A method for constructing a park-enterprise association relationship, a medium and equipment

Abstract

The present application relates to the technical field of data processing, and more particularly to a park and enterprise association relationship construction method, medium and equipment, which identifies and merges the same park from different data sources, solves the problem of incomplete coverage of a single data source and repeated data conflict, improves the identification accuracy of the park entity, determines the target generation strategy through the data source type and data completeness, generates the initial boundary, so that the generation of the park boundary can be self-adapted to different types and richness of the data source, improves the robustness of the park boundary generation, calculates the minimum convex polygon or minimum circumscribed rectangle of the enterprise coordinate point which has been determined to be located inside the initial boundary or in the boundary buffer area, and fine-tunes the buffer outward, improves the smoothness and inclusiveness of the target boundary, and establishes the association relationship according to the target boundary and the to-be-matched enterprise coordinate data and generates the association confidence, and realizes the accurate association of the enterprise and the park.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a method, medium, and equipment for constructing a relationship between a park and an enterprise. Background Technology

[0002] The construction of relationships between industrial parks and enterprises typically relies on single data sources such as business registration addresses and map POIs, establishing the connection by determining whether the spatial distance between the enterprise's coordinates and the park's center point is less than a preset threshold. However, a single data source cannot fully cover the actual geographical scope of the park, and there is a lack of effective fusion mechanisms between different data sources such as official boundary documents, enterprise location coordinates, and remote sensing imagery. This results in low accuracy of the generated park boundaries, failing to accurately reflect the park's physical spatial distribution. Furthermore, existing methods often use circular or rectangular areas with fixed radii when generating park boundaries, failing to dynamically adjust the boundary generation strategy based on the actual data corresponding to the park. When data sources are insufficient, the boundaries are too coarse; when data sources are abundant, high-precision data cannot be fully utilized. In addition, the matching of enterprise coordinates with the park relies solely on point-area inclusion or a single distance threshold, lacking a multi-level progressive matching strategy. This can easily lead to missed matches for enterprises whose coordinates deviate from the park boundaries due to surveying errors or GPS offsets, making it difficult to support the subsequent business decision-making requirements for assessing the reliability of the association.

[0003] Therefore, how to make full use of multi-source heterogeneous data to dynamically generate high-precision park boundaries in order to achieve accurate association between enterprises and parks has become an urgent problem to be solved. Summary of the Invention

[0004] To address the aforementioned technical problems, the present invention provides a method for constructing a relationship between industrial parks and enterprises. This method includes the following steps: S1. Based on the park name data and park spatial reference data within the target geographical area, identify and merge the same park from different data sources to obtain at least one undetermined park.

[0005] S2, for each pending zone, determine the target generation strategy from the preset boundary generation strategy based on the data source type and data completeness corresponding to the pending zone, and generate the initial boundary of the pending zone according to the target generation strategy.

[0006] S3. For each enterprise coordinate point in the coordinate data of the park-related enterprises corresponding to the undetermined park, calculate the spatial distance from the enterprise coordinate point to the initial boundary, and determine whether the enterprise coordinate point is located inside the initial boundary or within the boundary buffer area based on the spatial distance.

[0007] S4. Based on the coordinates of all enterprises located inside the initial boundary or within the boundary buffer area, the initial boundary is spatially adjusted to obtain the target boundary of the proposed park. The target boundary is a closed polygon composed of an ordered sequence of boundary vertex coordinates, which is used to characterize the actual geographical range of the proposed park.

[0008] S5: Based on the target boundaries of all pending parks and the coordinate data of the enterprises to be matched, establish a relationship between each enterprise to be matched and at least one pending park, and generate a relationship confidence score.

[0009] The present invention also provides a non-transitory computer-readable storage medium storing at least one instruction or at least one program, wherein the at least one instruction or at least one program is loaded and executed by a processor to realize the above-mentioned method for constructing the relationship between the park and the enterprise.

[0010] The present invention also provides an electronic device, including a processor and the aforementioned non-transitory computer-readable storage medium.

