Management system, decision support system and equipment based on digital government big data mining analysis

Abstract

A management system, a decision support system and equipment based on digital government big data mining analysis belong to the technical field of data analysis and management. In order to solve the problem that the keyword extraction of the existing management system exists in the submerged common problem text and causes the serious problem to be weakened, the present application firstly carries out word segmentation and cleaning on the complaint data, then carries out clustering based on the cleaned data, determines the keyword selection index and extracts the keyword based on the number of the word x in the text corresponding to the problem data, the number of the corresponding text and the total number of the problem data corresponding to the text, takes the keyword as a node, uses a directed hypergraph generation tree algorithm to obtain a directed association rule graph of the keyword, and provides keyword query and display. On this basis, based on the keyword reflecting the problem and the personnel database, the jurisdiction area and the organization and position responsibility matching are carried out, and the information of the future occurrence of similar problems is given based on the Bayesian network prediction.

Description

Technical Field

[0001] This invention belongs to the field of data analysis and management technology, specifically designing an information management system, a decision support system, and equipment. Background Technology

[0002] "Internet Plus Government Services" has now become the most powerful tool for modernizing urban governance. Data mining based on "Internet Plus Government Services" data can promptly identify problems in various aspects of urban management, thereby quickly identifying the most widespread and impactful issues. This provides a scientific and reasonable basis for decision-making, enabling managers to address problems in a timely manner, prioritizing them from most important to least important, based on limited resources and time. Alternatively, it can facilitate the allocation of resources and resources to address specific problems.

[0003] With the development of internet technology, most cities have established management platforms, even grid-based management platforms, forming a five-level closed-loop linkage of grids, communities, streets, counties / districts, and the city level. This achieves closed-loop management across seven stages: information collection, case establishment, task assignment, task processing, processing feedback, verification and case closure, and performance evaluation. While this approach provides city managers with a data foundation for problem-solving and improves efficiency, it's undeniable that many issues remain to be addressed in urban management, such as transportation, environmental sanitation, and noise pollution. Furthermore, with rapid urban development, the number of problems requiring urban management is increasing, and the scope of these problems is expanding, sometimes even to unprecedented levels. This increases the difficulty of urban management, leading to challenges involving multiple districts and departments. The inability of different districts and departments to quickly establish coordinated workflows not only results in lengthy problem-solving times but also significantly increases the difficulty of resolving issues.

[0004] With rapid urban development, the surge in grid-based management data and its increasing complexity have led to a rapid increase in data volume. The amount of data on urban management platforms is staggering. If relying solely on manual labor to extract and report problems one by one, requiring managers to allocate resources and departments to resolve them, the sheer volume of reported data would be overwhelming, placing a tremendous workload on management. More importantly, some issues that seem urgent to a few individuals may not be considered urgent or even problems at all by others. For example, some people might complain about noise pollution near a morning market, but for the majority, it might not be a problem. This means that some situations are significantly influenced by personal factors. If these issues are indiscriminately reported and resolved on the management platform, it could not only waste administrative resources but also fail to address situations where individuals perceive them as problems due to personal biases. More importantly, this approach fails to address the degree and scope of impact of specific issues, preventing managers from prioritizing and allocating resources accordingly. Consequently, problems with a large and widespread impact cannot be addressed promptly, unnecessarily prolonging resolution time and damaging the manager's competence and credibility. While this could be addressed by assigning severity levels to those collecting and reporting information, this method relies solely on the subjective assessment of the "problem's" importance. This subjective bias and limited understanding lead to an unobjective approach, and it fails to reflect the scope of the problem's impact (the number of people affected). Consequently, problems with a wide impact and urgent need for resolution are downgraded in importance, unnecessarily extending resolution time and further impacting the manager's competence and credibility. Furthermore, this method of reporting and resolving issues one by one is not conducive to the unified resolution of similar problems. For example, in a certain district, there is a sewage pollution problem in area A, and area B has the same or similar problems. If we rely on manpower to report them one by one or select them to report according to their importance, regardless of the method, it is impossible for manpower to collect and report all the information in one day. Therefore, it may be possible that problem A is reported on one day, approved by the manager and the relevant departments and resources are arranged to resolve it. However, if problem B is found on another day and needs to be reported again for the same or similar issues, then the manager will need to approve and arrange the relevant departments and resources to resolve the same or similar issues again. This not only wastes administrative resources to a certain extent, but also seriously affects the efficiency of resource allocation and prolongs the time for problem resolution.

[0005] Of course, with the development of data mining technology, computer technology can be used to cluster and rank the importance of problem data, which can alleviate the above problems to some extent. However, current information importance ranking methods are not well applied to the "problem feedback / complaint" function / system / module of urban management platforms. Among them, the TF-IDF (term frequency–inverse document frequency) method, which is commonly used in information processing systems, is often used to mine information importance. However, it is worth noting that this information mining method is more suitable for mining important information in documents. This is because the IDF statistical method regards documents that do not appear more frequently as documents with higher importance. This method can cause information with more complaints to be buried to a certain extent. However, it is precisely the problems with more complaints that have a greater impact on people's lives and are more serious. In other words, this method can lead to serious problems being weakened in handling. This may seriously affect the information processing of the "problem feedback / complaint" function / system / module of the management platform, and even cause a certain degree of deviation from the actual situation. Summary of the Invention

[0006] To address the problem that keyword extraction in existing management systems often results in the overloading of common problematic texts, leading to the weakening of critical processing, this invention proposes a management system and a corresponding decision support system based on big data mining and analysis in digital government.

[0007] A management system based on big data mining and analysis in digital government includes:

[0008] Data acquisition and word segmentation module: Based on the time period information to be analyzed, the module acquires problem data from the management platform database, with each problem data corresponding to a problem data ID; then, the acquired problem data is segmented and part-of-speech tagging is performed.

[0009] Data cleaning module: used to clean the word segmentation results;

[0010] Location area determination module: Based on the cleaned data, the module determines the location area of ​​each data point reflecting a problem, and "binds" the location area to the data point reflecting the problem, that is, establishes a mapping relationship between the data point reflecting the problem and the location area corresponding to the problem, and stores it.

