An enterprise intelligent matching recommendation method and system for park investment

Abstract

The present application relates to the technical field of big data analysis, in particular to an enterprise intelligent matching recommendation method and system for park investment, comprising the following steps: selecting a multi-dimensional correlation characteristic combination algorithm of enterprises; designing a multi-weight matching degree calculation function; constructing a dynamic weight adjustment mechanism; executing an enterprise intelligent matching recommendation algorithm process; automatically matching and recommending park investment enterprises according to the designed process; the beneficial effect is that: by comprehensively considering the four core dimensions of enterprise development demand, park resource supply, industrial collaborative adaptation degree and regional economic contribution potential, a multi-weight matching degree calculation function is constructed, replacing the traditional single matching logic which only relies on industry type or policy preferential treatment. The accuracy and efficiency of park investment are improved, data support is provided for park industrial layout optimization, and decision-making assistance is provided for accurate connection between investment personnel and enterprises, thereby reducing the cost of investment, improving the industrial agglomeration effect and sustainable development ability of the park.

Description

Technical Field

[0001] This invention relates to the field of big data analytics, specifically to a method and system for intelligent matching and recommendation of enterprises for investment promotion in industrial parks. Background Technology

[0002] With the rapid development of my country's economy, the construction of industrial parks across the country has accelerated, with the number of parks increasing from hundreds in the early days to tens of thousands today, covering a wide range of industries. Simultaneously, the number of enterprises in the market has also experienced explosive growth, with the number of small and medium-sized enterprises (SMEs) nationwide alone exceeding tens of millions. Against this backdrop, industrial park investment promotion faces unprecedented challenges. Traditional investment matching models are no longer adequate to meet the needs of the new era. Specific problems include: (1) Low matching efficiency: Currently, when screening companies, the investment promotion teams of industrial parks often have to deal with a massive amount of company information (including company registration information, operating data, industry sectors, development plans, etc.) and park information (such as the park's industrial positioning, land resources, supporting facilities, policy system, etc.). Taking a certain provincial economic development zone as an example, when its investment promotion team conducts a key industry investment promotion campaign, it needs to manually review the information of more than 5,000 companies and compare the matching conditions between the companies and the park one by one. The entire screening process takes as long as 2-3 weeks, which not only consumes a lot of manpower and time costs, but also leads to slow progress in investment promotion work and misses many opportunities to cooperate with high-quality companies.

[0003] (2) Insufficient accuracy: Traditional matching of enterprises with industrial parks mainly relies on the experience and judgment of investment promotion personnel, which is easily affected by subjective factors. For example, some investment promotion personnel may not have a deep understanding of a certain industry sector and therefore overlook the fit between the core advantages of the enterprise and the industrial positioning of the park; or they may focus too much on the size and revenue of the enterprise during the screening process, while ignoring the enterprise's development potential and synergy with the existing industries in the park. Such subjective judgment bias often leads to unsatisfactory matching results. Some enterprises that settle in the park cannot make full use of the park's resources and policy advantages and develop slowly; while some high-quality enterprises that are highly compatible with the park are excluded, resulting in a waste of park resources and loss of investment opportunities.

[0004] (3) Lack of dynamic adaptability: The development status of enterprises and the industrial policies of the park are constantly changing. On the one hand, the business direction, market demand, and core technologies of enterprises may be adjusted with changes in the market environment. For example, a manufacturing enterprise may transform from traditional manufacturing to intelligent manufacturing, and its demand for park resources and policies will also change accordingly. On the other hand, in order to adapt to the development trend of the industry, the park may adjust its industrial positioning and optimize its policy system. For example, a park may add new support policies for the new energy industry to attract relevant enterprises to settle in. However, the traditional matching method cannot capture these changes in real time. It usually performs a one-time matching according to fixed standards and procedures, which cannot update the information of enterprises and parks in a timely manner, resulting in a lag in the matching results and failing to meet the dynamic investment promotion needs.

[0005] (4) Neglecting supply chain synergy: In the traditional investment promotion matching process, investment promotion personnel often only focus on the matching of individual attributes between enterprises and parks, such as whether the enterprise's industrial field is consistent with the park's industrial positioning and whether the enterprise's resource needs can be met by the park, while ignoring the enterprise's position in the supply chain and its synergistic relationship with the park's existing industries. This single matching perspective is not conducive to the formation of industrial agglomeration effect. For example, a certain park has already attracted many automobile manufacturing enterprises, but in the investment promotion process, it has not focused on attracting automobile parts manufacturers, which has led to automobile manufacturing enterprises having to purchase parts from other places, increasing production costs and reducing industrial competitiveness; at the same time, it is also impossible to form a complete supply chain, affecting the overall development quality and scale of the park's industries.

[0006] In existing technologies, although some industrial parks have introduced investment promotion systems, these systems are relatively simple in function, mostly using keyword matching or single-dimensional indicators (such as enterprise revenue, registered capital, etc.) for screening. For example, one investment promotion system only matches by "industry keywords," listing any enterprise whose name or business scope contains keywords related to the park's industry positioning as a potential partner. This approach fails to comprehensively consider multiple attributes such as the enterprise's core capabilities, development needs, and supply chain relationships, and also lacks analysis of industrial chain synergy. As a result, the accuracy and practicality of the matching results are low, failing to meet the park's needs for precise and high-quality investment promotion. Summary of the Invention

[0007] The purpose of this invention is to provide a method and system for intelligent matching and recommendation of enterprises for investment promotion in industrial parks, so as to solve the problems mentioned in the background art.

[0008] To achieve the above objectives, the present invention provides the following technical solution: a method for intelligent matching and recommendation of enterprises for investment promotion in industrial parks, comprising the following steps: 101: Select an algorithm that combines multi-dimensional related features of an enterprise; 102: Design a multi-weighted matching degree calculation function; 103: Construct a dynamic weight adjustment mechanism; 104: Execute the enterprise intelligent matching and recommendation algorithm process; 105: The system automatically matches and recommends companies for investment in the park according to the designed process, including the following seven sub-steps: I. Data entry and preprocessing: The system connects to the park's basic database and the enterprise investment database to complete data cleaning, standardization and feature extraction; II. Enterprise Feature Expansion: Generate an extended enterprise feature set E' through a multi-dimensional association feature combination algorithm; III. Park Weight Setting: Based on the park's investment attraction target tags, determine the weight coefficients of the matching degree calculation function; IV. Batch Matching Calculation: Perform batch matching between the park and the enterprise database, and calculate the overall matching degree for each enterprise; V. Results Filtering and Sorting: Filter out companies with low matching scores and generate a recommendation list by matching score; VI. Recommendation Results Output: Push recommendation lists and visual reports to investment promotion personnel in the park, allowing them to view company details and matching reasons; VII. Dynamic Data Updates: Data and recommendation results are updated periodically or triggered on a regular basis.

