Intelligent matching method and system for entering enterprises in park

Abstract

The invention discloses a park entering enterprise intelligent matching method and system, and relates to the technical field of data intelligent matching, and the method comprises the following steps: constructing an industrial knowledge graph based on industrial data, calculating a first matching degree of a target enterprise and a candidate park according to the industrial knowledge graph and the attribute information of the target enterprise, generating a structured demand vector of the target enterprise based on the demand information of the target enterprise, generating a structured supply vector based on the service supply information of the candidate park, and calculating the similarity between the demand vector and the supply vector to obtain a second matching degree between the target enterprise and the candidate park, and comprehensively processing the service supply information, the first matching degree and the second matching degree of the candidate parks to generate a comprehensive matching result, thereby maximizing the dual benefits of industrial collaboration and service adaptation on the premise of ensuring that the matching result meets the rigid demand of an enterprise, and improving the user experience. And the investment attracting precision, the enterprise satisfaction and the park industry ecological quality are improved, and comprehensive exceeding of the prior art is realized.

Description

Technical Field

[0001] This invention relates to the field of intelligent data matching technology, and in particular to an intelligent matching method and system for enterprises entering a park. Background Technology

[0002] In recent years, with the large-scale and professional development of industrial parks in my country, the precise matching of investment attraction and enterprise site selection has become a key link in improving the operational efficiency of parks and promoting industrial agglomeration. Data-driven intelligent matching technology has shown great potential in solving such supply and demand matching problems. By introducing big data analysis, knowledge graphs and machine learning methods, it aims to replace the traditional matching model that relies on human experience and extensive recommendations, and achieve rapid screening and accurate recommendations from massive amounts of information. The importance of this technology is not only reflected in improving the success rate of investment attraction and enterprise satisfaction, but also in its ability to promote the optimization of the industrial ecosystem and the synergy of the industrial chain through scientific analysis, thereby driving high-quality regional economic development.

[0003] However, existing matching technologies still have significant limitations. On the one hand, most methods only perform simple screening based on shallow attributes such as enterprise classification and scale, lacking in-depth modeling of complex upstream and downstream, technological collaboration and other ecological relationships between industries, making it difficult to assess the potential for industrial synergy after enterprises move in. On the other hand, existing technologies usually treat enterprise demand and park supply as isolated labels for comparison, failing to transform unstructured demand descriptions and supply information into quantifiable and calculable structured vectors, thus making it impossible to accurately measure the degree of matching between service supply and demand, and easily overlooking the core needs of enterprises and the unique service capabilities of the park. Summary of the Invention

[0004] The technical problem addressed by this invention is that existing matching technologies still have significant limitations. On the one hand, most methods only perform simple screening based on shallow attributes such as enterprise classification and scale, lacking in-depth modeling of complex upstream and downstream, technological collaboration and other ecological relationships between industries, making it difficult to assess the potential for industrial synergy after enterprises settle in. On the other hand, existing technologies usually treat enterprise demand and park supply as isolated labels for comparison, failing to transform unstructured demand descriptions and supply information into quantifiable and calculable structured vectors, thus making it impossible to accurately measure the degree of matching between service supply and demand, and easily overlooking the core needs of enterprises and the unique service capabilities of the park.

[0005] To solve the above-mentioned technical problems, the present invention provides the following technical solution: A method for intelligent matching of enterprises entering a park includes the following steps: Step S1: Construct an industry knowledge graph based on industry data, and calculate the first matching degree between the target enterprise and the candidate park based on the industry knowledge graph and the attribute information of the target enterprise; Step S2: Generate a structured demand vector for the target enterprise based on its demand information, and generate a structured supply vector based on the service supply information of the candidate parks. Calculate the similarity between the demand vector and the supply vector to obtain the second matching degree between the target enterprise and the candidate park; Step S3: The service supply information of the candidate parks, the first matching degree and the second matching degree are comprehensively processed to generate a comprehensive matching result.

[0006] As a preferred embodiment of the intelligent matching method for enterprises entering a park as described in this invention, the method involves acquiring the target enterprise's attribute information, demand information, industry data, and service supply information of the candidate parks. Attribute information includes the target company's industry attributes and business attributes; Among them, the industry attribute includes the target company's main industry classification code, and the business attribute includes the target company's registered capital and years of establishment; The requirements information includes a description of the target company's service needs; The demand description includes the demand for physical space conditions, policy services, and financial services; The physical space requirements include the target company's required area, net height, and load capacity, and the target company must specify the necessary physical space requirements, policy services, and financial services in its requirements description. Service supply information includes the resource conditions, service items, supported industry scope, and minimum capital requirements for candidate parks; The resource conditions include the available physical space conditions of the candidate park, which include the leasable area of ​​the candidate park, the net floor height of the candidate park, and the floor load of the candidate park. The service items include policy services, financial services, talent services, and marketing services, and cover the needs for physical space conditions, policy services, and financial services.

[0007] As a preferred embodiment of the intelligent matching method for enterprises entering a park as described in this invention, step S1 involves constructing an industry knowledge graph based on industry data, and calculating the first matching degree between the target enterprise and the candidate park in the industrial ecosystem dimension based on the industry knowledge graph and the attribute information of the target enterprise. The industry data includes enterprise business registration information, industry classification standards, and macroeconomic relationship data. Step S1 includes steps S101, S102, S103 and S104; Step S101: Based on the industry classification standard data, construct an industry classification node with a hierarchical structure; Based on macroeconomic data, establish relationships between industry classification nodes. These relationships include upstream and downstream supply chain relationships, technological cooperation relationships, or service support relationships. Based on enterprise registration information data, extract enterprise entity and industry attributes, and create corresponding enterprise entity nodes; Step S102: Connect the enterprise entity node to the corresponding industry classification node according to the subordinate relationship between the enterprise entity node and the industry classification node; Integrate industry classification nodes, enterprise entity nodes, and the relationships between industry classification nodes to form an industry knowledge graph; Step S103: Map the industry attributes of the target enterprise to the industry knowledge graph to obtain the corresponding industry classification node, which serves as the target industry node; The industry attributes of all resident enterprises in the candidate park are mapped to the industry knowledge graph, resulting in a set of industry classification nodes corresponding to all resident enterprises, which serves as the industry classification node set of the candidate park. Step S104: In the industry knowledge graph, calculate the shortest path length of the industry knowledge graph between the target industry node and each node in the industry node set; Based on the shortest path length, the corresponding industry correlation is obtained through a preset inverse proportional function; Based on the industry correlation between the target industry node and all nodes in the industry classification node set, and combined with the weight of each node, the first matching degree between the target enterprise and the candidate park is obtained through weighted calculation.

