Policy combination calculation method based on multi-dimensional interaction of policy objectives and tools

Abstract

The present invention discloses a policy combination calculation method based on the multi-dimensional interaction of policy objectives and tools, comprising the following steps: (1) collecting policy documents, collecting and cleaning them; (2) constructing a dictionary of policy objectives and tools in the target field; (3) extracting two types of elements, policy objectives and tools; (4) establishing a policy objective-tool co-occurrence association; (5) analyzing the policy objective-tool multi-dimensional interaction pattern and identifying four complex policy combination design mechanisms; (6) quantifying four policy combinations; (7) exploring the spatiotemporal evolution law of regional policy combinations in combination with the regional attributes of policy issuing agencies. The present invention; by systematically constructing a relationship network between policy objectives and policy tools, it is possible to design policy combinations more effectively, thereby reducing the uncertainty of the combination of policy objectives and policy tools.

Description

Technical Field

[0001] The present invention relates to the technical field of data representation and computer system, and in particular to a policy combination calculation method based on multi-dimensional interaction of policy objectives and tools. Background Art

[0002] With the continuous changes in the policy environment, the world has entered a new stage of complex policy formulation and implementation. In this stage, multiple policy tools are combined to achieve specific policy goals, and policy mix design becomes the core element to achieve policy effects. Policy mix design and adoption not only involves continuous dynamic adjustments in the process of setting and achieving policy goals, but also includes orderly evolution along the predetermined policy path under the guidance of a specific policy framework. Although the policy mix design mechanism may be complex, it usually follows a specific evolutionary pattern. Therefore, it is particularly important to effectively identify and trace the evolutionary mechanism of policy mix design and evaluate the implementation effect of the policy mix design mechanism.

[0003] Early policy portfolio identification mainly relied on qualitative analysis, and the Delphi method and interview method were widely used. These methods are not only time-consuming and labor-intensive, but the results are easily affected by subjective factors. With the introduction of quantitative analysis technology, policy research has gradually used bibliometrics to conduct in-depth analysis of policy documents, implementation cases, and co-occurrence and citation relationships of policy elements. At present, based on the analysis of the multidimensional interactive relationship between policy objectives and tools, the relationship between policy objectives and policy tools in policy texts can be effectively identified. However, past policy portfolio identification methods often rely only on single policy information, and fail to fully consider the dynamic interaction between policy objectives and policy tools and their impact on the effectiveness of policy implementation.

[0004] At present, the existing technology uses genetic algorithms to optimize the combination of subsidy policies to reduce manual calculation time or recommend suitable talent policies through deep learning, improve the accuracy and applicability of policy recommendations, and thus improve the overall decision-making efficiency. However, starting from a single policy tool combination, the dynamic evolution characteristics of policy combination elements are not considered, and there is a lack of exploration and evidence collection in terms of the multidimensional association between policy goals and tools. Some existing technologies also use a constructed vocabulary to retrieve and identify the initial policy goals and policy tools in policy documents, and manually identify and collect the remaining policy tools with the support of expert knowledge. However, this policy tool identification method requires a lot of expert knowledge in specific policy fields and cannot be expanded to policy element identification in other industrial fields. Or the existing technology cases are relatively single, and it is difficult to analyze the multidimensional interactive relationship between policy goals and policy tools in a fine-grained manner, and lack the measurement of the three-dimensional relationship between policy goals and tools and the mining of dynamic design mechanism characteristics from an interactive perspective. Summary of the invention

[0005] Purpose of the invention: The purpose of the present invention is to provide a policy combination calculation method based on the multidimensional interaction of policy objectives and tools, to establish a multidimensional association between policy objectives and policy tools from the perspective of policy objective-tool combination, to detect the multidimensional interaction mode of policy objectives and tools, to define and identify the four types of drift, conversion, stratification and fusion between policy combination design mechanisms, to construct and analyze the evolution of policy combination design mechanisms in a fine-grained manner, so as to solve the technical problems in the existing technical methods regarding the granularity, depth and accuracy of policy combination design mechanism detection, and to reduce the dependence on expert knowledge and human labor.