[0009] Preferably, the multi-dimensional association feature combination algorithm for enterprises proposed in step 101 combines adjacent features under the four dimensions of "industry attributes - development needs - resource dependence - contribution potential" in the original enterprise data in pairs to form new association features, thus solving the problems of insufficient descriptive power and low matching accuracy of traditional single features.

[0010] Preferably, the multi-weighted matching degree calculation function designed in step 102 is a quantitative calculation function for park-enterprise matching, which is constructed by integrating four core indicators: park resource supply matching degree, industrial synergy adaptation degree, economic contribution potential, and regional policy fit. The formula is as follows: M(P,E)=α×S(P,E)+β×C(P,E)+γ×V(P,E)+δ×P(P,E) In the formula: M(P,E): the comprehensive matching degree between park P and enterprise E, with a value range of [0,1]. The higher the value, the stronger the fit; α, β, γ, δ: weight coefficients for each dimension, satisfying α+β+γ+δ=1, which can be dynamically adjusted according to the park's investment attraction priority. The default values ​​are α=0.35, β=0.3, γ=0.2, and δ=0.15; S(P,E): the matching degree of park resource supply, calculated as (the number of enterprise demand characteristics that match the park's resources / the total number of enterprise demand characteristics); C(P,E): the industrial synergy fit degree, calculated as (the number of existing industrial characteristics in the park that match the park's resources / the total number of enterprise demand characteristics); V(P,E): Economic contribution potential, calculated as min((expected annual tax revenue of enterprises / average annual tax revenue of newly added enterprises in the park), 1) × 0.5 + min((expected investment scale of enterprises / total annual investment target of the park), 1) × 0.5; P(P,E): Regional policy fit, calculated as (number of park policies that enterprises can enjoy / total number of special investment policies that the park has issued).

[0011] Preferably, the dynamic weight adjustment mechanism constructed in step 103 is a mechanism that automatically adjusts the values ​​of α, β, γ, and δ in the multi-weight matching degree calculation function based on the park's investment attraction target labels. The execution steps are as follows: I. Establish a target label system for investment attraction in the park, including four core labels: "industry clustering", "tax contribution", "innovation-driven", and "employment-creating". Each park can choose one main label and one to two auxiliary labels. II. Establish a label-weight mapping rule base. Specific rules include: For the main label "Industry Agglomeration": β (industry synergy matching weight) is increased to 0.4, α (resource supply matching degree) is adjusted to 0.3, γ (economic contribution potential) is 0.2, and δ (regional policy fit) is 0.1; For the main label "Innovation-Driven": α (resource supply matching degree, focusing on matching R&D land and scientific research platforms) is increased to 0.4, β (industry synergy matching degree) is 0.3, γ (economic contribution potential) is 0.15, and δ (regional policy fit, focusing on matching R&D subsidies) is 0.15. III. If the park does not actively set labels, the system will use "balanced" weights by default (α=0.35, β=0.3, γ=0.2, δ=0.15), while also supporting investment promotion personnel to manually fine-tune the weight coefficients.

[0012] Preferably, the enterprise intelligent matching and recommendation algorithm process executed in step 104 includes the following three sub-steps: I. Data Preprocessing and Feature Extraction: The preprocessing objects include two categories: "Park Basic Data" and "Enterprise Investment Promotion Data": Park Basic Data: Extracting four core information categories: "Industry Positioning (leading industries, supporting industries)," "Resource Supply (land use indicators, factory area, supporting facilities)," "Policy System (tax reduction and exemption, R&D subsidies, talent policies)," and "Existing Enterprises (enterprise list, industry type, output value scale)," forming the park feature set P; Enterprise Investment Promotion Data: Extracting four core information categories: "Basic Attributes (industry, registered capital, years of establishment)," "Development Needs (land use needs, financing needs, supporting needs)," "Operating Status (annual output value, tax contribution, R&D investment)," and "Development Planning (investment plan, capacity target, regional layout)," forming the enterprise original feature set E; Data Cleaning: Removing invalid data, unifying data format, and standardizing feature labels; II. Combination of multi-dimensional related features of enterprises: The original feature set E of enterprises after cleaning is divided into four dimensional subsets according to "industry attributes - development needs - resource dependence - contribution potential"; adjacent features within the same dimension and related features between different dimensions are combined in pairs to generate related features; the original features and related features are merged to form the expanded feature set E' of enterprises; III. Park-Enterprise Matching Degree Calculation and Result Processing: Park Tag Setting: Park managers select target tags for investment attraction, and the system generates weight coefficients α, β, γ, and δ based on the tag-weight mapping rule library; Multi-dimensional Matching Degree Calculation: For a single park P and all enterprises E in the enterprise database, the multi-weight matching degree calculation function is substituted one by one to obtain the comprehensive matching degree M(P,E); Outlier Filtering: Enterprises with M(P,E) < 0.3 are removed, and the list of high-potential matching enterprises is retained; Result Sorting: The retained high-potential enterprises are sorted from high to low according to M(P,E) to form a recommended enterprise list for park P; Recommendation Reason Generation: For the top 10 enterprises, the 2-3 core dimensions with the highest matching degree are automatically extracted to generate a visual recommendation report; Dynamic Update: When park data or enterprise data is updated, the system automatically re-executes the above process to update the recommendation list.

