Enterprise policy customization report generation method

Abstract

The invention relates to the technical field of data analysis, in particular to an enterprise policy customization report generation method, which comprises the following steps: acquiring a policy file, constructing a policy file library, and dynamically updating the policy file library; disassembling the policy file in the policy file library, generating a service variety and a service item corresponding to the policy file, forming a policy product based on the service variety, the service item and a declaration condition corresponding to the service item, and constructing a policy product library; constructing a corresponding logic expression for the business item of each policy product in the policy product library; obtaining enterprise data, substituting the enterprise data into the logic expression corresponding to each business item in the policy product for matching, and generating a matching result; and generating an enterprise policy customization report based on the matching result. According to the method and the device, an enterprise can be semi-automatically and quickly assisted to complete matching and evaluation of the declaration conditions of the policy document, and meanwhile, detailed reports and improvement suggestions are generated.

Description

Technical Field

[0001] The present application relates to the field of data analysis technology, and in particular to a method for generating a customized report on enterprise policies. Background Art

[0002] When matching the published corporate policy documents, enterprises need to meet a series of diverse and detailed matching conditions. Matching conditions usually involve the company's years of establishment, operating income, R&D investment, employee composition and other aspects, aiming to ensure that policy resources can effectively support qualified enterprises.

[0003] At present, when matching policy documents and applying for policy benefits, enterprises mainly rely on their internal staff to analyze a large number of policy documents one by one, and compare their own situation to determine whether they meet the requirements of the specific policy documents for the enterprise. Today's policy documents usually contain a large number of complex terms, conditions and requirements, involving segmented policy content in different fields, industries and enterprise categories. For enterprises, the above method is not only time-consuming and labor-intensive, but also prone to misjudgment by staff due to the complexity and timeliness of policy documents. This phenomenon is particularly obvious in the context of accelerated policy document updates, which makes enterprises consume a lot of human resources in the policy application process and increase the risk of missing policy benefits due to misunderstandings.

[0004] To sum up, when enterprises match a large number of policy documents, the existing technology has the shortcomings of insufficient analysis of complex logical relationships and low degree of automation, making it difficult to match policy documents quickly and accurately, which leads to enterprises facing the risk of inefficiency and misjudgment when applying for preferential treatment of policy documents. Summary of the invention

[0005] This application provides a method for generating customized enterprise policy reports, which can semi-automatically and quickly assist enterprises in matching and evaluating the application conditions of policy documents, and generate detailed reports and improvement suggestions. This application provides the following technical solutions: In a first aspect, the present application provides a method for generating a customized enterprise policy report, the method comprising: obtaining a policy document and building a policy document library, and dynamically updating the policy document library; Disassemble the policy documents in the policy document library to generate business types and business items corresponding to the policy documents, form policy products based on the business types, business items and declaration conditions corresponding to the business items, and build a policy product library; Constructing corresponding logical expressions for the business items of each policy product in the policy product library; Acquire enterprise data, bring the enterprise data into the logical expression corresponding to each business item in the policy product for matching, and generate matching results; Generate a customized report on enterprise policies based on the matching results.

[0006] In a specific possible implementation scheme, the policy documents in the policy document library are disassembled to generate business types and business items corresponding to the policy documents, and the policy products are formed based on the business types, the business items and the declaration conditions corresponding to the business items, including: Constructing a business variety vocabulary set, generating an extended set of each business variety vocabulary in the business variety vocabulary set; Calculating the relevance of the extended set of each of the business variety words in the policy document, and selecting the business variety words corresponding to the relevance greater than or equal to a preset threshold as the business variety; By manually confirming the business items under the business types, the declaration conditions in the policy documents are associated with each business item one by one, and policy products are formed based on the business types, business items and the declaration conditions corresponding to the business items.

[0007] In a specific possible implementation scheme, calculating the relevance of each extended set of the business variety vocabulary in the policy document and selecting the business variety vocabulary corresponding to the relevance greater than or equal to a preset threshold as the business variety includes: Use the following formula to calculate the business category vocabulary An extended collection of In the policy document The correlation in : ; in, , indicating the business type vocabulary An extended collection of In Extension words, For extended words and policy documents The average cosine similarity of each sentence in , indicating the expansion word and policy documents The overall file similarity, Indicates expansion word Frequency of occurrence in policy documents, is the frequency weighting function, ; After calculating the correlation of the extended set of each business variety vocabulary in the policy document, the calculated correlation is compared with the preset threshold, and the business variety vocabulary corresponding to the correlation greater than or equal to the preset threshold is selected as the business variety, which is reviewed and confirmed by policy experts. If correct, the business variety vocabulary is determined as the business variety of the policy document.

[0008] In a specific possible implementation scheme, constructing a corresponding logical expression for the business items of each policy product in the policy product library includes: Extract the conditional variables associated with each business item; Determine whether a corresponding judgment formula needs to be established for each conditional variable; The condition variables are combined based on the logical relationship between the declaration conditions, and all condition variables are combined into a complete logical expression according to the logical structure in the policy document.

