dimensional modeling feature data of the corresponding government platforms includes:
[0020] Set characteristic indicators corresponding to each government affairs platform. The characteristic indicators are used to perform indicator analysis corresponding to each government affairs platform. The characteristic indicators set for each government affairs platform include a first indicator, a second indicator, a third indicator and a fourth indicator. The first indicator maps out the first dimension modeling characteristic data corresponding to each government affairs platform, the second indicator maps out the second dimension modeling characteristic data corresponding to each government affairs platform, the third indicator maps out the third dimension modeling characteristic data corresponding to each government affairs platform, and the fourth indicator maps out the fourth dimension modeling characteristic data corresponding to each government affairs platform;
[0021] The first dimension modeling feature data, the second dimension modeling feature data, the third dimension modeling feature data and the fourth dimension modeling feature data corresponding to each government affairs platform are aggregated to generate the corresponding multi-dimensional modeling feature data for each government affairs platform.
[0022] Furthermore, the process of constructing a corresponding demand mining model based on the multi-dimensional modeling feature data of each government platform includes:
[0023] A modeling program is set up, and the program environment is initialized, modeling items are configured, and data filtering fields are entered for the modeling program. After the program environment initialization, modeling item configuration, and data filtering field entry of the modeling program are completed, the multi-dimensional modeling feature data corresponding to each government affairs platform is input into the modeling program, and then the modeling program constructs a demand mining model corresponding to each government affairs platform through deep learning technology;
[0024] The first dimension modeling feature data, second dimension modeling feature data, third dimension modeling feature data and fourth dimension modeling feature data of each government platform are respectively processed with feature vectors to obtain the corresponding modeling feature vectors. The frequency of the terms corresponding to each modeling feature vector in the bag-of-words model is used as the modeling weight, and the modeling weight is annotated to the corresponding modeling feature vector. The modeling program sets the mining parameters of the corresponding type of each modeling feature vector and the computing power allocation ratio of the mining parameters according to the modeling weight, and then constructs the demand mining model corresponding to each government platform.
[0025] Furthermore, the process of using the demand mining model to mine the government resource data of the corresponding government platforms and then obtaining government feedback demand information includes:
[0026] Set a data mining period for each government platform's corresponding demand mining model. During each data mining period, the demand mining model of each government platform conducts data mining on its corresponding government resource data, and obtains several sub-feedback demands corresponding to each government platform through data mining. The sub-feedback demands include administrative feedback demands, public safety demands, public opinion monitoring demands, and people's livelihood service demands.
[0027] Among them, administrative feedback needs include administrative appeal time, administrative appeal events and administrative appealers; public safety needs include safety incident reporters, safety incidents and safety incident reporting time; public opinion monitoring needs include public opinion reporters, public opinion events and public opinion reporting time; people's livelihood service needs include different types of people's livelihood service events, the scope of people's livelihood event services and the list of people benefiting from people's livelihood services;
[0028] The administrative feedback needs, public safety needs, public opinion monitoring needs and people's livelihood service needs of each government resource data platform after data mining by the corresponding demand mining model are summarized as the government feedback demand information corresponding to each government platform.
[0029] Furthermore, the process of matching corresponding government work order information according to the government feedback demand information of each government platform includes:
[0030] Set up a work order database, which is used to store several types of work order information, including administrative problem handling work orders, public safety problem handling work orders, public opinion problem handling work orders, and livelihood problem handling work orders;
[0031] The administrative problem handling work order is used to solve administrative appeal events corresponding to a number of administrative feedback needs, the public security problem handling work order is used to solve security events corresponding to a number of public security needs, the public opinion problem handling work order is used to solve public opinion events corresponding to a number of public opinion monitoring needs, and the livelihood problem handling work order is used to solve different types of livelihood service events corresponding to a number of livelihood service needs;
[0032] The government feedback demand information corresponding to each government platform is input into the work order database, and then the administrative problem handling work orders, public safety problem handling work orders, public opinion problem handling work orders and people's livelihood problem handling work orders corresponding to the administrative feedback demand, public safety demand, public opinion monitoring demand and people's livelihood service demand in the government feedback demand information are matched from the work order database;
[0033] The administrative problem handling work orders, public safety problem handling work orders, public opinion problem handling work orders and people's livelihood problem handling work orders matched by each government platform from the work order database are merged to generate the government work order information corresponding to each government platform.
[0034] Furthermore, a government data analysis system includes:
[0035] The government data collection module is used to send data extraction applications to several government platforms and collect government resource data from the corresponding government platforms after the data extraction applications are approved;
[0036] The word meaning feature extraction module is used to perform word meaning analysis on the government resource data of each government platform, and then construct a word meaning feature set for each government platform;
[0037] A modeling feature data generation module is used to perform index analysis on the semantic feature sets of several government platforms and generate multi-dimensional modeling feature data of the corresponding government platforms;
[0038] The mining model building module is used to build the corresponding demand mining model based on the multi-dimensional modeling feature data of each government platform;
[0039] The data mining module is used to perform data mining of government resource data of the corresponding government platforms through the demand mining model, thereby obtaining government feedback demand information;
[0040] The feedback matching module is used to match the corresponding government work order information according to the government feedback demand information of each government platform.