[0008] As a preferred embodiment of the intelligent matching method for enterprises entering a park as described in this invention, step S2 involves generating a structured demand vector for the target enterprise based on the demand information of the target enterprise, and generating a structured supply vector based on the service supply information of the candidate park. Step S2 includes steps S201, S202, S203, S204 and S205; Step S201: Construct a feature list that includes multiple dimensions of matching features, including physical space features, policy service features, and financial service features; Each matching feature is assigned a standardized name, data type, a preset set of quantized values, and a regular expression for extracting information. Step S202: For each matching feature in the feature list, use the regular expression corresponding to the matching feature to match the demand description of the target enterprise; If a match is successful and a text fragment is extracted, a value is obtained from the quantization set corresponding to the matching feature based on the extracted text fragment, which is used as the demand intensity value on the matching feature. Step S203: Assign weight coefficients to each matching feature based on the business attributes of the target enterprise, and use the weight coefficients to weight all demand intensity values ​​to generate a structured demand vector for the target enterprise.

[0009] As a preferred embodiment of the intelligent matching method for enterprises entering a park as described in this invention, step S204 involves using the regular expression corresponding to each matching feature in the feature list to match the service supply information of the candidate park. If a match is successful and a text fragment is extracted, a value is obtained from the quantization set corresponding to the matching feature based on the extracted text fragment, which is used as the supply capacity value on the matching feature. Step S205: Based on the maximum supply capacity value of the same matching feature of all candidate parks, normalize the supply capacity value of each candidate park to generate a structured supply vector for each candidate park.

[0010] As a preferred embodiment of the intelligent matching method for enterprises entering a park as described in this invention, a second matching degree is obtained by calculating the demand vector and the supply vector using the cosine similarity formula.

[0011] As a preferred embodiment of the intelligent matching method for enterprises entering a park as described in this invention, step S3 involves comprehensively processing the service supply information of the candidate park, the first matching degree, and the second matching degree to generate a comprehensive matching result. Step S3 includes steps S301, S302 and S303; Step S301: Based on historical matching data and the park's investment promotion strategy, determine the recommendation level according to the first matching degree and the second matching degree; The recommendation levels include top recommendation, worth considering, and not recommended; The recommendation level is determined by the following logic: if the first matching degree is greater than or equal to the first threshold and the second matching degree is greater than or equal to the second threshold, it is determined to be a priority recommendation. If the first matching degree is greater than or equal to the third threshold or the second matching degree is greater than or equal to the fourth threshold, it is considered acceptable. If the first matching degree is less than or equal to the fifth threshold and the second matching degree is less than or equal to the sixth threshold, it is determined as not recommended.

[0012] As a preferred embodiment of the intelligent matching method for enterprises entering a park as described in this invention, step S302 involves screening candidate parks based on preset constraints, removing candidate parks that do not meet any of the constraints, and obtaining the final set of candidate parks. The constraints include physical space constraints, qualification compatibility constraints, and service coverage constraints. Physical space constraints include the target company's required area being less than or equal to the leasable area of ​​the candidate park, the candidate park's net floor height being greater than or equal to the target company's required net height, and the candidate park's floor load being greater than or equal to the target company's required load. The qualification matching constraint is that the target company's main industry classification code belongs to the industry scope supported by the candidate park and the target company's registered capital is greater than or equal to the minimum entry capital requirement of the candidate park. The service coverage constraint is that the coverage rate of policy services and financial services in the candidate parks is greater than or equal to the coverage rate threshold.

[0013] As a preferred embodiment of the intelligent matching method for enterprises entering a park as described in this invention, in step S303, if the final candidate park set is empty, a comprehensive matching result with no recommended parks is generated. If the final candidate park set is not empty, the candidate parks in the final candidate park set are divided according to the recommendation level determined in step S301. Within the same recommendation level, sort in descending order based on the weighted sum of the first and second matching degrees; If the weighted summation results are the same, they are sorted in descending order according to the first matching degree; If the weighted summation result is the same as the first matching degree, then the coverage of the target enterprise's unmarked needs for the service items of the candidate parks is sorted in descending order, and the sorted candidate park list is output as the comprehensive matching result.

[0014] A smart matching system for companies entering a park includes a construction module, a calculation module, and a matching module; The module constructs an industry knowledge graph based on industry data, and calculates the first matching degree between the target enterprise and the candidate park based on the industry knowledge graph and the attribute information of the target enterprise. The calculation module generates a structured demand vector for the target enterprise based on its demand information, and generates a structured supply vector based on the service supply information of the candidate parks. Calculate the similarity between the demand vector and the supply vector to obtain the second matching degree between the target enterprise and the candidate park; The matching module comprehensively processes the service supply information of the candidate parks, the first matching degree, and the second matching degree to generate a comprehensive matching result.

[0015] The beneficial effects of this invention are as follows: Traditional methods often rely on static labels of enterprise classification and scale for shallow screening, which cannot assess the potential for industrial synergy after enterprises settle in, nor can they quantify the degree of matching between service supply and demand. This solution, by constructing an industrial knowledge graph and calculating the first matching degree, achieves for the first time a quantitative assessment of the industrial ecosystem correlation of the park, solving the pain point of existing technologies lacking in-depth analysis of the industrial chain dimension. At the same time, by converting unstructured text into structured vectors and calculating the second matching degree, it achieves refined matching of personalized and diversified service needs, overcoming the defect of traditional label comparison methods that cannot capture core demands. Finally, through hard constraints and multi-dimensional comprehensive ranking, this solution ensures that the matching results maximize the dual benefits of industrial synergy and service adaptation while meeting the rigid needs of enterprises. Thus, it achieves a comprehensive leap over existing technologies in improving the accuracy of investment attraction, enterprise satisfaction, and the quality of the park's industrial ecosystem. Attached Figure Description

[0016] Figure 1 This is a flowchart illustrating the steps of an intelligent matching method for enterprises entering a park, as provided in one embodiment of the present invention.

[0017] Figure 2 This is a basic flowchart of an intelligent matching system for enterprises entering a park, provided as an embodiment of the present invention. Detailed Implementation

[0018] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0019] Example 1, referring to Figure 1 As an embodiment of the present invention, a method for intelligent matching of enterprises entering a park is provided, including the following steps: Step S1: Construct an industry knowledge graph based on industry data, and calculate the first matching degree between the target enterprise and the candidate park based on the industry knowledge graph and the attribute information of the target enterprise; Step S2: Generate a structured demand vector for the target enterprise based on its demand information, and generate a structured supply vector based on the service supply information of the candidate parks. Calculate the similarity between the demand vector and the supply vector to obtain the second matching degree between the target enterprise and the candidate park; Step S3: The service supply information, first matching degree and second matching degree of the candidate parks are comprehensively processed to generate a comprehensive matching result.

