Intelligent policy matching recommendation method

Abstract

The invention relates to the technical field of policy matching, in particular to an intelligent policy matching recommendation method, which comprises the following steps of: extracting policy conditions and policy rewards based on a policy file, acquiring target enterprise information, comparing information corresponding to the elastic condition and the rigid condition in the target enterprise information with the elastic condition and the rigid condition in each policy file one by one; and when the non-conforming elastic condition is greater than or equal to 1, calculating the ratio of the target enterprise information data to the quantifiable index in the corresponding non-conforming elastic condition, and when the ratio is greater than 80%, pushing the corresponding policy file and the corresponding policy reward to the user. According to the policy recommendation method, policy recommendation is performed to the user in a manner of matching the elastic condition and the rigid condition, and when the non-conforming elastic condition is greater than or equal to 1, the recommendation of the corresponding policy can be continued, so that the industry pain point that enterprises approaching to the standard are mistakenly screened is solved.

Description

Technical Field

[0001] The present invention relates to the field of policy matching technology, and in particular to an intelligent policy matching recommendation method. Background Art

[0002] In today's society, governments formulate and implement numerous policies to promote economic development, advance social progress, and safeguard people's well-being. These policies, covering a wide range of areas, including the economy, science and technology, education, healthcare, and environmental protection, are numerous and complex. With the rapid development of information technology, while many government departments have established e-government systems to digitally store policy information, the efficiency of policy information utilization remains low. For businesses, timely and accurate access to relevant policy information is crucial. However, due to the fragmented and complex nature of policy information, businesses often struggle to find the policies that meet their needs within this vast sea of ​​policies. For example, a technology-based enterprise may qualify for multiple policies supporting technological innovation. However, due to a lack of effective policy matching tools, the enterprise may not fully understand these policies, potentially missing out on opportunities to apply for funding, tax incentives, and other benefits. While some policy matching methods exist to address this issue, they typically only match fully eligible policies, potentially screening out nearly eligible enterprises and resulting in low matching coverage. Summary of the Invention

[0003] In view of this, the purpose of the present invention is to propose an intelligent policy matching recommendation method to solve the problem that currently only fully compliant policies can be matched, and there is a situation where "enterprises that are close to meeting the standards are mistakenly screened out", resulting in a low matching coverage rate.

[0004] Based on the above objectives, the present invention provides an intelligent policy matching recommendation method, comprising:

[0005] Extract policy conditions and policy incentives based on policy documents. The policy conditions include flexible conditions and rigid conditions. The flexible conditions are quantifiable indicators, while the rigid conditions are non-quantifiable indicators.

[0006] Obtain target enterprise information, compare the information in the target enterprise information corresponding to the flexible conditions and rigid conditions with the flexible conditions and rigid conditions in each policy document one by one, and obtain the flexible conditions that the target enterprise meets, the flexible conditions that it does not meet, the rigid conditions that it meets, and the rigid conditions that it does not meet in each policy document;

[0007] When the number of unqualified flexible conditions and unqualified rigid conditions in the policy document reaches zero, the corresponding policy document and the corresponding policy reward will be pushed to the user;

[0008] When the number of non-compliant rigid conditions in the policy document is zero and the number of non-compliant elastic conditions is ≥1, the ratio of the target enterprise information data to the quantifiable indicators in the corresponding non-compliant elastic conditions is calculated. When the ratio is greater than 80%, the corresponding policy document and the corresponding policy reward will be pushed to the user.

[0009] Optionally, when the ratio is greater than 80%, the actual benefit is calculated based on the input cost and the generated benefit of achieving the non-compliant elasticity condition, and the result is pushed to the user.

[0010] Optionally, when the number of unqualified flexible conditions in the policy document is zero and the number of unqualified rigid conditions is 1, the actual benefits are calculated based on the input costs and benefits generated by achieving the unqualified rigid conditions, and the results are pushed to the user.