[0011] Basic Information Database: Stores and manages administrative geographical factors and corresponding regional boundaries, natural geographical factors and corresponding regional boundaries, streets and alleys and corresponding segmented regional boundaries, as well as infrastructure and corresponding geographical locations within the administrative region.

[0012] Keyword extraction module: Based on the cleaned data, feature words are obtained by clustering nouns; then, keywords are extracted based on the clustered feature words; during the keyword extraction process, keyword selection metrics are used to select keywords, as follows:

[0013]

[0014] in, This represents the number of times word x appears in the text corresponding to the data reflecting the problem, and N represents the total number of texts corresponding to the data reflecting the problem. This indicates the number of texts corresponding to the problem data containing the word x; This is the proportionality coefficient.

[0015] Directed Association Rule Graph Generation Module: Using keywords as nodes, the module generates a directed association rule graph of keywords using a directed hypergraph spanning tree algorithm.

[0016] Keyword Query and Display Module: This module receives query terms input by administrator users, retrieves the most matching keywords based on the query terms, and prominently displays the directed relationships of the confirmed keyword nodes in the directed association rule graph of the keywords; alternatively, it receives pointer action information from administrator users in the directed association rule graph of the keywords, confirms the keyword node pointed to by the pointer based on the pointer action information, and prominently displays the directed relationships of the confirmed keyword nodes in the directed association rule graph of the keywords.

[0017] Furthermore, in determining the location area where each piece of problematic data occurred, if the respondent who reported the problematic data filled in the corresponding location area, the location area is directly determined; if the respondent who reported the problematic data reported the problem directly without filling in the determined location area, the location area is determined by information extraction.

[0018] Furthermore, the method of determining the location region through information extraction employs information matching; the specific process of determining the location region using information matching is as follows:

[0019] Based on the order of expression of the data reflecting the problem, each word in the word segmentation result is matched with the address feature word. If a word in the word segmentation result can be directly and completely matched with the address feature word, or if multiple consecutive words in the word segmentation result can be completely matched with the address feature word, then the corresponding address feature word is used as the location information word. Based on the location information word, the geographical location and spatial range corresponding to the problem are determined.

[0020] Furthermore, the data cleaning module's cleaning process includes a step of removing stop words.

[0021] A management device based on big data mining and analysis of digital government is disclosed. The device includes a processor and a memory. The memory stores at least one instruction, which is loaded and executed by the processor.

[0022] A decision support system based on big data mining, analysis, and management in digital government includes:

[0023] Organization and Personnel Database: Stores the organizational relationships of all departments in each jurisdiction, as well as their corresponding job responsibilities.

[0024] Demand information receiving module: used to obtain keywords and directed association rule graphs obtained by the directed association rule graph generation module of the management system based on digital government big data mining and analysis as described in any one of claims 1 to 4; at the same time, it provides keyword options to users and provides corresponding directed association rule graphs according to the keyword options;

[0025] The first support suggestion module: Based on the keywords confirmed by the user and the corresponding directed association rule graph, it first identifies the keywords corresponding to the keyword nodes and determines whether the words belong to the "problem" related to the reflected problem or the "location" related to the reflected problem; then, based on the words related to the "location" of the reflected problem and the words related to the "problem", it matches the jurisdiction, organization, and job responsibilities in the institution and personnel database.

[0026] Furthermore, based on words related to the "location" of the problem and words related to the "problem," during the matching of jurisdiction, organization, and job responsibilities in the database of institutions and personnel, for words related to the "location" of the problem, the ID corresponding to the problem data is determined according to the problem data corresponding to the keywords reflected by the nodes belonging to the "problem," and the location area where the problem data occurred is obtained from the location area determination module. Based on the location area where the problem occurred, a jurisdiction matching recommendation corresponding to the solution that can be implemented is given.

[0027] Furthermore, the system also includes a second support suggestion module: based on the keywords confirmed by the user and the corresponding directed association rule graph, it extracts topic and scenario information, obtains all data of the corresponding complaint case according to the topic and scenario information determined by the scenario and topic determination module, analyzes the data and provides information on the future occurrence of similar problems, including whether it will occur in the future and / or the expected location, thereby serving as potential decision support suggestions.

[0028] Furthermore, the second support suggestion module uses a Bayesian network to analyze the data and provide information on the future occurrence of similar problems.

[0029] A decision support device based on big data mining, analysis and management of digital government is disclosed. The device includes a processor and a memory. The memory stores at least one instruction, which is loaded and executed by the processor.

[0030] Beneficial effects:

[0031] This invention provides a management system based on digital government big data mining and analysis. It segments and cleans complaint data, then clusters and extracts keywords based on the cleaned data. During this process, keyword selection indicators are determined based on the number of instances of word x in the text corresponding to the problem data, the total number of texts corresponding to the problem data, and the number of texts corresponding to word x. This effectively solves the problem of existing keyword extraction methods that overwhelm common problem texts, leading to the weakening of the processing of serious issues. This invention effectively "reflects" the importance of feature words, thus effectively extracting problems with a wide impact and greater severity. Furthermore, because this digital government big data mining and analysis management system can accurately and effectively identify the complaint issues, it provides effective decision support for decision-making. It improves the information processing capabilities of the management platform's "problem feedback / complaint" function / system / module, enabling targeted and efficient matching and response to these issues, thereby providing a foundation for improving the management efficiency and level of managers. Attached Figure Description

[0032] Figure 1 The image shows screenshots of real complaint cases collected from the urban grid management platform of a certain district in a certain city between July 2022 and July 2023.

[0033] Figure 2 This is a part-of-speech distribution chart of complaint case data for a certain district on the management platform.

[0034] Figure 3 This document describes a vocabulary list and part-of-speech bar chart for complaint case data in a certain district of the management platform.

[0035] Figure 4 This is a word cloud of keywords for complaint case data in a certain district of the management platform.