[0006] Technical solution: The policy combination calculation method based on the multi-dimensional interaction of policy objectives and tools described in the present invention comprises the following steps:

[0007] (1) Select the target policy research field, collect policy documents, and then clean them up;

[0008] (2) Constructing a dictionary of policy objectives and tools in the target area: Using manual coding combined with verb phrases and gerund combination structures to construct the initial policy objectives and tool vocabulary;

[0009] (3) Extracting two types of elements: policy objectives and tools: Using natural language processing (NLP) to automatically identify policy objectives and policy tools in text content based on the policy objectives or tool vocabulary;

[0010] (4) Establishing the policy goal-tool co-occurrence association: Based on the co-occurrence matrix of “policy goal-tool” in each policy text, the policy goal and policy tool are co-occurrence associated;

[0011] (5) Analyze the multidimensional interaction patterns of policy objectives and tools, and identify four complex policy mix design mechanisms: drift, transformation, stratification, and integration;

[0012] (6) Quantification of four policy combinations: Using information entropy and similarity indicators, quantify the degree of drift, transformation, stratification and integration of regional policy combination designs in each year;

[0013] (7) Based on the regional attributes of policy-issuing agencies, explore the temporal and spatial evolution of regional policy combinations.

[0014] Furthermore, step (1) includes the following steps:

[0015] (11) Determine the relevant policy database according to the target policy area and construct a search term list, search and collect policy documents based on the term list, and obtain the initial text data set;

[0016] (12) Eliminate irrelevant policy documents and use text matching algorithms to filter out documents that are irrelevant to the target policy area;

[0017] (13) Clean the issuing agencies of policy documents, build a dictionary of issuing agencies and corresponding regions, and match the regions to which policy documents belong based on the dictionary to generate an integrated dataset containing valid policy documents and their issuing years, agency information, and corresponding regions.

[0018] Furthermore, step (2) includes the following steps:

[0019] (21) Preliminary extraction of policy objectives: Based on the policy objective coding of existing studies, a batch of policy text data is randomly selected for manual coding, and an initial policy objective dictionary is constructed based on the expression characteristics of verb phrases;

[0020] (22) Preliminary extraction of policy tools: Based on the policy tool coding of existing studies, a batch of policy text data is randomly selected for manual coding, and an initial policy tool dictionary is constructed based on the expression characteristics of verb and noun combinations;

[0021] Furthermore, step (3) includes the following steps:

[0022] (31) Preliminary extraction of policy elements: Using Python program to perform automatic indexing process, using natural language processing technology (NLP) to identify policy objectives and tools embedded in the document based on the initial policy objectives or tool vocabulary;

[0023] (32) Improve the policy element dictionary: Identify and collect policy texts that do not conform to the predefined elements, classify and mark them, manually code the screened texts again to extract policy objectives and tools, and add the newly extracted elements to the policy element dictionary; repeat the automatic indexing process to identify updated policy elements; repeat step (31) until no new policy elements are generated, ensuring the improvement of the dictionary;

[0024] (33) Extract and improve policy elements: The extracted policy elements should be verified to ensure the uniqueness and validity of each element. If the same policy objective or tool appears multiple times, they should be merged into one entry and the redundant duplicates should be deleted.

[0025] Furthermore, step (4) includes the following steps:

[0026] (41) Constructing the co-occurrence relationship of policy combinations: Extracting the text data of policy objectives and policy tools from policy documents, and using text analysis technology to construct a co-occurrence matrix to record the co-occurrence frequency between each policy objective and its related policy tools. If there is co-occurrence, it is recorded as 1, otherwise it is 0.

[0027] (42) Identify the policy mix: Based on the “policy objective-policy tool” common matrix, identify policy objectives and policy tool pairs to form a preliminary policy mix;

[0028] (43) Delete combinations with less than 2 policy instruments. After deleting invalid combinations, reconstruct the remaining data.

