Policy searching and disassembling method and device for policy redemption platform and medium
By automating the acquisition, classification, and decomposition of policy documents, the problems of low efficiency and high information error rate in existing technologies have been solved, and unified decomposition standards and efficient information filling have been achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- 天元大数据信用管理有限公司
- Filing Date
- 2026-01-07
- Publication Date
- 2026-04-21
AI Technical Summary
The existing policy implementation service platform relies on manual operation, which leads to low efficiency in obtaining policy documents, inconsistent breakdown, cumbersome and error-prone information filling, and affects the efficiency of subsequent processes.
The system automatically retrieves policy documents from government websites via data interfaces, and uses classification and information extraction models to automatically classify and break them down, generating granular items and populating key information into the platform's details page.
It has achieved fully automated processing of policy documents, unified decomposition standards, reduced information error rate, improved processing efficiency, and avoided errors and discrepancies caused by human intervention.
Smart Images
Figure CN121903556A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the fields of e-government and policy management technology, and in particular to a policy search and decomposition method, equipment and medium for a policy implementation platform. Background Technology
[0002] The policy implementation service platform undertakes core functions such as policy document management, item breakdown, departmental assignment, and fund disbursement; however, its existing policy processing procedures rely on manual operation, which has obvious shortcomings. First, the efficiency of obtaining policy documents is low. The existing policy implementation service platform requires operators to manually search for policy documents from the official websites of various government departments and then upload them to the platform's "policy document library". This not only consumes a lot of manpower, but also easily leads to problems such as missing documents and delayed acquisition, resulting in untimely policy updates on the platform.
[0003] Second, policy decomposition relies on manual judgment. Operations personnel need to manually add documents to be decomposed and confirm the decomposition results. Moreover, there is no uniform standard for decomposition dimensions (such as the target audience of the matter and the scope of rewards and subsidies). The format and information completeness of the decomposed matters vary greatly among different operations personnel, requiring repeated adjustments, which affects the efficiency of subsequent processes such as "matter release follow-up" and "service guide release follow-up".
[0004] Third, the information filling is cumbersome: After manual breakdown, key information such as the leading department, subsidy standards, and application conditions of the item needs to be filled in manually, which is prone to data entry errors. Summary of the Invention
[0005] This application provides a policy search and decomposition method, device, and medium for a policy implementation platform to address the following technical issues: how to improve policy processing efficiency, unify policy decomposition standards, and reduce information error rates.
[0006] In a first aspect, embodiments of this application provide a policy search and decomposition method for a policy implementation platform. The method includes: obtaining policy documents from a government department's website via a data interface and uploading the policy documents to the policy implementation platform; inputting the policy documents into a classification model to obtain classification results output by the classification model, wherein the classification model is used to classify the government documents and generate classification results; when the classification results indicate that the policy document is a policy implementation document, decomposing the policy document into granular items according to a preset decomposition dimension, wherein the granular items correspond to the preset decomposition dimension; inputting the granular items into an information extraction model to obtain key information of the granular items output by the information extraction model, wherein the information extraction model is used to extract information from the granular items based on the item information of the policy implementation platform to obtain key information; and filling the key information into the item details page of the policy implementation platform according to the item information.
[0007] Secondly, embodiments of this application also provide a policy search and dismantling device for a policy fulfillment platform. The device includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, which are executed by the at least one processor to enable the at least one processor to perform a policy search and dismantling method for a policy fulfillment platform as described in the first aspect above.
[0008] Thirdly, embodiments of this application also provide a computer storage medium storing computer-executable instructions, which, when executed, implement a policy search and decomposition method for a policy fulfillment platform as described in the first aspect above.
