Policy recommendation method and device, electronic equipment and storage medium

Abstract

The invention provides a policy recommendation method and device, electronic equipment and a storage medium, and the method comprises the steps: obtaining a policy file, extracting policy label data from the policy file through a policy label model, and generating a policy label database; obtaining enterprise information, extracting enterprise label data from the enterprise information through the enterprise label model, and generating an enterprise label database; wherein the enterprise label model is constructed based on policy label model mapping; and performing matching processing on the policy label database and the enterprise label database, obtaining a matched policy file according to a matching result, and performing policy recommendation to an enterprise. According to the invention, policy declaration information and enterprise information can be rapidly and efficiently obtained, a policy supply side and a policy demand side are accurately matched, the policy implementation effect is ensured, and the government affair service quality of the government is improved.

Description

Technical Field

[0001] The present invention relates to the field of intelligent recommendation technology, and in particular to a policy recommendation method, device, electronic equipment and storage medium. Background Art

[0002] At present, most government departments release new policies and support information through their official websites.

[0003] In the existing technology, policy recommendation systems mostly use relevant information such as active registration and policy browsing by enterprises, and use basic enterprise information databases such as Qichacha to unilaterally profile enterprise users, and then use natural language processing technology to extract keywords related to policy texts as policy tags, and finally calculate the matching degree based on policy tags and enterprise tags. However, this recommendation system has the following defects: it is heavily dependent on enterprise registration and reporting information, and the audience range is limited; it does not establish a detailed label model for policy application conditions and corresponding enterprise objects, which can easily lead to misalignment of policy and enterprise matching; corresponding label matching often only considers the successful matching rate, but ignores the decisive role of key policy conditions in matching.

[0004] It should be noted that this section is intended to provide a background or context to the embodiments of the present disclosure as recited in the claims. No admission is made that the description herein is prior art by virtue of its inclusion in this section. Summary of the invention

[0005] The embodiments of the present invention provide a policy recommendation method, device, electronic device and storage medium to solve the problems in the prior art that the policy recommendation method has a limited audience range, does not establish a matching model between the policy application conditions and the corresponding enterprise objects in a targeted manner, and ignores the decisive role of key policy conditions in matching.

[0006] In a first aspect, an embodiment of the present invention provides a policy recommendation method, comprising:

[0007] Obtaining a policy document, extracting policy label data from the policy document through a policy label model, and generating a policy label database;

[0008] Acquire enterprise information, extract enterprise tag data from the enterprise information through an enterprise tag model, and generate an enterprise tag database; wherein the enterprise tag model is constructed based on the policy tag model mapping;

[0009] The policy tag database and the enterprise tag database are matched, a matching policy document is obtained according to the matching result, and a policy recommendation is made to the enterprise.

[0010] Preferably, before the step of obtaining the policy document, the following steps are included:

[0011] Classify the policy document samples and set the first-level policy labels corresponding to each category;

[0012] Setting the second-level policy tags included in each of the first-level policy tags, processing the policy document samples corresponding to the second-level policy tags, and obtaining the corresponding third-level policy tags;

[0013] Based on the three-level policy labels, the policy label model is constructed.

[0014] Preferably, the step of mapping and constructing the enterprise tag model based on the policy tag model includes:

[0015] Constructing enterprise basic labels and enterprise special labels of the enterprise according to the enterprise objects required in the policy label model;

[0016] The enterprise tag model is constructed based on the enterprise basic tags and the enterprise special tags.

[0017] Preferably, the enterprise information includes enterprise numerical information and enterprise text information.

[0018] Preferably, the step of matching the policy tag database with the enterprise tag database, obtaining a matching policy document according to the matching result and recommending a policy to the enterprise includes:

[0019] Calculating the text similarity value of the policy tag database and the enterprise tag database, and comparing the text similarity value with a similarity threshold to obtain a matching score; the matching score is 1 when the text similarity value is greater than or equal to the similarity threshold, otherwise it is 0;

[0020] The weight of each policy tag data in the policy tag database is set, and the comprehensive matching degree is calculated according to the weight of the policy tag data and the matching score; the calculation formula of the comprehensive matching degree is: comprehensive matching degree = weight of policy tag data × matching score × 100%;

[0021] If the comprehensive matching degree is greater than or equal to the matching threshold, the policy document corresponding to the comprehensive matching degree is recommended to the enterprise.