[0029] Furthermore, step (5) includes the following steps:

[0030] (51) Identify the dynamic evolution path of policy mix design and compare the policy mix design mechanisms in year t and year t+1;

[0031] (52) Identify the policy mix drift mechanism and measure the degree of change in the policy mix in the direction of policy objectives; go through the policy mix of year t+1 one by one and conduct a detailed analysis of each policy mix. Check whether the policy tool elements in the policy mix of year t+1 are consistent with those of year t; if they are consistent, continue to check the policy objective elements; if the policy objective elements of year t+1 are inconsistent with those of year t and contain new policy objective elements, or the number of objective elements in year t+1 is less than that of year t, then it is identified as policy mix drift;

[0032] (53) Identify the policy mix switching mechanism and measure the degree of change in the policy mix in terms of policy tools; traverse the policy mix in year t+1 and compare it with the mix in year t; check whether the policy target elements in the policy mix in year t+1 are consistent with those in year t; if they are consistent, continue to check the policy tool elements; if the policy tool elements in year t+1 are not consistent with those in year t and contain policy tool elements that did not appear in year t, or the number of tool elements in year t+1 is less than that in year t, then identify it as a policy mix switching;

[0033] (54) Identify the policy mix stratification mechanism, measure its similarity to existing combinations and the degree to which it can be optimized within the existing framework; check whether the policy target elements in the policy mix of year t+1 are consistent with those of year t; if not, record the current target status; if the policy mix of year t+1 completely contains the target elements of year t and adds new policy target elements; at the same time, check the consistency of the current policy tool elements in the policy mix of year t; if not, and the tool mix of year t+1 completely contains the policy tool elements of year t and adds new tools, then identify it as a policy mix stratification;

[0034] (55) Identify the policy portfolio fusion mechanism and measure the innovation of the policy portfolio design that introduces new elements to replace old elements; traverse the policy portfolio of year t+1 and check the policy target elements one by one; confirm whether the current policy target element has appeared in the policy portfolio of year t. If not, record the policy target as a newly introduced target; similarly, check the policy tool elements in the policy portfolio of year t+1; check whether the current policy tool element appears in the policy portfolio of year t; if not, record the tool as a newly introduced tool; if the policy target and policy tool elements that did not appear in year t both appear in the policy portfolio of year t+1, and are inconsistent with the target and tool elements in the policy portfolio of year t, and the policy portfolio of year t+1 does not contain any target and tool elements in the policy portfolio of year t, then it is identified as a policy portfolio fusion.

[0035] Furthermore, step (6) includes the following steps:

[0036] (61) Constructing panel data of cities and years, including list data of policy combinations adopted by cities;

[0037] (62) The entropy index is used to analyze the situation in which cities introduce new policy objectives while keeping the original policy tools unchanged when implementing policy combinations, so as to assess the degree of drift and adaptability of policy combinations;

[0038] (63) Using the entropy index to analyze the introduction of new tools while keeping the policy objectives unchanged and to assess the degree of transformation and adaptability of the policy mix;

[0039] (64) Calculate the similarity between the policy objectives and policy tool elements in the city’s policy portfolio of the current year and the previous year through similarity, so as to measure the stratification of the city’s policy portfolio each year;

[0040] (65) The joint entropy index is used to measure the diversity of the joint changes in policy objectives and policy tools in a city’s policy mix compared with the previous year, in order to measure the degree of integration of a city’s policy mix each year.

[0041] Furthermore, step (7) includes the following steps:

[0042] (71) Based on the panel data of cities and years, a panel dataset containing data on multiple regions and their policy portfolio designs is constructed;

[0043] (72) Use time series analysis methods to evaluate the changing trends of policy mix design mechanisms in different regions over time. Analyze the starting time, frequency and changing characteristics of policy mix design mechanisms in different regions, and identify the time nodes and key events of policy mix design adoption;

[0044] (73) Calculate the frequency of occurrence of each policy mix design mechanism and its distribution characteristics in different regions and time periods to quantify the evolution of policy mix design;

[0045] (74) Summarize the spatiotemporal evolution of the policy mix design mechanism and extract the key findings of the regional policy mix design mechanism; put forward targeted policy recommendations based on the implementation effects of the policy mix design in each region.