[0020] In one embodiment, this solution first constructs an industry knowledge graph and calculates the first matching degree in the industry chain collaboration dimension based on the graph and the attribute information of the target enterprise. It then transforms the enterprise's unstructured needs and the park's service supply information into structured demand and supply vectors, respectively, and calculates their similarity to obtain the second matching degree in the service supply and demand dimension. Finally, it performs global processing on the comprehensive service supply information, the first matching degree, and the second matching degree to generate a comprehensive matching result. The core of this design is to conduct a dual evaluation of integrated industry ecosystem adaptation and service supply and demand matching, overcoming the limitations of traditional methods that only focus on single static labels. By quantifying the enterprise's industry connections and niche value through the industry knowledge graph, it achieves a deeper matching; by vectorizing, it achieves refined matching of diversified and personalized service needs, improving the accuracy of the matching; and the final comprehensive decision ensures that the matching result satisfies the rigid constraints of the enterprise while maximizing industry collaboration and comprehensive service satisfaction. This results in intelligent, precise, and ecological investment attraction matching, effectively improving the park's operational efficiency and the quality of industrial agglomeration.

[0021] Obtain attribute information, demand information, industry data of target enterprises, and service supply information of candidate parks; Attribute information includes the target company's industry attributes and business attributes; Among them, the industry attribute includes the target company's main industry classification code, and the business attribute includes the target company's registered capital and years of establishment; The requirements information includes a description of the target company's service needs; The demand description includes the demand for physical space conditions, policy services, and financial services; The physical space requirements include the target company's required area, net height, and load capacity, and the target company must specify the necessary physical space requirements, policy services, and financial services in its requirements description. Service supply information includes the resource conditions, service items, supported industry scope, and minimum capital requirements for candidate parks; Among them, the resource conditions include the available physical space conditions of the candidate park, which include the leasable area of ​​the candidate park, the net floor height of the candidate park, and the floor load of the candidate park. The service items include policy services, financial services, talent services, and marketing services, and cover the needs for physical space conditions, policy services, and financial services.

[0022] In one embodiment, raw data is obtained through multiple channels. The target company's attribute information is obtained by parsing standardized application forms. Attribute information specifically includes the main industry classification code as an industry attribute, and registered capital and years of establishment as business attributes. Industry data, such as national standard industry classifications and supply chain relationship data, are collected from publicly available macroeconomic databases and industry reports through crawling and ETL tools. Service supply information for candidate parks is entered by the park management through a dedicated backend, providing structured forms to ensure data standardization. For subsequent calculations, all information must undergo standardized conversion. Industry codes uniformly adopt national standard codes such as I651; capital units are uniformly in ten thousand yuan; physical space parameters include the area of ​​demand and supply, net area, etc. The system specifies high-level requirements and loads, with clear numerical values ​​and units. Service demand and supply are mapped to a predefined, standardized service item library. For example, a company's described need for loan support is matched with debt financing services in the item library. Crucially, companies are required to clearly mark essential conditions, such as minimum area and specific policy support, and these markings are stored as independent structured identifiers. For example, a company's structured demand data might be: Industry Code: I651, Registered Capital: 500, Required Area: 200, Required Services: High-tech Enterprise Certification Guidance, Credit Matching. Meanwhile, a park's supply data might be: Supported Industries: I65, I64, Leasable Area: 1500, Services Provided: High-tech Certification Guidance, Bank Loan Matching, Talent Recruitment.

[0023] Step S1: Construct an industry knowledge graph based on industry data, and calculate the first matching degree between the target enterprise and the candidate park in the industrial ecosystem dimension based on the industry knowledge graph and the attribute information of the target enterprise. The industry data includes enterprise business registration information, industry classification standards, and macroeconomic relationship data. Step S1 includes steps S101, S102, S103 and S104; Step S101: Based on the industry classification standard data, construct an industry classification node with a hierarchical structure; Based on macroeconomic data, establish relationships between industry classification nodes. These relationships include upstream and downstream supply chain relationships, technological cooperation relationships, or service support relationships. Based on enterprise registration information data, extract enterprise entity and industry attributes, and create corresponding enterprise entity nodes; Step S102: Connect the enterprise entity node to the corresponding industry classification node according to the subordinate relationship between the enterprise entity node and the industry classification node. Integrate industry classification nodes, enterprise entity nodes, and the relationships between industry classification nodes to form an industry knowledge graph; Step S103: Map the industry attributes of the target enterprise to the industry knowledge graph to obtain the corresponding industry classification nodes, which serve as the target industry nodes; The industry attributes of all resident enterprises in the candidate park are mapped to the industry knowledge graph, resulting in a set of industry classification nodes corresponding to all resident enterprises, which serves as the set of industry classification nodes for the candidate park. Step S104: In the industry knowledge graph, calculate the shortest path length of the industry knowledge graph between the target industry node and each node in the industry node set. Based on the shortest path length, the corresponding industry correlation is obtained through a preset inverse proportional function; Based on the industrial relevance of the target industry node and all nodes in the set of industry classification nodes, and combined with the weight of each node, the first matching degree between the target enterprise and the candidate park is obtained through weighted calculation.

[0024] In one embodiment, an industry classification node with a parent-child hierarchical structure is constructed based on the national standard "National Economic Industry Classification". Based on imported macroeconomic relationship data, such as input-output tables, upstream and downstream supply chain and technological cooperation relationships are established between these industry classification nodes. An initial strength value is attached to each relationship edge. This value is specifically set by directly using the direct consumption coefficient between corresponding industry sectors in the national input-output table as the initial strength value of the relationship edge. For example, when the direct consumption coefficient of one industry on another industry is 0.15, the initial strength value of the corresponding relationship edge is set to 0.15. Simultaneously, entity nodes are created for each enterprise based on enterprise business registration information data, thus forming the skeleton of the industry knowledge graph. Subsequently, according to the main industry classification code attribute of each enterprise entity node, the enterprise entity node is attached as a child node to the corresponding industry classification node, completing the graph integration. Then, the main industry code of the target enterprise is mapped to the target industry node T in the graph, and the industry classification nodes corresponding to all resident enterprises in the candidate park are constructed into a set S. Finally, in the industry knowledge graph, the relationship between the target industry node T and each node in set S is calculated. The shortest path length d between the nodes; here, the shortest path length d is defined as the minimum number of relation edges required to connect two nodes in the knowledge graph. All types of relation edges are considered as unit edges of length 1 when calculating the path length. The initial strength value attached to the relation edge is not included in the calculation of this shortest path length. Relationship edges include upstream and downstream supply chains, technical collaboration, service support, and parent-child category relationships. Subsequently, through a preset inverse proportional function R(T, The formula ) = 1 / (d+1) converts the path length into industry relevance. For example, if d = 1 for a direct relevance, then R = 0.5. Simultaneously, based on each node in set S... The registered capital of the corresponding enterprises is now uniformly expressed in ten thousand yuan, which is used to calculate their weight. , The calculation process is as follows: calculate the sum of the registered capital of all enterprises in set S, and the ratio of the registered capital of an enterprise to the sum is the weight. The first match degree between the target company and the candidate park is calculated using the formula: First Match Degree The calculation shows that this value quantifies the potential for collaboration between the target enterprise and the existing industrial ecosystem of the park based on the knowledge graph topology. This process encodes macro-industrial relationships and micro-enterprise entities into a computable graph structure and defines a clear association quantification algorithm and weighted synthesis formula, which transforms the traditional qualitative industrial collaboration judgment into an objective and interpretable numerical score, providing a solid industrial ecosystem dimension basis for core matching decisions.