[0013] A smart matching and recommendation system for attracting businesses to industrial parks includes the following modules: Algorithm selection module: Used to select algorithms that combine multi-dimensional related features of an enterprise; Function design module: Used to design multi-weighted matching degree calculation functions; Mechanism building module: used to build dynamic weight adjustment mechanisms; Algorithm execution module: Used to execute the enterprise intelligent matching and recommendation algorithm process; Matching and Recommendation Module: Used to automatically match and recommend companies for investment in the park according to the designed process, including: Data entry and preprocessing unit: The system connects to the park's basic database and the enterprise investment database to complete data cleaning, standardization and feature extraction; Enterprise Feature Extension Unit: Generates an extended feature set E' of the enterprise through a multi-dimensional association feature combination algorithm; Park weight setting unit: Based on the park's investment attraction target tags, determine the weight coefficients of the matching degree calculation function; Batch matching calculation unit: performs batch matching between the park and the enterprise database, and calculates the overall matching degree of each enterprise; Results filtering and sorting unit: Filters companies with low matching degree and generates a recommendation list by matching degree; Recommendation results output unit: Pushes a recommendation list and visual report to the park's investment promotion personnel, supporting the viewing of company details and matching reasons; Dynamic data update unit: Updates data and recommendation results periodically or triggered by events.

[0014] Preferably, the multi-dimensional association feature combination algorithm for enterprises proposed in the algorithm selection module combines adjacent features in the four dimensions of "industry attributes - development needs - resource dependence - contribution potential" in the original enterprise data in pairs to form new association features, so as to solve the problems of insufficient descriptive power and low matching accuracy of traditional single features.

[0015] Preferably, the multi-weighted matching degree calculation function designed in the function design module is a quantitative calculation function for park-enterprise matching, which is constructed by comprehensively considering four core indicators: park resource supply matching degree, industrial synergy adaptation degree, economic contribution potential, and regional policy fit degree. The formula is as follows: M(P,E)=α×S(P,E)+β×C(P,E)+γ×V(P,E)+δ×P(P,E) In the formula: M(P,E): the comprehensive matching degree between park P and enterprise E, with a value range of [0,1]. The higher the value, the stronger the fit; α, β, γ, δ: weight coefficients for each dimension, satisfying α+β+γ+δ=1, which can be dynamically adjusted according to the park's investment attraction priority. The default values ​​are α=0.35, β=0.3, γ=0.2, and δ=0.15; S(P,E): the matching degree of park resource supply, calculated as (the number of enterprise demand characteristics that match the park's resources / the total number of enterprise demand characteristics); C(P,E): the industrial synergy fit degree, calculated as (the number of existing industrial characteristics in the park that match the park's resources / the total number of enterprise demand characteristics); V(P,E): Economic contribution potential, calculated as min((expected annual tax revenue of enterprises / average annual tax revenue of newly added enterprises in the park), 1) × 0.5 + min((expected investment scale of enterprises / total annual investment target of the park), 1) × 0.5; P(P,E): Regional policy fit, calculated as (number of park policies that enterprises can enjoy / total number of special investment policies that the park has issued).

[0016] Preferably, the dynamic weight adjustment mechanism built in the mechanism construction module is a mechanism that automatically adjusts the values ​​of α, β, γ, and δ in the multi-weight matching degree calculation function based on the park's investment attraction target tags. The specific execution steps are as follows: Establish a target label system for investment attraction in the park, including four core labels: "industry clustering", "tax contribution", "innovation-driven", and "employment-creating". Each park can choose one main label and one or two auxiliary labels. A label-weight mapping rule base is established, with specific rules including: for the main label "industry cluster type": β (industry synergy matching degree weight) is increased to 0.4, α (resource supply matching degree) is adjusted to 0.3, γ (economic contribution potential) is 0.2, and δ (regional policy fit) is 0.1; for the main label "innovation-driven type": α (resource supply matching degree, focusing on matching R&D land and scientific research platforms) is increased to 0.4, β (industry synergy matching degree) is 0.3, γ (economic contribution potential) is 0.15, and δ (regional policy fit, focusing on matching R&D subsidies) is 0.15. If the park does not actively set labels, the system will use a "balanced" weight by default (α=0.35, β=0.3, γ=0.2, δ=0.15), while also allowing investment promotion personnel to manually fine-tune the weight coefficients.

[0017] Preferably, the enterprise intelligent matching and recommendation algorithm process executed in the algorithm execution module includes the following steps: Data preprocessing and feature extraction: The preprocessing objects include two categories: "basic data of the industrial park" and "enterprise investment data". Basic data of the park: Extract four core information categories: "Industrial positioning (leading industries, supporting industries)", "Resource supply (land use indicators, factory area, supporting facilities)", "Policy system (tax reduction and exemption, R&D subsidies, talent policies)" and "Existing enterprises (list of enterprises, industry type, output value scale)" to form the park feature set P; Enterprise investment data: Extract four core information categories: "basic attributes (industry, registered capital, years of establishment)," "development needs (land use needs, financing needs, supporting needs)," "operating status (annual output value, tax contribution, R&D investment)," and "development plan (investment plan, capacity target, regional layout)" to form the original set of enterprise characteristics E; Data cleaning: removing invalid data, standardizing data format, and standardizing feature labels; Enterprise multi-dimensional correlation feature combination: The original feature set E of enterprises after cleaning is divided into 4 dimensional subsets according to "industry attributes - development needs - resource dependence - contribution potential"; adjacent features within the same dimension and correlation features between different dimensions are combined in pairs to generate correlation features; the original features and correlation features are merged to form the expanded enterprise feature set E'. Park-Enterprise Matching Degree Calculation and Result Processing: Park tag setting: Park managers select target tags for investment attraction, and the system generates weight coefficients α, β, γ, δ based on the tag-weight mapping rule library; Multi-dimensional matching degree calculation: For a single park P and all enterprises E in the enterprise database, substitute them one by one into the multi-weight matching degree calculation function to obtain the comprehensive matching degree M(P,E); Outlier filtering: Remove companies with M(P,E) < 0.3 and retain the list of high-potential matching companies; Results ranking: The high-potential enterprises that are retained are ranked from high to low according to M(P,E) to form a recommended list of enterprises in park P; Recommendation reason generation: For the top 10 companies, automatically extract the 2-3 core dimensions with the highest matching degree and generate a visual recommendation report; Dynamic updates: When park data or enterprise data is updated, the system automatically re-executes the above process and updates the recommendation list.