[0036] Figure 5 This is a diagram showing the association rules for complaint cases in a certain district of the management platform.

[0037] Figure 6 This is for the management platform's keyword (Xipan Community) association rule graph node data.

[0038] Figure 7 This is a directed relationship graph for the keyword (Xipan Community).

[0039] Figure 8 This is a directed association rule graph of keywords for complaint cases in the Xipan residential area of ​​a certain district of the management platform.

[0040] Figure 9 This is a word cloud of 12345 complaint cases in a certain city from January to August 2023.

[0041] Figure 10 for Figure 9 A bar chart showing the number of keywords that appear most frequently in a word cloud.

[0042] Figure 11 This is a graph of directed association rules determined based on 12345 complaint cases.

[0043] Figure 12 This is a Bayesian network prediction diagram for pipelines and ducts.

[0044] Figure 13 The Bayesian network prediction graph corresponding to the reasons for gas-related complaints.

[0045] Figure 14 This is a Bayesian network prediction graph of gas-fired areas within a time frame. Detailed Implementation

[0046] This invention proposes a management system based on big data mining and analysis in digital government, and a decision support system for solving corresponding problems. This system is an embedded system / subsystem or functional unit of an urban grid management platform, providing a foundation for the urban management platform. Through big data analysis, it guides urban management services, assists in building smart cities, and improves the level of urban management. For the urban grid management platform, information collection is the starting point for all tasks of the urban governance supervision and command platform, and also the foundation of the system described in this invention. In the case of the urban grid management platform, the main subjects of information collection are the general public and / or urban management supervisors.

[0047] For ordinary citizens, issues can be reported through the user registration portal of the urban management platform or as a guest on the platform or hotline. For example, the public can report urban management problems through WeChat official accounts or citizen hotlines. In some cases, the platform's back-end center generates a case file based on the report, which can then be sent to the currently patrolling urban management supervisor for verification. The supervisor will decide whether to file a case based on the verification results. For urban management supervisors, they use their mobile phones to patrol their designated areas according to pre-defined rules. Once a problem is discovered, they use the app to collect and report information. After these issues are handled by government departments or enterprises, in some cases, urban management supervisors can conduct on-site verification and report the results back to the system.

[0048] The urban management platform falls under the category of the municipal-level digital urban management system. Therefore, this invention can be used as an embedded system / subsystem of a specially developed urban management platform. It can be integrated with the provincial-level digital urban management system as needed, and the provincial-level digital urban management platform can also serve as a data source for information collection.

[0049] It should be noted that when this invention is developed as an urban management platform, it is actually a functional unit within the platform, or a specially developed embedded system / subsystem of the urban management platform. The corresponding information is collected and stored in the problem reporting database. Depending on actual needs, if it is necessary to confirm the problem and generate a case file, a separate database can be designed to store the confirmed problem reporting data, denoted as the confirmed problem reporting database. Alternatively, the confirmed problem reporting data can be marked and stored within the problem reporting database. For ease of description, the database containing the problem reporting data to be processed will be uniformly referred to as the problem reporting database.

[0050] In addition, the present invention also designs a basic information database, which stores information such as administrative geographical factors and corresponding regional boundaries, natural geographical factors and corresponding regional boundaries, streets and alleys and corresponding segmented regional boundaries, and infrastructure names and corresponding geographical locations and boundaries within the administrative region.

[0051] The following description uses a specific implementation method, taking a city's urban grid management platform as an example. Specific implementation method one:

[0053] This implementation method is a management system based on big data mining and analysis in digital government, including:

[0054] Data acquisition and word segmentation module: Based on the time period information to be analyzed, the module acquires the data reflecting the problem from the management platform database. Each data reflecting the problem corresponds to a problem data ID. Then, the acquired data reflecting the problem is segmented and part-of-speech tagging is performed.

[0055] This implementation method analyzes real complaint case data collected from the urban grid management platform of a certain district in a certain city from July 2022 to July 2023. The real complaint case data collected from the urban grid management platform of a certain district in a certain city from July 2022 to July 2023 is as follows: Figure 1 As shown in the figure. Regarding data analysis, the word segmentation and part-of-speech tagging results of the case data reflecting the issues are as follows: Figure 2 As shown in Table 1, the basic data information is as follows.

[0056] Table 1

[0057]

[0058] Data cleaning module: used to clean the word segmentation results; the cleaning process includes a step to remove stop words, which can be set and expanded / deleted according to actual needs.

[0059] Removing stop words is not only to reduce computation, but also to better facilitate the processing of the feature word extraction module.

[0060] Location area determination module: Based on the cleaned data, the module determines the location area of ​​each data point reflecting a problem, and "binds" the location area to the data point reflecting the problem, that is, establishes a mapping relationship between the data point reflecting the problem and the location area corresponding to the problem, and stores it.

[0061] In determining the location area of ​​each piece of data reflecting a problem, if the respondent has filled in the corresponding location area, the location area is determined directly; if the respondent has directly reported the problem without filling in a specific location area, the location area is determined by information extraction.

[0062] The method of determining the location region by information extraction can be either semantic recognition or information matching.

[0063] Considering that government service platforms such as urban grid management platforms have relatively short information and relatively fixed formats, some implementations use information matching to determine the location area. The specific process is as follows:

[0064] Based on the order of expression of the data reflecting the problem, each word in the word segmentation result is matched with the address feature word. If a word in the word segmentation result can be directly and completely matched with the address feature word, or if multiple consecutive words in the word segmentation result can be completely matched with the address feature word, then the corresponding address feature word is used as the location information word. Since the address feature words all have corresponding regional boundaries or geographical locations and boundaries, the geographical location and spatial range corresponding to the problem can be determined based on the location information word.

[0065] It should be noted that even a single data ID may correspond to an issue that involves two or even three jurisdictions, meaning the identified location area could be two or even three. For example, if a wastewater pollution problem occurs at the intersection of two or even three jurisdictions, then the identified location area of ​​the problem data could be two or even three.