[0041] Compared with the prior art, the beneficial effects of the present invention are: collecting the government resource data corresponding to several government platforms for word meaning analysis, and then constructing the word meaning feature set of each government platform, performing indicator analysis on the word meaning feature set to generate multi-dimensional modeling feature data of the corresponding government platform, and constructing a demand mining model corresponding to each government platform based on the multi-dimensional modeling feature data. The demand mining model performs data mining on the government resource data of each government platform to obtain government feedback demand information, and matches the corresponding government work order information according to the government feedback demand information of each government platform, and executes all events recorded in the government work order information. On the one hand, a corresponding demand mining model is constructed through the multi-dimensional modeling feature data corresponding to each government platform, and corresponding data mining is performed to improve the accuracy of government data analysis. On the other hand, the government feedback demand information obtained after data mining is matched with the corresponding government work order information, and the government work order information is executed, thereby promptly solving the government demand problems represented by the government data. BRIEF DESCRIPTION OF THE DRAWINGS
[0042] Figure 1 It is a flow chart of the present invention. DETAILED DESCRIPTION
[0043] like Figure 1 As shown, a government data analysis method includes the following steps:
[0044] Step S1: Send a data pull application to several government platforms, and collect government resource data of the corresponding government platforms after the data pull application is approved;
[0045] Step S2: performing word meaning analysis on the government resource data of each government platform, and then constructing a word meaning feature set of each government platform;
[0046] Step S3: performing index analysis on the semantic feature sets of several government platforms, and generating multi-dimensional modeling feature data of the corresponding government platforms;
[0047] Step S4: construct a corresponding demand mining model based on the multi-dimensional modeling feature data of each government affairs platform;
[0048] Step S5: The demand mining model performs data mining on the government resource data of the corresponding government platforms, thereby obtaining government feedback demand information;
[0049] Step S6: Match the corresponding government work order information according to the government feedback demand information of each government platform.
[0050] It should be further explained that, in the specific implementation process, a data pull application is sent to several government platforms, and after the data pull application is approved, the process of collecting government resource data of the corresponding government platforms includes:
[0051] Number several government service platforms and record the number as i, then i = 1, 2, 3, ..., n, where n is a natural number greater than 0, set the application period and record the application period as T 申请 , then T 申请 =[t1, t2], where t1 is the start time of the application period and t2 is the end time of the application period;
[0052] At the start time corresponding to t1, several data pulling applications are generated and sent to several corresponding government platforms. When several government platforms receive any data pulling application, it is determined whether the IP address of the data pulling application is in the preset legal address form. If so, a pulling file is constructed according to the data pulling application, and the pulling file collects the government resource data corresponding to the corresponding government platform. If not, the corresponding data pulling application is included in the preset blacklist, and a new data pulling application is generated to replace the data pulling application included in the blacklist. A new pulling file is generated by the replaced data pulling application to collect the government resource data corresponding to the government platform; until the government resource data corresponding to several government platforms are collected;
[0053] At the end time corresponding to t2, the government resource data corresponding to several government platforms are stored in the set database, and the government resource data includes administrative management data, public security data, public opinion monitoring data and people's livelihood service data.
[0054] It should be noted that administrative management data includes data on internal government management, personnel, finance, and assets; public security data involves data on criminal records, traffic management, emergency management, and fire safety; public opinion monitoring data involves social dynamic data such as online public opinion, public opinion, and media reports; people's livelihood service data includes residents' health records, education information, social welfare, and the use of public service facilities.