[0025] Step S2: Generate a structured demand vector for the target enterprise based on its demand information, and generate a structured supply vector based on the service supply information of the candidate parks. Step S2 includes steps S201, S202, S203, S204 and S205; Step S201: Construct a feature list that includes multiple dimensions of matching features, including physical space features, policy service features, and financial service features; Each matching feature is assigned a standardized name, data type, a preset set of quantized values, and a regular expression for extracting information. Step S202: For each matching feature in the feature list, use the regular expression corresponding to the matching feature to match the demand description of the target enterprise. If a match is successful and a text fragment is extracted, a value is obtained from the quantization set corresponding to the matching feature based on the extracted text fragment, which serves as the demand intensity value on the matching feature. Step S203: Assign weight coefficients to each matching feature based on the business attributes of the target enterprise, and use the weight coefficients to weight all demand intensity values ​​to generate a structured demand vector for the target enterprise.

[0026] Step S204: For each matching feature in the feature list, use the regular expression corresponding to the matching feature to match the service supply information of the candidate parks. If a match is successful and a text fragment is extracted, a value is obtained from the quantization set corresponding to the matching feature based on the extracted text fragment, which serves as the supply capacity value on the matching feature. Step S205: Based on the maximum supply capacity value of the same matching feature of all candidate parks, normalize the supply capacity value of each candidate park to generate a structured supply vector for each candidate park.

[0027] The second matching degree is obtained by calculating the demand vector and the supply vector using the cosine similarity formula.

[0028] In one embodiment, the core objective of step S2 is to transform the unstructured service demands of the target enterprise and the textual service supply information of the candidate park into calculable and comparable mathematical vectors, and to achieve a precise quantitative assessment of the degree of service supply and demand matching through vector similarity measurement. Specifically, step S201 involves constructing a standardized list of matching features as a unified framework for subsequent text information parsing and quantification. Each entry in this list corresponds to a matching feature, comprehensively covering key dimensions such as physical space conditions, policy services, and financial services. Each feature must clearly define the following four elements, with a standardized name used to uniquely identify the feature, such as demand area, high-tech enterprise certification services, and debt financing services; data type. To define the format of the feature value, such as floating-point number, Boolean value, or integer; the quantization value set and mapping rules are the conversion standards from text fragments to specific numerical values; information extraction regular expressions are used to identify and extract relevant information patterns from the text; the purpose of this step is to establish a central rule base, providing a deterministic basis for automated text parsing from the step design perspective; from the core design perspective, it ensures that heterogeneous text information from enterprises and parks can be mapped to the same multi-dimensional feature space, thus making their matching and comparison possible. The principle is based on information extraction technology using predefined rules. For example, for the policy service feature of high-tech enterprise certification guidance, a definition is made, specifically including: standardized name; data type; quantization... The value set and mapping rules are defined as follows: 1 is assigned when the keyword is matched and the service is confirmed to be provided or requested; otherwise, 0 is assigned. The regular expression is: =[High-tech Enterprise Certification|High-tech Enterprise Certification; when the keyword "High-tech Enterprise Certification" or "High-tech Enterprise Certification" is matched in the text, the feature value is 1; otherwise, it is 0. All feature regular expressions follow unified rules to ensure universality and compatibility: the numerical part uses [0-9]+[.]{0,1}[0-9], compatible with integers, single-digit and multi-digit decimal representations; the unit part covers full names, abbreviations, and industry common names, such as area units being uniformly [square meters|㎡|square meters|square], length units being [meters|m], and load units being [kN / ㎡|kilonewtons / square meter|kilonewtons]; fuzzy expressions... Processing rules: For ambiguous terms like "about" or "around", a filtering logic of [about|around|approximately]{0,1} is added to the regular expression to extract the core values; for expressions within the 200-300㎡ range, the upper limit of the range is used as the quantification basis; Abnormal scenario handling rules: If the regular expression matching fails or the company does not mention a certain feature, the quantification value defaults to 0; if multiple text fragments are matched, the one with the largest value is taken as the valid data; at the same time, the range division of the quantification value set is based on industry-standard data, such as the required area range referencing the common scale of leasing by electronic technology companies, and the floor load range referencing the industry standards for equipment installation load-bearing capacity; the weight coefficients are calibrated through nearly 3 years of park-company matching data to ensure that the priority of the demand is consistent with the actual needs of the company.