[0009] In a specific possible implementation scheme, the obtaining of enterprise data, bringing the enterprise data into the logical expression corresponding to each business item in the policy product for matching, and generating a matching result includes: After obtaining the enterprise data, extract the corresponding enterprise data according to the conditional variables required in each logical expression; After extracting part of the enterprise data, manually check whether the extracted part of the enterprise data is the same as the data required in the logical expression and delete the extracted redundant enterprise data; If they are the same, the extracted enterprise data will be substituted into the logical expression, and calculated item by item according to the rules of the logical expression to obtain the matching result. If part of the extracted enterprise data is different, it will be re-extracted until it is correct.

[0010] In a specific implementation scheme, extracting corresponding enterprise data according to the conditional variables required in each logical expression includes: By calculating the semantic similarity, the corresponding enterprise data is extracted and each condition variable is defined as , each item of enterprise data is , both are processed by the existing sentence vector model to obtain their embedding vectors respectively and , the semantic similarity between each conditional variable and each item of enterprise data is calculated through the embedding vectors of the two. The calculation formula is as follows: ; in, represents the inner product of two vectors, Represents the Euclidean distance between two vectors The power term, is an adjustable exponential parameter, representing the power term of the Euclidean distance, used to adjust the weight of the distance. Set to 2; For each conditional variable, the semantic similarity between it and each item of enterprise data is calculated, and each item of enterprise data whose semantic similarity is greater than or equal to a preset threshold is extracted, and the extracted part of the enterprise data is brought into the logical expression.

[0011] In a specific implementation scheme, extracting corresponding enterprise data according to the conditional variables required in each logical expression includes: The corresponding enterprise data is extracted by calculating the coverage of keywords, the description sentences of the conditional variables and the sentences in the enterprise data are segmented, keywords are extracted, and the core keywords in the conditional variables are screened out using the TF-IDF method to obtain a keyword set. , for each item of enterprise data , calculate the coverage of conditional variable keywords in the sentence as follows: ; in, Expressing sentences The coverage of conditional variable keywords in Indicates the enterprise data The word set of a sentence, Represents a set of conditional variable keywords and word set The number of elements in the intersection of How many condition variable keywords are included in Represents a set of conditional variable keywords The total number of keywords in the The weight of each matching word is smoothed as follows: ; in Indicates conditional variable keywords In the sentence The average frequency of occurrence in the sentence is used to calculate the smoothed weighted coverage of all sentences in the enterprise data. , sentences whose calculation results exceed the preset threshold are selected for data extraction.

[0012] In the second aspect, the present application provides a system for generating enterprise policy customized reports, which adopts the following technical solutions: A system for generating enterprise policy customized reports, comprising: A policy document acquisition module is used to acquire policy documents and build a policy document library, and dynamically update the policy document library; A policy document disassembly module is used to disassemble the policy documents in the policy document library, generate business types and business items corresponding to the policy documents, form policy products based on the business types, business items and declaration conditions corresponding to the business items, and build a policy product library; A logic expression building module, used to build a corresponding logic expression for the business items of each policy product in the policy product library; A policy matching module is used to obtain enterprise data, bring the enterprise data into the logical expression corresponding to each business item in the policy product for matching, and generate a matching result; A report generation module is used to generate a customized report on enterprise policies based on the matching results.

[0013] In a third aspect, the present application provides an electronic device, comprising a processor and a memory; the memory stores a program, and the program is loaded and executed by the processor to implement a method for generating a customized enterprise policy report as described in the first aspect.

[0014] In a fourth aspect, the present application provides a computer-readable storage medium, wherein the storage medium stores a program, and when the program is executed by a processor, the program is used to implement a method for generating a customized enterprise policy report as described in the first aspect.

[0015] In summary, the beneficial effects of this application include at least: 1) By breaking down the content in the policy document into business categories and business matters, and building a policy product library based on these breakdown results, this application can efficiently and systematically organize and store policy document information. Compared with the traditional manual analysis method, it can significantly improve the efficiency and accuracy of the breakdown process and reduce the risk of human error.

[0016] 2) By calculating the relevance of the extended set of business type words and combining semantic expansion and word frequency weighting functions, this application can achieve more accurate matching. This method considers multi-dimensional factors such as semantic similarity and word frequency, so that the relevant information in the policy document can be more accurately matched with the specific data of the enterprise, avoiding mismatches and omissions in traditional keyword matching methods.

[0017] 3) The need for manual intervention is reduced through semi-automatic processing of the matching process of policy documents and enterprise data. Through intelligent disassembly, calculation and matching methods, a large number of policy documents can be processed quickly, helping enterprises to complete policy preferential declarations in a shorter period of time, reducing the workload of manual operations and improving the automation level of the overall system.