[0046] An electronic device described in the present invention includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the computer program is loaded into the processor, it implements any one of the policy combination calculation methods based on multi-dimensional interaction of policy objectives and tools.

[0047] A storage medium described in the present invention stores a computer program, and when the computer program is executed by a processor, it implements any one of the policy combination calculation methods based on multi-dimensional interaction of policy objectives and tools.

[0048] Beneficial effects: Compared with the prior art, the present invention has the following significant advantages: by identifying the evolutionary path of the policy combination and four policy combination design mechanisms (drift, conversion, stratification and fusion), the present invention can monitor and evaluate the changes in the policy combination in real time. This dynamic identification mechanism makes the policy design process more flexible and can adapt to the ever-changing social environment and policy needs; the diversity of data improves the comprehensiveness and accuracy of policy combination analysis by integrating natural language processing and manual coding; by systematically constructing the relationship network between policy goals and policy tools, the policy combination can be designed more effectively, thereby reducing the uncertainty of the combination of policy goals and policy tools. This process not only improves the adaptability of the policy combination, but also enhances its implementation effect. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 It is a flow chart of the present invention. DETAILED DESCRIPTION

[0050] The technical solution of the present invention is further described below in conjunction with the accompanying drawings.

[0051] like Figure 1 As shown, the embodiment of the present invention provides a policy combination calculation method based on policy target-tool multi-dimensional interaction, including the following steps:

[0052] S1 selects the target policy research field, collects the released policy documents, and collects and cleans large-scale policy data texts; including the following steps:

[0053] S11 selects the target policy area, identifies the relevant policy text database and constructs a search word list, searches and collects policy documents based on the word list to obtain the initial text data set;

[0054] S12 removes irrelevant policy texts and uses a text matching algorithm to filter out texts that are irrelevant to the target policy area to ensure that the policy text data is valid;

[0055] S13 cleans the issuing agencies of policy texts, builds a dictionary of issuing agencies and corresponding regions, and matches the regions to which policy texts belong based on the dictionary to generate an integrated dataset containing valid policy texts and their agency information, issuing years, and corresponding regions.

[0056] S2 builds a dictionary of policy objectives and policy tools in the target area: uses manual coding to build an initial policy objective and tool vocabulary, and combines natural language processing technology to improve the initial policy element vocabulary; the details are as follows:

[0057] S21 Preliminary extraction of policy objectives. Combined with the existing research on policy objective coding, a batch of policy text data was randomly selected for manual coding, and the initial policy objective dictionary was constructed based on the expression characteristics of verb phrases, which served as the basic policy element relationship data for the subsequent period;

[0058] S22 Preliminary extraction of policy tools. Combining the existing research on policy tool coding, a batch of policy text data was randomly selected for manual coding, and an initial policy tool dictionary was constructed based on the expression characteristics of verb and noun combinations, which served as the subsequent basic policy element relationship data.

[0059] S3 extracts two types of elements: policy objectives and tools: Natural language processing (NLP) technology is used to identify policy objectives and tools in the text; the following steps are included:

[0060] S31 Preliminary extraction of policy elements. Use Python programs to perform the automatic indexing process and write programs to use natural language processing (NLP) techniques and identify policy objectives and tools embedded in documents based on the initial policy objectives or tool vocabulary;

[0061] S32 Policy element dictionary improvement. Identify and collect policy texts that do not conform to predefined elements, classify and mark them. Manually encode the screened texts again to extract policy objectives and policy tools, and add the newly extracted elements to the policy element dictionary to ensure the comprehensiveness and accuracy of the dictionary. Repeat the automatic indexing process to identify updated policy elements. Repeat this process until no new policy elements are generated to ensure the improvement of the dictionary.

[0062] S33 Policy elements extraction and improvement. The extracted policy elements are verified to ensure the uniqueness and validity of each element. If the same policy objective or tool appears multiple times, it should be merged into one entry and the redundant duplicates should be deleted to reduce redundancy.