[0029] Taking the demand description of an electronics technology company (main industry classification code I651, registered capital 4.5 million yuan) as an example, the specific demonstration of the regular expression extraction process and vector generation results includes: the company's demand description is that it needs to lease a factory building of about 300 square meters, with a floor height of not less than 4.5 meters and a floor load of not less than 5 kN / square meter, and needs to enjoy high-tech enterprise certification guidance and credit matching services within 5 million yuan. First, the matching features are extracted through regular expressions, specifically including the demand area, floor height, floor load, high-tech enterprise certification service and credit matching service. Among them, the demand area: matching regular expression [approximately|around|about]{0,1}[0-9]+[.]{0,1}[0-9] [square meters|㎡|square meters|square], extract the text fragment 300㎡, corresponding to the set of quantified values ​​{below 100: 1, 100-300: 2, 300-500: 3, above 500: 4}, 300㎡ belongs to the 300-500 range, and the quantified value is 3; Floor height: Matching regular expression [approximately|around|about]{0,1}[0-9]+[.]{0,1}[0-9] [meters|m], extract the text fragment 4.5 meters, corresponding to the quantization value set {below 3: 1, 3-4.5: 2, above 4.5: 3}, according to the interval boundary value, it is classified into the upper limit interval rule. 4.5 meters belongs to the interval above 4.5, and the quantization value is 3; Floor load: Matching regularity [approximately|around|about]{0,1}[0-9]+[.]{0,1}[0-9] [kN / ㎡|kN / m²|kN], extract the text fragment 5kN / ㎡, corresponding to the set of quantized values ​​{below 3: 1, 3-5: 2, above 5: 3}, classify it into the upper limit interval rule according to the interval boundary value, 5kN / ㎡ belongs to the interval above 5, and the quantized value is 3; High-tech enterprise certification service: Matching regular expression [High]tech [Technology] Enterprise Certification | High-tech Enterprise Certification, extracting text fragments for High-tech Enterprise Certification Guidance, corresponding quantitative value set {No demand: 0, Demand: 1}, because the demand text was extracted, the quantitative value is 1; Credit matching service: Matching regular credit | loan | financing | funding support, extracting text fragments for credit matching, corresponding to a quantitative value set {below 1 million: 1, 1 million-5 million: 2, 5 million-10 million: 3, above 10 million: 4}, with the value below 5 million belonging to the 1 million-5 million range, and a quantitative value of 2; According to the preset weight allocation rules (for enterprises with registered capital of less than 5 million yuan, the weight coefficient for financial service features is 1.5, and the weight coefficient for other features is 1.0), the registered capital of this enterprise is 4.5 million yuan, which is less than 5 million yuan. Therefore, the weight coefficient for financial service-related features (credit matching services) is 1.5, and the weight coefficients for other features (demand area, floor height, floor load, and high-tech enterprise certification services) are all 1.0. The dimensions of the structured demand vector correspond to demand area, floor height, floor load, high-tech enterprise certification services, and credit matching services, respectively. The values ​​of each dimension are calculated by quantifying the value and multiplying it by the weight coefficient: demand area dimension 3 × 1.0 = 3.0, floor height dimension 3 × 1.0 = 3.0, floor load dimension 3 × 1.0 = 3.0, high-tech enterprise certification services dimension 1 × 1.0 = 1.0, and credit matching services dimension 2 × 1.5 = 3.0. The final generated structured demand vector is [3.0, 3.0, 3.0, 1.0, 3.0]. If a company's requirement is described as needing to lease approximately 250-300 square meters of factory space and requiring financing support, then: the required area is extracted as 300 square meters using regularization, with a quantified value of 3; the floor height and floor load are not mentioned, so their quantified values ​​are both 0; the financing support is extracted from the credit matching service, with a quantified value of 2, defaulting to a value in the range of 1 million to 5 million. If the company does not specify the amount, it can be supplemented through subsequent interactions or calibrated according to the industry average; if the company's registered capital is 6 million yuan, the weight coefficient of the credit matching service is 1.0, and the final vector is [3.0, 0.0, 0.0, 0.0, 2.0].

[0030] Step S202: Iterate through the feature list defined in step S201. For each matching feature in the list, use its specific regular expression to scan and match the demand description text submitted by the target enterprise and the service supply information text maintained by the candidate park. If a relevant pattern is successfully matched in the text and a text fragment is extracted, the text fragment is converted into a specific numerical value according to the preset quantitative value set and mapping rules of the feature. For enterprise demand, this value is called the demand intensity value; for park supply, this value is called the supply capacity value. The purpose of this design is to realize the automatic and standardized conversion from natural language description to structured numerical values. Its basic principle is pattern matching and table lookup mapping. For example, for the feature of demand area, if the enterprise describes the demand area as approximately 200 square meters, 200 will be extracted and converted into the floating-point number 200.0 as the value of the demand vector in this dimension.

[0031] After generating the basic demand vector, step S203 assigns different weight coefficients to each matching feature dimension according to the business attributes of the target enterprise, such as the registered capital and years of establishment as defined above, and weights the basic vector to form a final weighted demand vector that reflects the characteristics of the enterprise. The purpose is to enable the matching model to dynamically adapt to the actual situation of different enterprises. For example, small and medium-sized enterprises with lower registered capital or shorter establishment period have a higher urgency for financial services and should be given higher weight. The principle is to adjust the relative importance of different demand dimensions by introducing weight coefficients and setting clear weight allocation rules. For example, if the registered capital of the target enterprise is less than 5 million yuan, the weight coefficient of all financial service-related features is set to 1.5; otherwise, the weight coefficient is 1.0. The value of the weighted demand vector in the j-th dimension is the product of the original demand intensity value of that dimension and the corresponding weight coefficient.

[0032] Step S205: After obtaining the original supply vectors of all candidate parks, this step performs normalization processing. For each matching feature dimension j, first find the maximum supply capacity value of all candidate parks in that dimension, and then divide the original value of each park in that dimension by this maximum value to obtain the normalized value. The purpose is to eliminate the influence caused by the difference in units and numerical ranges between different features, so that the values ​​of all feature dimensions are normalized to the [0,1] interval, thereby ensuring the fairness and effectiveness of subsequent similarity calculation. The principle is the maximum value normalization method.

[0033] The second matching degree is calculated by calculating the cosine similarity between the two vectors after obtaining the weighted demand vector of the target enterprise and the normalized supply vector of a candidate park. Specifically, the dot product of the two vectors is calculated and then divided by the product of their respective magnitudes, i.e., the vector lengths. The second matching degree value is between 0 and 1. The higher the value, the higher the degree of fit between the service demand of the target enterprise and the service supply of the candidate park.

[0034] The standard feature template established in step S201 is the cornerstone of the entire process in step S2, ensuring that the symmetrical processing in steps S202 and S204 can be carried out in the same semantic and mathematical space. This is a prerequisite for achieving comparability. The dynamic weighting based on enterprise attributes in step S203 and the supply capacity normalization based on global data in step S205 optimize the vector representation from the demand side and the supply side, respectively, so that the final cosine similarity calculation can more realistically and accurately reflect the matching degree between supply and demand. Step S2 transforms the vague service demands that rely on manual interpretation and subjective judgment in traditional investment promotion, such as the need for strong financial support, into objective, accurate and horizontally comparable quantitative matching scores. This not only greatly improves the efficiency and consistency of the evaluation, but also solves the core defect of existing technologies that cannot perform fine-grained matching of soft service conditions. This second matching degree, combined with the first matching degree generated in step S1, constitutes a comprehensive and multi-dimensional intelligent matching evaluation system from industrial ecosystem collaboration to service supply and demand matching.