[0018] By disassembling policy documents, extracting relevant business types and business items, and establishing a policy product library, we can help enterprises quickly and accurately match the conditions and requirements in policy documents. By building logical expressions to match enterprise data, we can determine whether the enterprise meets the policy requirements, improve the automation and accuracy of policy matching, and solve the errors and efficiency problems in manual analysis.

[0019] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application and implement it according to the contents of the specification, the following is a detailed description of the preferred embodiments of the present application in conjunction with the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 It is a flowchart of the method for generating a customized report of enterprise policies in an embodiment of the present application.

[0021] Figure 2 It is a flowchart of disassembling policy documents in the policy document library in an embodiment of the present application.

[0022] Figure 3 It is a schematic diagram of the overall process of the method for generating a customized report on enterprise policies in an embodiment of the present application.

[0023] Figure 4 It is a structural block diagram of the enterprise policy customized report generation system in the embodiment of the present application.

[0024] Figure 5 It is a block diagram of an electronic device for generating a customized report on enterprise policies in an embodiment of the present application. DETAILED DESCRIPTION

[0025] The specific implementation methods of the present application are further described in detail below in conjunction with the accompanying drawings and examples. The following examples are used to illustrate the present application but are not intended to limit the scope of the present application.

[0026] Optionally, the present application uses the enterprise policy customized report generation method provided in each embodiment as an example for use in an electronic device, where the electronic device is a terminal or a server. The terminal may be a computer, a tablet computer, etc. This embodiment does not limit the type of electronic device.

[0027] Reference Figure 1 , is a flow chart of a method for generating a customized enterprise policy report provided by an embodiment of the present application, the method comprising at least the following steps: Step S101, obtain policy documents and build a policy document library, and dynamically update the policy document library.

[0028] In step S101, first obtain policy documents issued by the government from public channels, such as government websites, government public accounts or other official government release channels. The obtained policy documents are then stored in the constructed policy document library. The policy document library in this application selects an existing database management system (such as MySQL, MongoDB or Elasticsearch, etc.) for storage, so as to efficiently handle the retrieval, update and storage requirements of policy documents. After obtaining the policy documents and building the policy document library, the policy document library needs to be dynamically updated, and the update frequency is usually once a day. When a new policy document is released, it is automatically stored in the policy document library. For policy documents that have expired or been replaced, the policy document library will automatically identify and mark them to ensure that the files in the policy library are always the latest and valid versions. Through regular synchronization or real-time updates, the policy document library can accurately reflect changes in the policy environment and ensure that enterprises use the latest policy data when matching.

[0029] Step S102: disassemble the policy documents in the policy document library, generate business types and business items corresponding to the policy documents, compose policy products based on the business types, business items and declaration conditions corresponding to the business items, and build a policy product library.

[0030] In step S102, each policy document in the policy document library will be deeply analyzed and disassembled, and the document content will be organized into two levels of "business types" and "business matters" that can reflect policy implementation and reporting requirements. Then, the reporting conditions corresponding to each business matter in the policy document will be associated with each other. After the disassembly is completed, a complete policy product is formed and stored in a pre-built policy product library. Optionally, the policy product library also selects an existing database management system (such as MySQL, MongoDB or Elasticsearch, etc.) for storage.

[0031] It should be noted that although sometimes they are under the same policy document and business type, the declaration conditions for each business item are different. Therefore, this application will correspond the declaration conditions for each business item so that enterprises can efficiently determine whether they meet the declaration conditions for the business item.

[0032] That is, in this application, business varieties are high-level classifications for industrial fields or application scenarios covered in specific policy documents, and usually reflect the core direction or theme of policy support. Business varieties can be regarded as the main field categories of policy incentive objects, such as "biomedicine", "intelligent manufacturing", "new materials", etc. Each business variety represents a major area or sector for the allocation of policy incentive resources, which is used to achieve high-level resource positioning and policy goal orientation in policy documents. Business matters are specific support matters or activity categories under business varieties, which are further subdivided into business varieties, covering specific projects or functions supported by policies. Business matters usually reflect the actual activities or application needs of enterprises in this field. Business matters can help enterprises clarify the specific application conditions, scope of application and operating details of different policy support projects, so as to apply for policy benefits in a more targeted manner.

[0033] During implementation, policy documents can be manually disassembled and classified one by one. The invited policy experts or policy analysts will manually disassemble each policy document in the policy document library into one or more business types based on the content in the policy document, and then subdivide the business types into one or more business items based on the specific terms of the policy. Finally, the declaration conditions in the policy document will be associated with each business item one by one, thereby completing the formation of policy products and storing them in the policy product library.

[0034] In addition, as preferably, in another feasible embodiment, referring to Figure 2 , you can also use the following method to disassemble the policy files in the policy file library: Step S1021: construct and dynamically update a business type vocabulary set, and generate an extended set of each business type vocabulary in the business type vocabulary set.