[0063] S4 establishes the policy goal-tool co-occurrence association: according to the co-occurrence matrix of "policy goal-policy tool" in the policy document, the policy goal is co-occurred with the policy tool; including the following steps:

[0064] The details are as follows:

[0065] S41 Construction of co-occurrence relationship of policy combination. Extract the text data of policy objectives and policy tools from the policy documents, and use text analysis technology to construct a co-occurrence matrix to record the co-occurrence frequency between each policy objective and its related policy tools. If there is co-occurrence, it is recorded as 1, otherwise it is 0.

[0066] S42 Policy combination identification. According to the co-occurrence matrix of “policy objectives-policy tools”, identify the policy objectives and policy tools to form a preliminary policy combination.

[0067] S43 performs a cleaning operation to delete combinations with less than 2 policy instruments to ensure the effectiveness and practicality of the policy combination. After deleting invalid combinations, the remaining data is reconstructed to ensure the integrity and consistency of the data set.

[0068] S5 identifies four complex policy mix design mechanisms based on the multi-dimensional interaction patterns between policy goals and tools in the network: drift, transformation, stratification, and fusion; the details are as follows:

[0069] S51 Identify the dynamic evolution path of policy mix design and compare t Year and t +1 year policy mix design mechanism;

[0070] S52 identifies the policy mix drift mechanism and measures the degree of change in the policy mix in the direction of policy objectives. t +1 year policy combinations are reviewed one by one, and each policy combination is analyzed in detail. t +1 Whether the policy tool elements in the policy mix are consistent with t If the policy target elements in year t+1 are the same as those in year t+2, then the policy target elements in year t+3 will be the same as those in year t+4. t 2 years and contain new policy objective elements, or t +1 Year target element number is less than t If the number of years is too large, it is identified as policy mix drift;

[0071] S53 identifies the policy mix conversion mechanism and measures the degree of change in the policy mix in terms of policy tool direction.t +1 year policy combinations are traversed, and t Year combinations are compared. Check t +1 Whether the policy target elements in the policy portfolio are consistent with t If the policy is consistent with the previous year, continue to check the policy tool elements; if t +1 year policy tool elements are not related t The year remains the same and contains t policy instrument elements that did not appear in 2017, or t +1 year The number of instrument elements is less than t If the number of years is too high, it is identified as a policy mix shift;

[0072] S54 Identify the policy mix layering mechanism and measure its similarity to existing mixes and how it can be optimized within the existing framework. t +1 Whether the policy target elements in the policy portfolio are consistent with t If the policy combination for year t+1 completely contains t elements of the target for the year and add new target elements; at the same time, check whether the current policy instrument elements are t policy portfolio; if not, and t +1 year tool kit fully included t If the policy tool elements of the year are replaced by new tools, it is identified as a policy portfolio layer;

[0073] S55 identifies the policy portfolio fusion mechanism and measures the innovation of the policy portfolio design by introducing new elements to replace old elements. Traverse the policy portfolio of year t+1 and check the policy target elements one by one. Confirm whether the current policy target element has appeared in the policy portfolio of year t. If not, record the policy target as a newly introduced target. Similarly, check the policy tool elements in the policy portfolio of year t+1. Check whether the current policy tool element has appeared in the policy portfolio of year t. t If it does not appear in the policy mix, the instrument is recorded as a newly introduced instrument. t Policy objectives and instrument elements that have not appeared in 2017 are t +1 year policy portfolio, and t The policy objectives and policy instrument elements in the policy mix remain inconsistent, and t +1 year policy portfolio does not contain t If any policy objective and policy tool element in the policy mix of the year is not included, it is identified as a policy mix fusion, as shown in Table 1.

[0074] Table 1 Types of policy mix design mechanisms

[0075]

[0076] S6 quantifies four policy portfolio designs: using information entropy and similarity indicators, quantifies the drift, transformation, stratification and integration of regional policy portfolio designs in different years; including the following steps:

[0077] S61 constructs city and year panel data and extracts the policy combination data of each city in a specific year from the relevant policy database;

[0078] S62 uses the entropy index to analyze the situation of cities introducing new policy goals while keeping the original policy tools unchanged when implementing policy combinations, in order to evaluate the drift degree and adaptability of policy combinations. The calculation formula for the drift degree is as follows:

[0079] ;

[0080] Indicates the city i No. t The increase in the annual policy entropy index reflects the diverse changes in policy objectives. It is a policy goal g In the t The frequency of occurrence in the year is the number of policy targets in the policy dataset.