[0009] The policy search and decomposition method, device, and medium for a policy implementation platform provided in this application have the following beneficial effects: In this embodiment, policy documents from government websites can be obtained through a data interface and uploaded to the policy implementation platform. The policy documents are then input into a classification model to obtain classification results. If the classification results indicate the policy document is a policy implementation document, it is broken down into granular items based on preset decomposition dimensions. These granular items are then input into an information extraction model to obtain key information, which is ultimately used to populate the item details page on the policy implementation platform. This fully automates the process from obtaining the policy document to uploading it to the policy implementation platform, as well as subsequent classification, decomposition, key information extraction, and population, eliminating the need for manual intervention and improving policy processing efficiency. Furthermore, decomposing the policy document into granular items based on preset decomposition dimensions standardizes policy decomposition, avoiding format confusion caused by human interpretation differences. The subsequent extraction of key information through the information extraction model and its subsequent population into the policy implementation platform avoids policy implementation disputes caused by human input errors, reducing the information error rate. Attached Figure Description
[0010] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments of this application and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings: Figure 1 A flowchart illustrating a policy search and decomposition method for a policy implementation platform provided in this application embodiment; Figure 2 A flowchart illustrating another policy search and decomposition method for a policy fulfillment platform provided in this application embodiment; Figure 3 This is a schematic diagram of the internal structure of a policy search and dismantling device for a policy implementation platform provided in an embodiment of this application. Detailed Implementation
[0011] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be clearly and completely described below in conjunction with specific embodiments and corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.
[0012] This application provides a method for policy search and decomposition on a policy implementation platform. The technical solution proposed in this application will be described in detail below with reference to the accompanying drawings.
[0013] Figure 1This is a flowchart illustrating a policy search and decomposition method for a policy implementation platform provided in this application embodiment. Figure 1 As shown in the embodiment of this application, a policy search and decomposition method for a policy implementation platform specifically includes the following steps: Step 101: Obtain policy documents from government websites through data interfaces and upload the policy documents to the policy implementation platform.
[0014] In this embodiment of the application, policy documents from government department websites can be obtained through a data interface and uploaded to the policy implementation platform, eliminating the need for operators to manually search for policy documents from the official websites of various government departments. This enables automatic policy document retrieval, which not only improves the efficiency of document retrieval but also saves manpower and resources. It also minimizes the problems of document omissions and retrieval delays, ensuring that the policies on the policy implementation platform are updated in a timely manner.
[0015] Step 102: Input the policy document into the classification model and obtain the classification result output by the classification model.
[0016] The classification model is used to classify the government documents and generate classification results.
[0017] In this embodiment, the acquired policy documents can be input into the classification model to obtain classification results. In practical applications, the specific categories and structure of the classification model are not specifically limited and can be determined according to the actual situation. This saves human resources and improves work efficiency.
[0018] Step 103: If the classification result indicates that the policy document is a policy implementation document, the policy document is broken down into granular items according to the preset decomposition dimensions.
[0019] The granular items correspond to the preset decomposition dimensions.
[0020] In this embodiment, when the classification result indicates that the policy document is a policy implementation document, the policy document can be broken down into corresponding granular items according to preset decomposition dimensions. The preset decomposition dimensions can be "target audience" or "scope of subsidies," which is consistent with the decomposition logic when manually decomposing the policy implementation platform. Thus, the policy document can be decomposed into corresponding granular items. For example, the decomposition dimension of "target audience" can yield a corresponding granular item. In this way, the policy decomposition standard can be unified, which not only adapts to the decomposition logic of the policy implementation platform but also avoids the confusion of item formats caused by differences in human understanding, thereby improving the overall work efficiency.
[0021] Step 104: Input the granular items into the information extraction model to obtain the key information of the granular items output by the information extraction model.
[0022] The information extraction model is used to extract key information from the granular items based on the information of the policy implementation platform.
[0023] In this embodiment, the obtained granular items can be input into an information extraction model to obtain key information about the granular items extracted from the policy implementation platform. In practical applications, the specific category and structure of the information extraction model are not specifically limited in this application and can be determined according to the actual situation. This eliminates the need for manual extraction, improving efficiency and reducing the error rate of information extraction.
[0024] Step 105: Based on the aforementioned information, fill the key information into the details page of the policy implementation platform.
[0025] In this embodiment of the application, after obtaining the key information of granular matters, the corresponding key information can be filled into the details page of the policy implementation platform according to the corresponding matter information. In this way, the error rate of key information of matters can be reduced and policy implementation disputes caused by manual input errors can be avoided.