[0055] It should be further explained that, in the specific implementation process, the process of performing word meaning analysis on the government resource data of each government platform and then constructing the word meaning feature set of each government platform includes:
[0056] Sequentially perform word meaning analysis on administrative management data, public security data, public opinion monitoring data, and people's livelihood service data in the government resource data corresponding to each government platform, wherein the word meaning analysis includes word meaning information extraction, word bag model construction, and word meaning feature matching;
[0057] The word meaning information of administrative management data, public security data, public opinion monitoring data and people's livelihood service data in the government resource data corresponding to each government platform is obtained through the word meaning information extraction, and the extracted word meaning information content is as follows:
[0058] Set extraction words corresponding to administrative management data, public safety data, public opinion monitoring data and people's livelihood service data, respectively process administrative management data, public safety data, public opinion monitoring data and people's livelihood service data into several vocabulary segments, mark the parts of the respective vocabulary segments that are consistent with the extracted vocabulary as their respective corresponding word meaning information, and remove the parts of the vocabulary segments that are inconsistent with the extracted vocabulary;
[0059] The extracted vocabulary is a high-frequency keyword vocabulary corresponding to each of the administrative management data, public safety data, public opinion monitoring data and people's livelihood service data. The high-frequency keyword vocabulary represents the effective data information part of each. After the administrative management data, public safety data, public opinion monitoring data and people's livelihood service data are processed into a number of vocabulary segments corresponding to each of them, the extracted vocabulary and the vocabulary segments are converted into binary numbers. If the extracted vocabulary and the vocabulary segment are equal in each bit of the binary number, it means that the extracted vocabulary and the vocabulary segment are consistent, otherwise, they are inconsistent;
[0060] The contents of constructing the bag-of-words model are as follows: inputting the semantic information of each government platform into the pre-constructed initial neural network model, classifying the semantic information of the same frequency in each government platform into a group of terms, summarizing the terms of different frequencies into a preset blank statistical table, and then forming a term statistical table corresponding to each government platform, constructing a corresponding term sparse matrix according to the frequencies corresponding to several groups of terms in each term statistical table, vectorizing the text data of each term sparse matrix, and then obtaining several term feature vectors corresponding to each term sparse matrix, summarizing all the term feature vectors of each government platform, and then constructing a bag-of-words model of the government resource data corresponding to each government platform;
[0061] The construction content of the term sparse matrix corresponding to each term statistical table is as follows: set a blank initial term distribution matrix, record the number of matrix rows and matrix columns corresponding to the initial term distribution matrix as p and q respectively, that is, the blank initial term distribution matrix is the blank matrix corresponding to p*q, set the term sparse determination threshold, and record it as τ, according to the relationship between the term sparse determination threshold and the corresponding frequency of each term, fill each term in the blank initial term distribution matrix with different matrix values, the matrix values include "0" and "1", record the frequency of each term included in the term statistical table as f, if f≥τ, the filled matrix value is 1, if f<τ, the filled matrix value is 0, and then when the matrix value filling of all terms corresponding to each term statistical table is completed, the term sparse matrix corresponding to the term statistical table is constructed, and the term sparse matrix effectively compresses the term statistical table.
[0062] The bag-of-words model is composed of several types of sub-models, including administrative management sub-model, public security sub-model, public opinion monitoring sub-model and people's livelihood service sub-model;
[0063] Among them, the administrative management sub-model is constructed by several entry feature vectors of the entry statistics table corresponding to the administrative management data. Similarly, the public security sub-model, the public opinion monitoring sub-model and the people's livelihood service sub-model are constructed by several entry feature vectors of the entry statistics table corresponding to the public security data, the public opinion monitoring data and the people's livelihood service data respectively.
[0064] The word bag model corresponding to each government platform is used to match the semantic features of the government resource data corresponding to each government platform. The matching content is: the word semantic features include administrative management semantic features, public safety semantic features, public opinion monitoring semantic features and people's livelihood service semantic features. The administrative management sub-model, public safety sub-model, public opinion monitoring sub-model and people's livelihood service sub-model in the word bag model corresponding to each government platform obtain the administrative management semantic features, public safety semantic features, public opinion monitoring semantic features and people's livelihood service semantic features of the government resource data in the government platform;
[0065] The administrative management semantic features, public safety semantic features, public opinion monitoring semantic features and people's livelihood service semantic features corresponding to each government platform are summarized and constructed into their corresponding semantic feature sets.
[0066] It should be further explained that, in the specific implementation process, the process of performing index analysis on the semantic feature sets of several government platforms and generating multi-dimensional modeling feature data of the corresponding government platforms includes:
[0067] The semantic feature set of the government platform numbered i is marked as Data[i], and the feature index corresponding to each government platform is set. The feature index is used to perform the index analysis corresponding to each government platform. The content of the index analysis is as follows:
[0068] The characteristic indicators set for each government affairs platform include a first indicator, a second indicator, a third indicator and a fourth indicator. The first indicator maps out the first dimension modeling characteristic data corresponding to each government affairs platform, the second indicator maps out the second dimension modeling characteristic data corresponding to each government affairs platform, the third indicator maps out the third dimension modeling characteristic data corresponding to each government affairs platform, and the fourth indicator maps out the fourth dimension modeling characteristic data corresponding to each government affairs platform;
[0069] The first indicator is used to describe the characteristic word sense data of different word sense dimensions corresponding to different administrative management word sense features in the word sense feature set, and the word sense dimension value corresponding to the characteristic word sense data of the administrative management word sense feature under the standard word sense dimension is set, and the word sense dimension value is recorded as γ 标准, obtain the word meaning dimension values corresponding to other administrative management word meaning features except the standard word meaning dimension, and record them as γother;