[0035] Specifically, first construct a business category vocabulary set The business category vocabulary set stores a large number of business category vocabulary pre-collected by policy experts or policy analysts. , such as biomedicine, technological innovation, etc. At the same time, the business variety vocabulary set is updated every once in a while, and the extended set corresponding to the new business variety vocabulary is regenerated after the update .

[0036] Optionally, the present application uses an existing word vector model (such as Word2Vec or BERT) to generate an extended set of each business variety vocabulary in the business variety vocabulary set. The extended set contains words semantically related to the business variety vocabulary, such as synonyms, associated words, and related field terms, thereby covering a variety of expressions that may appear in policy documents.

[0037] Step S1022: Calculate the relevance of the extended set of each business variety vocabulary in the policy document, and select the business variety vocabulary corresponding to the relevance greater than or equal to a preset threshold as the business variety.

[0038] Specifically, the business category vocabulary is calculated using the following formula An extended collection of In the policy document The correlation in : ; in, , indicating the business type vocabulary An extended collection of In Extension words, For extended words and policy documents The average cosine similarity of each sentence in , indicating the expansion word and policy documents The overall file similarity. Indicates expansion word Frequency of occurrence in policy documents. is the frequency weighting function, which in this application is , which can smooth the influence of word frequency and avoid excessive influence of extremely frequent words on the results. In the above formula, matching based solely on vocabulary may not cover all relevant expressions. The extended words in the vocabulary extension set may be synonyms, related terms, or terms used in different expressions, and these words may appear in different ways in policy documents. In this way, even if the vocabulary of the business variety does not appear directly in the policy document, associations can be established through semantically similar extended words, thereby improving the accuracy of matching.

[0039] In implementation, after calculating the correlation of the extended set of each business variety vocabulary in the policy document, the calculated correlation is compared with the preset threshold, and the business variety vocabulary corresponding to the correlation greater than or equal to the preset threshold is selected as the business variety. At this time, the policy expert will review and confirm. If it is correct, the business variety vocabulary will be determined as the business variety of the policy document. If it is wrong, it will be recalculated or directly selected by the policy expert.

[0040] Step S1023, manually confirm the business items under the business types, and associate the declaration conditions in the policy document with each business item one by one, and form a policy product based on the business types, business items and the declaration conditions corresponding to the business items.

[0041] Specifically, after confirming the business type, policy experts or policy analysts will confirm the business items under the business type, and associate the declaration conditions in the policy document with each business item one by one, thereby completing the formation of the policy product and storing it in the policy product library.

[0042] In step S102, traditional manual disassembly is often prone to misunderstanding due to the complexity and diversity of policy documents, especially when judging the details and semantics of the policy. The introduction of automated expansion and correlation calculation can greatly reduce such misjudgment, improve accuracy, and further ensure the quality of policy products through expert review. Since the declaration conditions of each business item are matched with the business variety through extended vocabulary, it is possible to more accurately connect the field of policy support with the actual situation of the enterprise, reducing the risk of enterprises missing opportunities due to unclear understanding or misjudgment of the scope of policy support. In addition, this method makes it possible to analyze a large number of policy documents, especially when policy documents are frequently updated and the content is complicated, the automated disassembly and matching process can efficiently process large quantities of documents, ensuring that each policy document can be quickly and accurately included in the system. In summary, the combination of automation and expert review can provide enterprises with more efficient and accurate policy matching and analysis services in a complex and dynamically changing policy environment. This not only reduces the workload of manual intervention, but also reduces the risk of policy misunderstandings and missed policy benefits, providing greater protection for enterprises to obtain policy support.

[0043] Step S103: construct a corresponding logical expression for the business items of each policy product in the policy product library.

[0044] In step S103, a corresponding logical expression is manually constructed for the business items of each policy product that is completed by combination. The logical expression is formed based on the business types, business items and their associated declaration conditions of the policy product, and is intended to express the specific combination of conditions that the enterprise must meet to obtain the matching result of the policy product. Specifically, the logical expression is composed of business types, business items and the conditional elements under them according to a certain logical relationship, so that the expression can accurately reflect the compliance requirements in the policy document.

[0045] Specifically, first extract the conditional variables associated with each business item , that is, the application conditions, such as minimum operating income, employee composition ratio, R&D investment, etc., and then determine whether it is necessary to establish a corresponding judgment formula for each condition variable, such as the assumption condition Corresponding to a numerical indicator of the enterprise Minimum value must be met or the maximum value , the judgment formula is as follows: ; Among them, the value is 1 when the condition is met, and the value is 0 when it is not met, thus forming a Boolean logic judgment mechanism.

[0046] Then combine the condition variables based on the logical relationship between the declaration conditions. For example, if the policy document requires certain conditions to "coexist", use the logical "and" relationship to connect; if it is required to meet any one of the conditions, use the logical "or" relationship to connect. According to the logical structure in the policy document, combine all condition variables into a complete logical expression. Construct the logical expression of the business items of each policy product in the policy product library in the above way. For example, the logical expression of a policy product may be: ; The above expression means that only when and or in accordance with And in accordance with or Only then does the enterprise meet the declaration requirements for a certain business item in the policy product.