[0026] In this embodiment, policy documents from government websites can be obtained through a data interface and uploaded to the policy implementation platform. The policy documents are then input into a classification model to obtain classification results. If the classification results indicate the policy document is a policy implementation document, it is broken down into granular items based on preset decomposition dimensions. These granular items are then input into an information extraction model to obtain key information, which is ultimately used to populate the item details page on the policy implementation platform. This fully automates the process from obtaining the policy document to uploading it to the policy implementation platform, as well as subsequent classification, decomposition, key information extraction, and population, eliminating the need for manual intervention and improving policy processing efficiency. Furthermore, decomposing the policy document into granular items based on preset decomposition dimensions standardizes policy decomposition, avoiding format confusion caused by human interpretation differences. The subsequent extraction of key information through the information extraction model and its subsequent population into the policy implementation platform avoids policy implementation disputes caused by human input errors, reducing the information error rate.
[0027] In one possible implementation, the step of obtaining policy documents from government department websites via a data interface and uploading the policy documents to the policy implementation platform includes: Receive the government department websites set by the user and the acquisition frequency corresponding to the government department websites; At a preset time point, policy documents from the government department's website are obtained through a data interface, wherein the preset time point is related to the acquisition frequency; Based on the format of the policy document, parse the policy document and extract its file metadata; Upload the policy document and its metadata to the policy implementation platform.
[0028] In the above embodiments, the system can receive government department websites set by the user, along with the corresponding acquisition frequency for each website. The user can be an operator of the policy implementation platform or other staff. Each government department website can have a different acquisition frequency. In practical applications, the user can preset the URLs of the government department websites to be acquired and set the corresponding acquisition frequency (e.g., synchronized with the policy implementation platform's round-robin time at 00:00 daily, or incremental acquisition every 2 hours to avoid duplicate acquisition). In the above embodiments, policy documents from the government department websites can be acquired through a data interface at preset time points corresponding to the acquisition frequency. The acquisition of policy documents is conducted with the permission of the government department website and complies with relevant laws and regulations. The type of data interface is not specifically limited. In practical applications, policy documents in HTML, PDF, and Word formats can be obtained, all of which are supported by the policy implementation platform's file library. Subsequently, the BeautifulSoup algorithm is used to parse HTML files, the PyPDF2 algorithm to extract PDF text, and the python-docx algorithm to read Word files, extracting the file metadata, including the issuing department, release date, and document title, ensuring consistency with the fields displayed on the policy implementation platform. Then, the policy documents and their metadata obtained through the data interface can be automatically uploaded to the policy implementation platform, for example, to its policy document library, and marked as automatically obtained to avoid confusion with manually uploaded files.
[0029] In one possible implementation, inputting the policy document into a classification model and obtaining the classification result output by the classification model includes: The policy document is preprocessed to obtain the preprocessed policy document, wherein the preprocessing includes Jieba word segmentation and stop word removal; Keyword extraction is performed on the preprocessed policy document; If the policy document includes the keywords, the policy document is input into the classification model to obtain the classification result output by the classification model.
[0030] In the above embodiments, text preprocessing can be performed on the body of the obtained policy document. For example, Jieba word segmentation and stop word removal (such as meaningless words like "of", "and", etc.) can be carried out. Then, keywords related to policy implementation such as "award and subsidy", "subsidy", "declaration", "cashing" (referring to the core words judged manually by the policy cashing platform) are extracted. In the case where the above keywords are included, further judgment can be made through a classification model. In the case where they are not included, it can be determined that the policy document is not a policy cashing type document. In practical applications, the classification model can adopt a pre-trained BERT model and be fine-tuned in combination with the historical policy classification data of the policy cashing platform (such as policy documents marked as "yes / no"), and then output a classification result of "yes / no". In practical applications, the model accuracy is not less than 92%. At the same time, the classification result can be synchronized to the "whether it is a policy cashing type document" field of the policy cashing platform. At the same time, the original functions of "mark as a policy cashing type document" and "mark as a non-policy cashing type document" on the policy cashing platform need to be retained. After the operator adjusts the classification result, the adjusted policy document can be fed back to the classification model as a sample to optimize the model adaptability.