[0011] Optionally, the method further includes:

[0012] Establish a database containing the matching results of multiple enterprises and various policy conditions, and mark the position of each enterprise in the industrial chain node;

[0013] When the target enterprise cannot meet the target policy because the elasticity ratio is lower than 80% or more than one rigidity condition is not met, the following steps are performed:

[0014] Industrial chain node matching: Compare the target enterprise's industrial chain nodes with those of enterprises in the database to screen out related enterprises at upstream, same-level or downstream nodes;

[0015] Data fusion verification: Fuse the selected related enterprise data with the target enterprise data to generate a simulated merged data set;

[0016] Policy compliance determination: If the simulated merged data set meets all the conditions of the target policy, the corresponding associated enterprise information in the database will be output.

[0017] Optionally, when there are multiple related companies that meet the requirements in the database, the output is based on the following rules:

[0018] Dynamic weight setting: allows users to set differentiated first-level weight values ​​W for different industry chain node positions: upstream / same level / downstream C ;

[0019] Priority calculation: Calculate the score of each associated enterprise based on the first-level weight value. The calculation formula is:

[0020] Score = 100 × W C ;

[0021] Sorting output: Output the list of related companies in descending order of comprehensive scores.

[0022] When there are multiple affiliated companies that meet the requirements, in order to further screen out more suitable affiliated companies, users can set first-level weight values ​​for affiliated companies at upstream, same-level or downstream nodes according to their own needs, and then calculate the score of each affiliated company based on the first-level weight value. The list of affiliated companies is output in descending order of the comprehensive score. In this way, affiliated companies that better meet user expectations can be further screened out.

[0023] Optionally, when the target enterprise cannot meet the target policy because the elasticity condition ratio is lower than 80%, the allocation of the first-level weight value is dynamically generated by an intelligent weight allocation algorithm, including the following steps:

[0024] Calculate the synergy effect weight component:

[0025] Based on the target enterprise A and the affiliated enterprise B i The topological distance d(L A ,L Bi ), the synergistic effect value S is calculated according to the exponential decay model:

[0026]

[0027] d(L A ,L Bi ) represents target companies A and B i In the topological distance of the industrial chain graph, k represents the attenuation factor;

[0028] Calculate the policy gain coefficient P gain :

[0029]

[0030] where R P is the policy reward amount, Cost (C) is the integration cost, and PPI is the policy compliance potential index;

[0031]

[0032] η represents the excess gain coefficient, For affiliated company B i Under the elastic condition e j The quantitative value of For target enterprise A under elastic condition e j The quantitative value of is the policy requirement threshold, is the policy sensitivity weight, E1 represents the set of elastic conditions that are not met, and E2 represents the number of elastic conditions that are not met;

[0033] First-level weight value W C =α·S+β·P gain, where α and β are secondary weight values, α+β=1.

[0034] Through the industrial chain synergy effect value S; policy compliance potential index PPI and policy gain coefficient P gain The three work together to drive the optimal decision. Compared with manual weight judgment, it solves the problem of weight subjectivity and significantly improves accuracy.

[0035] Optionally, the policy conditions and policy rewards extracted based on the policy document, wherein the flexible conditions are quantifiable indicators and the rigid conditions are non-quantifiable indicators, specifically include:

[0036] Policy text parsing: Using natural language processing models to identify structured fields in policy documents, including: policy names, extracted from the document title; policy rewards, extracting amounts or resource descriptions from the "reward content," "support standards," and "policy support" fields; and policy conditions, extracting conditional statements from the "application requirements," "applicable objects," and "policy support" fields.

[0037] Condition classification processing: Quantifiable indicator conditions including numerical ranges, percentages, or amounts are marked as flexible conditions; non-quantifiable indicator conditions including qualifications, regions, and ownership are marked as rigid conditions;

[0038] Structured storage: The extracted policy names, policy rewards, flexible condition lists, and rigid condition lists are stored in the policy knowledge graph database.

[0039] Optionally, the step of comparing the target enterprise information with the policy conditions one by one may specifically include:

[0040] Condition traversal matching: traverse each flexible condition and rigid condition in the policy file, and execute: extract the field data with the same name from the enterprise information database;

[0041] Differentiated verification logic: Flexible conditions: call the numerical comparator to verify whether the enterprise data meets the threshold; rigid conditions: call the text matcher to verify whether the enterprise attribute contains keywords.