[0047] Step S104: Acquire enterprise data, bring the enterprise data into the logical expression corresponding to each business item in the policy product for matching, and generate a matching result.

[0048] In step S104, first obtain enterprise data, which includes the company's establishment year, operating income, R&D investment, number of employees, tax information, qualification certificates, etc. On the one hand, enterprise data is automatically collected through the API interface, and on the other hand, it is obtained by uploading by the enterprise itself to ensure the reliability and real-time nature of the data source. After obtaining the enterprise data, the corresponding enterprise data is extracted according to the conditional variables required in each logical expression, rather than directly calling all the data of the enterprise. For example, if the business items of a policy product require that the "establishment year" and "operating income" of the enterprise meet specific conditions, then only the enterprise data about these two conditional variables will be extracted from the enterprise data, and other irrelevant enterprise data, such as "employee composition" or "R&D investment", will not be obtained. After extracting part of the enterprise data, manually check the extracted part of the enterprise data to check whether it is the same as the data required in the logical expression and delete the redundant extracted enterprise data. If it is the same, substitute the extracted enterprise data into the logical expression, calculate item by item according to the rules of the logical expression (such as logical AND, logical OR, etc.) to obtain the matching result. If the matching result 1, that is, the logical expression is true, indicating that the enterprise meets the conditions of the business item. If the matching result If it is 0, the logical expression is false, indicating that the enterprise does not meet the conditions of the business item. If some of the extracted enterprise data is incorrect, re-extract until it is correct.

[0049] The extracted enterprise data is substituted into the logical expression and the calculation process is as follows: for each logical expression of a business item, the conditional variables are searched one by one. For each conditional variable, the corresponding value is obtained from the extracted enterprise data, and the value is substituted into the corresponding position of the logical expression. For example, if the logical expression contains the conditional variable "establishment year ≥ 3 years", the establishment year data of the enterprise is brought into this expression, and it is judged whether it meets the condition "≥ 3 years". After the variable substitution is completed, the logical expression will be converted into an expression consisting of Boolean values ​​(i.e., whether the condition is met or not). The logical expression finally calculates a Boolean value (i.e., true or false) to indicate whether the enterprise meets the conditions of this business item. Calculating whether a Boolean value is satisfied by a logical expression is a relatively common technical means in this field, so it will not be described in detail here.

[0050] In implementation, this application specifically implements two methods for extracting corresponding enterprise data according to the conditional variables required in each logical expression as follows: In one embodiment, the present application extracts the corresponding enterprise data by calculating the semantic similarity. First, each condition variable is defined as , each item of enterprise data (i.e. each sentence) is , both are processed by the existing sentence vector model to obtain their embedding vectors respectively and , and then the semantic similarity between each conditional variable and each item of enterprise data is calculated through the embedding vectors of the two. The calculation formula is as follows: ; in, Represents the inner product of two vectors, which is used to capture the directional consistency of semantics. Represents the Euclidean distance between two vectors The power term is used to increase the weight of sentences with higher semantic similarity at the distance level. Is an adjustable exponential parameter, representing the power term of the Euclidean distance, used to adjust the weight of the distance. In this application, it is set to 2. The denominator is divided by 2 for normalization to keep the result in a reasonable range. In the above formula, when the gap between the conditional variable and the enterprise data sentence is large, the higher power will further reduce the similarity value of these sentences. This can effectively avoid the interference of low-relevance data sentences on the matching results, thereby improving the accuracy and reliability of the screening and ensuring that qualified sentences are selected first. Since the same concept in enterprise data may be presented in different ways of expression, direct keyword matching is prone to miss qualified sentences. Semantic similarity calculation can transcend this difference in expression, capture the deep semantics of the sentence through vector expression, and make the match more inclusive. This design ensures the semantic consistency between the conditional variable and the enterprise data, rather than relying solely on surface matching of words.

[0051] Then, for each condition variable, the semantic similarity between it and each item of enterprise data is calculated, and each item of enterprise data whose semantic similarity is greater than or equal to a preset threshold is extracted, and the extracted part of the enterprise data is brought into the subsequent logical matching.

[0052] In another embodiment, the present application attempts to extract the corresponding enterprise data by calculating the coverage of keywords. First, the description sentences of the condition variables and the sentences in the enterprise data are segmented to extract keywords. Then, the TF-IDF method in the prior art is used to filter out the core keywords in the condition variables to obtain a keyword set. , and then for each item of enterprise data (i.e., each sentence), the coverage of the conditional variable keywords in the sentence is calculated as follows: ; in, Expressing sentences The coverage of conditional variable keywords in the sentence reflects the degree of matching conditions in the sentence. Indicates the enterprise data The word set of a sentence, Represents a set of conditional variable keywords and word set The number of elements in the intersection of The number of condition variable keywords included. Represents a set of conditional variable keywords The total number of keywords in the .