[0035] Step S3: The service supply information, first matching degree and second matching degree of the candidate parks are comprehensively processed to generate a comprehensive matching result; Step S3 includes steps S301, S302 and S303; Step S301: Based on historical matching data and the park's investment promotion strategy, determine the recommendation level according to the first matching degree and the second matching degree; The recommendation levels include top recommendation, worth considering, and not recommended; The recommendation level is determined by the following logic: if the first matching degree is greater than or equal to the first threshold and the second matching degree is greater than or equal to the second threshold, it is determined to be a priority recommendation. If the first matching degree is greater than or equal to the third threshold or the second matching degree is greater than or equal to the fourth threshold, it is considered acceptable. If the first matching degree is less than or equal to the fifth threshold and the second matching degree is less than or equal to the sixth threshold, it is determined as not recommended.

[0036] Step S302: Based on preset constraints, candidate parks are screened, and candidate parks that do not meet any of the constraints are removed to obtain the final set of candidate parks. The constraints include physical space constraints, qualification compatibility constraints, and service coverage constraints. Physical space constraints include the target company's required area being less than or equal to the leasable area of ​​the candidate park, the candidate park's net floor height being greater than or equal to the target company's required net height, and the candidate park's floor load being greater than or equal to the target company's required load. The qualification matching constraint is that the target company's main industry classification code belongs to the industry scope supported by the candidate park and the target company's registered capital is greater than or equal to the minimum entry capital requirement of the candidate park. The service coverage constraint is that the service items in the candidate park are marked as necessary physical space conditions for the target enterprise, and the coverage rate of policy services and financial services is greater than or equal to the coverage rate threshold.

[0037] Step S303: If the final candidate park set is empty, generate a comprehensive matching result with no recommended parks. If the final candidate park set is not empty, the candidate parks in the final candidate park set are divided according to the recommendation level determined in step S301. Within the same recommendation level, sort in descending order based on the weighted sum of the first and second matching degrees; If the weighted sums are the same, sort them in descending order according to the first matching degree; If the weighted summation result is the same as the first matching degree, then the coverage of the target enterprise's unmarked needs for the service items of the candidate parks is sorted in descending order, and the sorted list of candidate parks is output as the comprehensive matching result.

[0038] In one embodiment, the core objective of this step is to comprehensively screen, classify, and sort the preliminary matching results calculated in steps S1 and S2, combined with rigid admission criteria and investment promotion strategies, and finally generate an executable, prioritized intelligent recommendation list.

[0039] Step S301: The processing method in this step is to conduct a preliminary rating of candidate parks that simultaneously possess both a first matching degree and a second matching degree according to preset threshold rules. The recommendation level is divided into three levels: priority recommendation, consideration, and not recommended. The specific judgment logic and threshold setting basis are as follows: The judgment logic is based on the statistical quantile of the distribution of the two matching degrees in historical successful matching data and the strategic orientation of park investment promotion, such as placing more emphasis on industrial ecosystem or service facilities. For example, if the first matching degree is greater than or equal to 0.7 (the first threshold is set to 0.7) and the second matching degree is greater than or equal to 0.7 (the second threshold is set to 0.7), it is judged as a priority recommendation. This threshold setting aims to screen out high-quality candidates with both high industrial synergy and service fit. If the first matching degree is large... If the first matching degree is less than or equal to 0.4 (the third threshold is set to 0.4) or the second matching degree is greater than or equal to 0.4 (the fourth threshold is set to 0.4), it is considered acceptable. This rule ensures that parks with a certain degree of matching in any dimension are not overlooked. If the first matching degree is less than or equal to 0.2 (the fifth threshold is set to 0.2) and the second matching degree is less than or equal to 0.2 (the sixth threshold is set to 0.2), it is not recommended. This measure aims to directly exclude candidates with extremely low matching degrees in both aspects, thereby improving processing efficiency. From a design perspective, this step provides coarse-grained grouping for subsequent sorting. From a core design perspective, it introduces rule-based judgment based on historical data and strategic orientation, transforming the abstract matching quality into a clear business level, providing decision-making reference for investment promotion personnel.

[0040] Step S302: The processing method of this step is to filter all candidate parks based on a series of preset, rigid conditions that must be met. These constraints include physical space constraints, qualification matching constraints, and service coverage constraints. Physical space constraints apply: the target company's required area must be less than or equal to the leasable area of ​​the candidate park, the company's required net height must be less than or equal to the park's floor net height, and the company's required load must be less than or equal to the park's floor load. These constraints are designed to ensure basic physical space feasibility.

[0041] Qualification matching constraints: the target company's main industry classification code is ∈ the candidate park, the supported industry range is set, and the company's registered capital is greater than or equal to the park's minimum entry capital requirement, in order to meet the park's industrial positioning and entry threshold.

[0042] Service coverage constraint: For the physical space conditions, policy services, and financial services that the target enterprise marks as necessary in its demand description, the coverage rate of the corresponding service items provided by the candidate park from its service supply information must be greater than or equal to a preset coverage threshold. For example, the coverage threshold is set at 90%. The basis for setting this threshold is to ensure that the core demands of the enterprise are almost completely met. This value can be determined based on the tolerance analysis of the unmet necessary services in the satisfaction survey of historical resident enterprises. This constraint is used to ensure that the core demands of the enterprise are basically met. The service coverage constraint explicitly sets a coverage threshold of 90%. This threshold was determined through a retrospective analysis of 1,000 successful cases of matching between parks and enterprises over the past three years. When the coverage threshold is set at 90%, the satisfaction rate of enterprises after moving in reaches over 92%, which is significantly higher than the 78% satisfaction rate corresponding to the 80% threshold. This balances the protection of core needs with the flexibility of park supply. After removing candidate parks that do not meet any of the above conditions, the remaining parks constitute the final candidate park set. This step implements a veto to ensure that all parks entering the final recommendation list fully meet the hard conditions, thus guaranteeing the feasibility and reliability of the recommendation results.

[0043] Step S303 involves refining and organizing the final candidate park set selected through hard constraints. Based on the recommendation level determined in step S301, the parks are divided into priority recommendation and consideration groups. Then, within the same recommendation level, a multi-level ranking rule is implemented. Specifically, this includes calculating a comprehensive score for each park. The calculation method involves setting a weight coefficient α for the first matching degree and a weight coefficient β for the second matching degree. The product of the first matching degree and weight coefficient α, and the product of the second matching degree and weight coefficient β are added to obtain the comprehensive score, and the parks are ranked in descending order of this score. The weight coefficients α and β are used to balance the importance of industry matching and service matching. For example, based on the investment promotion strategy, α = 0.6 and β = 0.4 are set. The setting is based on the long-term strategic orientation of park investment promotion. If more emphasis is placed on building an industrial ecosystem, α is increased; if more emphasis is placed on service empowerment, β is increased. The conversion rate of historical matching results can be optimized and calibrated through A / B testing, and the matching degree of the two dimensions is weighted and integrated.