[0042] Optionally, the target enterprise may meet the following flexible conditions, does not meet the flexible conditions, meets the following rigid conditions, and does not meet the following rigid conditions:

[0043] Eligible elastic conditions: Entries with enterprise data ≥ threshold value;

[0044] Unsatisfied elasticity conditions: Entries with enterprise data less than the threshold;

[0045] Rigid conditions that meet the following criteria: Entries with successful text matching;

[0046] Unmatched rigid conditions: Entries for which the text matching fails.

[0047] When making policy matching recommendations using this method, firstly, based on the policy documents, policy conditions and policy rewards are extracted, wherein the policy conditions include flexible conditions and rigid conditions, wherein the flexible conditions are quantifiable indicators and the rigid conditions are non-quantifiable indicators, and then the target enterprise information is obtained, which can be obtained through user input, and the information corresponding to the flexible conditions and rigid conditions in the target enterprise information is compared one by one with the flexible conditions and rigid conditions in each policy document, and the flexible conditions that the target enterprise meets, the elastic conditions that do not meet, the rigid conditions that meet, and the rigid conditions that do not meet in each policy document are obtained. When the number of flexible conditions that do not meet and the number of rigid conditions that do not meet in the policy document are zero, it means that the target enterprise's information is not satisfactory. If the information meets the policy conditions in the policy document, the corresponding policy document and the corresponding policy reward will be pushed to the user. When the number of non-compliant rigid conditions in the policy document is zero, and the non-compliant elastic conditions are ≥1, the ratio of the target enterprise information data to the quantifiable indicators in the corresponding non-compliant elastic conditions is calculated. When the ratio is greater than 80%, the corresponding policy document and the corresponding policy reward will be pushed to the user. For example, the R&D investment of the target enterprise is 4.2 million, and the elastic condition of the corresponding policy is R&D investment ≥ 5 million yuan. At this time, (420 / 500)×100%=84%, which means that the target enterprise is close to meeting the policy conditions and can meet the relevant requirements through appropriate improvements. At this time, the corresponding policy will be pushed to the user.

[0048] From the above, it can be seen that the present invention recommends policies to users by matching elastic conditions and rigid conditions. When both the elastic conditions and the rigid conditions are met, it means that the information of the target enterprise meets the policy conditions in the policy document. At this time, the corresponding policy document and the corresponding policy reward are pushed to the user, achieving accurate policy recommendations. At the same time, when the non-compliant elastic condition is ≥1, the corresponding policy recommendation can continue to be made according to the above requirements, thereby improving the matching coverage rate, solving the industry pain point of "enterprises that are close to meeting the standards being mistakenly screened out", and providing users with more comprehensive information to help users make decisions. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only for the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0050] Figure 1 This is a flowchart of a policy matching method according to an embodiment of the present invention. DETAILED DESCRIPTION

[0051] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below with reference to specific embodiments.

[0052] It should be noted that, unless otherwise defined, the technical or scientific terms used in the present invention should have the usual meanings understood by people with ordinary skills in the field to which the present invention belongs. The "first", "second" and similar words used in the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. "Include" or "comprise" and similar words mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the object being described changes, the relative positional relationship may also change accordingly.

[0053] like Figure 1 As shown, an intelligent policy matching recommendation method includes:

[0054] S1: Extract policy conditions and policy incentives based on policy documents. The policy conditions include flexible conditions and rigid conditions. The flexible conditions are quantifiable indicators, while the rigid conditions are non-quantifiable indicators.

[0055] Optional, specifically including (a) policy text parsing: identifying structured fields in policy documents through natural language processing models, including: policy name, extracted from the document title; policy rewards, extracting the amount or resource description from the "reward content", "support standards", and "policy support" fields; policy conditions, extracting conditional statements from the "application requirements", "applicable objects", and "policy support" fields;

[0056] (b) Condition classification: Quantifiable indicator conditions that include numerical ranges, percentages, or amounts are marked as flexible conditions (e.g., “R&D investment ≥ 5 million yuan”); non-quantifiable indicator conditions such as qualifications, region, and ownership are marked as rigid conditions (e.g., “registered in a high-tech zone”);

[0057] (c) Structured storage: The extracted policy names, policy rewards, flexible condition lists, and rigid condition lists are stored in the policy knowledge graph database;