[0022] In a second aspect, an embodiment of the present invention provides a policy recommendation device, including:

[0023] A policy label database module is used to obtain a policy document, extract policy label data from the policy document through a policy label model, and generate a policy label database;

[0024] An enterprise tag database module is used to obtain enterprise information, extract enterprise tag data from the enterprise information through an enterprise tag model, and generate an enterprise tag database; wherein the enterprise tag model is constructed based on the policy tag model mapping;

[0025] The policy matching and recommendation module is used to match the policy tag database and the enterprise tag database, obtain matching policy files according to the matching results, and make policy recommendations to the enterprise.

[0026] Preferably, the policy label database module is specifically used for:

[0027] Classify the policy document samples and set the first-level policy labels corresponding to each category;

[0028] Setting the second-level policy tags included in each of the first-level policy tags, processing the policy document samples corresponding to the second-level policy tags, and obtaining the corresponding third-level policy tags;

[0029] Based on the three-level policy labels, the policy label model is constructed.

[0030] Preferably, the policy matching recommendation module is specifically used for:

[0031] Calculating the text similarity value of the policy tag database and the enterprise tag database, and comparing the text similarity value with a similarity threshold to obtain a matching score; the matching score is 1 when the text similarity value is greater than or equal to the similarity threshold, otherwise it is 0;

[0032] The weight of each policy tag data in the policy tag database is set, and the comprehensive matching degree is calculated according to the weight of the policy tag data and the matching score; the calculation formula of the comprehensive matching degree is: comprehensive matching degree = weight of policy tag data × matching score × 100%;

[0033] If the comprehensive matching degree is greater than or equal to the matching threshold, the policy document corresponding to the comprehensive matching degree is recommended to the enterprise.

[0034] In a third aspect, an embodiment of the present invention provides a computing device, comprising a processor and a memory, wherein the memory stores execution instructions, and the processor reads the execution instructions in the memory to execute the steps described in the above-mentioned policy recommendation method.

[0035] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium comprising computer-executable instructions, wherein the computer-executable instructions are used to execute the steps described in the above-mentioned policy recommendation method.

[0036] The policy recommendation method and device provided by the embodiment of the present invention scientifically divides the enterprise policies, and then constructs a policy label model after extracting various policy label templates, and constructs a corresponding enterprise label model on this basis. Automatic labeling tools, policy label models and enterprise label models are used to automatically extract policy and enterprise label data into the corresponding database, and then the comprehensive matching degree of labels is calculated using similarity to provide online and offline policy push services for government departments. The present invention can quickly and efficiently obtain policy declaration information and enterprise information, and accurately match the policy supply side and demand side, ensure the effect of policy implementation, and improve the quality of government services. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0038] Figure 1 A flowchart of a policy recommendation method provided by an embodiment of the present invention;

[0039] Figure 2 Another implementation flow chart of a policy recommendation method provided by an embodiment of the present invention;

[0040] Figure 3 A schematic diagram of the structure of a policy recommendation device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0041] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0042] It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, further definition and explanation thereof is not required in subsequent drawings.

[0043] Figure 1 A flowchart of a policy recommendation method provided by an embodiment of the present invention is shown in FIG. Figure 1 As shown, the policy recommendation method mainly includes:

[0044] Step 101: Obtain a policy document, extract policy label data from the policy document through a policy label model, and generate a policy label database.

[0045] In the embodiment, relevant policy documents, such as various new policies, support information, etc., can be automatically collected from official government websites, etc., and the collection method is not limited.

[0046] The collected policy documents need to be extracted through the policy label model to facilitate the subsequent matching with the corresponding enterprises. Specifically, the construction of the policy label model in this embodiment includes the following steps:

[0047] S1. Classify the policy document samples and set the first-level policy labels corresponding to each category.

[0048] In this step, after obtaining some policy document samples, the policy documents are divided from the perspective of policy tools according to the corporate policies involved in the policy document samples. This embodiment divides the policy document samples into three categories, and sets the first-level policy labels corresponding to each category, namely supply policy, demand policy and environmental policy. The policy tools described in this embodiment refer to the means to achieve policy goals.

[0049] S2. Set the second-level policy tags included in each first-level policy tag, process the policy document samples corresponding to the second-level policy tags, and obtain the corresponding third-level policy tags.

[0050] In this step, according to the policy tools, the obtained policy document samples are encoded and the policy themes are extracted, and finally the secondary policy tags included in each first-level policy tag are summarized. For example, the second-level policy tags included in the first-level policy tag supply policy are finance and finance, investment and credit, talent training, foreign investment and foreign trade, and industrial development. The second-level policy tags included in the first-level policy tag demand policy are scientific and technological innovation and tax reduction and fee reduction, and the second-level policy tags included in the first-level policy tag environmental policy are market supervision, entrepreneurship support, and comprehensive. In this embodiment, there are 10 second-level policy tags in total. Of course, those skilled in the art can set other numbers according to actual conditions.