[0066] Address keywords refer to the names of administrative geographical factors, natural geographical factors, streets and alleys and corresponding segmented areas, and infrastructure in the basic information database;

[0067] Basic Information Database: Stores and manages basic information such as administrative geographical factors and corresponding regional boundaries, natural geographical factors and corresponding regional boundaries, streets and alleys and corresponding segmented regional boundaries, and infrastructure and corresponding geographical locations and boundaries within the administrative region; among which,

[0068] Administrative geographical factors include districts, counties, towns, and villages;

[0069] Natural geographical factors include rivers, lakes, and mountains;

[0070] The infrastructure includes residential areas (including hotels, guesthouses, restaurants, etc.) and landmarks.

[0071] The term "community" not only includes the address range of the entire community, but also includes the addresses and ranges of individual building numbers.

[0072] The address of the marker includes the following:

[0073] a) Memorial sites and buildings with geographical significance, including buildings, squares, sports facilities, parks and green spaces, memorial sites, scenic spots and historical sites, etc.

[0074] b) Units and compounds with geographical significance, including hospitals, schools, and other organizations;

[0075] c) Transportation facilities with geographical significance, including bridges, road roundabouts, and transportation hubs.

[0076] In this embodiment, the geographic information data (administrative geographic factors, natural geographic factors, streets and alleys, and infrastructure, etc.) in the basic information database are designed based on the national standards "Classification and Code of 1:500, 1:1000, and 1:2000 Topographic Map Elements" and "1:500, 1:1000, and 1:2000 Topographic Map Symbols," combined with the requirements of the Ministry of Housing and Urban-Rural Development and relevant provincial standards. The information in the basic information database is encoded and stored according to a unit grid.

[0077] Unit grid data refers to data on basic urban management jurisdictional responsibility units divided according to certain rules. Unit grid data can include geographic unit data of human activities, natural geographic unit data such as undeveloped mountains or river basins, patchy geographic unit data of residential communities under the jurisdiction of urban residential areas or commercial areas, and strip-shaped geographic unit data of special jurisdiction such as highways and railways, etc. The unit grid is the spatial location and jurisdictional responsibility division unit for urban management components and events, and is one of the benchmarks for locating urban management objects. In its division, the unit grid needs to cover every inch of urban land, with no blind spots. The formation of unit grid data includes the formation of geospatial division information data and grid coding data.

[0078] The process of dividing the unit grid follows the requirements of the national industry standard "Unit Grid Division and Coding Rules for Urban Administrative Comprehensive Supervision Information System" (CT / T213-2005) issued by the Ministry of Construction. It combines local conditions and the differentiated needs of various departments to divide the geographic unit grid and draw the unit grid atlas.

[0079] Based on the principles of unit grid division and considering the actual situation of a certain city, the following principles are mainly followed:

[0080] 1. Principle of local jurisdiction

[0081] The division and coding of unit grids follow the existing administrative jurisdiction hierarchy, with three levels: street office, community, and unit grid. The maximum boundary of a unit grid is the community boundary, and it should generally not cross the community boundary.

[0082] 2. Geographical Layout Principles

[0083] The boundaries are divided according to the layout of natural units such as roads, residential land, public facilities land, public green spaces, squares, bridges, open spaces, rivers, hills, and lakes in the city; the outer boundary line of the community is used as the boundary; the center line of the road is used as the boundary line; and if it is a mountain, the ridge line is used as the boundary.

[0084] 3. Principle of Current Status Management

[0085] To ensure the integrity of buildings and management components and facilitate effective management, the boundaries of the unit grid should not cross buildings or management components.

[0086] 4. Load balancing principle

[0087] The grid division should take into account the manageable unit range of a single supervisor's workload, as well as the differences in the density of urban components and the frequency of events. For grids located in built-up areas and patrolled on foot, the grid boundary and area should be determined based on the total length of patrol roads within the grid, which is about 8 kilometers. For grids in non-built-up areas or those patrolled by vehicle, the area range of a unit grid can be expanded to about 2 million square meters.

[0088] 5. Principle of convenient accessibility

[0089] To facilitate management implementation, management paths should be made as convenient as possible, while also considering easy accessibility by foot or other means of transportation. The design and division of the unit grid, based on the principle of uniformity and openness, should not only be suitable for urban management but also provide a compatible operating platform for extended applications in multiple fields such as public security management, emergency medical services, special surveys, and urban emergency response. Each unit grid serves as both a geographical unit for citizens' daily lives and a geographical unit for decision-making and analysis applications in urban management. The principle of seamless splicing applies; the division of unit grids should be neither gap-bound nor overlapping. The principle of relative stability applies; the division of unit grids should maintain stable verification.

[0090] The unit grids corresponding to the information in the basic information database are designed according to the requirements of the geocoding census. Each code in a unit grid is unique in both time and space, and its original code should not be used when a unit grid is changed. Since the city is a developing urban area, the boundaries and number of unit grids may be adjusted appropriately as the built-up area expands. Therefore, when coding unit grids, it is necessary to consider the possibility of grid splitting and merging in the future, and reserve codes for any newly added grids.

[0091] The sequence codes of the grid cells are spatially encoded according to a left-to-right, top-to-bottom principle. The grid cell atlas is based on topographic maps and divided according to specific principles. Each grid cell is assigned a grid code and then printed as a map. Each grid cell in the atlas includes attribute descriptions.

[0092] The matching method in the location area determination module not only ensures the accuracy of matching regions reflected by administrative and geographical factors, but also effectively avoids address information matching errors caused by word segmentation. In particular, it can effectively avoid incorrect matching for residential communities in infrastructure. For example, when segmenting the Riverside Community, it will be divided into "Riverside" + "Community". If the location is matched using "Riverside" or "Community", errors may occur. In addition, this method can more effectively avoid incorrect matching of address feature words composed of verbs + nouns. For example, Guanhu Community, Guanhu International, Guanjiang International, etc., will be split into "Guan" + other nouns when segmenting words. If the matching is performed only based on the noun part of speech, errors may occur.