[0058] S2: Obtain target enterprise information, compare the information in the target enterprise information corresponding to the flexible conditions and rigid conditions with the flexible conditions and rigid conditions in each policy document one by one, and obtain the flexible conditions that the target enterprise meets, the flexible conditions that it does not meet, the rigid conditions that it meets, and the rigid conditions that it does not meet in each policy document;

[0059] Optional, specifically including (a) condition traversal matching: traversing each flexible condition and rigid condition in the policy document, executing: extracting data of fields with the same name from the enterprise information database (e.g., the flexible condition "R&D investment" corresponds to "R&D expenses" in the enterprise financial table);

[0060] (b) Differentiated verification logic: Flexible conditions: calling a numerical comparator to verify whether the enterprise data meets the threshold (such as enterprise R&D expenses ≥ 5 million yuan); rigid conditions: calling a text matcher to verify whether the enterprise attributes contain keywords (such as whether the enterprise qualification list contains "high-tech enterprise").

[0061] Eligible elastic conditions: Entries with enterprise data ≥ threshold value;

[0062] Unsatisfied elasticity conditions: Entries with enterprise data less than the threshold;

[0063] Rigid conditions that meet the following criteria: Entries with successful text matching;

[0064] Unsatisfied rigid conditions: Entries that failed text matching;

[0065] S3: When the number of unqualified flexible conditions and unqualified rigid conditions in the policy document reaches zero, the corresponding policy document and the corresponding policy reward are pushed to the user;

[0066] S4: When the number of non-compliant rigid conditions in the policy document is zero and the number of non-compliant elastic conditions is ≥1, calculate the ratio of the target enterprise information data to the quantifiable indicators in the corresponding non-compliant elastic conditions. When the ratio is greater than 80%, push the corresponding policy document and the corresponding policy reward to the user.

[0067] When making policy matching recommendations using this method, firstly, based on the policy documents, policy conditions and policy rewards are extracted, wherein the policy conditions include flexible conditions and rigid conditions, wherein the flexible conditions are quantifiable indicators and the rigid conditions are non-quantifiable indicators, and then the target enterprise information is obtained, which can be obtained through user input, and the information corresponding to the flexible conditions and rigid conditions in the target enterprise information is compared one by one with the flexible conditions and rigid conditions in each policy document, and the flexible conditions that the target enterprise meets, the elastic conditions that do not meet, the rigid conditions that meet, and the rigid conditions that do not meet in each policy document are obtained. When the number of flexible conditions that do not meet and the number of rigid conditions that do not meet in the policy document are zero, it means that the target enterprise's information is not satisfactory. If the information meets the policy conditions in the policy document, the corresponding policy document and the corresponding policy reward will be pushed to the user. When the number of non-compliant rigid conditions in the policy document is zero, and the non-compliant elastic conditions are ≥1, the ratio of the target enterprise information data to the quantifiable indicators in the corresponding non-compliant elastic conditions is calculated. When the ratio is greater than 80%, the corresponding policy document and the corresponding policy reward will be pushed to the user. For example, the R&D investment of the target enterprise is 4.2 million, and the elastic condition of the corresponding policy is R&D investment ≥ 5 million yuan. At this time, (420 / 500)×100%=84%, which means that the target enterprise is close to meeting the policy conditions and can meet the relevant requirements through appropriate improvements. At this time, the corresponding policy will be pushed to the user.

[0068] From the above, it can be seen that the present invention recommends policies to users by matching elastic conditions and rigid conditions. When both the elastic conditions and the rigid conditions are met, it means that the information of the target enterprise meets the policy conditions in the policy document. At this time, the corresponding policy document and the corresponding policy reward are pushed to the user, achieving accurate policy recommendations. At the same time, when the non-compliant elastic condition is ≥1, the corresponding policy recommendation can continue to be made according to the above requirements, thereby improving the matching coverage rate, solving the industry pain point of "enterprises that are close to meeting the standards being mistakenly screened out", and providing users with more comprehensive information to help users make decisions.