[0011] This invention has at least the following beneficial effects: By identifying and merging the same park from different data sources based on park name data and spatial reference data, it effectively integrates multi-source heterogeneous data, solves the problems of incomplete coverage and duplicate data conflicts of a single data source, and improves the accuracy of park entity identification; By determining the target generation strategy and generating the initial boundary from the preset boundary generation strategy based on the data source type and data completeness, the generation of park boundaries can adaptively adapt to data sources of different types and richness, improving the robustness of park boundary generation; By calculating the minimum convex hull polygon or minimum outer rectangle of the enterprise coordinate points that have been determined to be located inside the initial boundary or within the boundary buffer area and finely adjusting the buffer outward, the smoothness and inclusiveness of the target boundary are improved, and the association relationship is established between the target boundary and the coordinate data of the enterprise to be matched and the association confidence is generated, realizing the accurate association between enterprises and parks. Attached Figure Description

[0012] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0013] Figure 1 This is a flowchart illustrating a method for constructing a relationship between a park and an enterprise, as provided in Embodiment 1 of the present invention. Detailed Implementation

[0014] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on 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.

[0015] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It is understood that, where appropriate, the terms used to distinguish similar objects can be interchanged so that the invention can also be implemented in other embodiments besides the illustrated or described embodiments. Furthermore, the terms "including," "having," and any variations are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or server that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to these processes, methods, products, or devices.

[0016] Example 1 This first embodiment provides a method for constructing the relationship between a park and its enterprises, such as... Figure 1 As shown, the method for constructing the relationship between the park and enterprises includes the following steps: S1. Based on the park name data and park spatial reference data within the target geographical area, identify and merge the same park from different data sources to obtain at least one undetermined park.

[0017] The target geographic area is the specific spatial range within which the relationship between the park and the enterprise needs to be established. This includes, but is not limited to, administrative division codes, latitude and longitude rectangles, or polygonal boundaries in GeoJSON format, used to define the spatial scope for subsequent collection and processing of park and enterprise data. A park is a concentrated area entity within the target geographic area with certain geographical boundaries, providing a space for enterprises to locate and conduct production and business activities. This includes, but is not limited to, high-tech industrial development zones, economic and technological development zones, industrial parks, science parks, software parks, logistics parks, and cultural and creative industrial parks.

[0018] In this embodiment, park-related information is obtained through various data sources, including official data, map POI data, business registration text data, remote sensing interpretation data, and OSM road network data. By integrating these heterogeneous data sources, the incomplete coverage and positioning errors of a single data source can be mitigated. The redundancy and complementarity between data sources improve the accuracy of park identification and boundary generation. Specifically, government official data, such as park lists and official boundary documents published by departments like the Natural Resources Bureau and Science and Technology Bureau, have high authority and credibility, with an initial credibility of 0.9. Map POI data, provided by various commercial map open platforms, includes park points of interest data such as names, addresses, coordinates, and category tags, with an initial credibility of 0.7. Business registration text data, consisting of registered and operating addresses from enterprise credit databases, allows for the extraction of park names through natural language processing, with an initial credibility of 0.5. Remote sensing interpretation data, extracted from satellite imagery combined with a semantic segmentation model, shows contiguous building outlines and can serve as a reference for park boundaries, with an initial credibility of 0.3. OSM road network data is road network vector data from open-source maps. It serves as a constraint benchmark for boundary clipping and is not included in the confidence assignment in this embodiment.

[0019] Correspondingly, park name data used to identify the park is collected from various data sources. Each park name data entry contains at least one name text string, and may also include variations, aliases, abbreviations, or the official full name. Park name data can come from different data sources such as official directories, map POIs, and park fields in business registration addresses. Park name data from different data sources undergoes standardization processing, such as removing redundant suffixes like "industrial park" and "science park," and removing parenthetical remarks, to eliminate differences in expression between different data sources and improve the accuracy of text similarity calculations.

[0020] Spatial reference data representing the park's geographical location is collected from various data sources, including at least the coordinates of a spatial reference point, which may be the geographic center or a boundary vertex. Optionally, a sequence of vertex coordinates of the park's boundary polygon may also be included. The geographic center coordinates of the park can be derived from a map POI location or the center location published in an official directory. Boundary vertices are the coordinates of one or more vertices on the park's boundary polygon, used to accurately describe the park's geometry.