[0053] Then, in order to avoid excessive influence of high-frequency words on the results as much as possible, the weight of each matching word can be smoothed as follows: ; in Indicates conditional variable keywords In the sentence The average frequency of occurrence in . Calculate the smoothed weighted coverage for sentences in all enterprise data , select sentences whose calculation results exceed the preset threshold for data extraction. In the above formula, compared with the complex semantic similarity method, this calculation method based on keyword coverage significantly reduces the computational complexity. It only requires word segmentation and keyword matching, does not rely on deep models, and is suitable for large-scale sentence screening and matching tasks. Keyword coverage provides an intuitive measure of sentence matching, and smoothed weighted coverage further improves the accuracy of matching, ensuring that not only the number of vocabulary matches, but also the actual weight of the vocabulary in the sentence is taken into account.

[0054] In step S104, by extracting only part of the enterprise data that matches the logical expression, redundant calculation of all enterprise data is avoided. This can significantly improve data processing efficiency, because it is not necessary to perform matching operations on all enterprise data every time, and only process part of the data related to specific conditions. Both policy documents and enterprise data may change, and the extraction and review mechanism can flexibly respond to these changes. If certain conditions or data fields in the policy change, manual review can ensure that these changes will not cause errors or omissions in system processing, and ensure the accuracy of policy matching. In addition, automated systems may make misjudgments when processing complex logic, especially in legal and policy-related fields, and incorrect judgments may bring serious compliance risks. Manual review can play a corrective role, reducing the risk of enterprises missing preferential policies or violating compliance requirements due to incorrect data judgments.

[0055] Step S105: Generate a customized enterprise policy report based on the matching results.

[0056] In step S105, a customized policy report is generated for the enterprise based on the matching results generated in step S104. This report presents the matching results between the enterprise and various policy documents to the enterprise, helping the enterprise to clearly understand the types of business, business items and corresponding application conditions that it can apply for. The report is not only intuitive and practical, but also provides a convenient function to further view policy details through technical means (such as QR codes).

[0057] Specifically, first, the overview part of the policy report is generated for the enterprise. In this part, the report will first show the total number of business varieties that the enterprise currently meets and the corresponding number of business items. This section lists the "business variety" categories of all policies that the enterprise meets during the matching process. Each business variety represents a certain type of policy support field (such as "biomedicine", "intelligent manufacturing", "scientific and technological innovation", etc.). Under each business variety, the system will also list the specific number of "business items" that the enterprise meets. Business items are policy conditions that are directly related to the specific funding or support projects that the enterprise can actually apply for. Through this design, enterprises can understand at a glance in which areas they have the opportunity to apply for policy benefits, and how many specific projects they meet.

[0058] In the main body of the report, the report will list in detail the types of business that each enterprise meets and their corresponding business items, and will also attach the judgment basis for each business item, that is, the policy document behind the business item. The judgment basis not only shows the specific policy terms, but also indicates the specific content that the enterprise meets under these terms.

[0059] In addition, in order to facilitate corporate users to further understand the declaration conditions and policy details of each business item, a corresponding QR code is provided for each business item in the report. Corporate users can directly access the policy documents and declaration conditions related to the business item by scanning the QR code. This design greatly improves the practicality and convenience of the report and reduces the difficulty of enterprises in finding specific declaration requirements.

[0060] In summary, combined with Figure 3 This technical solution provides a method for automatically generating customized enterprise policy reports, aiming to solve the problems of inefficiency, misjudgment risk and manual parsing of complex policy documents faced by enterprises when matching and applying for policy benefits. By disassembling policy documents, extracting relevant business varieties and business matters, and establishing a policy product library, it helps enterprises quickly and accurately match the conditions and requirements in policy documents. By building logical expressions to match enterprise data, it can determine whether the enterprise meets the policy requirements, improve the automation and accuracy of policy matching, and solve the errors and efficiency problems in manual analysis.

[0061] It should be noted that today's policy documents usually contain a large number of complex clauses, conditions and requirements, involving subdivided policy content in different fields, industries and enterprise categories. For enterprises, manually extracting key information from these complex documents and judging whether they meet the conditions is a tedious and error-prone process. Therefore, breaking down policy documents into two levels, "business varieties" and "business matters", has the following significant benefits: First, policy documents are often lengthy and involve multiple fields and different applicable conditions. By breaking them down into two levels, "business varieties" and "business matters", complex policy content can be structured and decomposed to extract clear core support directions. In this way, when matching policies, enterprises can quickly identify policy varieties and matters related to their own businesses, avoiding getting lost in the complex policy clauses. Second, in the process of breaking down, business varieties represent high-level classifications in policy documents, such as "biomedicine", "intelligent manufacturing" and other fields, while business matters are specific support projects, such as "clinical trial funding" or "equipment purchase subsidies". By classifying policy documents according to these levels, enterprises can accurately match them based on their own industries and actual needs. This classification method can greatly narrow the matching scope, improve the efficiency of policy matching, and avoid the tedious work of manual comparison. In addition, policy documents are constantly changing, especially when regulations are updated or new policy support is added. Breaking down policies into "business types" and "business matters" can help companies quickly connect with new policy changes. Because only a certain business type or matter needs to be updated and reviewed, rather than re-examining the entire policy document. In this way, companies can adjust their policy declaration strategies in a timely manner and reduce misjudgments and omissions caused by policy changes.