[0070] Among them, γ 标准 =Num 标准 / Num 总 , where Num 标准 Num is the number of feature word meaning data corresponding to the administrative management word meaning feature under the standard word meaning dimension. 总 The total number of feature word sense data corresponding to the administrative management word sense feature is set to the lower threshold and upper threshold corresponding to the administrative management word sense feature, and are recorded as λ1 and λ2 respectively, where λ1<λ2;
[0071] Will γ 其他 -γ 标准 The word meaning dimensions of all feature word meaning data of >λ2 are summarized as high-frequency word meaning dimensions;
[0072] Let λ1<γ 其他 -γ 标准 The word meaning dimensions of all feature word meaning data with ≤λ2 are summarized as low-frequency word meaning dimensions;
[0073] Then, all characteristic word meaning data in the standard word meaning dimension, high-frequency word meaning dimension and low-frequency word meaning dimension in the administrative management word meaning feature are integrated to generate the first dimension modeling characteristic data corresponding to the current government affairs platform;
[0074] It should be noted that the administrative management word meaning features in the administrative management data include several characteristic word meaning data that comprehensively consider different word meaning dimensions such as standard word meaning dimension, high-frequency word meaning dimension and low-frequency word meaning dimension, enriching the feature extraction dimension and laying the foundation for the data breadth of subsequent data analysis;
[0075] Similarly, the second indicator is used to describe the characteristic word sense data of different word sense dimensions corresponding to different public safety word sense features in the word sense feature set, set the word sense dimension value of the characteristic word sense data corresponding to the public safety word sense feature under the standard word sense dimension, and record the word sense dimension value as α 标准 , obtain the word sense dimension values corresponding to other public safety word sense features except the standard word sense dimension, and record them as α 其他 ;
[0076] Among them, α 标准 =Num1 标准 / Num1 总 , where Num1 标准 Num1 is the number of feature word meaning data corresponding to the public safety word meaning feature under the standard word meaning dimension, 总The total number of feature word meaning data corresponding to the public safety word meaning feature is set to the lower threshold and upper threshold corresponding to the public safety word meaning feature, and are recorded as μ1 and μ2 respectively, where μ1<μ2;
[0077] α 其他 -α 标准 The word meaning dimensions of all feature word meaning data of >μ2 are summarized as high-frequency word meaning dimensions;
[0078] μ1<α 其他 -α 标准 The word meaning dimensions of all feature word meaning data ≤μ2 are summarized as low-frequency word meaning dimensions;
[0079] Then, all characteristic word meaning data in the standard word meaning dimension, high-frequency word meaning dimension and low-frequency word meaning dimension in the public security word meaning feature are integrated to generate the second dimension modeling characteristic data corresponding to the current government affairs platform;
[0080] The third indicator is used to describe the characteristic word sense data of different word sense dimensions corresponding to different public opinion monitoring word sense features in the word sense feature set, and the word sense dimension value corresponding to the characteristic word sense data of the public opinion monitoring word sense feature under the standard word sense dimension is set, and the word sense dimension value is recorded as η 标准 , obtain the word meaning dimension values corresponding to other public opinion monitoring word meaning features except the standard word meaning dimension, and record them as η 其他 ;
[0081] Among them, η 标准 =Num2 标准 / Num2 总 , where Num2 标准 Num2 is the number of feature word meaning data corresponding to the public opinion monitoring word meaning feature under the standard word meaning dimension. 总 The total number of feature word meaning data corresponding to the public opinion monitoring word meaning feature is set to the lower threshold and upper threshold of distinction corresponding to the public opinion monitoring word meaning feature, and are recorded as δ1 and δ2 respectively, where δ1<δ2;
[0082] η 其他 -η 标准 The word meaning dimensions of all feature word meaning data of >δ2 are summarized as high-frequency word meaning dimensions;
[0083] δ1<η 其他 -η 标准 The word meaning dimensions of all feature word meaning data ≤δ2 are summarized as low-frequency word meaning dimensions;
[0084] Then, all characteristic word meaning data in the standard word meaning dimension, high-frequency word meaning dimension and low-frequency word meaning dimension in the word meaning features of public opinion monitoring are integrated to generate the third dimension modeling characteristic data corresponding to the current government affairs platform;
[0085] The fourth indicator is used to describe the characteristic word meaning data of different word meaning dimensions corresponding to different livelihood service word meaning features in the word meaning feature set, set the word meaning dimension value of the characteristic word meaning data corresponding to the livelihood service word meaning feature under the standard word meaning dimension, and record the word meaning dimension value as ρ 标准 , obtain the corresponding word meaning dimension values of other livelihood service word meaning features except the standard word meaning dimension, and record them as ρ 其他 ;
[0086] Among them, ρ 标准 =Num3 标准 / Num3 总 , where Num3 标准 Num3 is the number of feature word meaning data corresponding to the livelihood service word meaning feature under the standard word meaning dimension. 总 The total number of feature word meaning data corresponding to the livelihood service word meaning feature is set to the lower threshold and upper threshold corresponding to the livelihood service word meaning feature, and are recorded as ω1 and ω2 respectively, where ω1<ω2;
[0087] ρ 其他 -ρ 标准 The word meaning dimensions of all feature word meaning data of >ω2 are summarized as high-frequency word meaning dimensions;
[0088] Let ω1<ρ 其他 -ρ 标准 The word meaning dimensions of all feature word meaning data ≤ω2 are summarized as low-frequency word meaning dimensions;
[0089] Then, all characteristic word meaning data in the standard word meaning dimension, high-frequency word meaning dimension and low-frequency word meaning dimension of the word meaning characteristics of people's livelihood services are integrated to generate the fourth dimension modeling characteristic data corresponding to the current government affairs platform;
[0090] The first dimension modeling feature data is used to describe the common features of different types of administrative-related events in administrative management data, as well as the unique features of each type of administrative-related event;
[0091] The second dimension modeling feature data is used to describe the common features of different types of public safety events in the public safety data, as well as the unique features of each type of public safety event;
[0092] The third dimension modeling feature data is used to express the common features of different types of public opinion events in the public opinion monitoring data, as well as the unique features of each type of public opinion event;
[0093] The fourth dimension modeling feature data is used to describe the common features of different types of livelihood-related events in the livelihood service data, as well as the unique features of each type of livelihood-related event;
[0094] It should be noted that common features are used to reflect the similar features of different types of events and to quickly classify all events with the same features. Unique features are used to reflect the personalized features of each type of event. By integrating all modeling feature data of different dimensions, the adaptability of subsequent model construction is enriched.