[0093] Keyword extraction module: Based on the cleaned data, clustering is performed according to nouns to obtain feature words; then, keywords are extracted based on the clustered feature words.

[0094] Taking the complaint case data of a certain district in a certain city's urban grid management platform from July 2022 to July 2023 as an example, this dataset is referred to as the case dataset. Analysis revealed that the cleaned data contained a large number of nouns; "organization / group name + noun / verb + noun" accounted for 70.47% of the data. Therefore, this invention uses nouns as keywords for analysis, which reduces ambiguity. Custom terms, due to their strong personalization and inconsistent standards, are not analyzed in detail. The basic information of the real complaint case data collected from the urban grid management platform of a certain district in a certain city from July 2022 to July 2023 is as follows: Figure 3 As shown, the feature word list ultimately yielded 8250 entries.

[0095] In some embodiments, the existing TF-IDF (term frequency–inverse document frequency) method has been commonly used to extract keywords. However, manual verification revealed that the effectiveness of this keyword extraction method differs significantly from the content that the actual problem data aims to express. Further research and analysis showed that this method is unsatisfactory primarily because the way the actual problem data is presented differs considerably from the traditional IDF statistical method. This is because IDF is an inverse text frequency index. , where N represents the total number of documents, and DF(x) represents the number of documents containing the term x. Through research, it is found that the fewer the number of documents with a certain keyword, or the relatively fixed number of documents with a certain keyword while the total number of documents is larger, the greater the corresponding IDF value. However, for government affair processing platforms / systems, especially the function of feeding back municipal problems, the more times the same problem is reflected, the greater the impact of the problem on the lives of the masses and the more serious the problem is, which means the number of such problems (which can be understood as the corresponding number of documents) will be larger. If the IDF method is followed, common problems may be submerged to a certain extent, which is inappropriate. The fundamental reason is that IDF itself is a measure of the general importance of a word, and its specific formula form is designed for documents to determine the importance of certain words. Because for documents, a higher word frequency is not always better, as there are many words such as "de" and "le" in documents, and almost every document contains these words. To eliminate the influence of these unimportant high-frequency words, some scholars use TF-IDF to extract keywords. Although the present invention may not perform stop word removal processing to cooperate with TF-IDF for keyword extraction, this will still cause the problem of submerging common documents (which are actually texts reflecting problems) to a certain extent.

[0096] After analyzing the real complaint case data collected on the urban grid management platform of a certain district in a certain city from July 2022 to July 2023, the present invention finds that the texts reflecting problems are generally not too long, among which the texts with 3-5 sentences account for the largest proportion in the problem reflection data, and the number of words contained in each sentence is not too large, with an average sentence length of 12.36 words, which is about 10 words after removing stop words. On this basis, the present invention uses a keyword selection indicator to select keywords, and the keyword selection indicator is as follows:

[0097]

[0098] wherein, represents the number of term x in the text corresponding to the problem reflection data where term x is located, and N represents the total number of texts corresponding to the problem reflection data, represents the number of texts corresponding to the problem reflection data where term x is located. is a proportional coefficient, which is used to adjust the keyword selection indicator and can be determined according to the actual situation of the problem reflection dataset. For example, if the population of some cities is too large or the city scale is too large, the amount of problem reflection data will increase correspondingly, and the scope of the reflected problems will be wider, which will lead to a relatively smaller proportion of some problems. When the keyword selection indicator is relatively small, it is not conducive to the expression of the overall data, so adjustment through the proportional coefficient is required. For the case dataset described in this embodiment, the proportional coefficient here is set to 2.

[0099] Based on the aforementioned complaint case data, and through research and analysis of the text corresponding to each reported issue and its word count, it was found that the text corresponding to each reported issue generally contains 3-10 words, rarely exceeding 15. Therefore, this invention utilizes... To determine the amount of expression of the qualifier x in the text that reflects the problem data. This is equivalent to the average expression level of word frequencies in the question data, used to measure the importance of a word in the question data. The formula utilizes... The format can control the expression of 0-15 words in the corresponding text between 0-1, which is equivalent to a probability-like processing. At the same time, it limits 3-7 words to a more "linear" expression, and makes the words with 1-10 words close to the two ends "prefer" compressed, thus playing a certain limiting role. For the words with 11-15 words close to the two ends, the expression is compressed "to a greater extent". Keyword selection metrics are used to measure the frequency of occurrence of data related to a particular issue within all data, reflecting the importance and scope of an issue. It can effectively "reflect" the importance of feature words.

[0100] For each noun after noun clustering, sort them according to the selection index corresponding to the word, and extract keywords in the sorting order.

[0101] Based on the noun clustering results above, the most frequent keywords extracted were market, streamside, property management, disturbance, rights protection, and noise. The keyword cloud of the case dataset is shown below. Figure 4 As shown, a clear clustering effect can be seen, with the key communities being the sources of complaints and the main issues being readily apparent.

[0102] Directed Association Rule Graph Generation Module: Using keywords as nodes, the module generates a directed association rule graph of keywords using a directed hypergraph spanning tree algorithm.

[0103] For the case dataset described in this embodiment, keywords are treated as nodes, and edges are connected between keywords appearing in the same case. The direction of the edges is marked according to the attributes of the left and right neighboring words, thus completing the transformation from keyword data to a directed hypergraph, resulting in a directed association rule graph of keywords. The association rule graph of complaint cases in a certain district of the management platform is as follows: Figure 5 As shown, the communities with the most complaints in a certain district are Cultural Community, Daqiao Community, and Xipan Community. The department with the most complaints is the Market Supervision Bureau. The most concentrated issues are rights protection issues, disturbance issues, noise issues, and property management issues. Figure 6 To manage the node data of a keyword (Xipan Community) in the management platform, the directed relationship graph is as follows: Figure 7As shown, the strongest association rule is the noise, disturbance, and property management issues reported in the Xipan community. Below, using Xipan as an example, we continue data mining; the basic data information is shown in Table 2.