[0031] In a possible implementation manner, the disassembling of the policy document into granulated matters according to a preset disassembling dimension includes: Obtain a preset disassembling dimension set by the user and matching the policy cashing platform; Obtain the chapter in the policy document corresponding to the preset disassembling dimension through a regular expression; Use a sentence vector clustering algorithm for the chapter to obtain granulated matters corresponding to the preset disassembling dimension.
[0032] In the above embodiments, policy cashing type documents can be disassembled into granulated matters. In practical applications, it supports the user to preset a disassembling dimension matching the policy cashing platform, including "object faced by the matter" (such as travel agencies, high-tech enterprises, individual industrial and commercial households, etc.), "award and subsidy scope" (such as tourism team rewards, interest subsidies for technological transformation special loans, talent subsidies, etc.), "matter type" (fund subsidies, qualification certifications, etc.). Then, match the chapter in the policy document corresponding to the disassembling dimension through a regular expression (such as "reward object" corresponding to "object faced by the matter", "reward standard" corresponding to "award and subsidy scope"), so as to adapt to the chapter reference logic during manual disassembling on the policy cashing platform; then use a sentence vector clustering algorithm to disassemble the chapter into independent granulated matters. The disassembling result needs to display "number of disassembled matters", "disassembler (system automatic)", "disassembling time", and be synchronized to the policy cashing platform, and the disassembling time does not exceed 10 minutes. In this way, not only can the disassembling standard be unified, but also the efficiency can be improved.
[0033] In practical applications, for example, a policy document might state: Enterprises must submit application materials by date A. The subsidy standard is a maximum of 5 million yuan, and applicants must be high-tech enterprises. Application materials include a business license and a research and development report. Enterprises that submit false materials will have their subsidy eligibility revoked and the funds recovered. The preset decomposition dimensions are ["subsidy standard", "application time", "eligibility conditions", "application materials", "regulatory measures"]. By matching sections using regular expressions, the following sentences can be matched: "Companies must submit their application materials by December 31, 2025." (Application deadline) "The subsidy standard is up to 5 million yuan, and applicants must be high-tech enterprises." (Subsidy standard + eligibility requirements) "The application materials include a business license and a research and development report." "Companies that submit false information will have their subsidy eligibility revoked and the funds recovered." (Regulatory Measures) Sentence vector clustering algorithm can be used to further decompose the data.
[0034] In one possible implementation, the step of using a sentence vector clustering algorithm to break down the policy document into independent granular items according to the preset decomposition dimensions includes: Determine whether the chapter corresponds to only one of the preset decomposition dimensions; If a chapter corresponds to only one preset disassembly dimension, then the dimension of the chapter is determined to be the preset disassembly dimension. When the chapter corresponds to at least one of the preset decomposition dimensions, a sentence vector clustering algorithm is used to decompose the chapter into clauses and match the clauses with the preset decomposition dimensions; Organize the chapters or sentences of the same preset decomposition dimension to obtain the granular items corresponding to the preset decomposition dimension.
[0035] In practical applications, to improve efficiency, when using sentence vector clustering algorithms on chapters, it's advisable to first determine whether the chapter corresponds to only one preset decomposition dimension. In practice, this can be achieved by counting the dimensions to which keywords appearing in the chapter belong; if all keywords belong to the same dimension, it's considered a single dimension. Other methods can also be used, without specific limitations. Sentence vector clustering algorithms can be employed when the chapter corresponds to at least one dimension. For example, if it's determined that only the chapter "The subsidy standard is a maximum of 5 million yuan, and it must be a high-tech enterprise." corresponds to two preset decomposition dimensions, then the chapter can be segmented into sentences: Clause 1: "The subsidy standard is a maximum of 5 million yuan." Clause 2: "And it must be a high-tech enterprise." Then, generate sentence vectors (assuming Sentence-BERT is used): Clause 1 vector: [0.12, -0.05, 0.78, ...] Clause 2 vector: [0.11, -0.06, 0.77, ...] Then, using clustering and post-processing, K-Means (K=5) clustering is performed: Clause 1 → Cluster 0 (Subsidy Standard) Clause 2 → Cluster 1 (Eligibility Requirements) At this point, other sentences are directly assigned to the corresponding clusters.