[0018] Compared with the prior art, the beneficial effects of the present invention are: This invention proposes an intelligent matching and recommendation method and system for attracting investment to industrial parks. By comprehensively considering four core dimensions—enterprise development needs, park resource supply, industrial synergy and suitability, and regional economic contribution potential—it constructs a multi-weighted matching degree calculation function, replacing the traditional single-matching logic that relies solely on industry type or policy incentives. The improved intelligent matching and recommendation method significantly enhances the accuracy and efficiency of investment attraction in industrial parks, provides data support for optimizing the park's industrial layout, and offers decision-making assistance to investment promotion personnel in accurately connecting with enterprises. This reduces investment attraction costs and enhances the park's industrial agglomeration effect and sustainable development capabilities. Attached Figure Description

[0019] Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of the present invention clear and complete, the embodiments of the present invention will be further described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are only some, not all, embodiments of the present invention, and are merely illustrative of the embodiments of the present invention. They are not intended to limit the embodiments of the present invention. All other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] Example 1: This invention provides a technical solution: a method for intelligent matching and recommendation of enterprises for investment promotion in industrial parks, comprising the following steps: (1) Selection of multi-dimensional correlation feature combination algorithm for enterprises The core starting point of this invention is feature extraction from the original enterprise data. Traditional industrial park investment promotion often limits the processing of enterprise features to a single dimension (such as independent indicators like "industry," "registered capital," and "annual output value"), making it difficult to reflect the correlation between enterprise development needs and park resources. For example, a "new energy vehicle battery R&D enterprise" has independent features including "manufacturing - automobile manufacturing," "registered capital of 50 million yuan," and "R&D personnel accounting for 35%." However, these independent features alone cannot accurately match a park that "possesses supporting facilities for the new energy industry, preferential policies for R&D land, and a cluster of upstream and downstream enterprises in the industrial chain." Therefore, this invention proposes a multi-dimensional correlation feature combination algorithm for enterprises. By combining and associating the original features of enterprises across different dimensions, it generates composite features with greater matching value, solving the problems of insufficient descriptive power and low matching accuracy of traditional single features.

[0022] The algorithm's core idea is to pairwise associate and combine adjacent features across four dimensions—"industry attributes, development needs, resource dependence, and contribution potential"—in a company's original data to form new associated features. For example, a company's original feature set might be: E=(e1,e2,e3,e4,e5,e6); Among them, e1 = "Manufacturing - New Energy Battery" (industry attribute), e2 = "R&D land demand" (development demand), e3 = "Annual R&D investment ≥ 20%" (resource dependence), e4 = "Requires supporting enterprises in the industrial chain" (resource dependence), e5 = "Expected annual tax revenue ≥ 5 million yuan" (contribution potential), and e6 = "Drives employment ≥ 200 people" (contribution potential).

[0023] After the multi-dimensional correlation feature combination algorithm is executed, the feature set is expanded to: E'=(e1,e2,e3,e4,e5,e6,e1e2,e1e3,e2e4,e3e5,e4e5,e5e6); Among them, the original features e1-e6 remain unchanged, while the newly added related features are: e1e2 = "Manufacturing - New Energy Battery + R&D Land Demand", e1e3 = "Manufacturing - New Energy Battery + Annual R&D Investment ≥ 20%", e2e4 = "R&D Land Demand + Need for Supporting Enterprises in the Industrial Chain", etc., which are all core features that directly reflect the matching logic of "Enterprise Demand - Park Resources".

[0024] (2) Design of multi-weighted matching degree calculation function In the matching process between industrial parks and enterprises, determining whether a particular enterprise is suitable for a specific park requires a comprehensive assessment of three core dimensions: "the degree of fit between the enterprise's needs and the park's supply," "the degree of synergy between the enterprise and the park's existing industries," and "the enterprise's potential contribution to the park's economy." A high match in a single dimension (such as only matching industry type) cannot represent overall suitability. For example, a "biopharmaceutical R&D enterprise" is matched with Park A (matching industry type, but lacking R&D land quota) and Park B (matching industry type + having R&D land + already having 3 biopharmaceutical manufacturing enterprises). Traditional methods only determine "both are matched," while this invention, through a multi-weighting function, can quantify that the matching degree of Park B is significantly higher than that of Park A.

[0025] To this end, this invention integrates four core indicators (matching degree of park resource supply, adaptability of industrial synergy, potential for economic contribution, and alignment with regional policies) to design a multi-weighted park-enterprise matching degree calculation function, as shown in the following formula: M(P,E)=α×S(P,E)+β×C(P,E)+γ×V(P,E)+δ×P(P,E); In the formula: M(P,E): The overall matching degree between park P and enterprise E, with a value range of [0,1]. The higher the value, the stronger the suitability. α, β, γ, δ: Weight coefficients for each dimension, satisfying α+β+γ+δ=1, which can be dynamically adjusted according to the park's investment attraction priorities (such as "priority for industrial agglomeration" and "priority for tax contribution") (default α=0.35, β=0.3, γ=0.2, δ=0.15). S(P,E): Park resource supply matching degree. It calculates the overlap rate between the "development demand characteristics" (such as land use, financing, and supporting facilities) of enterprises and the existing resources of the park. The formula is S(P,E) = (number of enterprise demand characteristics that match park resources / total number of enterprise demand characteristics). C(P,E): Industrial synergy fit, which calculates the correlation between the enterprise's industry and the park's leading industry and upstream and downstream industries. The formula is C(P,E) = (number of existing industries in the park that form industrial chain synergy with the enterprise / total number of leading industries in the park) × 0.6 + (the degree of fit between the enterprise's technology direction and the park's industrial upgrading needs) × 0.4; V(P,E): Economic contribution potential, quantified based on the company's historical tax revenue, output growth rate, and expected investment scale. The formula is V(P,E)=min((Expected annual tax revenue of the company / Average annual tax revenue of newly added companies in the park),1)×0.5+min((Expected investment scale of the company / Total annual investment target of the park),1)×0.5; P(P,E): Regional policy fit, which calculates the coverage ratio of enterprises enjoying park-specific policies (such as R&D subsidies and tax reductions). The formula is P(P,E) = (number of park policies that enterprises can enjoy / total number of special investment promotion policies issued by the park).

[0026] (3) Construction of dynamic weight adjustment mechanism Different industrial parks have varying investment attraction needs at different stages of development (e.g., newly built parks focus on "industrial scale agglomeration," while mature parks focus on "industrial upgrading and high added value"), making fixed weights unsuitable for diverse investment attraction scenarios. To address this issue, this invention proposes a dynamic weight adjustment mechanism that automatically adjusts the values ​​of α, β, γ, and δ in the matching degree calculation function based on the park's investment attraction target labels (pre-set by park managers), ensuring that the recommended results are consistent with the actual needs of the park.