[0021] Because the same park entity may present different but highly similar names and slightly offset spatial coordinates in different data sources, relying solely on a single dimension such as name or distance can easily lead to incorrect or missed merging. Therefore, this embodiment integrates park name data and park spatial reference data, employing a dual constraint of name similarity and spatial similarity for park identification and merging. Combined with an iterative merging strategy, several initial parks describing the same park from multi-source heterogeneous data are grouped into the same pending park. Furthermore, after multiple rounds of iterative merging, each pending park finally determined corresponds to a unique park entity, and is associated with multi-source fusion data source information, reference spatial location, and collected enterprise coordinate data, serving as the basic unit for subsequent boundary generation and enterprise matching.

[0022] In one specific implementation, the park name data includes the name text of several initial parks, and the park spatial reference data includes the coordinates of several spatial reference points of the initial parks. S1 includes the following steps: S11, build a set of parks based on all the initial parks.

[0023] S12, for any two parks in the park set, calculate the text similarity between the name texts of the two parks, and the spatial distance between the spatial reference point coordinates of the two parks.

[0024] S13, when the text similarity is greater than the first preset threshold and the spatial distance is less than the second preset threshold, the two parks are merged into one undetermined park, and the two parks in the park set are replaced with the undetermined park to obtain the updated park set.

[0025] S14. For the updated set of parks, repeat steps S12 to S13 until no two parks in the updated set of parks simultaneously satisfy the condition that the corresponding text similarity is greater than the first preset threshold and the corresponding spatial distance is less than the second preset threshold. When the repeated execution stops, each park in the set of parks is determined as a pending park.

[0026] In this process, all initial park records collected from multiple data sources are placed into a park set, and each record contains its name text and spatial reference point coordinates.

[0027] Those skilled in the art will recognize that any existing text similarity calculation method falls within the protection scope of this invention, such as edit distance similarity and the Jaccard coefficient, which will not be elaborated upon here. Similarly, those skilled in the art will recognize that any existing spatial distance calculation method falls within the protection scope of this invention, such as the Havesing formula, which will not be elaborated upon here.

[0028] When both the name similarity and spatial proximity conditions are met, two parks are considered to point to the same park entity with a very high probability and can be merged. Therefore, if the text similarity is greater than a first preset threshold and the spatial distance is less than a second preset threshold, the two parks are merged into a new pending park. The merging method is as follows: a new park record is created, whose name can be any one of the options or a longer name, and whose spatial reference point can be one of the options or the center of both. The merged data source list is also recorded. Then, the two parks in the original park set are replaced with the new pending park to obtain an updated park set.

[0029] The process iterates until no pair of parks meets the merging condition in a complete traversal. At this point, each remaining park in the park set is a pending park, which serves as the basic unit for subsequent boundary generation and matching with enterprises.

[0030] The specific values ​​of the first and second preset thresholds can be set by the implementer according to the actual situation. For example, in this embodiment, the first preset threshold is 0.85 and the second preset threshold is 200 meters.

[0031] The above-mentioned method identifies the same park from different data sources based on dual constraints of name similarity and spatial distance, and adopts an iterative merging strategy to gradually merge the initial parks that meet the conditions. This method accurately aggregates records pointing to the same real-world park from multi-source heterogeneous data, eliminates data redundancy and conflicts, ensures the completeness and uniqueness of the set of parks to be determined, and provides a reliable basic entity for subsequent boundary generation and enterprise matching.

[0032] S2, for each pending zone, determine the target generation strategy from the preset boundary generation strategy based on the data source type and data completeness corresponding to the pending zone, and generate the initial boundary of the pending zone according to the target generation strategy.

[0033] Among them, the data source type refers to the variety of sources of the original data associated with the proposed park. Different types of data sources provide different boundary precision and initial reliability. The data completeness is the number of coordinate data of park-related enterprises collected for the proposed park, that is, the number of enterprises that have been identified as having settled in the park. It is used to measure whether the enterprise coordinate clustering method can be used to generate boundaries. Correspondingly, the more data there is, the more reliable the clustering results are.

[0034] This embodiment adaptively selects the most suitable boundary generation strategy based on the data source type of the proposed park and the amount of coordinate data of the park-related enterprises, thereby generating reasonable initial boundaries under various data conditions.

[0035] In one specific embodiment, S2 includes the following steps: S21, for each pending zone, determine whether the data source type corresponding to the pending zone is an official boundary file.