[0104] Table 2

[0105]

[0106] Figure 8 This is a directed association rule graph of keywords for complaints about the Xipan residential community in a certain district of the management platform. As shown in the graph, the strongest association rule is: noise, disturbance, and property management issues in Xipan community complaints. Further investigation by our staff revealed the following: 1. The noise problem stems from Xipan community's proximity to Wanda Plaza, where three dance teams practice square dancing every night using loudspeakers regardless of the weather. 2. Xipan community comprises three sections: A, B, and C. Sections B and C are older sections, while section A is a newly renovated area with a low occupancy rate, resulting in incomplete coverage of property management fees. Furthermore, the property management company's offices are located in section B, leading to slower responses to certain requests from section A. Complainants are primarily residents of section A, and the issues raised regarding property management inaction mainly focus on community sanitation, construction work, clutter, and inadequate garden management. 3. The noise pollution issues raised by residents of Xipan Community mainly concern urban appearance and sanitation. This is because Xipan Community is adjacent to Wanda Plaza and a pedestrian street, where businesses frequently encroach on the road. Additionally, the excessive smoke from roadside barbecues is a major source of complaints they have made to the urban management department. Other issues, such as road maintenance, are also supported by facts. Furthermore, the missing manhole covers on the main road near the community, causing excessive noise from vehicles driving over them at night, are indeed present. In conclusion, communication with business experts indicates that our data mining results are largely consistent with the actual situation.

[0107] Based on platform data of the "Internet + government services" type, the system of this invention can directly use the platform database or establish a data sharing library. It can build a basic information database according to real business scenarios, generate high-dimensional big data based on platform data / shared database data and self-built thesaurus, and perform data mining. This can effectively discover problems reflected by the public in urban management, provide data support for solving urban management problems, effectively improve the efficiency of urban emergency response or problem handling, and thus improve the level of urban management.

[0108] The management system based on big data mining and analysis of digital government described in this embodiment, as an embedded system / subsystem or functional unit of the urban grid management platform, can effectively provide a data analysis foundation for the entire management platform. It is one of the most widely influential governance models and policy tools in urban management and grassroots governance, enabling the platform as a whole to solve grassroots problems in urban governance, respond to the ever-changing public expectations, and alleviate the increasing pressure of urban governance. It plays an important role and exerts a significant influence in urban facility maintenance, community safety, public services, community administration, employment assistance, grassroots security, information collection and maintenance, and other tasks.

[0109] Keyword Query and Display Module: This module receives query terms input by administrator users, retrieves the most matching keywords based on the query terms, and highlights the directed relationships of the confirmed keyword nodes in the directed association rule graph of the keywords; alternatively, it receives pointer action information from administrator users in the directed association rule graph of the keywords, confirms the keyword node pointed to by the pointer based on the pointer action information, and highlights the directed relationships of the confirmed keyword nodes in the directed association rule graph of the keywords.

[0110] During the process of confirming the keyword node pointed to by the pointer based on the pointer information, the drag information of the pointer (the drag action and direction of the pointer) is collected; in the directed association rule graph of the keyword, according to the drag information of the pointer (the drag action and direction of the pointer), the directed association rule graph of the keyword is scrolled according to the connection relationship of the nodes, and the final confirmation of the keyword node pointed to by the pointer is received. Specific Implementation Method Two:

[0112] This embodiment is a management device based on big data mining and analysis of digital government. The device includes a processor and a memory. It should be understood that it includes any device including a processor and a memory described in this invention. The device may also include other units and modules that perform display, interaction, processing, control and other functions through signals or instructions.

[0113] The memory stores at least one instruction, which is loaded and executed by the processor to form the management system based on big data mining and analysis of digital government.

[0114] It should be understood that the instructions include computer program products, software, or computerized methods corresponding to any method described in this invention; the instructions can be used to program computer systems or other electronic devices. The memory may include a readable medium on which instructions are stored, and may include, but is not limited to, magnetic storage media, optical storage media; magneto-optical storage media include read-only memory (ROM), random access memory (RAM), erasable programmable memory (e.g., EPROM and EEPROM), and flash memory layers, or other types of media suitable for storing electronic instructions. Specific implementation method three:

[0116] This embodiment is a decision support system based on digital government big data mining and analysis management. It is the decision support system corresponding to the management system based on digital government big data mining and analysis described in Specific Embodiment 1. That is, it generates decision support suggestions based on the information provided by the directed association rule graph generation module of the management system based on digital government big data mining and analysis.

[0117] This embodiment describes a decision support system based on big data mining, analysis, and management in digital government, comprising:

[0118] Organization and Personnel Database: Stores the organizational relationships and corresponding job responsibilities of all departments within each jurisdiction. Specifically, it includes: Department Table, Department Relationship Table, Job Title Table, Role Table, Employee Table, Employee and Role Relationship Table, Department Employee Table, Employee Job Relationship Table, and Role Relationship Table. It describes the composition of departments and personnel, depicting the relationships and changes between departments, personnel, jobs, and roles, such as job and personnel transfers. Roles determine employee permissions, facilitating information authorization. Organization and personnel data can be independent of process management while simultaneously providing a foundation for it.

[0119] Demand information receiving module: used to obtain keywords and directed association rule graphs obtained by the directed association rule graph generation module of the management system based on digital government big data mining and analysis; at the same time, it provides keyword options to users and provides corresponding directed association rule graphs according to the keyword options.

[0120] The first support suggestion module: Based on the keywords confirmed by the user and the corresponding directed association rule graph, it first identifies the keywords corresponding to the keyword nodes and determines whether the words belong to the "problem" related words or the "location" related words of the problem reflected; then, based on the "location" related words and the "problem" related words, it matches the jurisdiction, organization, and job responsibilities in the organization and personnel database.