[0069] In some embodiments, when the ratio is greater than 80%, the actual benefits are calculated based on the input costs and generated benefits of achieving the non-compliant elasticity conditions, and the results are pushed to the user. The input costs can be calculated by the difference between the elasticity conditions and the target enterprise information. The generated benefits can include policy incentives and projected benefits derived from relevant historical data. By pushing the actual benefits to the user, users can be provided with more intuitive data for decision-making.

[0070] In some embodiments, when the number of unqualified flexible conditions in a policy document is zero and the number of unqualified rigid conditions is one, the actual benefits are calculated based on the investment costs and benefits of achieving the unqualified rigid conditions, and the results are pushed to the user. For example, when the registered location is unqualified, the actual benefits are calculated based on the investment costs and benefits of migrating the registered location. By pushing the actual benefits to the user, users can be provided with more intuitive data for decision-making.

[0071] When a target enterprise cannot meet the target policy because its elasticity ratio is lower than 80% or more than one rigid condition is not met, it is not easy to meet the target policy based solely on the enterprise's own conditions. To improve the coverage of policy matching and provide users with a solution, in some embodiments, the method further includes:

[0072] Establish a database containing the matching results of multiple enterprises and various policy conditions, and mark the position of each enterprise in the industrial chain node;

[0073] When the target enterprise cannot meet the target policy because the elasticity ratio is lower than 80% or more than one rigidity condition is not met, the following steps are performed:

[0074] Industrial chain node matching: Compare the target enterprise's industrial chain nodes with those of enterprises in the database to screen out related enterprises at upstream, same-level or downstream nodes;

[0075] Data fusion verification: Fuse the selected related enterprise data with the target enterprise data to generate a simulated merged data set;

[0076] Policy compliance determination: If the simulated merged data set meets all the conditions of the target policy, the corresponding associated enterprise information in the database will be output.

[0077] In this embodiment, when the target enterprise cannot meet the target policy because the elasticity condition ratio is lower than 80% or more than one rigid condition is not met, the industrial chain node of the target enterprise is compared with the industrial chain nodes of the enterprises in the database, and the related enterprises located at the upstream, same-level or downstream nodes are screened out. Then, the screened related enterprise data is merged with the target enterprise data to generate a simulated merged data set. Finally, if the simulated merged data set meets all the conditions of the target policy, the corresponding related enterprise information in the database is output. Through the related enterprise information, the user can communicate with the related enterprise to determine whether there is a possibility of subsequent merger or acquisition, thereby solving the user's policy-related issues and other enterprise operation issues, thereby improving the coverage of policy matching and further providing users with solutions. The policy matching dimension is expanded from a single enterprise to the industrial chain, opening up a new path for resource integration.

[0078] In some embodiments, when there are multiple related companies that meet the requirements in the database, the following rules are used for output:

[0079] Dynamic weight setting: Allow users to set differentiated first-level weight values ​​W for different industry chain node positions (upstream / same level / downstream) C ;

[0080] Priority calculation: Calculate the score of each associated enterprise based on the first-level weight value. The calculation formula is:

[0081] Score = 100 × W C ;

[0082] Sorting output: Output the list of related companies in descending order of comprehensive scores.

[0083] When there are multiple affiliated companies that meet the requirements, in order to further screen out more suitable affiliated companies, users can set first-level weight values ​​for affiliated companies at upstream, same-level or downstream nodes according to their own needs, and then calculate the score of each affiliated company based on the first-level weight value. The list of affiliated companies is output in descending order of the comprehensive score. In this way, affiliated companies that better meet user expectations can be further screened out.

[0084] In some embodiments, when the target enterprise cannot meet the target policy because the elasticity ratio is lower than 80%, the first-level weight value W C The distribution of is dynamically generated through an intelligent weight distribution algorithm, which includes the following steps:

[0085] (a) Calculate the synergistic effect weight component:

[0086] Based on the target enterprise A and the affiliated enterprise B i The topological distance d(L A ,L Bi ), the synergistic effect value S is calculated according to the exponential decay model:

[0087]

[0088] d(L A ,L Bi ) represents target companies A and B i Topological distance in the industry chain graph (e.g. upstream → downstream = 3 hops, same level = 1 hop);

[0089] k represents the decay factor (the default is 0.5, and the weight decays by 40% for every additional hop of distance);

[0090] (b) Calculate the policy gain coefficient P gain :

[0091]

[0092] where R P is the policy reward amount, Cost (C) is the integration cost, and PPI is the policy potential index (PPI);

[0093]

[0094] η: excess gain coefficient (default 0.3, user adjustable); For affiliated company B i In the elastic condition e j The quantitative value of For target enterprise A under elastic condition e j The quantitative value of is the policy requirement threshold, is the policy sensitivity weight (default 1, configurable: core condition weight increases), E1 represents the set of elastic conditions that are not met, and E2 represents the number of elastic conditions that are not met;

[0095] First-level weight value W C =α·S+β·P gain , where α and β are secondary weight values, α+β=1.