[0036] Finally, extract cluster keywords and determine the corresponding preset decomposition dimensions.
[0037] Cluster 0: ["Subsidy", "Standard", "5 million yuan"] → Tag: "Subsidy Standard" Cluster 1: ["Required", "High-tech Enterprise"] → Tag: "Eligibility Requirements".
[0038] This ensures that each output item is semantically independent and strictly corresponds to a preset dimension.
[0039] Vector clustering automatically discovers semantically similar sentences by converting sentences into vectors and grouping them. In policy text decomposition, it can refine the granularity of decomposition, breaking down chapters into independent issues and avoiding information mixing. It can also adapt to complex expressions, handling multiple expressions of the same dimension through semantic similarity clustering. Furthermore, it can improve efficiency, reduce manual annotation, and support automated parsing of large-scale policy documents. In practical applications, the sentence vector model and clustering algorithm can be determined based on business needs to balance accuracy and efficiency.
[0040] In one possible implementation, inputting the granular items into an information extraction model to obtain key information about the granular items output by the information extraction model includes: Obtain the extraction rules corresponding to the aforementioned item information; Information is extracted from the granular items according to the extraction rules to obtain the key fields in the granular items that correspond to the extraction rules; If the key field matches the extraction rule, extract key information from the key field; If the key field is inconsistent with the extraction rule, the key field and the context corresponding to the key field are input into the information extraction model to obtain the key information output by the information extraction model.
[0041] In the above embodiments, extraction rules matching the information of the policy implementation platform can be obtained, such as "subsidy amount rules" (matching "maximum XX million yuan" and "subsidy per household XX yuan"), "lead department rules" (matching "issuing department is XX bureau" or "responsible by XX department" in the document), and "application deadline rules" (matching "application deadline is XXXX year XX month XX day"). Then, these extraction rules are used to extract the key fields corresponding to the extraction rules in the granular items. Subsequent determination can be made based on the relationship between the key fields and the extraction rules. For example, the subsidy amount rule matches "maximum 500,000 yuan." In this case, when the key fields and extraction rules are consistent, key information is extracted from the key fields. However, if the matched result is "a reward of 20-30% of the investment amount," where the key fields and extraction rules are inconsistent, the key fields and their corresponding context can be input into the information extraction model to obtain the key information output by the model. For example, the information extraction model can use the BERT-NER model to extract key information such as the lead department, subsidy amount, application deadline, and application conditions of the item. In practical applications, determining whether a key field matches the extraction rule can be done by comparing the extracted key field with the extraction rule. If the similarity between the two is greater than a certain value, they are considered to be consistent. Alternatively, if they are completely consistent, they are considered to be consistent. There are no specific restrictions.
[0042] In practical applications, extraction rules can handle explicit, fixed, and high-frequency expression patterns, while information extraction models can handle fuzzy, variable, and complex semantic expressions. This ensures an optimal balance between efficiency (extraction rules quickly handle most aspects) and effectiveness (information extraction models accurately handle the difficult parts), while avoiding unnecessary waste of computational resources through clear conditional judgments, making it highly practical.
[0043] In one possible implementation, after filling the key information into the details page of the policy implementation platform based on the matter information, the method further includes: Perform an integrity check on the key information populated into the details page of the item; If any part of the item details page is missing, the missing part of the item details page will be marked and the item details page will be highlighted.
[0044] In practical applications, key information can be located and filled in based on the information provided. For example, the lead department can be filled in the "Responsible Unit" column, the application deadline in the "Application Deadline" column, and the application conditions in the "Application Conditions" column. At the same time, the format should be consistent when filling in the information: all information should be filled in according to the standard format, such as using "ten thousand yuan" as the unit for the amount, "YYYY-MM-DD" for the date, and the full name of the department (e.g., "Municipal Science and Technology Bureau"). Source information can also be recorded, such as marking the information as "AI automatically extracted", recording the extraction time, and recording whether a rule or a model was used.