[0044] If the overall scores are the same, they will be sorted in descending order according to the first matching degree. The principle is that, in the case of the same total score, priority will be given to parks with stronger industrial ecosystem synergy. If the overall score and the first matching degree are the same, the coverage rate of the candidate park service items for other needs not marked as necessary by the target enterprise is calculated, and they are sorted in descending order according to this coverage rate. The principle is that when the core conditions and the main matching degree indicators, namely the overall score and the first matching degree, are the same and the priority cannot be distinguished, the ability to meet the additional needs (non-essential services) is used as the final deciding factor. Finally, the output is a list of candidate parks sorted according to this rule as the comprehensive matching result. If the final set of candidate parks is empty, then no recommended parks will be output.

[0045] Steps S301, S302, and S303 constitute a progressive decision-making pipeline of hierarchical > filtering > fine-grained ranking. Step S301 uses soft scoring for rapid initial screening and classification, improving the intuitiveness and strategic nature of the decision. Step S302 applies hard conditions for absolute filtering, ensuring the minimum feasibility of the results. The order of these two steps avoids wasting ranking resources on obviously non-compliant candidates. Step S303, within the final pool of high-quality candidates, introduces a weighted comprehensive score and a multi-level parallel ranking system to ensure the output list has the highest matching degree. In terms of superiority and stability, step S3 deeply integrates the quantitative indicators of first and second matching, which are calculated in the previous steps and focus on a single dimension, with the rigid constraints and business strategies that are indispensable in business practice, as well as the weight coefficients α and β. This generates a result that is both scientifically quantifiable and practical, and reflects business priorities. This completely solves the one-sided problem of traditional recommendation methods that either only focus on hard conditions and ignore matching degree, or only calculate matching degree and ignore feasibility or strategic orientation. Ultimately, it realizes the leap of intelligent matching from calculation to reliable decision-making.

[0046] Example 2, refer to Figure 2 In another embodiment of the present invention, which differs from the first embodiment, an intelligent matching system for enterprises entering a park is provided, including a construction module, a calculation module and a matching module; The module constructs an industry knowledge graph based on industry data, and calculates the first matching degree between the target enterprise and the candidate park based on the industry knowledge graph and the attribute information of the target enterprise. The calculation module generates a structured demand vector for the target enterprise based on its demand information, and generates a structured supply vector based on the service supply information of the candidate parks. Calculate the similarity between the demand vector and the supply vector to obtain the second matching degree between the target enterprise and the candidate park; The matching module comprehensively processes the service supply information, first matching degree, and second matching degree of the candidate parks to generate a comprehensive matching result.

[0047] The service items are marked as necessary physical space conditions for the target enterprise and In one embodiment, an industry knowledge graph is constructed, and the first matching degree in the industrial chain collaboration dimension is calculated based on the graph and the attribute information of the target enterprise. The unstructured demand of the enterprise and the service supply information of the park are transformed into structured demand vectors and supply vectors, respectively, and the second matching degree in the service supply and demand dimension is obtained by calculating their similarity. The comprehensive service supply information, the first matching degree, and the second matching degree are then processed globally to generate a comprehensive matching result. The core of this design is to conduct a dual evaluation of industrial ecosystem adaptation and service supply and demand fit, which solves the limitation of traditional methods that only focus on a single static label. The industry knowledge graph quantifies the industrial association and niche value of enterprises, realizing the depth of matching; vectorization enables refined matching of diversified and personalized service needs, improving the accuracy of matching; and the final comprehensive decision ensures that the matching result not only meets the rigid constraints of enterprises, but also maximizes industrial collaboration and comprehensive service satisfaction. Thus, the overall intelligent, precise, and ecological investment promotion matching is achieved, effectively improving the park's operational efficiency and the quality of industrial agglomeration.

[0048] This invention addresses the shortcomings of traditional methods that rely heavily on static labels based on enterprise classification and size for superficial screening. These methods fail to assess the potential for industrial synergy after enterprise entry and struggle to quantify the matching degree between service supply and demand. This solution, by constructing an industry knowledge graph and calculating a first-degree matching, achieves a quantitative assessment of the industrial ecosystem's relevance for the first time, resolving the pain point of existing technologies lacking in-depth analysis along the industrial chain. Furthermore, by converting unstructured text into structured vectors and calculating a second degree matching, it enables refined matching of personalized and diversified service needs, overcoming the limitation of traditional label comparison methods in capturing core demands. Finally, through rigid constraints and multi-dimensional comprehensive ranking, this solution ensures that the matching results maximize the dual benefits of industrial synergy and service adaptation while meeting the rigid needs of enterprises. This comprehensively surpasses existing technologies in improving the accuracy of investment attraction, enterprise satisfaction, and the quality of the industrial ecosystem within the park.

[0049] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media containing computer-usable program code. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0050] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the protection scope of the present invention.

Claims

1. A method for intelligent matching of enterprises entering a park, characterized in that, Includes the following steps, Step S1: Construct an industry knowledge graph based on industry data, and calculate the first matching degree between the target enterprise and the candidate park based on the industry knowledge graph and the attribute information of the target enterprise; Step S2: Generate a structured demand vector for the target enterprise based on its demand information, and generate a structured supply vector based on the service supply information of the candidate parks. Calculate the similarity between the demand vector and the supply vector to obtain the second matching degree between the target enterprise and the candidate park; Step S3: Perform comprehensive processing on the service supply information of the candidate parks, as well as the first matching degree and the second matching degree, to generate a comprehensive matching result.

2. The intelligent matching method for enterprises entering a park as described in claim 1, characterized in that, Obtain attribute information, demand information, industry data of target enterprises, and service supply information of candidate parks; Attribute information includes the target company's industry attributes and business attributes; Among them, the industry attribute includes the target company's main industry classification code, and the business attribute includes the target company's registered capital and years of establishment; The requirements information includes a description of the target company's service needs; The demand description includes the demand for physical space conditions, policy services, and financial services; The physical space requirements include the target company's required area, net height, and load capacity, and the target company must specify the necessary physical space requirements, policy services, and financial services in its requirements description. Service supply information includes the resource conditions, service items, supported industry scope, and minimum capital requirements for candidate parks; The resource conditions include the available physical space conditions of the candidate park, which include the leasable area of ​​the candidate park, the net floor height of the candidate park, and the floor load of the candidate park. The service items include policy services, financial services, talent services, and marketing services, and cover the needs for physical space conditions, policy services, and financial services.