[0096] For example: target enterprise A defect set E1 = {R&D investment};

[0097] Policy conditions: R&D requirements: ≥1 million;

[0098] Target Company A's current R&D: 600,000;

[0099] Candidate combinations:

[0100] B1: R&D 500,000 → 1.1 million after integration;

[0101] B2: R&D 700,000 → 1.3 million after integration;

[0102] S1=e -0.5x2 =0.37;

[0103] S2=e -0.5x1 =0.61;

[0104] At η = 0.3, In the case of PPI B1 =1.036, PPI B2 =1.108;

[0105] Let R P = 2 million yuan, B1 combination cost: 800,000 yuan; B2 combination cost: 1 million yuan; (In actual operation, relevant data can be obtained based on historical data of the same type);

[0106] P gainB1 =0.555; P gainB2 =0.507;

[0107] When α=0.4 and β=0.6,

[0108] W CB1 =0.4802, W CB2 =0.5468; 100×W CB2 >100×W CB1

[0109] Therefore, the affiliate B2 is more recommended for subsequent consideration.

[0110] As can be seen from the above, in this embodiment, the synergy cost is objectively quantified by the industrial chain synergy effect value S; the policy compliance potential index PPI dynamically evaluates the excess value and policy gain coefficient P gain To suppress resource mismatch, the three work together to drive optimal decision-making. Compared with manual weight judgment, it solves the problem of weight subjectivity and significantly improves accuracy.

[0111] Those skilled in the art should understand that the discussion of any of the above embodiments is merely illustrative and is not intended to imply that the scope of the present invention (including the claims) is limited to these examples. Within the scope of the present invention, the technical features in the above embodiments or different embodiments may be combined, the steps may be implemented in any order, and there are many other variations of the above aspects of the present invention, which are not provided in detail for the sake of simplicity.

[0112] The present invention is intended to cover all such substitutions, modifications and variations that fall within the broad scope of the appended claims. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. An intelligent policy matching recommendation method, characterized in that: include: Extract policy conditions and policy incentives based on policy documents. The policy conditions include flexible conditions and rigid conditions. The flexible conditions are quantifiable indicators, while the rigid conditions are non-quantifiable indicators. Obtain target enterprise information, compare the information in the target enterprise information corresponding to the flexible conditions and rigid conditions with the flexible conditions and rigid conditions in each policy document one by one, and obtain the flexible conditions that the target enterprise meets, the flexible conditions that it does not meet, the rigid conditions that it meets, and the rigid conditions that it does not meet in each policy document; When the number of unqualified flexible conditions and unqualified rigid conditions in the policy document reaches zero, the corresponding policy document and the corresponding policy reward will be pushed to the user; When the number of non-compliant rigid conditions in the policy document is zero and the number of non-compliant elastic conditions is ≥1, the ratio of the target enterprise information data to the quantifiable indicators in the corresponding non-compliant elastic conditions is calculated. When the ratio is greater than 80%, the corresponding policy document and the corresponding policy reward will be pushed to the user.

2. The intelligent policy matching recommendation method according to claim 1, characterized in that: When the ratio is greater than 80%, the actual benefit is calculated based on the input cost and the generated benefit of achieving the non-compliant elasticity conditions, and the result is pushed to the user.

3. The intelligent policy matching recommendation method according to claim 1, characterized in that: When the number of unqualified flexible conditions in the policy document is zero and the number of unqualified rigid conditions is 1, the actual benefits are calculated based on the input costs and benefits generated by achieving the unqualified rigid conditions, and the results are pushed to the user.