[0045] In practical applications, after the information is filled in, it can be validated. If problems such as "leading department not matched" or "missing subsidy amount" exist, it can be marked as "pending manual supplementation" and highlighted on the details page of that item to prompt operations personnel to handle it. Simultaneously, a reasonableness check can be performed (whether the information is reasonable), such as abnormal amounts (e.g., a subsidy of 100 million yuan, which may be a misidentification), contradictory dates (the deadline is earlier than the current date), invalid departments, or department names not in the standard directory). A consistency check can also be performed (whether there is internal consistency), determining whether information within the same item conflicts, ensuring consistent monetary units (not mixing "yuan" and "ten thousand yuan"), and consistent time descriptions. In cases of anomalies, they can also be marked and highlighted. This achieves automated and standardized processing of policy information, improving efficiency and minimizing errors in data entry.
[0046] In one possible implementation, the method further includes: The document status of the disassembled policy documents will be synchronized to the policy implementation platform. The number of policy documents disassembled within a preset time period is counted, and the number is synchronized to the policy implementation platform; If the status of a granular item changes, the item status will be synchronized to the policy implementation platform to update the item release progress.
[0047] In practical applications, the file status (e.g., "disassembled" or "disassembly failed") of automatically disassembled policy documents can be synchronized to the corresponding module of the policy implementation platform, updating the "status" field. Simultaneously, the number of policy documents disassembled within a preset time period (e.g., 1 minute) can be automatically counted, along with data on operational push items, departmental claimed items, and departmental pending confirmation items, synchronizing this data to the policy implementation platform in real time. This avoids manual aggregation and ensures that statistical data is consistent with the disassembly progress. When the status of a granular item changes to "pushed," "pending," or "abandoned," it can be synchronized to the policy implementation platform in real time, updating the item's release progress. In this way, various data in the policy implementation platform are synchronized in real time, eliminating the need for manual aggregation. The delay between statistical results and disassembly progress is short (generally no more than 5 minutes, improving the timeliness of statistical data), ensuring that platform administrators can promptly grasp the policy disassembly and claim status, supporting decision-making.
[0048] Figure 2 This diagram illustrates the policy search and granular breakdown process of a policy implementation platform in an application scenario provided by this application embodiment. Figure 2 As shown, the policy search and granular decomposition of the policy implementation platform in this application embodiment may further include the following execution process: During the automatic policy search phase, the following steps are performed: Step 11, Information Source Configuration: Preset government department websites and acquisition frequency.
[0049] Step 12, Obtaining Multiple File Formats: Parsing HTML / PDF / WORD.
[0050] Step 13, Incremental Deduplication: Compare the MD5 value of the file / publication time.
[0051] Step 14: Automatically upload to the policy implementation document library and mark it as AI upload.
[0052] During the policy classification and screening stage, the following steps are performed: Step 21: Text preprocessing, including Jieba word segmentation and stop word removal.
[0053] Step 22, BERT model classification: Based on historical classification data, the classification results are synchronized to the policy implementation platform.
[0054] Step 23, manual adjustment: retain the category marked as policy fulfillment / non-policy fulfillment.
[0055] During the granulation and disassembly stage, the following steps are performed: Step 31, Decompose Dimension Configuration: Preset Item-Oriented Objects.
[0056] Step 32, Chapter Positioning: Regular expression matching reward object / standard.
[0057] Step 33: Use vector clustering to break down the chapters into granular items.
[0058] During the information filling and validation phase, the following steps are performed: Step 41, rule base call: match the maximum reward of X yuan.
[0059] Step 41, BERT-NER Extraction: Automatically fill in the lead department / reward.
[0060] Step 42, Verify the flag: If information is missing or conflicting, mark it as abnormal.
[0061] During the platform data mining and synchronization phase, the following steps are performed: Step 51: The disassembly task is synchronized and updated to the disassembly task management.