[0051] The policy document samples corresponding to the second-level policy labels are summarized to obtain the corresponding third-level policy labels. The third-level policy labels include: general labels: policy source, policy object, policy theme, applicable industry, business department, etc., application conditions: policy applicable area, application time, support method, enterprise scale, enterprise registration time, enterprise business scope, enterprise registration status, enterprise qualifications, special recognition information, special recognition time, enterprise intangible assets, credit record, business ability, enterprise awards / honors, history of similar policies, policy goals, reward amount, application materials, etc.

[0052] S3. Build a policy label model based on the three-level policy labels.

[0053] In this step, after the three-level policy labels are summarized through the above steps, a policy label model is constructed based on the three-level policy labels.

[0054] Using the three-level policy labels in the policy label model and a small amount of training samples, the policy label model is trained with the help of UIE (Universal Information Extraction) integrated open source annotation tool doccano. Then, the policy label data about the fields such as "policy source, policy object, policy theme, applicable industry, business department, policy applicable area, application time, support method, enterprise scale, enterprise registration time, enterprise business scope, enterprise registration status, enterprise qualification, special recognition information, special recognition time, enterprise intangible assets, credit record, operating ability, enterprise awards / honors, non-duplicate enjoyment of policies, policy goals, reward amount, application materials" in the three-level policy labels are extracted from the automatically collected policy document annotations, and the policy label data are uniformly stored in the policy label database to complete the information extraction of policy documents.

[0055] Step 102: Obtain enterprise information, extract enterprise tag data from the enterprise information through an enterprise tag model, and generate an enterprise tag database.

[0056] In the embodiment, the enterprise information may be automatically collected through an enterprise information database, an enterprise basic information query platform, or an enterprise credit open platform, and the collection method is not limited.

[0057] The collected policy information needs to be extracted through the enterprise tag model to facilitate matching with the corresponding policy documents. Specifically, the steps of building the enterprise tag model in this embodiment include the following:

[0058] P1. Construct the enterprise basic labels and enterprise special labels of the enterprise according to the enterprise objects required in the policy label model.

[0059] In this step, according to the enterprise objects required in the policy label model, the basic enterprise labels and enterprise special labels are constructed. The basis for formulating these two types of labels is that the basic enterprise labels are symbols that reflect the common characteristics of each enterprise, while the enterprise special labels are intended to highlight the uniqueness and differences between different enterprises. Of course, those skilled in the art can adjust the number and types of these two types of labels according to actual conditions.

[0060] Among them, the basic enterprise tags constructed include: company name, company ID, operating status, industry, enterprise type, enterprise scale, establishment date, registered capital, tax record, enterprise scale, and registered address; the special enterprise tags include: credit record, special identification information, revenue, profit, intellectual property rights, and pledge information. The detailed indicators within these two types of tags can be flexibly adjusted.

[0061] P2. Build an enterprise label model based on enterprise basic labels and enterprise special labels.

[0062] In this step, an enterprise label model can be constructed based on the enterprise basic labels and enterprise special labels constructed in the above steps.

[0063] The enterprise tag model is built based on the policy tag model, which realizes the mapping between the policy tag model and the enterprise tag model. The enterprise basic tags and enterprise special tags are generated, so that the enterprise information can be extracted. The extracted enterprise tag data will be uniformly stored in the enterprise tag database.

[0064] Preferably, the enterprise information includes enterprise numerical information and enterprise text information. For enterprise numerical information, the enterprise label model and regular tools are used to automatically extract the enterprise label data, and the extracted enterprise label data are uniformly stored in the enterprise label database; for enterprise text information, the enterprise label model and a small number of training samples are used to train the enterprise label model with the help of UIE integrated open source annotation tool doccano, and then automatically extract the extracted enterprise label data, and uniformly store the extracted enterprise label data in the enterprise label database.

[0065] Step 103: Match the policy tag database and the enterprise tag database, obtain matching policy documents based on the matching results, and make policy recommendations to the enterprise.

[0066] In the embodiment, based on the multi-dimensional similarity policy-enterprise matching algorithm model, multi-dimensional intelligent matching of the enterprise tag database and the policy tag database is realized, and the comprehensive matching results are recorded. According to the matching results and the threshold, the matching results that meet the threshold conditions are compared, and the corresponding matching policy files are intelligently pushed to the enterprise.