[0095] The first dimension modeling feature data, second dimension modeling feature data, third dimension modeling feature data and fourth dimension modeling feature data corresponding to each government platform are aggregated to generate multi-dimensional modeling feature data corresponding to each government platform, and the multi-dimensional modeling feature data corresponding to the government platform numbered i is marked as Data`[i].
[0096] It should be further explained that, in the specific implementation process, the process of building a corresponding demand mining model based on the multi-dimensional modeling feature data of each government platform includes:
[0097] Set up a modeling program, initialize the program environment, configure modeling items, and enter data filtering fields for the modeling program. The program environment initialization content is: obtain the environment log of the modeling program, and determine the amount of alienated information in the environment log, and record the amount of information as C 信息 ;
[0098] Set the upper threshold of the amount of information corresponding to the alienated information, and record the upper threshold as C 阈值 , when C 信息 ≥C 阈值 When C 信息 <C 阈值 When the modeling environment of the modeling program is determined to be a safe environment, no operation is performed;
[0099] The content of modeling item configuration is: obtain the amount of multi-dimensional modeling feature data that the modeling program needs to process, and configure the modeling computing power, modeling hyperparameters and model evaluation parameters corresponding to the current modeling program according to the amount of data. The modeling hyperparameters include learning rate, regularization parameter, number of model layers for deep learning and number of neurons in each layer. The model evaluation parameters include model normalization rate, model recall rate and F1 value;
[0100] The amount of multi-dimensional modeling feature data is recorded as D 建模 , the initial modeling power of the modeling program is recorded as SL 初始 , set the data volume per unit D 单位 The corresponding computing power increase value is recorded as SL 增加 , and the modeling power corresponding to the current modeling program is recorded as SL建模 , there is SL 建模 =(D 建模 ÷D 单位 )×SL 增加 +SL 初始 ;
[0101] It should be noted that the learning rate refers to a hyperparameter used to adjust the amplitude of model parameter updates during model training. It determines the size of parameter updates in each iteration. Too high or too low learning rates may lead to unstable training process or slow convergence. It is usually necessary to select a suitable learning rate through experiments. Regularization is used to control the complexity of the model in deep learning to prevent overfitting (that is, the model performs well on training data but poorly on unseen data) and to help balance the relationship between the model's fitting and generalization capabilities. The model normalization rate measures the proportion of samples that are truly positive among the samples predicted by the model to be positive. The model recall rate measures the proportion of samples that are truly positive among the samples predicted by the model to be positive. The F1 value is the harmonic mean of the model normalization rate and the recall rate, which takes both the normalization rate and the recall rate into consideration.
[0102] The content entered in the data filtering field is: define several data filtering fields, set the data entry period of the modeling program, and enter several data filtering fields into the modeling program during the data entry period. The data filtering field is used to filter out illegal field information in the modeling program. The illegal field information is the vulnerability details information corresponding to several environmental vulnerability points located when the modeling environment of the modeling program is in an abnormal environment. The data filtering field is matched with the vulnerability details information corresponding to the environmental vulnerability point. After the match is successful, the filtering relationship is successfully constructed, and the environmental vulnerability point is eliminated by the data filtering field to repair the modeling environment to a safe environment;
[0103] After the modeling program's program environment initialization, modeling item configuration, and data filtering field entry are completed, the multi-dimensional modeling feature data corresponding to each government platform is input into the modeling program, and then the modeling program uses deep learning technology to build a demand mining model corresponding to each government platform.