[0081] S63 uses the entropy index to analyze the situation of introducing new tools and evaluate the conversion degree and adaptability of policy combinations when the policy objectives remain unchanged. The conversion degree is calculated as follows:

[0082] ;

[0083] Indicates the city i The increase in the entropy index of policy tools in year t reflects the changes in the diversity of policy tools. It is a policy tool e In the t The frequency of occurrence in a year is the number of policy instruments in the policy dataset.

[0084] S64 calculates the similarity between the policy objectives and policy tool elements in the city's policy portfolio of the current year and the previous year through similarity, so as to measure the stratification of the city's policy portfolio each year. The calculation formula of the stratification is as follows:

[0085] ;

[0086] in, , They are t -1 year, t Vector representation of annual policy objectives and policy tools.

[0087] S65 measures the diversity of the changes in policy objectives and policy tools in a city's policy portfolio compared to the previous year by using the joint entropy index to measure the integration of the city's policy portfolio each year. The calculation formula for the integration is as follows:

[0088] ;

[0089] in, It is t Policy goals for the year g and policy tools e The joint frequency of , n is the number of policy objectives in the policy dataset, and m is the number of policy instruments in the policy dataset.

[0090] S7 analyzes the temporal and spatial evolution of regional policy combinations based on the regions where policy issuing agencies are located. This includes the following steps:

[0091] S71 constructs a panel data set containing multiple regions and their policy portfolio design data based on panel data of cities and years;

[0092] S72 uses time series analysis methods to evaluate the changing trends of policy mix design mechanisms in different regions over time. Analyze the starting time, frequency and changing characteristics of policy mix design mechanisms in different regions, and identify the time nodes and key events of policy mix design adoption;

[0093] S73 calculates the frequency of occurrence of each policy combination design mechanism and its distribution characteristics in different regions and time periods to quantify the evolution of policy combination design.

[0094] S74 summarizes the spatiotemporal evolution of the policy mix design mechanism and extracts the key findings of the regional policy mix design mechanism. According to the implementation effects of the policy mix design in each region, targeted policy recommendations are put forward. These recommendations aim to optimize the regional policy mix design, improve the effectiveness of policy implementation, and promote the sustainable development of regional economy and society.

[0095] An embodiment of the present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the computer program is loaded into the processor, the policy combination calculation method based on multi-dimensional interaction of policy objectives and tools is implemented.

[0096] An embodiment of the present invention further provides a storage medium storing a computer program, wherein the computer program, when executed by a processor, implements any one of the policy combination calculation methods based on multi-dimensional interaction of policy objectives and tools.