4. The intelligent policy matching recommendation method according to claim 1, characterized in that: The method further comprises: Establish a database containing the matching results of multiple enterprises and various policy conditions, and mark the position of each enterprise in the industrial chain node; When the target enterprise cannot meet the target policy because the elasticity ratio is lower than 80% or more than one rigidity condition is not met, the following steps are performed: Industrial chain node matching: Compare the target enterprise's industrial chain nodes with those of enterprises in the database to screen out related enterprises at upstream, same-level or downstream nodes; Data fusion verification: Fuse the selected related enterprise data with the target enterprise data to generate a simulated merged data set; Policy compliance determination: If the simulated merged data set meets all the conditions of the target policy, the corresponding associated enterprise information in the database will be output.

5. The intelligent policy matching recommendation method according to claim 4, characterized in that: When there are multiple related companies that meet the requirements in the database, they are output according to the following rules: Dynamic weight setting: allows users to set differentiated first-level weight values ​​W for different industry chain node positions: upstream / same level / downstream C ; Priority calculation: Calculate the score of each associated enterprise based on the first-level weight value. The calculation formula is: Score = 100 × W C ; Sorting output: Output the list of related companies in descending order of comprehensive scores.

6. The intelligent policy matching recommendation method according to claim 5, characterized in that: When the target enterprise cannot meet the target policy because the elasticity condition ratio is lower than 80%, the allocation of the first-level weight value is dynamically generated by an intelligent weight allocation algorithm, including the following steps: Calculate the synergy effect weight component: Based on the target enterprise A and the affiliated enterprise B i The topological distance d(L A ,L Bi ), the synergistic effect value S is calculated according to the exponential decay model: d(L A ,L Bi ) represents target companies A and B i In the topological distance of the industrial chain graph, k represents the attenuation factor; Calculate the policy gain coefficient P gain : where R P is the policy reward amount, Cost (C) is the integration cost, and PPI is the policy compliance potential index; η represents the excess gain coefficient, For affiliated company B i In the elastic condition e j The quantitative value of For target enterprise A under elastic condition e j The quantitative value of is the policy requirement threshold, is the policy sensitivity weight, E1 represents the set of elastic conditions that are not met, and E2 represents the number of elastic conditions that are not met; First-level weight value W C =α·S+β·P gain , where α and β are secondary weight values, α+β=1.

7. The intelligent policy matching recommendation method according to claim 1, characterized in that: The policy conditions and policy rewards extracted based on the policy documents, the flexible conditions are quantifiable indicators, and the rigid conditions are non-quantifiable indicators, specifically include: Policy text parsing: This uses a natural language processing model to identify structured fields in policy documents. For example, the policy name is extracted from the document title; for policy rewards, the amount or resource description is extracted from the "Reward Content," "Support Standards," and "Policy Support" fields; and for policy conditions, the conditional statements are extracted from the "Application Requirements," "Applicable Targets," and "Policy Support" fields. Condition classification processing: Quantifiable indicator conditions including numerical ranges, percentages, or amounts are marked as flexible conditions; non-quantifiable indicator conditions including qualifications, regions, and ownership are marked as rigid conditions; Structured storage: The extracted policy names, policy rewards, flexible condition lists, and rigid condition lists are stored in the policy knowledge graph database.

8. The intelligent policy matching recommendation method according to claim 1, characterized in that: The aforementioned comparison of target enterprise information with policy conditions specifically includes: Condition traversal matching: traverse each flexible condition and rigid condition in the policy file, and execute: extract the field data with the same name from the enterprise information database; Differentiated verification logic: Flexible conditions: call the numerical comparator to verify whether the enterprise data meets the threshold; rigid conditions: call the text matcher to verify whether the enterprise attribute contains keywords.

9. The intelligent policy matching recommendation method according to claim 1, characterized in that: The target enterprise meets the following flexible conditions, does not meet the flexible conditions, meets the rigid conditions, and does not meet the rigid conditions: Eligible elastic conditions: Entries with enterprise data ≥ threshold value; Unsatisfied elasticity conditions: Entries with enterprise data less than the threshold; Rigid conditions that meet the following criteria: Entries with successful text matching; Unmatched rigid conditions: Entries for which the text matching fails.

Images