[0027] The execution steps of the dynamic weight adjustment mechanism are as follows: Establish a target label system for investment attraction in the park, including four core labels: "industry clustering", "tax contribution", "innovation-driven", and "employment-creating". Each park can choose one main label and one or two auxiliary labels. Establish a label-weight mapping rule base, for example: The main label is "industrial cluster type": β (industrial synergy and fit weight) is increased to 0.4, α (resource supply matching degree) is adjusted to 0.3, γ (economic contribution) is 0.2, and δ (policy) is 0.1; The main label is "innovation-driven": α (resource supply, with a focus on matching R&D land and scientific research platforms) increased to 0.4, β (industrial synergy) 0.3, γ (economic contribution) 0.15, δ (policy, with a focus on matching R&D subsidies) 0.15; If the park does not actively set labels, the system will use a "balanced" weight by default (α=0.35, β=0.3, γ=0.2, δ=0.15), while also allowing investment promotion personnel to manually fine-tune the weight coefficients.

[0028] (4) Enterprise intelligent matching and recommendation algorithm process The algorithm of this invention mainly includes the following four steps: I. Data preprocessing and feature extraction; II. Combination of multi-dimensional related features of enterprises; III. Calculation of matching degree between industrial park and enterprise; IV. Ranking and recommendation of matching results.

[0029] I. Data Preprocessing and Feature Extraction The preprocessing objects include two categories: "basic data of the industrial park" and "enterprise investment promotion data". Basic data of the park: Extract four core information categories: "Industrial positioning (leading industries, supporting industries)", "Resource supply (land use indicators, factory area, supporting facilities)", "Policy system (tax reduction and exemption, R&D subsidies, talent policies)" and "Existing enterprises (list of enterprises, industry type, output value scale)" to form the park feature set P; Enterprise investment data: Extract four core information categories: "basic attributes (industry, registered capital, years of establishment)," "development needs (land use needs, financing needs, supporting needs)," "operating status (annual output value, tax contribution, R&D investment)," and "development plan (investment plan, capacity target, regional layout)" to form the original set of enterprise characteristics E; Data cleaning: Remove invalid data (such as missing ≥50% of enterprise operating data), unify data format (such as converting land use requirements to "mu"), and standardize feature labels (such as adopting the "National Economic Industry Classification" GB / T4754-2017 for industry classification).

[0030] II. Combination of multi-dimensional related characteristics of enterprises For the original set of enterprise features E after cleaning, the "enterprise multi-dimensional association feature combination algorithm" is used for feature expansion: The original characteristics of enterprises are divided into four subsets based on "industry attributes, development needs, resource dependence, and contribution potential"; By combining adjacent features within the same dimension and related features between different dimensions, we can generate related features (such as "industry attributes + development needs" and "resource dependence + contribution potential"). The original features and associated features are merged to form an expanded set of enterprise features E', which provides richer dimensional support for subsequent matching calculations.

[0031] III. Park-Enterprise Matching Degree Calculation Park tag setting: The park manager selects the target tags for investment attraction (such as "innovation-driven + industry cluster"). The system automatically generates the weight coefficients α, β, γ, δ of the matching degree calculation function based on the tag-weight mapping rule library. Multi-dimensional matching degree calculation: For a single park P and all enterprises E in the enterprise database, substitute them one by one into the "multi-weight matching degree calculation function" to obtain the comprehensive matching degree M(P,E) between each enterprise and the park; Outlier filtering: Remove companies with M(P,E) < 0.3 (low match), and retain the list of companies with high potential for matching.

[0032] IV. Matching Results Ranking and Recommendation Results ranking: The high-potential enterprises that are retained are ranked from high to low according to M(P,E) to form a recommended list of enterprises in park P; Recommendation reason generation: For the top 10 companies, automatically extract the 2-3 core dimensions with the highest matching degree (such as "industry synergy fit degree 0.92 + resource supply fit degree 0.88") and generate a visual recommendation report; Dynamic updates: When park data (such as resource supply and policy adjustments) or enterprise data (such as operating conditions and development needs) is updated, the system automatically re-executes the above process to update the recommendation list and ensure the timeliness of the recommendation results.

[0033] (5) Matching and recommending enterprises for investment in the park The process of intelligent matching and recommendation of enterprises for investment attraction in the park using the method of this invention is as follows: I. Data entry and preprocessing: The system connects to the park's basic database and the enterprise investment database to complete data cleaning, standardization and feature extraction; II. Enterprise Feature Expansion: Generate an extended enterprise feature set E' using the "Enterprise Multi-Dimensional Related Feature Combination Algorithm"; III. Park Weight Setting: Based on the park's investment attraction target tags, determine the weight coefficients of the matching degree calculation function; IV. Batch Matching Calculation: Perform batch matching between the park and the enterprise database, and calculate the overall matching degree for each enterprise; V. Results Filtering and Sorting: Filter out companies with low matching scores and generate a recommendation list by matching score; VI. Recommendation Results Output: Push recommendation lists and visual reports to investment promotion personnel in the park, allowing them to view company details and matching reasons; VII. Dynamic Data Updates: Data and recommendation results are updated regularly (e.g., monthly) or triggered by factors such as changes in park policies.

[0034] Example 2, based on Example 1, proposes an intelligent matching and recommendation system for attracting investment to industrial parks, including the following modules: Algorithm selection module: used to select multi-dimensional related feature combination algorithms for enterprises; the multi-dimensional related feature combination algorithm proposed in the algorithm selection module combines adjacent features under the four dimensions of "industry attributes - development needs - resource dependence - contribution potential" in the original enterprise data in pairs to form new related features, so as to solve the problems of insufficient descriptive power and low matching accuracy of traditional single features.