[0062] Step 52, Quantity Statistics Synchronization: Automatically summarize the AI-disassembled quantities.
[0063] Step 53, Item status synchronization: Pushed / Pending / Abandoned.
[0064] Step 54, process complete: the item is then assigned to a department / service guide.
[0065] The above are embodiments of the method proposed in this application. Based on the same inventive concept, embodiments of this application also provide a policy search and granular decomposition system. Exemplarily, this policy search and granular decomposition system includes an automatic policy search and upload module, a policy classification and adaptation module, a granular decomposition module, a rule understanding and filling module, a platform data synchronization module, and a system management module. Each module seamlessly integrates with the existing functional modules of the policy implementation platform. The specific structure and functions are as follows: The platform comprises several modules: an automatic policy search and upload module, which automatically retrieves policy documents from government websites via a data interface and uploads them to the "Policy Document Library" of the policy implementation platform; a policy classification and adaptation module, which filters policy implementation documents and adapts to the "Is it a policy implementation document?" judgment function of the "Task Decomposition Management" module on the policy implementation platform; a granular decomposition module, which decomposes policy implementation documents into granular items and adapts to the "Decomposition Item Management" module on the policy implementation platform; a rule understanding and filling module, which supports the "View Details" and modification functions of the "Decomposition Item Management" module on the policy implementation platform; a platform data synchronization module, which synchronizes the decomposed data to relevant modules on the policy implementation platform; and a system management module, which manages system configuration and permissions and adapts to the administrator permission system of the policy implementation platform. In practical applications, the system management module includes a parameter configuration unit, which allows users to configure the acquisition frequency, classification model threshold, decomposition dimensions, and rule base content, consistent with the configuration permissions of the "administrator" role on the policy implementation platform. The system management module also includes a permission management unit, employing the RBAC model to set "system administrator" and "operation personnel" roles, with permissions matching those on the policy implementation platform (e.g., operation personnel can only handle "awaiting manual supplementation" items and cannot modify acquisition parameters). Furthermore, the system management module includes a log query unit, storing operation logs for policy acquisition, classification, decomposition, and synchronization, supporting queries by module dimensions such as "decomposition task management" and "decomposition item management," facilitating troubleshooting.
[0066] Based on the same inventive concept, this application also provides a policy search and dismantling device for a policy implementation platform, the structure of which is as follows: Figure 3 As shown.
[0067] Figure 3 This is a schematic diagram of the internal structure of a policy search and dismantling device for a policy implementation platform provided in this application embodiment. (See diagram for example.) Figure 3 As shown, the device includes: At least one processor 301; And a memory 302 that is communicatively connected to at least one processor; The memory 302 stores instructions that can be executed by at least one processor. The instructions are executed by at least one processor 301 so that at least one processor 301 can: execute the policy search and granular decomposition method of the policy fulfillment platform described above.
[0068] Some embodiments of this application provide corresponding to Figure 1 A non-volatile computer storage medium stores computer-executable instructions, which are configured to execute the policy search and granular decomposition method of the aforementioned policy implementation platform.
[0069] The various embodiments in this application are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the embodiments for IoT devices and media are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions of the method embodiments.
[0070] The systems, media, and methods provided in this application are one-to-one correspondences. Therefore, the systems and media also have similar beneficial technical effects as their corresponding methods. Since the beneficial technical effects of the methods have been described in detail above, the beneficial technical effects of the systems and media will not be repeated here.
[0071] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.
[0072] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.
[0073] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.
[0074] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.
[0075] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.
[0076] Memory may include non-persistent storage in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.
[0077] Computer-readable media include both permanent and non-permanent, removable and non-removable media that can store information by any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.
[0078] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.
[0079] The above description is merely an embodiment of this application and is not intended to limit this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principle of this application should be included within the scope of the claims of this application.