[0062] Figure 4 This is a structural block diagram of an enterprise policy customized report generation system provided by an embodiment of the present application. The system includes at least the following modules: The policy document acquisition module is used to acquire policy documents and build a policy document library, and dynamically update the policy document library; The policy document disassembly module is used to disassemble the policy documents in the policy document library, generate the business types and business items corresponding to the policy documents, form policy products based on the business types, business items and the declaration conditions corresponding to the business items, and build a policy product library; A logic expression building module is used to build corresponding logic expressions for the business items of each policy product in the policy product library; The policy matching module is used to obtain enterprise data, bring the enterprise data into the logical expression corresponding to each business item in the policy product for matching, and generate matching results; The report generation module is used to generate enterprise policy customized reports based on the matching results.

[0063] For relevant details, refer to the above method embodiment.

[0064] Figure 5 4 is a block diagram of an electronic device provided by an embodiment of the present application. The device at least includes a processor 401 and a memory 402.

[0065] The processor 401 may include one or more processing cores, such as a 4-core processor, an 8-core processor, etc. The processor 401 may be implemented in at least one hardware form of DSP (Digital Signal Processing), FPGA (Field-Programmable Gate Array), and PLA (Programmable Logic Array). The processor 401 may also include a main processor and a coprocessor. The main processor is a processor for processing data in the awake state, also known as a CPU (Central Processing Unit); the coprocessor is a low-power processor for processing data in the standby state. In some embodiments, the processor 401 may be integrated with a GPU (Graphics Processing Unit), which is responsible for rendering and drawing the content to be displayed on the display screen. In some embodiments, the processor 401 may also include an AI (Artificial Intelligence) processor, which is used to process computing operations related to machine learning.

[0066] The memory 402 may include one or more computer-readable storage media, which may be non-transitory. The memory 402 may also include a high-speed random access memory and a non-volatile memory, such as one or more disk storage devices and flash storage devices. In some embodiments, the non-transitory computer-readable storage medium in the memory 402 is used to store at least one instruction, which is used to be executed by the processor 401 to implement the enterprise policy customized report generation method provided in the method embodiment of the present application.

[0067] In some embodiments, the electronic device may further optionally include: a peripheral device interface and at least one peripheral device. The processor 401, the memory 402 and the peripheral device interface may be connected via a bus or a signal line. Each peripheral device may be connected to the peripheral device interface via a bus, a signal line or a circuit board. Schematically, the peripheral devices include but are not limited to: a radio frequency circuit, a touch display screen, an audio circuit, and a power supply.

[0068] Of course, the electronic device may also include fewer or more components, which is not limited in this embodiment.

[0069] Optionally, the present application also provides a computer-readable storage medium, in which a program is stored, and the program is loaded and executed by a processor to implement the enterprise policy customized report generation method of the above method embodiment.

[0070] Optionally, the present application also provides a computer product, which includes a computer-readable storage medium, in which a program is stored, and the program is loaded and executed by a processor to implement the enterprise policy customized report generation method of the above method embodiment.

[0071] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0072] The above embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention patent. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the patent of the present application shall be subject to the attached claims.

Claims

1. A method for generating a customized enterprise policy report, characterized in that: The method comprises: acquiring policy documents and constructing a policy document library, and dynamically updating the policy document library; Disassemble the policy documents in the policy document library to generate business types and business items corresponding to the policy documents, form policy products based on the business types, business items and declaration conditions corresponding to the business items, and build a policy product library; Constructing corresponding logical expressions for the business items of each policy product in the policy product library; Acquire enterprise data, bring the enterprise data into the logical expression corresponding to each business item in the policy product for matching, and generate matching results; Generate a customized report on enterprise policies based on the matching results.

2. The method for generating a customized enterprise policy report according to claim 1, characterized in that: The policy documents in the policy document library are disassembled to generate business types and business items corresponding to the policy documents, and the policy products are formed based on the business types, the business items and the declaration conditions corresponding to the business items, including: Constructing a business variety vocabulary set, generating an extended set of each business variety vocabulary in the business variety vocabulary set; Calculating the relevance of the extended set of each of the business variety words in the policy document, and selecting the business variety words corresponding to the relevance greater than or equal to a preset threshold as the business variety; By manually confirming the business items under the business types, the declaration conditions in the policy documents are associated with each business item one by one, and policy products are formed based on the business types, business items and the declaration conditions corresponding to the business items.