Claims

1. A policy combination calculation method based on multi-dimensional interaction of policy objectives and tools, characterized in that: The following steps are involved: (1) Select the target policy research field, collect policy documents, and then clean them up; (2) Constructing a dictionary of policy objectives and tools in the target area: Using manual coding combined with verb phrases and gerund combination structures to construct the initial policy objectives and tool vocabulary; (3) Extracting two types of elements: policy objectives and tools: Using natural language processing (NLP) to automatically identify policy objectives and policy tools in text content based on the policy objectives or tool vocabulary; (4) Establishing the co-occurrence association between policy objectives and tools: Based on the co-occurrence matrix of "policy objectives and tools" in each policy text, the policy objectives and policy tools are co-occurred; (5) Analyze the multi-dimensional interaction model between policy objectives and tools, and identify four complex policy mix design mechanisms, namely drift, transformation, stratification, and integration; the details are as follows: S51 Identify the dynamic evolution path of policy mix design and compare t Year and t +1 year policy mix design mechanism; S52 identifies the policy mix drift mechanism and measures the degree of change of the policy mix in the direction of policy objectives; t +1 year policy combinations are reviewed one by one, and each policy combination is analyzed in detail; check t +1 Whether the policy tool elements in the policy mix are consistent with t If consistent, continue to check policy objective elements; If the policy target elements in year t+1 are t 2 years and contain new policy objective elements, or t +1 Year target element number is less than t If the number of years is too large, it is identified as policy mix drift; S53 identifies the policy mix conversion mechanism and measures the degree of change in the policy mix in terms of policy tool direction; t +1 year policy combinations are traversed, and t Compare year combinations; check t +1 Whether the policy target elements in the policy portfolio are consistent with t If the policy is consistent with the previous year, continue to check the policy tool elements; if t +1 year policy tool elements are not related t The year remains the same and contains t policy instrument elements that did not appear in 2017, or t +1 year The number of instrument elements is less than t If the number of years is too high, it is identified as a policy mix shift; S54 Identify the policy mix layering mechanism, measure its similarity to existing combinations and how it can be optimized within the existing framework; check t +1 Whether the policy target elements in the policy portfolio are consistent with t If the policy combination in year t+1 completely contains t elements of the target for the year and add new target elements; at the same time, check whether the current policy instrument elements are t policy portfolio; if not, and t +1 year tool kit fully included t If the policy tool elements of the year are replaced by new tools, it is identified as a policy portfolio layer; S55 identifies the policy portfolio fusion mechanism and measures the innovation of the policy portfolio design by introducing new elements to replace old elements; traverses the policy portfolio of year t+1 and checks the policy target elements one by one; confirms whether the current policy target elements have appeared in the policy portfolio of year t; If it does not appear, record the policy objective as a newly introduced objective; Similarly, the policy tool elements in the policy portfolio in year t+1 are checked; check whether the current policy tool elements are in t It has appeared in the policy mix of the year; If it does not appear, the tool is recorded as a newly introduced tool; If in t Policy objectives and instrument elements that have not appeared in 2017 are t +1 year policy portfolio, and t The policy objectives and policy instrument elements in the policy mix remain inconsistent, and t +1 year policy portfolio does not contain t Any policy objective and policy instrument element in the policy mix of the year is identified as policy mix integration; (6) Quantification of the four policy combinations: Using information entropy and similarity indicators, quantify the degree of drift, transformation, stratification and integration of regional policy combination designs in each year; including the following steps: S61 constructs city and year panel data and extracts the policy combination data of each city in a specific year from the relevant policy database; S62 uses the entropy index to analyze the situation of cities introducing new policy goals while keeping the original policy tools unchanged when implementing policy combinations, so as to evaluate the drift degree and adaptability of policy combinations; the calculation formula of drift degree is as follows: ； in, Indicates the city i No. t The increase in the annual policy entropy index reflects the diversity of policy goals; It is a policy goal g In the t The frequency of occurrence in the year is the number of policy targets in the policy dataset; S63 uses the entropy index to analyze the situation of introducing new tools and evaluate the degree of conversion and adaptability of policy combinations when the policy objectives remain unchanged; the calculation formula for the conversion degree is as follows: ； in, Indicates the city i The increase in the entropy index of policy tools in year t reflects the changes in the diversity of policy tools; It is a policy tool e In the t The frequency of occurrence in the year is the number of policy instruments in the policy dataset; S64 calculates the similarity between the policy objectives and policy tool elements in the city's policy portfolio of the current year and the previous year through similarity, so as to measure the stratification degree of the city's policy portfolio each year; the calculation formula of stratification degree is as follows: ； in, , They are t -1 year, t Vector representation of annual policy objectives and policy tools; S65 measures the diversity of the joint changes in policy objectives and policy tools in a city’s policy portfolio compared to the previous year by using the joint entropy index to measure the integration of the city’s policy portfolio each year. The calculation formula for the integration is as follows: ； in, It is t Policy goals for the year g and policy tools e The joint frequency of , n is the number of policy objectives in the policy dataset, and m is the number of policy instruments in the policy dataset; (7) Based on the regional attributes of policy-issuing agencies, explore the temporal and spatial evolution of regional policy combinations.