Claims
1. A method for policy search and decomposition on a policy implementation platform, characterized in that, include: The policy documents from government websites are obtained through data interfaces, and then uploaded to the policy implementation platform. The policy document is input into the classification model to obtain the classification result output by the classification model, wherein the classification model is used to classify the policy document and generate the classification result; If the classification result indicates that the policy document is a policy implementation document, the policy document is broken down into granular items according to the preset decomposition dimensions, wherein the granular items correspond to the preset decomposition dimensions; The granular items are input into the information extraction model to obtain the key information of the granular items output by the information extraction model. The information extraction model is used to extract information from the granular items based on the item information of the policy implementation platform to obtain key information. Based on the aforementioned information, the key information is populated into the details page of the policy implementation platform.
2. The method according to claim 1, characterized in that, The step of obtaining policy documents from government websites through a data interface and uploading the policy documents to the policy implementation platform includes: Receive the government department websites set by the user and the acquisition frequency corresponding to the government department websites; At a preset time point, policy documents from the government department's website are obtained through a data interface, wherein the preset time point is related to the acquisition frequency; Based on the format of the policy document, parse the policy document and extract its file metadata; Upload the policy document and its metadata to the policy implementation platform.
3. The method according to claim 1, characterized in that, The step of inputting the policy document into the classification model and obtaining the classification result output by the classification model includes: The policy document is preprocessed to obtain the preprocessed policy document, wherein the preprocessing includes Jieba word segmentation and stop word removal; Keyword extraction is performed on the preprocessed policy document; If the policy document includes the keywords, the policy document is input into the classification model to obtain the classification result output by the classification model.
4. The method according to claim 1, characterized in that, The step of breaking down the policy document into granular items according to preset decomposition dimensions includes: Obtain the preset breakdown dimensions set by the user that match the policy implementation platform; Using regular expressions, the chapters in the policy document corresponding to the preset decomposition dimensions are obtained; The sentence vector clustering algorithm is used on the chapter to obtain granular items corresponding to the preset decomposition dimension.
5. The method according to claim 1, characterized in that, The process of using sentence vector clustering algorithm to break down the policy document into independent granular items according to the preset decomposition dimensions includes: Determine whether the chapter corresponds to only one of the preset decomposition dimensions; If a chapter corresponds to only one preset disassembly dimension, then the dimension of the chapter is determined to be the preset disassembly dimension. When the chapter corresponds to at least one of the preset decomposition dimensions, a sentence vector clustering algorithm is used to decompose the chapter into clauses and match the clauses with the preset decomposition dimensions; Organize the chapters or sentences of the same preset decomposition dimension to obtain the granular items corresponding to the preset decomposition dimension.
6. The method according to claim 1, characterized in that, The step of inputting the granular items into the information extraction model to obtain the key information of the granular items output by the information extraction model includes: Obtain the extraction rules corresponding to the aforementioned item information; Information is extracted from the granular items according to the extraction rules to obtain the key fields in the granular items that correspond to the extraction rules; If the key field matches the extraction rule, extract key information from the key field; If the key field matches the extraction rule, the key field and the context corresponding to the key field are input into the information extraction model to obtain the key information output by the information extraction model.
7. The method according to claim 1, characterized in that, After filling the key information into the details page of the policy implementation platform based on the matter information, the method further includes: Perform an integrity check on the key information populated into the details page of the item; If any part of the item details page is missing, the missing part of the item details page will be marked and the item details page will be highlighted.
8. The method according to claim 1, characterized in that, The method further includes: The file status of the disassembled documents will be synchronized to the policy implementation platform. The number of documents disassembled within a preset time period is counted, and the number is synchronized to the policy implementation platform; If the status of a granular item changes, the item status will be synchronized to the policy implementation platform to update the item release progress.
9. A policy search and decomposition device for a policy implementation platform, characterized in that, The device includes: At least one processor; And, a memory communicatively connected to the at least one processor; The memory stores instructions that can be executed by the at least one processor, which are executed by the at least one processor to enable the at least one processor to perform a policy search and decomposition method for a policy fulfillment platform as described in any one of claims 1-8.
10. A computer storage medium storing computer-executable instructions, characterized in that, When the computer-executable instructions are executed, they implement the policy search and decomposition method for a policy implementation platform as described in any one of claims 1-8.