[0121] For example, for keywords related to "location" reflecting a problem, based on the problem data corresponding to the keywords in the "problem" node, the ID of the problem data is determined, and the location region determination module is used to retrieve the location region where the problem data occurred. Based on the location region, a jurisdictional matching recommendation for a feasible solution is provided. It should be noted that even for a single problem data ID, the location region determination module may determine two or even three location regions where the problem data occurred. Therefore, the jurisdictional recommendation result may also be two or three, matching different jurisdictions and departments within those jurisdictions simultaneously. Furthermore, for keywords related to the "problem" reflected in the problem, such as "noise" or "environment," the corresponding department for solving the problem is matched, along with the responsible person who provides suggestions based on their job duties.

[0122] When there is more than one recommended district, the corresponding resolution task (after approval) is assigned to the corresponding district by the task allocation system on the management platform, and then the corresponding district assigns it to the relevant resolution department for resolution. The task allocation system can also be other embedded systems / subsystems or functional units on the management platform besides the system described in this invention. Since this invention does not focus on task allocation, it will not be described in detail here.

[0123] The second support suggestion module extracts topic and scenario information based on user-confirmed keywords and corresponding directed association rule graphs. It then obtains all data for the corresponding complaint cases based on the topic and scenario information determined by the scenario and topic determination module. The module analyzes the data and provides information on the future occurrence of similar problems, including whether they will occur in the future and / or the expected location, thus serving as potential decision support suggestions.

[0124] The process of analyzing data and providing information on the future occurrence of similar problems can employ methods such as Bayesian networks and neural networks. This implementation uses Bayesian networks as an example to provide potential decision support suggestions. This not only addresses the aftermath of a problem's occurrence but also provides suggestions for future resource allocation to mitigate its impact. Furthermore, it allows for proactive resource screening to reduce the likelihood of such problems arising.

[0125] To illustrate the effectiveness of the management system and decision support system based on digital government big data mining and analysis described in this invention, particularly the accuracy, effectiveness, and universal applicability of keyword extraction in the management system (which is the foundation for the effectiveness of decision support recommendations), this invention uses data from Embodiment 2. Embodiment 2 addresses approximately 910,000 12345 complaint cases in a certain city from January to August 2023. After effective data cleaning, a total of 115,734 words were retained, with 14,030 common feature words. A word cloud was created using nouns as keywords, as shown below. Figure 9 As shown in the image, the number of keywords that appear most frequently in the word cloud is as follows: Figure 10 As shown in the figure. The directed association rule graph obtained based on a management system using big data mining and analysis of digital government is as follows. Figure 11 As shown.

[0126] This embodiment provides decision support for pipeline systems. It uses a Bayesian network to analyze and predict complaints related to pipeline systems in a specific area, providing decision support suggestions. Bayesian network structure learning determines the optimal network structure based on data samples, prior knowledge, or a combination of both. The challenge lies in finding a model with the highest data fit. If the model contains redundant variable relationships (directed arcs), it increases the probability parameters that need to be learned and may also reveal erroneous causal relationships in the domain. This embodiment chooses a hybrid modeling approach: it can utilize experience or expert knowledge to provide subjective guidance, while also using data-driven methods to obtain hidden causal relationships between variables, thus establishing a relatively perfect Bayesian network model. The leaky modeling method integrates variable relationships, rules, and order information from domain knowledge into the data learning algorithm. Because this method can determine some parameter relationships and a simple model structure before data learning and remove meaningless causal relationships, it significantly reduces the space for optimal model search, thereby improving learning efficiency and the fitting accuracy between the optimal model and the data.

[0127] Then, the constructed Bayesian network is used for prediction, and past data is used for prediction testing. The results of using the Bayesian network to predict similar problems are as follows: Figure 12 , Figure 13 and Figure 14 As shown. Figure 12 To predict the type of pipeline in a certain region using the Bayesian network, the prediction of pipeline in a certain region is mainly based on the four types shown in the figure (the actual pipeline problem types are at least 11). Figure 13The diagram shows the correlation between the reasons for gas complaints. Gas complaints mainly revolve around routine construction and renovation, and are concentrated in the vicinity of Xingda Road, Xinfa Community, Bile Street, Xuanxi Community, Yonghe Street, Haxi Avenue, and Xuanxi Community. After manual screening and comparison with the data of complaint cases, it was found that the predicted results are very consistent with the actual situation. Figure 14 This is a correlation diagram of gas-related events occurring in a certain area within a time frame. After manual screening and comparison with complaint data, it was found that the predicted results are consistent with the actual situation. Based on the complaint analysis, similar complaints are likely to occur again, and manual judgment indicates a high probability of occurrence. For example, an investigation revealed that the gas pipeline in Building B06 of a certain district was corroded. However, due to the proximity of the gas pipeline in the stairwell to telecommunications and heating pipelines, construction could not be carried out, and the gas renovation was stalled. In August 2023, construction was hindered by some residents and some on-site construction workers, and the work was unsuccessful. In December 2023, complaints were still filed regarding the unfinished construction. This problem remains unresolved, and related complaints are predicted to continue in 2024. Specific implementation method four:

[0129] This embodiment is a decision support device based on big data mining, analysis and management of digital government. The device includes a processor and a memory. It should be understood that it includes any device including a processor and a memory described in this invention. The device may also include other units and modules that perform display, interaction, processing, control and other functions through signals or instructions.

[0130] The memory stores at least one instruction, which is loaded and executed by the processor to form the decision support system based on big data mining, analysis and management of digital government.

[0131] It should be understood that the instructions include computer program products, software, or computerized methods corresponding to any method described in this invention; the instructions can be used to program computer systems or other electronic devices. The memory may include a readable medium on which instructions are stored, and may include, but is not limited to, magnetic storage media, optical storage media; magneto-optical storage media include read-only memory (ROM), random access memory (RAM), erasable programmable memory (e.g., EPROM and EEPROM), and flash memory layers, or other types of media suitable for storing electronic instructions.

[0132] The above examples of the present invention are merely illustrative of the computational model and process of the present invention, and are not intended to limit the implementation of the present invention. Those skilled in the art will recognize that other variations or modifications can be made based on the above description. It is impossible to exhaustively list all possible implementations here. Any obvious variations or modifications derived from the technical solutions of the present invention are still within the scope of protection of the present invention.