[0036] S22, If it is an official boundary file, then the target generation strategy is determined to be the first preset strategy, wherein the first preset strategy is used to directly import the boundary data in the official boundary file as the initial boundary.

[0037] S23. If it is not an official boundary file, obtain the coordinate data of the park-related enterprises collected by the park to be determined, and count the number of enterprise coordinate data in the park-related enterprise coordinate data as the data completeness.

[0038] S24. If the data completeness is greater than the preset quantity threshold, the target generation strategy is determined to be the second preset strategy. The second preset strategy is used to cluster the coordinate data of related enterprises in the park and generate the initial boundary based on the convex hull of the clustering results.

[0039] S25, if the data completeness is less than or equal to the preset quantity threshold, the target generation strategy is determined to be the third preset strategy, wherein the third preset strategy is used to generate the initial boundary by boundary fusion based on remote sensing building outline data and / or map POI data.

[0040] If the data source for the proposed park is an official boundary file, the initial boundary can be obtained directly using the highest precision data. If it is not an official boundary but there are enough coordinates of the resident companies, the actual location data of the companies can be used to generate an initial boundary that fits the actual distribution through clustering and convex hull. If the data of resident companies is insufficient, the initial boundary can be generated by boundary fusion using remote sensing building outlines and / or map POI data to ensure that a usable initial boundary can be generated regardless of the data source type and the completeness of the data.

[0041] Specifically, in the first preset strategy, boundary data, such as the polygon vertex coordinate sequence, is directly read from the official boundary file as the initial boundary.

[0042] The locations of businesses within the park accurately reflect its actual usage area, and clustering and convex hulls can generate a geometry that closely reflects the distribution of businesses. Outward buffering can accommodate peripheral businesses, while road network trimming prevents boundaries from crossing roads, conforming to real-world geographical logic. Therefore, when the number of business coordinate data exceeds a preset threshold, a second preset strategy is adopted. Specifically, in the second preset strategy, DBSCAN spatial clustering is performed on all business coordinate points associated with the proposed park. No preset cluster number is required, and noise points can be automatically identified. The convex hull polygon of each cluster is extracted, and the convex hull polygon is buffered outward by a first preset distance. The buffered convex hull intersects with the OSM road network data, and the portion crossing roads is trimmed to obtain the initial boundary.

[0043] In the absence of sufficient enterprise coordinates, remote sensing building outlines and map POIs can provide spatial clues. Fusion of these can generate a relatively reasonable boundary, avoiding missing boundaries. Therefore, when the amount of enterprise coordinate data does not exceed a preset threshold, a third preset strategy is adopted. Specifically, in this strategy, remote sensing building outline data and map POI data for the location of the to-be-determined park are collected, such as park name POIs and surrounding building faces. A boundary polygon is generated through voting or weighted fusion. Specifically, the remotely extracted building outlines are aggregated outwards, combined with POI locations to generate a Thiessen polygon or buffer zone, and then the union is taken to obtain the initial boundary.

[0044] The specific value of the preset quantity threshold can be set by the implementer according to the actual situation. For example, in this embodiment, the preset quantity threshold is 15.

[0045] The specific value of the first preset distance can be set by the implementer according to the actual situation. For example, in this embodiment, the first preset distance is 20 meters.

[0046] In one specific implementation, generating an initial boundary based on the convex hull of the clustering results includes: S241, extract the convex hull polygons of each cluster in the clustering results.

[0047] S242, acquire road network data, buffer the convex hull polygon outward by a first preset distance, and then intersect and trim it with the road network data to obtain the initial boundary.

[0048] Specifically, the minimum convex hull is calculated using the Graham scan method or the Andrew algorithm to obtain polygons composed of ordered vertices, and the convex hull polygons of each cluster are extracted. The GIS buffer algorithm is used to buffer the convex hull polygons outward by a first preset distance, and the road line features intersecting with the buffered convex hull in the OSM road network data are obtained. The buffered convex hull is then subjected to a difference operation with the OSM road network data, or the convex hull is cut by road network lines to remove unreasonable parts located on the other side of the road, retaining the areas that intersect with the road and are continuous, thus obtaining the initial boundary.