[0104] The specific contents of constructing the demand mining model are as follows: perform feature vector processing on the first dimension modeling feature data, the second dimension modeling feature data, the third dimension modeling feature data and the fourth dimension modeling feature data of each government platform respectively, and then obtain the corresponding modeling feature vectors, use the frequency of the terms corresponding to each modeling feature vector in the bag-of-words model as the modeling weight, and annotate the modeling weight to the corresponding modeling feature vector, and let the modeling program set the mining parameters corresponding to each modeling feature vector and the computing power allocation ratio of the mining parameters according to the modeling weight, and then construct the demand mining model corresponding to each government platform;
[0105] The mining parameters of the modeling feature vector include data mining depth, data mining breadth and data mining duration;
[0106] The data mining depth refers to the degree and complexity of analyzing and mining the first dimension modeling feature data, the second dimension modeling feature data, the third dimension modeling feature data, and the fourth dimension modeling feature data in the government affairs platform; the greater the depth, the more complex the technology and methods involved in the data mining process, which may include advanced statistical analysis, machine learning models or deep learning algorithms. For example, using complex neural networks for image recognition or natural language processing involves a high data mining depth;
[0107] The breadth of data mining refers to the number of data dimensions or variables explored in a data set. The greater the breadth, the more data variables or dimensions are involved, which may include multiple data sources, multiple data types, or multiple features. For example, in public opinion analysis, in addition to basic public opinion event data, it may also involve public opinion behavior data, public opinion diffusion trend data, and other aspects, which demonstrates the breadth of data mining.
[0108] The data mining duration refers to the time span for data mining and analysis, which can cover the entire process from data collection, cleaning, feature engineering, model selection to training and optimization. The length of the duration is affected by data complexity, processing technology and computing resources. For example, processing a large data set may take a long time to complete the data mining process, while a small data set may be completed faster.
[0109] The computing power allocation ratio of the mining parameters is set as follows: the modeling weight corresponding to each modeling feature vector is recorded as QZ, the modeling weights of all modeling feature vectors are accumulated, and the total weight value of the entire modeling weight is obtained and recorded as QZ 总和 , the total computing power required to build the demand mining model is recorded as R 总 , the modeling computing power corresponding to each modeling feature vector is recorded as R 分 , then it is expressed as follows:
[0110] R 分 / R 总 =QZ / QZ 总和 ;
[0111] It should be noted that after constructing the demand mining model corresponding to each government platform, the demand mining model is trained, and the normalization rate, recall rate and F1 value corresponding to each demand mining model after model training are obtained. The model training is completed until the model normalization rate, model recall rate and F1 value set during the modeling item configuration are met. Otherwise, the model training continues.
[0112] It should be further explained that, in the specific implementation process, the demand mining model conducts data mining of the government resource data of the corresponding government platforms, and then obtains the government feedback demand information in the following process:
[0113] Set a data mining period for each government platform's corresponding demand mining model. During each data mining period, the demand mining model of each government platform conducts data mining on its corresponding government resource data;
[0114] Through the data mining, several sub-feedback demands corresponding to each government platform are obtained. Specifically, the sub-feedback demands include administrative feedback demands, public safety demands, public opinion monitoring demands, and people's livelihood service demands;
[0115] Among them, administrative management data corresponds to administrative feedback needs, which include administrative appeal time, administrative appeal events, and administrative appealers; public safety data corresponds to public safety needs, which include safety incident reporters, safety incidents, and safety incident reporting time; public opinion monitoring data corresponds to public opinion monitoring needs, which include public opinion reporters, public opinion events, and public opinion reporting time; people's livelihood service data corresponds to people's livelihood service needs, which include different types of people's livelihood service events, the scope of people's livelihood event services, and the list of people benefiting from people's livelihood services;
[0116] The administrative feedback needs, public safety needs, public opinion monitoring needs and people's livelihood service needs of each government resource data platform after data mining by the corresponding demand mining model are summarized as the government feedback demand information corresponding to each government platform.
[0117] It should be further explained that, in the specific implementation process, the process of matching the corresponding government work order information according to the government feedback demand information of each government platform includes:
[0118] Setting a work order library, wherein the work order library is used to store several types of work order information, including administrative problem handling work orders, public safety problem handling work orders, public opinion problem handling work orders, and people's livelihood problem handling work orders;
[0119] The administrative problem handling work order is used to solve administrative appeal events corresponding to a number of administrative feedback needs, the public security problem handling work order is used to solve security events corresponding to a number of public security needs, the public opinion problem handling work order is used to solve public opinion events corresponding to a number of public opinion monitoring needs, and the livelihood problem handling work order is used to solve different types of livelihood service events corresponding to a number of livelihood service needs;
[0120] The government feedback demand information corresponding to each government platform is input into the work order database, and then the administrative problem handling work orders, public safety problem handling work orders, public opinion problem handling work orders and people's livelihood problem handling work orders corresponding to the administrative feedback demand, public safety demand, public opinion monitoring demand and people's livelihood service demand in the government feedback demand information are matched from the work order database;
[0121] The administrative problem handling work orders, public safety problem handling work orders, public opinion problem handling work orders and people's livelihood problem handling work orders matched by each government platform from the work order database are merged to generate the government work order information corresponding to each government platform;
[0122] The government work order information is sent to the administrator, who arranges government personnel to handle all events recorded in the government work order information, namely administrative appeal events, security events, public opinion events, and people's livelihood service events, thereby achieving efficient analysis of government data and timely processing of events included in the government data, which improves the efficiency of government office work to a certain extent.