[0035] The function design module is used to design a multi-weighted matching degree calculation function. This function is a quantitative calculation function for park-enterprise matching, built upon four core indicators: park resource supply matching degree, industrial synergy adaptation degree, economic contribution potential, and regional policy fit. The formula is as follows: M(P,E)=α×S(P,E)+β×C(P,E)+γ×V(P,E)+δ×P(P,E) In the formula: M(P,E): the comprehensive matching degree between park P and enterprise E, with a value range of [0,1]. The higher the value, the stronger the fit; α, β, γ, δ: weight coefficients for each dimension, satisfying α+β+γ+δ=1, which can be dynamically adjusted according to the park's investment attraction priority. The default values ​​are α=0.35, β=0.3, γ=0.2, and δ=0.15; S(P,E): the matching degree of park resource supply, calculated as (the number of enterprise demand characteristics that match the park's resources / the total number of enterprise demand characteristics); C(P,E): the industrial synergy fit degree, calculated as (the number of existing industrial characteristics in the park that match the park's resources / the total number of enterprise demand characteristics); V(P,E): Economic contribution potential, calculated as min((expected annual tax revenue of enterprises / average annual tax revenue of newly added enterprises in the park), 1) × 0.5 + min((expected investment scale of enterprises / total annual investment target of the park), 1) × 0.5; P(P,E): Regional policy fit, calculated as (number of park policies that enterprises can enjoy / total number of special investment policies that the park has issued).

[0036] Mechanism Construction Module: Used to construct a dynamic weight adjustment mechanism. The dynamic weight adjustment mechanism constructed in the mechanism construction module is a mechanism that automatically adjusts the values ​​of α, β, γ, and δ in the multi-weight matching degree calculation function based on the park's investment attraction target tags. The specific execution steps are as follows: Establish a target label system for investment attraction in the park, including four core labels: "industry clustering", "tax contribution", "innovation-driven", and "employment-creating". Each park can choose one main label and one or two auxiliary labels. A label-weight mapping rule base is established, with specific rules including: for the main label "industry cluster type": β (industry synergy matching degree weight) is increased to 0.4, α (resource supply matching degree) is adjusted to 0.3, γ (economic contribution potential) is 0.2, and δ (regional policy fit) is 0.1; for the main label "innovation-driven type": α (resource supply matching degree, focusing on matching R&D land and scientific research platforms) is increased to 0.4, β (industry synergy matching degree) is 0.3, γ (economic contribution potential) is 0.15, and δ (regional policy fit, focusing on matching R&D subsidies) is 0.15. If the park does not actively set labels, the system will use a "balanced" weight by default (α=0.35, β=0.3, γ=0.2, δ=0.15), while also allowing investment promotion personnel to manually fine-tune the weight coefficients.

[0037] Algorithm Execution Module: Used to execute the enterprise intelligent matching and recommendation algorithm process; the enterprise intelligent matching and recommendation algorithm process executed in the algorithm execution module includes the following steps: Data preprocessing and feature extraction: The preprocessing objects include two categories: "basic data of the industrial park" and "enterprise investment data". Basic data of the park: Extract four core information categories: "Industrial positioning (leading industries, supporting industries)", "Resource supply (land use indicators, factory area, supporting facilities)", "Policy system (tax reduction and exemption, R&D subsidies, talent policies)" and "Existing enterprises (list of enterprises, industry type, output value scale)" to form the park feature set P; Enterprise investment data: Extract four core information categories: "basic attributes (industry, registered capital, years of establishment)," "development needs (land use needs, financing needs, supporting needs)," "operating status (annual output value, tax contribution, R&D investment)," and "development plan (investment plan, capacity target, regional layout)" to form the original set of enterprise characteristics E; Data cleaning: removing invalid data, standardizing data format, and standardizing feature labels; Enterprise multi-dimensional correlation feature combination: The original feature set E of enterprises after cleaning is divided into 4 dimensional subsets according to "industry attributes - development needs - resource dependence - contribution potential"; adjacent features within the same dimension and correlation features between different dimensions are combined in pairs to generate correlation features; the original features and correlation features are merged to form the expanded enterprise feature set E'. Park-Enterprise Matching Degree Calculation and Result Processing: Park tag setting: Park managers select target tags for investment attraction, and the system generates weight coefficients α, β, γ, δ based on the tag-weight mapping rule library; Multi-dimensional matching degree calculation: For a single park P and all enterprises E in the enterprise database, substitute them one by one into the multi-weight matching degree calculation function to obtain the comprehensive matching degree M(P,E); Outlier filtering: Remove companies with M(P,E) < 0.3 and retain the list of high-potential matching companies; Results ranking: The high-potential enterprises that are retained are ranked from high to low according to M(P,E) to form a recommended list of enterprises in park P; Recommendation reason generation: For the top 10 companies, automatically extract the 2-3 core dimensions with the highest matching degree and generate a visual recommendation report; Dynamic updates: When park data or enterprise data is updated, the system automatically re-executes the above process and updates the recommendation list.

[0038] Matching and Recommendation Module: Used to automatically match and recommend companies for investment in the park according to the designed process, including: Data entry and preprocessing unit: The system connects to the park's basic database and the enterprise investment database to complete data cleaning, standardization and feature extraction; Enterprise Feature Extension Unit: Generates an extended feature set E' of the enterprise through a multi-dimensional association feature combination algorithm; Park weight setting unit: Based on the park's investment attraction target tags, determine the weight coefficients of the matching degree calculation function; Batch matching calculation unit: performs batch matching between the park and the enterprise database, and calculates the overall matching degree of each enterprise; Results filtering and sorting unit: Filters companies with low matching degree and generates a recommendation list by matching degree; Recommendation results output unit: Pushes a recommendation list and visual report to the park's investment promotion personnel, supporting the viewing of company details and matching reasons; Dynamic data update unit: Updates data and recommendation results periodically or triggered by events.

[0039] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for intelligent matching and recommendation of enterprises for investment promotion in industrial parks, characterized in that: Includes the following steps: 101: Select an algorithm that combines multi-dimensional related features of an enterprise; 102: Design a multi-weighted matching degree calculation function; 103: Construct a dynamic weight adjustment mechanism; 104: Execute the enterprise intelligent matching and recommendation algorithm process; 105: The system automatically matches and recommends companies for investment in the park according to the designed process, including the following seven sub-steps: I. Data entry and preprocessing: The system connects to the park's basic database and the enterprise investment database to complete data cleaning, standardization and feature extraction; II. Enterprise Feature Expansion: Generate an extended enterprise feature set E' through a multi-dimensional association feature combination algorithm; III. Park Weight Setting: Based on the park's investment attraction target tags, determine the weight coefficients of the matching degree calculation function; IV. Batch Matching Calculation: Perform batch matching between the park and the enterprise database, and calculate the overall matching degree for each enterprise; V. Results Filtering and Sorting: Filter out companies with low matching scores and generate a recommendation list by matching score; VI. Recommendation Results Output: Push recommendation lists and visual reports to investment promotion personnel in the park, allowing them to view company details and matching reasons; VII. Dynamic Data Updates: Data and recommendation results are updated periodically or triggered on a regular basis.