Claims

1. A management system based on big data mining and analysis in digital government, characterized in that, include: Data acquisition and word segmentation module: Based on the time period information to be analyzed, it retrieves the data reflecting the problem from the management platform database. Each data reflecting the problem corresponds to a problem data ID. Then, it performs word segmentation and part-of-speech tagging on the retrieved data reflecting the problem. Data cleaning module: used to clean the word segmentation results; Location area determination module: Based on the cleaned data, the module determines the location area of ​​each data point reflecting a problem, and "binds" the location area to the data point reflecting the problem, that is, it establishes a mapping relationship between the data point reflecting the problem and the location area corresponding to the problem, and stores it. Basic Information Database: Stores and manages administrative geographical factors and corresponding regional boundaries, natural geographical factors and corresponding regional boundaries, streets and alleys and corresponding segmented regional boundaries, as well as infrastructure and corresponding geographical locations within the administrative region; Keyword extraction module: Based on the cleaned data, feature words are obtained by clustering nouns; then, keywords are extracted based on the clustered feature words; during the keyword extraction process, keyword selection metrics are used to select keywords, as follows: in, This represents the number of times word x appears in the text corresponding to the data reflecting the problem, and N represents the total number of texts corresponding to the data reflecting the problem. This indicates the number of texts corresponding to the problem data containing the word x; This is a proportional coefficient used to adjust the keyword selection index, and it is determined based on the actual problem and the situation of the dataset. To determine the amount of expression of the qualifier x in the text that reflects the problem data. This represents the average frequency of words in the problem data, used to measure the importance of words in the problem data; This is used to measure the frequency of occurrence of the data reflecting a certain problem in all data, which is to say, the importance and scope of influence of certain problems; Directed Association Rule Graph Generation Module: Using keywords as nodes, the module generates a directed association rule graph of keywords using a directed hypergraph spanning tree algorithm. Keyword Query and Display Module: This module receives query terms input by administrator users, retrieves the most matching keywords based on the query terms, and prominently displays the directed relationships of the confirmed keyword nodes in the directed association rule graph of the keywords; alternatively, it receives pointer action information from administrator users in the directed association rule graph of the keywords, confirms the keyword node pointed to by the pointer based on the pointer action information, and prominently displays the directed relationships of the confirmed keyword nodes in the directed association rule graph of the keywords.

2. The management system based on big data mining and analysis of digital government according to claim 1, characterized in that, In determining the location area of ​​each piece of data reflecting a problem, if the person who reported the problem has filled in the corresponding location area, then the location area is determined directly. If the person reporting the problem does so by directly stating the problem without specifying a location area, then the location area will be determined by information extraction.

3. The management system based on big data mining and analysis of digital government according to claim 2, characterized in that, The method for determining the location region by information extraction employs information matching; the specific process for determining the location region by information matching is as follows: Based on the order of expression of the data reflecting the problem, each word in the word segmentation result is matched with the address feature word. If a word in the word segmentation result can be directly and completely matched with the address feature word, or if multiple consecutive words in the word segmentation result can be completely matched with the address feature word, then the corresponding address feature word is used as the location information word. Based on the location information word, the geographical location and spatial range corresponding to the problem are determined.

4. The management system based on big data mining and analysis of digital government according to claim 3, characterized in that, The data cleaning module's cleaning process includes a stop word removal step.

5. A management device based on big data mining and analysis in digital government, characterized in that, The device includes a processor and a memory, the memory storing at least one instruction, which is loaded and executed by the processor as described in any one of claims 1 to 4, a management system based on big data mining and analysis of digital government.

6. A decision support system based on big data mining, analysis, and management in digital government, characterized in that: include: Organization and Personnel Database: Stores the organizational structure and corresponding job responsibilities of all departments in each jurisdiction; Demand information receiving module: used to obtain keywords extracted by the keyword extraction module and the directed association rule graph obtained by the directed association rule graph generation module of the management system based on digital government big data mining and analysis as described in any one of claims 1 to 4; at the same time, it provides keyword options to users and provides corresponding directed association rule graphs according to the keyword options; The first support suggestion module: Based on the keywords confirmed by the user and the corresponding directed association rule graph, it first identifies the keywords corresponding to the keyword nodes and determines whether the keywords are words related to the "problem" or the "location" of the problem reflected; then, based on the words related to the "location" of the problem reflected and the words related to the "problem", it matches the jurisdiction, organization, and job responsibilities in the organization and personnel database.

7. A decision support system based on big data mining, analysis, and management in digital government, as described in claim 6, is characterized in that... Based on words related to "location" and words related to "problem," during the matching of jurisdiction, organization, and job responsibilities in the database of institutions and personnel, for words related to "location" reflecting the problem, the ID corresponding to the problem data is determined according to the problem data reflected by the keywords of the node belonging to "problem," and the location area where the problem data occurred is obtained from the location area determination module. Based on the location area where the problem occurred, a jurisdiction matching recommendation corresponding to the solution that can be implemented is given.

8. A decision support system based on big data mining, analysis, and management in digital government, as described in claim 6 or 7, characterized in that: The system also includes a second support suggestion module: based on user-confirmed keywords and corresponding directed association rule graphs, it extracts topic and scenario information, obtains all data of the corresponding complaint cases according to the determined topic and scenario information, analyzes the data and provides information on the future occurrence of similar problems, including whether they will occur in the future and / or the expected location, thereby serving as potential decision support suggestions.

9. A decision support system based on big data mining, analysis, and management in digital government, as described in claim 8, is characterized in that... The second support suggestion module uses a Bayesian network to analyze data and provide information on how similar problems might occur in the future.

10. A decision support device based on big data mining, analysis, and management in digital government, characterized in that: The device includes a processor and a memory, the memory storing at least one instruction, which is loaded and executed by the processor as described in claim 8, a decision support system based on big data mining, analysis and management of digital government.

Images