2. The policy combination calculation method based on policy target-tool multi-dimensional interaction according to claim 1 is characterized in that: Step (1) includes the following steps: (11) Determine the relevant policy database according to the target policy area and construct a search term list, search and collect policy documents based on the term list, and obtain the initial text data set; (12) Eliminate irrelevant policy documents and use text matching algorithms to filter out documents that are irrelevant to the target policy area; (13) Clean the issuing agencies of policy documents, build a dictionary of issuing agencies and corresponding regions, and match the regions to which policy documents belong based on the dictionary to generate an integrated dataset containing valid policy documents and their issuing years, agency information, and corresponding regions.

3. The policy combination calculation method based on policy target-tool multi-dimensional interaction according to claim 1 is characterized in that: Step (2) includes the following steps: (21) Preliminary extraction of policy objectives: Based on the policy objective coding of existing studies, a batch of policy text data is randomly selected for manual coding, and an initial policy objective dictionary is constructed based on the expression characteristics of verb phrases; (22) Preliminary extraction of policy tools: Based on the policy tool coding of existing studies, a batch of policy text data is randomly selected for manual coding, and an initial policy tool dictionary is constructed based on the expression characteristics of verb and noun combinations.

4. The policy combination calculation method based on policy target-tool multi-dimensional interaction according to claim 1 is characterized in that: Step (3) includes the following steps: (31) Preliminary extraction of policy elements: Using Python program to perform automatic indexing process, using natural language processing technology (NLP) to identify policy objectives and tools embedded in the document based on the initial policy objectives or tool vocabulary; (32) Improve the policy element dictionary: Identify and collect policy texts that do not conform to the predefined elements, classify and mark them, manually code the screened texts again to extract policy objectives and tools, and add the newly extracted elements to the policy element dictionary; repeat the automatic indexing process to identify updated policy elements; repeat step (31) until no new policy elements are generated, ensuring the improvement of the dictionary; (33) Extract and improve policy elements: The extracted policy elements should be verified to ensure the uniqueness and validity of each element. If the same policy objective or tool appears multiple times, they should be merged into one entry and the redundant duplicates should be deleted.

5. The policy combination calculation method based on policy target-tool multi-dimensional interaction according to claim 1 is characterized in that: Step (4) includes the following steps: (41) Constructing the co-occurrence relationship of policy combinations: Extracting the text data of policy objectives and policy tools from policy documents, using text analysis technology to construct a co-occurrence matrix, recording the co-occurrence frequency between each policy objective and its related policy tools, where if there is co-occurrence, it is recorded as 1, otherwise it is recorded as 0; (42) Identify the policy mix: Based on the current matrix of "policy objectives-policy tools", identify policy objectives and policy tools to form a preliminary policy mix; (43) Delete combinations with less than 2 policy instruments. After deleting invalid combinations, reconstruct the remaining data.

6. The policy combination calculation method based on policy target-tool multi-dimensional interaction according to claim 1 is characterized in that: Step (7) includes the following steps: (71) Based on the panel data of cities and years, a panel dataset containing data on multiple regions and their policy portfolio designs is constructed; (72) Use time series analysis methods to evaluate the changing trends of policy mix design mechanisms in different regions over time; analyze the starting time, frequency and changing characteristics of policy mix design mechanisms in different regions, and identify the time nodes and key events for the adoption of policy mix design; (73) Calculate the frequency of occurrence of each policy mix design mechanism and its distribution characteristics in different regions and time periods to quantify the evolution of policy mix design; (74) Summarize the spatiotemporal evolution of the policy mix design mechanism and extract the key findings of the regional policy mix design mechanism; put forward targeted policy recommendations based on the implementation effects of the policy mix design in each region.

7. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the computer program is loaded into a processor, the policy combination calculation method based on multi-dimensional interaction of policy objectives and tools according to any one of claims 1 to 6 is implemented.

8. A storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the policy combination calculation method based on multi-dimensional interaction of policy objectives and tools according to any one of claims 1 to 6 is implemented.

Images