2. The intelligent matching and recommendation method for enterprises used in industrial park investment promotion according to claim 1, characterized in that: The multi-dimensional association feature combination algorithm for enterprises proposed in step 101 combines adjacent features under the four dimensions of "industry attributes, development needs, resource dependence, and contribution potential" in the original enterprise data in pairs to form new association features, thus solving the problems of insufficient descriptive power and low matching accuracy of traditional single features.

3. The intelligent matching and recommendation method for enterprises used in industrial park investment promotion according to claim 2, characterized in that: The multi-weighted matching degree calculation function designed in step 102 is a quantitative calculation function for park-enterprise matching, which is constructed by integrating four core indicators: park resource supply matching degree, industrial synergy adaptation degree, economic contribution potential, and regional policy fit. The formula is as follows: M(P,E)=α×S(P,E)+β×C(P,E)+γ×V(P,E)+δ×P(P,E) In the formula: M(P,E): the comprehensive matching degree between park P and enterprise E, with a value range of [0,1]. The higher the value, the stronger the suitability; α, β, γ, δ: weight coefficients of each dimension, satisfying α+β+γ+δ=1, which can be dynamically adjusted according to the priority of park investment promotion. The default values ​​are α=0.35, β=0.3, γ=0.2, and δ=0.15; S(P,E): the matching degree of park resource supply, calculated as (the number of enterprise demand characteristics that match park resources / the total number of enterprise demand characteristics). C(P,E): Industrial synergy fit, calculated as (number of existing industries in the park that form industrial chain synergy with enterprises / total number of leading industries in the park) × 0.6 + (the degree of fit between enterprise technology direction and the park's industrial upgrading needs) × 0.4; V(P,E): Economic contribution potential, calculated as min((expected annual tax revenue of the enterprise / average annual tax revenue of newly added enterprises in the park), 1)×0.5+min((expected investment scale of the enterprise / total annual investment target of the park), 1)×0.5; P(P,E): Regional policy fit, calculated as (number of park policies that the enterprise can enjoy / total number of special investment policies issued by the park).

4. The intelligent matching and recommendation method for enterprises used in industrial park investment promotion according to claim 3, characterized in that: The dynamic weight adjustment mechanism constructed in step 103 is a mechanism that automatically adjusts the values ​​of α, β, γ, and δ in the multi-weight matching degree calculation function based on the park's investment attraction target labels. The execution steps are as follows: I. Establish a target label system for investment attraction in the park, including four core labels: "industry clustering", "tax contribution", "innovation-driven", and "employment-creating". The park can choose one main label and one to two auxiliary labels. II. Establish a label-weight mapping rule base. Specific rules include: For the main label "Industry Agglomeration": β (industry synergy matching weight) is increased to 0.4, α (resource supply matching degree) is adjusted to 0.3, γ (economic contribution potential) is 0.2, and δ (regional policy fit) is 0.1; For the main label "Innovation-Driven": α (resource supply matching degree, focusing on matching R&D land and scientific research platforms) is increased to 0.4, β (industry synergy matching degree) is 0.3, γ (economic contribution potential) is 0.15, and δ (regional policy fit, focusing on matching R&D subsidies) is 0.

15. III. If the park does not actively set labels, the system will use "balanced" weights by default (α=0.35, β=0.3, γ=0.2, δ=0.15), while also supporting investment promotion personnel to manually fine-tune the weight coefficients.

5. The intelligent matching and recommendation method for enterprises used in industrial park investment promotion according to claim 4, characterized in that: The enterprise intelligent matching and recommendation algorithm process executed in step 104 includes the following three sub-steps: I. Data Preprocessing and Feature Extraction: The preprocessing objects include two categories: "Park Basic Data" and "Enterprise Investment Promotion Data". Park Basic Data: Extract four core information categories: "Industry Positioning (leading industries, supporting industries)", "Resource Supply (land use indicators, factory area, supporting facilities)", "Policy System (tax reduction and exemption, R&D subsidies, talent policies)" and "Existing Enterprises (enterprise list, industry type, output value scale)" to form the park feature set P; Enterprise Investment Promotion Data: Extract four core information categories: "Basic Attributes (industry, registered capital, years of establishment)", "Development Needs (land use needs, financing needs, supporting needs)", "Operating Status (annual output value, tax contribution, R&D investment)" and "Development Plan (investment plan, capacity target, regional layout)" to form the enterprise original feature set E; Data cleaning: removing invalid data, standardizing data format, and standardizing feature labels; II. Combination of multi-dimensional related features of enterprises: The original feature set E of enterprises after cleaning is divided into four dimensional subsets according to "industry attributes - development needs - resource dependence - contribution potential"; adjacent features within the same dimension and related features between different dimensions are combined in pairs to generate related features; the original features and related features are merged to form the expanded feature set E' of enterprises; III. Park-Enterprise Matching Degree Calculation and Result Processing: Park Tag Setting: Park managers select investment target tags, and the system generates weight coefficients α, β, γ, and δ based on the tag-weight mapping rule base; Multi-dimensional matching degree calculation: For a single park P and all companies E in the enterprise database, the multi-weighted matching degree calculation function is substituted one by one to obtain the comprehensive matching degree M(P,E); Outlier filtering: Companies with M(P,E) < 0.3 are removed, and the list of high-potential matching companies is retained; Result sorting: The retained high-potential companies are sorted from high to low according to M(P,E) to form the enterprise recommendation list of park P; Recommendation reason generation: For the top 10 companies, the 2-3 core dimensions with the highest matching degree are automatically extracted to generate a visual recommendation report; Dynamic update: When park data or enterprise data is updated, the system automatically re-executes the above process to update the recommendation list.