3. The method for generating a customized enterprise policy report according to claim 2, characterized in that: The step of calculating the relevance of the extended set of each of the business variety words in the policy document and selecting the business variety words corresponding to the relevance being greater than or equal to a preset threshold as the business variety includes: Use the following formula to calculate the business category vocabulary An extended collection of In the policy document The correlation in : ； in, , indicating the business type vocabulary An extended collection of In Extension words, For extended words and policy documents The average cosine similarity of each sentence in , indicating the expansion word and policy documents The overall file similarity, Indicates expansion word Frequency of occurrence in policy documents, is the frequency weighting function, ; After calculating the correlation of the extended set of each business variety vocabulary in the policy document, the calculated correlation is compared with the preset threshold, and the business variety vocabulary corresponding to the correlation greater than or equal to the preset threshold is selected as the business variety, which is reviewed and confirmed by policy experts. If correct, the business variety vocabulary is determined as the business variety of the policy document.

4. The method for generating a customized enterprise policy report according to claim 1, characterized in that: The constructing of a corresponding logical expression for the business items of each policy product in the policy product library comprises: Extract the conditional variables associated with each business item; Determine whether a corresponding judgment formula needs to be established for each conditional variable; The condition variables are combined based on the logical relationship between the declaration conditions, and all condition variables are combined into a complete logical expression according to the logical structure in the policy document.

5. The method for generating a customized enterprise policy report according to claim 1, characterized in that: The obtaining of enterprise data, bringing the enterprise data into the logical expression corresponding to each business item in the policy product for matching, and generating a matching result includes: After obtaining the enterprise data, extract the corresponding enterprise data according to the conditional variables required in each logical expression; After extracting part of the enterprise data, manually check whether the extracted part of the enterprise data is the same as the data required in the logical expression and delete the extracted redundant enterprise data; If they are the same, the extracted enterprise data will be substituted into the logical expression, and calculated item by item according to the rules of the logical expression to obtain the matching result. If part of the extracted enterprise data is different, it will be re-extracted until it is correct.

6. The method for generating a customized enterprise policy report according to claim 5, characterized in that: The extracting corresponding enterprise data according to the conditional variables required in each logical expression includes: By calculating the semantic similarity, the corresponding enterprise data is extracted and each condition variable is defined as , each item of enterprise data is , both are processed by the existing sentence vector model to obtain their embedding vectors respectively and , the semantic similarity between each conditional variable and each item of enterprise data is calculated through the embedding vectors of the two. The calculation formula is as follows: ； in, represents the inner product of two vectors, Represents the Euclidean distance between two vectors The power term, is an adjustable exponential parameter, representing the power term of the Euclidean distance, used to adjust the weight of the distance. Set to 2; For each conditional variable, the semantic similarity between it and each item of enterprise data is calculated, and each item of enterprise data whose semantic similarity is greater than or equal to a preset threshold is extracted, and the extracted part of the enterprise data is brought into the logical expression.

7. The method for generating a customized enterprise policy report according to claim 5, characterized in that: The extracting corresponding enterprise data according to the conditional variables required in each logical expression includes: The corresponding enterprise data is extracted by calculating the coverage of keywords, the description sentences of the conditional variables and the sentences in the enterprise data are segmented, keywords are extracted, and the core keywords in the conditional variables are screened out using the TF-IDF method to obtain a keyword set. , for each item of enterprise data , calculate the coverage of conditional variable keywords in the sentence as follows: ； in, Expressing sentences The coverage of conditional variable keywords in Indicates the enterprise data The word set of a sentence, Represents a set of conditional variable keywords and word set The number of elements in the intersection of How many condition variable keywords are included in Represents a set of conditional variable keywords The total number of keywords in the The weight of each matching word is smoothed as follows: ； in Indicates conditional variable keywords In the sentence The average frequency of occurrence in the sentence is used to calculate the smoothed weighted coverage of all sentences in the enterprise data. , sentences whose calculation results exceed the preset threshold are selected for data extraction.

8. A system for generating enterprise policy customized reports, characterized in that: include: A policy document acquisition module is used to acquire policy documents and build a policy document library, and dynamically update the policy document library; A policy document disassembly module is used to disassemble the policy documents in the policy document library, generate business types and business items corresponding to the policy documents, form policy products based on the business types, business items and declaration conditions corresponding to the business items, and build a policy product library; A logic expression building module, used to build a corresponding logic expression for the business items of each policy product in the policy product library; A policy matching module is used to obtain enterprise data, bring the enterprise data into the logical expression corresponding to each business item in the policy product for matching, and generate a matching result; A report generation module is used to generate a customized report on enterprise policies based on the matching results.

9. An electronic device, characterized in that: The device includes a processor and a memory; a program is stored in the memory, and the program is loaded and executed by the processor to implement a method for generating a customized enterprise policy report as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The storage medium stores a program, which, when executed by a processor, is used to implement a method for generating a customized enterprise policy report as described in any one of claims 1 to 7.

Images