[0067] The steps of matching the policy tag database and the enterprise tag database, obtaining a matching policy document according to the matching result and recommending a policy to the enterprise include the following steps:

[0068] Q1. Calculate the text similarity values ​​of the policy label database and the enterprise label database, and compare the text similarity values ​​with the similarity threshold to obtain the matching score.

[0069] In this step, the text similarity values ​​of the word vector policy tag database and the enterprise tag database are used and compared with a pre-given similarity threshold. The matching score is 1 when the text similarity value is greater than or equal to the similarity threshold, otherwise it is 0.

[0070] The similarity threshold is set by technicians in this field based on the verification results and conclusions of the text similarity algorithm model. The specific determination principles include: (1) Different specific application scenarios will have different similarity requirements and different similarity thresholds; (2) The setting of the similarity threshold needs to comprehensively consider the balance between the false positive rate and the false negative rate; (3) Experimental verification and dynamic adjustment are used to ensure that the setting and effect of the similarity threshold are always in the optimal state.

[0071] Q2. Set the weight of each policy label data in the policy label database, and calculate the comprehensive matching degree based on the weight and matching score of the policy label data.

[0072] In this step, the weight of each policy label data in the policy label database is first set, and then the comprehensive matching is calculated based on the weight and matching score of the policy label data. For example, those skilled in the art can evaluate the relative importance of each policy label data according to the analytic hierarchy process (AHP), and integrate the matrix algorithm to accurately calculate the weight distribution of each policy label data.

[0073] Among them, the calculation formula for comprehensive matching degree is: comprehensive matching degree = weight of policy label data × matching score × 100%.

[0074] According to the above formula, the matching degree between each policy label data and the enterprise can be accurately calculated, and the overall matching degree can be automatically adjusted using the weight and matching score of the policy label data.

[0075] Q3. If the comprehensive matching degree is greater than or equal to the matching threshold, the policy document corresponding to the comprehensive matching degree will be recommended to the enterprise.

[0076] In this step, a matching threshold is preset, such as 75% or 85%. If the calculated comprehensive matching degree is greater than or equal to the matching threshold, the policy document corresponding to the comprehensive matching degree is recommended to the enterprise.

[0077] Regarding the setting of the matching threshold, those skilled in the art can make appropriate adjustments based on different policy label models or scenarios. The specific principles for adjusting the matching threshold range include: (1) different specific application scenarios; (2) false positive rate and false negative rate; (3) model performance and effect; (4) universality of model parameters.

[0078] According to the policy recommendation method described in the above embodiment, the "capital support" policy involved in the "Qin Chuangyuan Comprehensive Service Platform" is used as an example to illustrate its specific implementation method. The platform brings together multi-channel and multi-type policies in order to enable various types of enterprises in Shaanxi Province to intuitively and accurately obtain "capital support" policy documents and obtain more detailed application conditions; at the same time, it assists government units to achieve intelligent matching of support policies and push them to enterprise users efficiently and accurately. According to step 101, the annotation and extraction of the original policy documents of the platform are completed, the data cleaning and fusion processing of the three-level policy labels is realized, and the policy label database is generated. According to step 102, the mapping of the policy label model and the enterprise label model is realized, the basic enterprise label and the special enterprise label are generated, the extraction of the original enterprise information is completed, and the enterprise label database is generated. According to the matching processing of the policy label database and the enterprise label database in step 103, the multi-dimensional intelligent matching of the policy label database and the enterprise label database is realized, and the comprehensive matching degree is recorded, and then the comprehensive matching degree is compared with the matching threshold, and the matching results that meet the threshold conditions are intelligently pushed.

[0079] Figure 2 Another implementation flow chart of a policy recommendation method provided by an embodiment of the present invention, refer to Figure 2 As shown, the policy recommendation method mainly includes:

[0080] Step 201: classify the policy document samples, set the first-level policy labels corresponding to each category; set the second-level policy labels included in each of the first-level policy labels, process the policy document samples corresponding to the second-level policy labels to obtain the corresponding third-level policy labels; and construct a policy label model based on the third-level policy labels;

[0081] Step 202: Obtain a policy document, extract policy label data from the policy document using a policy label model, and generate a policy label database;

[0082] Step 203: construct the basic enterprise label and the enterprise special label of the enterprise according to the enterprise object required in the policy label model; construct the enterprise label model based on the basic enterprise label and the enterprise special label;

[0083] Step 204: Acquire enterprise information, extract enterprise tag data from the enterprise information through the enterprise tag model, and generate an enterprise tag database;

[0084] Step 205: Match the policy tag database and the enterprise tag database, obtain matching policy documents based on the matching results, and make policy recommendations to the enterprise.