6. A smart matching and recommendation system for attracting investment to industrial parks, applied to the method described in claim 5, characterized in that: Includes the following modules: Algorithm selection module: Used to select algorithms that combine multi-dimensional related features of an enterprise; Function design module: Used to design multi-weighted matching degree calculation functions; Mechanism building module: used to build dynamic weight adjustment mechanisms; Algorithm execution module: Used to execute the enterprise intelligent matching and recommendation algorithm process; Matching and Recommendation Module: Used to automatically match and recommend companies for investment in the park according to the designed process, including: Data entry and preprocessing unit: The system connects to the park's basic database and the enterprise investment database to complete data cleaning, standardization and feature extraction; Enterprise Feature Extension Unit: Generates an extended feature set E' of the enterprise through a multi-dimensional association feature combination algorithm; Park weight setting unit: Based on the park's investment attraction target tags, determine the weight coefficients of the matching degree calculation function; Batch matching calculation unit: performs batch matching between the park and the enterprise database, and calculates the overall matching degree of each enterprise; Results filtering and sorting unit: Filters companies with low matching degree and generates a recommendation list by matching degree; Recommendation results output unit: Pushes a recommendation list and visual report to the park's investment promotion personnel, supporting the viewing of company details and matching reasons; Dynamic data update unit: Updates data and recommendation results periodically or triggered by events.

7. The intelligent matching and recommendation system for enterprise investment promotion in industrial parks according to claim 6, characterized in that: The multi-dimensional association feature combination algorithm for enterprises proposed in the algorithm selection module combines adjacent features in the four dimensions of "industry attributes, development needs, resource dependence, and contribution potential" in the original enterprise data in pairs to form new association features, so as to solve the problems of insufficient descriptive power and low matching accuracy of traditional single features.

8. The intelligent matching and recommendation system for enterprise investment promotion in industrial parks according to claim 7, characterized in that: The multi-weighted matching degree calculation function designed in the function design module is a quantitative calculation function for park-enterprise matching, which is constructed by integrating four core indicators: park resource supply matching degree, industrial synergy adaptation degree, economic contribution potential, and regional policy fit. The formula is as follows: M(P,E)=α×S(P,E)+β×C(P,E)+γ×V(P,E)+δ×P(P,E) In the formula: M(P,E): the comprehensive matching degree between park P and enterprise E, with a value range of [0,1]. The higher the value, the stronger the suitability; α, β, γ, δ: weight coefficients of each dimension, satisfying α+β+γ+δ=1, which can be dynamically adjusted according to the priority of park investment promotion. The default values ​​are α=0.35, β=0.3, γ=0.2, and δ=0.15; S(P,E): the matching degree of park resource supply, calculated as (the number of enterprise demand characteristics that match park resources / the total number of enterprise demand characteristics). C(P,E): Industrial synergy fit, calculated as (number of existing industries in the park that form industrial chain synergy with enterprises / total number of leading industries in the park) × 0.6 + (the degree of fit between enterprise technology direction and the park's industrial upgrading needs) × 0.4; V(P,E): Economic contribution potential, calculated as min((expected annual tax revenue of the enterprise / average annual tax revenue of newly added enterprises in the park), 1)×0.5+min((expected investment scale of the enterprise / total annual investment target of the park), 1)×0.5; P(P,E): Regional policy fit, calculated as (number of park policies that the enterprise can enjoy / total number of special investment policies issued by the park).

9. The intelligent matching and recommendation system for enterprise investment promotion in industrial parks according to claim 8, characterized in that: The dynamic weight adjustment mechanism built in the mechanism construction module is a mechanism that automatically adjusts the values ​​of α, β, γ, and δ in the multi-weight matching degree calculation function based on the park's investment attraction target tags. The specific execution steps are as follows: Establish a target label system for investment attraction in the park, including four core labels: "industry clustering", "tax contribution", "innovation-driven", and "employment-creating". Each park can choose one main label and one to two auxiliary labels. A label-weight mapping rule base is established, with specific rules including: for the main label "industry cluster type": β (industry synergy matching degree weight) is increased to 0.4, α (resource supply matching degree) is adjusted to 0.3, γ (economic contribution potential) is 0.2, and δ (regional policy fit) is 0.1; for the main label "innovation-driven type": α (resource supply matching degree, focusing on matching R&D land and scientific research platforms) is increased to 0.4, β (industry synergy matching degree) is 0.3, γ (economic contribution potential) is 0.15, and δ (regional policy fit, focusing on matching R&D subsidies) is 0.

15. If the park does not actively set labels, the system will use "balanced" weights by default (α=0.35, β=0.3, γ=0.2, δ=0.15), while also allowing investment promotion personnel to manually fine-tune the weight coefficients.

10. The intelligent matching and recommendation system for enterprise investment promotion in industrial parks according to claim 9, characterized in that: The enterprise intelligent matching and recommendation algorithm process executed in the algorithm execution module includes the following steps: Data preprocessing and feature extraction: The preprocessing objects include two categories: "basic data of the industrial park" and "enterprise investment data". Basic data of the park: Extract four core information categories: "industrial positioning (leading industries, supporting industries)," "resource supply (land use indicators, factory area, supporting facilities)," "policy system (tax reduction and exemption, R&D subsidies, talent policies)," and "existing enterprises (list of enterprises, industry type, output value scale)" to form the park feature set P; Enterprise investment data: Extract four core information categories: "basic attributes (industry, registered capital, years of establishment)," "development needs (land use needs, financing needs, supporting needs)," "operating status (annual output value, tax contribution, R&D investment)," and "development plan (investment plan, production capacity target, regional layout)" to form the original set of enterprise characteristics E; Data cleaning: removing invalid data, standardizing data format, and standardizing feature labels; Enterprise multi-dimensional correlation feature combination: The original feature set E of enterprises after cleaning is divided into 4 dimensional subsets according to "industry attributes - development needs - resource dependence - contribution potential"; adjacent features within the same dimension and correlation features between different dimensions are combined in pairs to generate correlation features; the original features and correlation features are merged to form the expanded enterprise feature set E'; Park-Enterprise Matching Degree Calculation and Result Processing: Park tag setting: Park managers select target tags for investment attraction, and the system generates weight coefficients α, β, γ, δ based on the tag-weight mapping rule library; Multi-dimensional matching degree calculation: For a single park P and all enterprises E in the enterprise database, substitute them one by one into the multi-weight matching degree calculation function to obtain the comprehensive matching degree M(P,E); Outlier filtering: Remove companies with M(P,E) < 0.3 and retain the list of high-potential matching companies; Results ranking: The high-potential enterprises that are retained are ranked from high to low according to M(P,E) to form a recommended list of enterprises in park P; Recommendation reason generation: For the top 10 companies, automatically extract the 2-3 core dimensions with the highest matching degree and generate a visual recommendation report; Dynamic updates: When park data or enterprise data is updated, the system automatically re-executes the above process and updates the recommendation list.

Images