[0123] The present invention also provides a government affairs data analysis system, which includes:
[0124] The government data collection module is used to send data extraction applications to several government platforms and collect government resource data from the corresponding government platforms after the data extraction applications are approved;
[0125] The word meaning feature extraction module is used to perform word meaning analysis on the government resource data of each government platform, and then construct a word meaning feature set for each government platform;
[0126] A modeling feature data generation module is used to perform index analysis on the semantic feature sets of several government platforms and generate multi-dimensional modeling feature data of the corresponding government platforms;
[0127] The mining model building module is used to build the corresponding demand mining model based on the multi-dimensional modeling feature data of each government platform;
[0128] The data mining module is used to perform data mining of government resource data of the corresponding government platforms through the demand mining model, thereby obtaining government feedback demand information;
[0129] The feedback matching module is used to match the corresponding government work order information according to the government feedback demand information of each government platform.
[0130] The above embodiments are only used to illustrate the technical method of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.
Claims
1. A government data analysis method, characterized in that: The following steps are involved: Step S1: Send a data pull application to several government platforms, and collect government resource data of the corresponding government platforms after the data pull application is approved; Step S2: performing word meaning analysis on the government resource data of each government platform, and then constructing a word meaning feature set of each government platform; Step S3: performing index analysis on the semantic feature sets of several government platforms, and generating multi-dimensional modeling feature data of the corresponding government platforms; Step S4: construct a corresponding demand mining model based on the multi-dimensional modeling feature data of each government affairs platform; Step S5: The demand mining model performs data mining on the government resource data of the corresponding government platforms, thereby obtaining government feedback demand information; Step S6: Match the corresponding government work order information according to the government feedback demand information of each government platform.
2. A government data analysis method according to claim 1, characterized in that: The process of sending data extraction applications to several government platforms and collecting government resource data from the corresponding government platforms after the data extraction applications are approved includes: Several government platforms are numbered and recorded as i, i=1, 2, 3, ..., n, where n is a natural number greater than 0. Several data pulling applications are generated in a set application period and sent to several government platforms. When several government platforms receive any data pulling application, it is determined whether the IP address of the data pulling application is in a preset legal address form. If so, a pulling file is constructed according to the data pulling application, and the pulling file collects the government resource data corresponding to the corresponding government platform. If not, the corresponding data pulling application is included in a preset blacklist, and a new data pulling application is generated to replace the data pulling application included in the blacklist. A new pulling file is generated by the replaced data pulling application to collect the government resource data corresponding to the government platform; until the government resource data corresponding to several government platforms are collected, the government resource data includes administrative management data, public security data, public opinion monitoring data and people's livelihood service data.
3. A government data analysis method according to claim 2, characterized in that: The process of performing word meaning analysis on the government resource data of each government platform and then constructing the word meaning feature set of each government platform includes: The word meaning analysis includes word meaning information extraction, word bag model construction and word meaning feature matching; The content of word meaning information extraction is: setting extraction words corresponding to administrative management data, public safety data, public opinion monitoring data and people's livelihood service data, processing administrative management data, public safety data, public opinion monitoring data and people's livelihood service data into several word segments respectively, marking the parts of each word segment that are consistent with the extracted words as their corresponding word meaning information, and removing the parts of the word segment that are inconsistent with the extracted words; The contents of constructing the bag-of-words model are as follows: inputting the semantic information of each government platform into the pre-constructed initial neural network model, classifying the semantic information of the same frequency in each government platform into a group of terms, summarizing the terms of different frequencies into a preset blank statistical table, and then forming a term statistical table corresponding to each government platform, constructing a corresponding term sparse matrix according to the frequencies corresponding to several groups of terms in each term statistical table, vectorizing the text data of each term sparse matrix, and then obtaining several term feature vectors corresponding to each term sparse matrix, summarizing all the term feature vectors of each government platform, and then constructing a bag-of-words model of the government resource data corresponding to each government platform; The bag-of-words model consists of an administrative management sub-model, a public safety sub-model, a public opinion monitoring sub-model and a people's livelihood service sub-model. The bag-of-words model constructed corresponding to each government affairs platform is used to match the semantic features of the government affairs resource data corresponding to each government affairs platform. The semantic features include administrative management semantic features, public safety semantic features, public opinion monitoring semantic features and people's livelihood service semantic features. The administrative management semantic features, public safety semantic features, public opinion monitoring semantic features and people's livelihood service semantic features corresponding to each government affairs platform are summarized and constructed into their respective corresponding semantic feature sets.
4. A government data analysis method according to claim 3, characterized in that: The process of performing index analysis on the semantic feature sets of several government platforms and generating multi-dimensional modeling feature data of the corresponding government platforms includes: Set characteristic indicators corresponding to each government affairs platform. The characteristic indicators are used to perform indicator analysis corresponding to each government affairs platform. The characteristic indicators set for each government affairs platform include a first indicator, a second indicator, a third indicator and a fourth indicator. The first indicator maps out the first dimension modeling characteristic data corresponding to each government affairs platform, the second indicator maps out the second dimension modeling characteristic data corresponding to each government affairs platform, the third indicator maps out the third dimension modeling characteristic data corresponding to each government affairs platform, and the fourth indicator maps out the fourth dimension modeling characteristic data corresponding to each government affairs platform; The first dimension modeling feature data, the second dimension modeling feature data, the third dimension modeling feature data and the fourth dimension modeling feature data corresponding to each government affairs platform are aggregated to generate the corresponding multi-dimensional modeling feature data for each government affairs platform.