[0085] In summary, the present invention scientifically divides corporate policies, and then extracts various policy label templates to construct a policy label model, and on this basis, constructs a corresponding corporate label model. Using automatic labeling tools, policy label models and corporate label models, the policy and corporate label data are automatically extracted into the corresponding database, and then the similarity is used to calculate the comprehensive matching degree of the labels to provide online and offline policy push services for government departments. The present invention can quickly and efficiently obtain policy declaration information and corporate information, and accurately match the policy supply side and demand side, ensure the effectiveness of policy implementation, and improve the quality of government services.

[0086] The specific implementation process of each step in the above embodiment can refer to the description of each method step in the above embodiment, which will not be repeated here.

[0087] It should be noted that, for the sake of simplicity, the embodiments of the above method are described as a series of action combinations, but those skilled in the art should know that the present invention is not limited by the described action sequence. Secondly, those skilled in the art should also know that the embodiments described in the specification are all preferred embodiments, and the actions involved are not necessarily required by the present invention.

[0088] Reference Figure 3 As shown, based on the same inventive concept, an embodiment of the present invention provides a policy recommendation device, which mainly includes:

[0089] The policy label database module 31 is used to obtain policy documents, extract policy label data from the policy documents through a policy label model, and generate a policy label database;

[0090] The enterprise tag database module 32 is used to obtain enterprise information, extract enterprise tag data from the enterprise information through an enterprise tag model, and generate an enterprise tag database; wherein the enterprise tag model is constructed based on the policy tag model mapping;

[0091] The policy matching and recommendation module 33 is used to match the policy tag database and the enterprise tag database, obtain matching policy documents according to the matching results, and make policy recommendations to the enterprise.

[0092] It should be noted here that the above-mentioned policy label database module 31, enterprise label database module 32 and policy matching recommendation module 33 correspond to steps 101 to 103 in the above-mentioned method embodiment, and the three units have the same instances and application scenarios implemented by the corresponding steps, but are not limited to the contents disclosed in the above-mentioned method embodiment.

[0093] Preferably, the policy label database module 31 is specifically used for:

[0094] Classify enterprise policies and set corresponding primary policy labels for each category;

[0095] Set the second-level policy tags included in each first-level policy tag, process the second-level policy tags, and obtain the corresponding third-level policy tags;

[0096] Based on the three-level policy labels, a policy label model is constructed.

[0097] Preferably, the policy matching recommendation module 32 is specifically used for:

[0098] Calculate the text similarity value of the policy label database and the enterprise label database, and compare the text similarity value with the similarity threshold to obtain a matching score; the matching score is 1 when the text similarity value is greater than or equal to the similarity threshold, otherwise it is 0;

[0099] The weight of each policy label data in the policy label database is set, and the comprehensive matching degree is calculated according to the weight and matching score of the policy label data; the calculation formula of the comprehensive matching degree is: comprehensive matching degree = weight of policy label data × matching score × 100%;

[0100] If the comprehensive matching degree is greater than or equal to the matching threshold, the policy document corresponding to the comprehensive matching degree will be recommended to the enterprise.

[0101] It should be noted that the policy recommendation device provided in the embodiment of the present invention and the policy recommendation method described in the aforementioned embodiment belong to the same technical concept, and its specific implementation process can refer to the description of the method steps in the aforementioned embodiment, which will not be repeated here.

[0102] It should be understood that the units included in the above policy recommendation device are only logical divisions based on the functions implemented by the device. In actual applications, the above units can be superimposed or split. In addition, the functions implemented by the policy recommendation device provided in this embodiment correspond one-to-one to the policy recommendation method provided in the above embodiment. The more detailed processing flow implemented by the device has been described in detail in the above method embodiment and will not be described in detail here.

[0103] Based on the same inventive concept, an embodiment of the present invention provides an electronic device, which mainly includes a processor and a memory, wherein the memory stores an execution instruction. The processor reads the execution instruction in the memory to execute the steps described in any embodiment of the above-mentioned policy recommendation method. Alternatively, the processor reads the execution instruction in the memory to implement the functions of each unit in any embodiment of the above-mentioned policy recommendation device.