3. The intelligent matching method for enterprises entering a park as described in claim 2, characterized in that, Step S1: Construct an industry knowledge graph based on industry data, and calculate the first matching degree between the target enterprise and the candidate park in the industrial ecosystem dimension based on the industry knowledge graph and the attribute information of the target enterprise. The industry data includes enterprise business registration information, industry classification standards, and macroeconomic relationship data. Step S1 includes steps S101, S102, S103 and S104; Step S101: Based on the industry classification standard data, construct an industry classification node with a hierarchical structure; Based on macroeconomic data, establish relationships between industry classification nodes. These relationships include upstream and downstream supply chain relationships, technological cooperation relationships, or service support relationships. Based on enterprise registration information data, extract enterprise entity and industry attributes, and create corresponding enterprise entity nodes; Step S102: Connect the enterprise entity node to the corresponding industry classification node according to the subordinate relationship between the enterprise entity node and the industry classification node; Integrate industry classification nodes, enterprise entity nodes, and the relationships between industry classification nodes to form an industry knowledge graph; Step S103: Map the industry attributes of the target enterprise to the industry knowledge graph to obtain the corresponding industry classification node, which serves as the target industry node; The industry attributes of all resident enterprises in the candidate park are mapped to the industry knowledge graph, resulting in a set of industry classification nodes corresponding to all resident enterprises, which serves as the industry classification node set of the candidate park. Step S104: In the industry knowledge graph, calculate the shortest path length of the industry knowledge graph between the target industry node and each node in the industry node set; Based on the shortest path length, the corresponding industry correlation is obtained through a preset inverse proportional function; Based on the industry correlation between the target industry node and all nodes in the industry classification node set, and combined with the weight of each node, the first matching degree between the target enterprise and the candidate park is obtained through weighted calculation.

4. The intelligent matching method for enterprises entering a park as described in claim 3, characterized in that, Step S2: Generate a structured demand vector for the target enterprise based on its demand information, and generate a structured supply vector based on the service supply information of the candidate parks. Step S2 includes steps S201, S202, S203, S204 and S205; Step S201: Construct a feature list that includes multiple dimensions of matching features, including physical space features, policy service features, and financial service features; Each matching feature is assigned a standardized name, data type, a preset set of quantized values, and a regular expression for extracting information. Step S202: For each matching feature in the feature list, use the regular expression corresponding to the matching feature to match the demand description of the target enterprise; If a match is successful and a text fragment is extracted, a value is obtained from the quantization set corresponding to the matching feature based on the extracted text fragment, which is used as the demand intensity value on the matching feature. Step S203: Assign weight coefficients to each matching feature based on the business attributes of the target enterprise, and use the weight coefficients to weight all demand intensity values ​​to generate a structured demand vector for the target enterprise.

5. The intelligent matching method for enterprises entering a park as described in claim 4, characterized in that, Step S204: For each matching feature in the feature list, use the regular expression corresponding to the matching feature to match the service supply information of the candidate park. If a match is successful and a text fragment is extracted, a value is obtained from the quantization set corresponding to the matching feature based on the extracted text fragment, which is used as the supply capacity value on the matching feature. Step S205: Based on the maximum supply capacity value of the same matching feature of all candidate parks, normalize the supply capacity value of each candidate park to generate a structured supply vector for each candidate park.

6. The intelligent matching method for enterprises entering a park as described in claim 5, characterized in that, The second matching degree is obtained by calculating the demand vector and the supply vector using the cosine similarity formula.

7. The intelligent matching method for enterprises entering a park as described in claim 6, characterized in that, Step S3: Perform comprehensive processing on the service supply information of the candidate parks, as well as the first matching degree and the second matching degree, to generate a comprehensive matching result; Step S3 includes steps S301, S302 and S303; Step S301: Based on historical matching data and the park's investment promotion strategy, determine the recommendation level according to the first matching degree and the second matching degree; The recommendation levels include top recommendation, worth considering, and not recommended; The recommendation level is determined by the following logic: if the first matching degree is greater than or equal to the first threshold and the second matching degree is greater than or equal to the second threshold, it is determined to be a priority recommendation. If the first matching degree is greater than or equal to the third threshold or the second matching degree is greater than or equal to the fourth threshold, it is considered acceptable. If the first matching degree is less than or equal to the fifth threshold and the second matching degree is less than or equal to the sixth threshold, it is determined as not recommended.

8. The intelligent matching method for enterprises entering a park as described in claim 7, characterized in that, Step S302: Based on preset constraints, candidate parks are screened, and candidate parks that do not meet any of the constraints are removed to obtain the final set of candidate parks. The constraints include physical space constraints, qualification compatibility constraints, and service coverage constraints. Physical space constraints include the target company's required area being less than or equal to the leasable area of ​​the candidate park, the candidate park's net floor height being greater than or equal to the target company's required net height, and the candidate park's floor load being greater than or equal to the target company's required load. The qualification matching constraint is that the target company's main industry classification code belongs to the industry scope supported by the candidate park and the target company's registered capital is greater than or equal to the minimum entry capital requirement of the candidate park. The service coverage constraint is that the service items in the candidate park are marked as necessary physical space conditions for the target enterprise, and the coverage rate of policy services and financial services is greater than or equal to the coverage rate threshold.

9. The intelligent matching method for enterprises entering a park as described in claim 8, characterized in that, Step S303: If the final candidate park set is empty, generate a comprehensive matching result with no recommended parks. If the final candidate park set is not empty, the candidate parks in the final candidate park set are divided according to the recommendation level determined in step S301. Within the same recommendation level, sort in descending order based on the weighted sum of the first and second matching degrees; If the weighted summation results are the same, they are sorted in descending order according to the first matching degree; If the weighted summation result is the same as the first matching degree, then the coverage of the target enterprise's unmarked needs for the service items of the candidate parks is sorted in descending order, and the sorted candidate park list is output as the comprehensive matching result.

10. A smart matching system for enterprises entering a park, which is applied to the smart matching method for enterprises entering a park as described in any one of claims 1-9, characterized in that, It includes a building module, a calculation module, and a matching module; The module constructs an industry knowledge graph based on industry data, and calculates the first matching degree between the target enterprise and the candidate park based on the industry knowledge graph and the attribute information of the target enterprise. The calculation module generates a structured demand vector for the target enterprise based on its demand information, and generates a structured supply vector based on the service supply information of the candidate parks. Calculate the similarity between the demand vector and the supply vector to obtain the second matching degree between the target enterprise and the candidate park; The matching module comprehensively processes the service supply information of the candidate parks, the first matching degree, and the second matching degree to generate a comprehensive matching result.

Images