5. A government data analysis method according to claim 4, characterized in that: The process of building a corresponding demand mining model based on the multi-dimensional modeling feature data of each government platform includes: A modeling program is set up, and the program environment is initialized, modeling items are configured, and data filtering fields are entered for the modeling program. After the program environment initialization, modeling item configuration, and data filtering field entry of the modeling program are completed, the multi-dimensional modeling feature data corresponding to each government affairs platform is input into the modeling program, and then the modeling program constructs a demand mining model corresponding to each government affairs platform through deep learning technology; The first dimension modeling feature data, second dimension modeling feature data, third dimension modeling feature data and fourth dimension modeling feature data of each government platform are respectively processed with feature vectors to obtain the corresponding modeling feature vectors. The frequency of the terms corresponding to each modeling feature vector in the bag-of-words model is used as the modeling weight, and the modeling weight is annotated to the corresponding modeling feature vector. The modeling program sets the mining parameters of the corresponding type of each modeling feature vector and the computing power allocation ratio of the mining parameters according to the modeling weight, and then constructs the demand mining model corresponding to each government platform.
6. A government data analysis method according to claim 5, characterized in that: The process of using the demand mining model to mine the government resource data of the corresponding government platforms and then obtaining government feedback demand information includes: Set a data mining period for each government platform's corresponding demand mining model. During each data mining period, the demand mining model of each government platform conducts data mining on its corresponding government resource data, and obtains several sub-feedback demands corresponding to each government platform through data mining. The sub-feedback demands include administrative feedback demands, public safety demands, public opinion monitoring demands, and people's livelihood service demands. Among them, administrative feedback needs include administrative appeal time, administrative appeal events and administrative appealers; public safety needs include safety incident reporters, safety incidents and safety incident reporting time; public opinion monitoring needs include public opinion reporters, public opinion events and public opinion reporting time; people's livelihood service needs include different types of people's livelihood service events, the scope of people's livelihood event services and the list of people benefiting from people's livelihood services; The administrative feedback needs, public safety needs, public opinion monitoring needs and people's livelihood service needs of each government resource data platform after data mining by the corresponding demand mining model are summarized as the government feedback demand information corresponding to each government platform.
7. A government data analysis method according to claim 6, characterized in that: The process of matching the corresponding government work order information according to the government feedback demand information of each government platform includes: Set up a work order database, which is used to store several types of work order information, including administrative problem handling work orders, public safety problem handling work orders, public opinion problem handling work orders, and livelihood problem handling work orders; The administrative problem handling work order is used to solve administrative appeal events corresponding to a number of administrative feedback needs, the public security problem handling work order is used to solve security events corresponding to a number of public security needs, the public opinion problem handling work order is used to solve public opinion events corresponding to a number of public opinion monitoring needs, and the livelihood problem handling work order is used to solve different types of livelihood service events corresponding to a number of livelihood service needs; The government feedback demand information corresponding to each government platform is input into the work order database, and then the administrative problem handling work orders, public safety problem handling work orders, public opinion problem handling work orders and people's livelihood problem handling work orders corresponding to the administrative feedback demand, public safety demand, public opinion monitoring demand and people's livelihood service demand in the government feedback demand information are matched from the work order database; The administrative problem handling work orders, public safety problem handling work orders, public opinion problem handling work orders and people's livelihood problem handling work orders matched by each government platform from the work order database are merged to generate the government work order information corresponding to each government platform.
8. A government data analysis system, used to implement the government data analysis method according to any one of claims 1 to 7, characterized in that: The system includes: The government data collection module is used to send data extraction applications to several government platforms, and collect government resource data of the corresponding government platforms after the data extraction applications are approved; The word meaning feature extraction module is used to perform word meaning analysis on the government resource data of each government platform, and then construct a word meaning feature set for each government platform; A modeling feature data generation module is used to perform index analysis on the semantic feature sets of several government platforms and generate multi-dimensional modeling feature data of the corresponding government platforms; The mining model building module is used to build the corresponding demand mining model based on the multi-dimensional modeling feature data of each government platform; The data mining module is used to perform data mining of government resource data of the corresponding government platforms through the demand mining model, thereby obtaining government feedback demand information; The feedback matching module is used to match the corresponding government work order information according to the government feedback demand information of each government platform.
Citation Information
Patent Citations
Government affair service hotline analysis method and system
CN110111084A
E-government project evaluation expert group recommendation method based on multi-dimensional feature balance
CN113254642A
Government affair service data platform detection system based on machine learning
CN114116955A
Artificial intelligence government affair data review method and system
CN116307765A
Government affair item dialogue recommendation method based on multi-dimensional vector fusion
CN118227894A