[0104] The embodiment of the present invention further provides a computer-readable storage medium, comprising computer-executable instructions, wherein the computer-executable instructions are used to execute the steps described in any of the above-mentioned policy recommendation methods or the functions of the various units in the above-mentioned policy recommendation device embodiments.

[0105] Optionally, in this embodiment, the above storage medium may be located in any computer terminal in a computer terminal group in a computer network, or in any mobile terminal in a mobile terminal group.

[0106] In the above embodiments of the present invention, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0107] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A policy recommendation method, characterized in that: include: Obtaining a policy document, extracting policy label data from the policy document through a policy label model, and generating a policy label database; Acquire enterprise information, extract enterprise tag data from the enterprise information through an enterprise tag model, and generate an enterprise tag database; wherein the enterprise tag model is constructed based on the policy tag model mapping; The policy tag database and the enterprise tag database are matched, a matching policy document is obtained according to the matching result, and a policy recommendation is made to the enterprise.

2. The method according to claim 1, characterized in that Prior to the step of obtaining the policy document, the following steps are included: Classify the policy document samples and set the first-level policy labels corresponding to each category; Setting the second-level policy tags included in each of the first-level policy tags, processing the policy document samples corresponding to the second-level policy tags, and obtaining the corresponding third-level policy tags; Based on the three-level policy labels, the policy label model is constructed.

3. The method according to claim 1, characterized in that: The step of mapping and constructing the enterprise tag model based on the policy tag model includes: Constructing enterprise basic labels and enterprise special labels of the enterprise according to the enterprise objects required in the policy label model; The enterprise tag model is constructed based on the enterprise basic tags and the enterprise special tags.

4. The method according to claim 1, characterized in that: The enterprise information includes enterprise numerical information and enterprise text information.

5. The method according to claim 1, characterized in that The step of matching the policy tag database with the enterprise tag database, obtaining a matching policy document according to the matching result and recommending a policy to the enterprise includes: Calculating the text similarity value of the policy tag database and the enterprise tag database, and comparing the text similarity value with a similarity threshold to obtain a matching score; the matching score is 1 when the text similarity value is greater than or equal to the similarity threshold, otherwise it is 0; The weight of each policy tag data in the policy tag database is set, and the comprehensive matching degree is calculated according to the weight of the policy tag data and the matching score; the calculation formula of the comprehensive matching degree is: comprehensive matching degree = weight of policy tag data × matching score × 100%; If the comprehensive matching degree is greater than or equal to the matching threshold, the policy document corresponding to the comprehensive matching degree is recommended to the enterprise.

6. A policy recommendation device, characterized in that: include: A policy label database module is used to obtain a policy document, extract policy label data from the policy document through a policy label model, and generate a policy label database; An enterprise tag database module is used to obtain enterprise information, extract enterprise tag data from the enterprise information through an enterprise tag model, and generate an enterprise tag database; wherein the enterprise tag model is constructed based on the policy tag model mapping; The policy matching and recommendation module is used to match the policy label database and the enterprise label database, obtain matching policy documents according to the matching results, and make policy recommendations to the enterprise.

7. The device according to claim 6, characterized in that The policy label database module is specifically used for: Classify the policy document samples and set the first-level policy labels corresponding to each category; Setting the second-level policy tags included in each of the first-level policy tags, processing the policy document samples corresponding to the second-level policy tags, and obtaining the corresponding third-level policy tags; Based on the three-level policy labels, the policy label model is constructed.

8. The device according to claim 1, characterized in that The policy matching recommendation module is specifically used for: Calculating the text similarity value of the policy tag database and the enterprise tag database, and comparing the text similarity value with a similarity threshold to obtain a matching score; the matching score is 1 when the text similarity value is greater than or equal to the similarity threshold, otherwise it is 0; The weight of each policy tag data in the policy tag database is set, and the comprehensive matching degree is calculated according to the weight of the policy tag data and the matching score; the calculation formula of the comprehensive matching degree is: comprehensive matching degree = weight of policy tag data × matching score × 100%; If the comprehensive matching degree is greater than or equal to the matching threshold, the policy document corresponding to the comprehensive matching degree is recommended to the enterprise.

9. An electronic device, characterized in that: The method comprises a processor and a memory, wherein the memory stores execution instructions, and the processor reads the execution instructions in the memory to execute the steps in the policy recommendation method according to any one of claims 1 to 5.

10. A computer-readable storage medium, characterized in that: The method comprises computer-executable instructions, wherein the computer-executable instructions are used to execute the steps in the policy recommendation method according to any one of claims 1 to 5.

Images