[0049] As described above, by adapting the target generation strategy from the preset boundary generation strategy based on the data source type and data completeness corresponding to the pending park, a smooth degradation from high-precision data to low-precision data is achieved, which enables the generation of park boundaries to flexibly adapt to different data conditions and improves the robustness and universality of park boundary generation.

[0050] S3. For each enterprise coordinate point in the coordinate data of the park-related enterprises corresponding to the undetermined park, calculate the spatial distance from the enterprise coordinate point to the initial boundary, and determine whether the enterprise coordinate point is located inside the initial boundary or within the boundary buffer area based on the spatial distance.

[0051] Among them, the coordinate data of enterprises associated with the park is a set of geographical coordinates of enterprises that have been confirmed to be located in the undetermined park.

[0052] By calculating the shortest spatial distance from the coordinates of each enterprise associated with the proposed park to the initial boundary, and based on the relationship between this shortest spatial distance and the preset value 0 and the preset buffer distance threshold, the positional relationship between the enterprise coordinates and the initial boundary is divided into three categories: inside the boundary, inside the boundary buffer, and outside the boundary buffer. This accurately determines which enterprises are truly located inside the park boundary, and identifies enterprises whose coordinates fall outside the park but should actually belong to the park due to coordinate acquisition errors, surveying deviations, or the enterprise being located at the edge of the boundary. These enterprises are then included in the boundary buffer for boundary correction in subsequent steps, avoiding the problems of missed matching and loss of boundary adjustment information.

[0053] In one specific embodiment, S3 includes the following steps: S31, For any enterprise coordinate point in the coordinate data of the park-related enterprises corresponding to the undetermined park, calculate the shortest spatial distance from the enterprise coordinate point to the initial boundary.

[0054] S32, when the shortest spatial distance is equal to 0, it is determined that the enterprise's coordinate point is located inside the initial boundary.

[0055] S33, when the shortest spatial distance is greater than 0 and less than or equal to the preset buffer distance threshold, it is determined that the enterprise coordinate point is located within the boundary buffer area of ​​the initial boundary.

[0056] S34, when the shortest spatial distance is greater than the buffer distance threshold, it is determined that the enterprise's coordinate point is outside the boundary buffer area of ​​the initial boundary.

[0057] Specifically, for each enterprise coordinate point, the perpendicular distance from that enterprise coordinate point to each edge of the initial boundary is calculated, and the minimum value among them is taken as the shortest spatial distance.

[0058] In this embodiment, since both the enterprise coordinates and the boundary vertices are latitude and longitude coordinates, the curvature of the Earth needs to be considered. The Havesing formula can be used to calculate the spherical distance, which represents the corresponding spatial distance.

[0059] The specific value of the preset buffer distance threshold can be set by the implementer according to the actual situation. For example, in this embodiment, the preset buffer distance threshold is 50 meters.

[0060] The above-mentioned method calculates the shortest spatial distance from each enterprise's coordinate point to the initial boundary, and makes hierarchical judgments based on the relationship between the distance value and the preset value 0 and the buffer distance threshold. This method accurately quantifies and classifies the positional relationship between the enterprise's coordinate points and the initial boundary of the park, providing clear hierarchical data basis for subsequent boundary correction and improving the flexibility and accuracy of boundary adjustment.

[0061] S4. Based on the coordinates of all enterprises located inside the initial boundary or within the boundary buffer area, the initial boundary is spatially adjusted to obtain the target boundary of the proposed park. The target boundary is a closed polygon composed of an ordered sequence of boundary vertex coordinates, which is used to characterize the actual geographical range of the proposed park.

[0062] By calculating the minimum convex hull polygon or minimum outer rectangle of the enterprise coordinate points that have been determined to be located inside the initial boundary or within the boundary buffer area, and finely adjusting the buffer outward, a closed polygon that closely matches the actual distribution of the enterprise and is smooth and reasonable is generated as the final target boundary, thus making up for the deviation and insufficient accuracy of the initial boundary.

[0063] In one specific embodiment, S4 includes the following steps: S41, collect all enterprise coordinate points that are determined to be located inside the initial boundary or within the boundary buffer area, and form a valid point set.

[0064] S42, calculates the minimum convex hull polygon or minimum bounding rectangle of the valid point set.

[0065] S43, buffer the smallest convex hull polygon or the smallest outer rectangle outward by a preset fine-tuning distance to obtain the adjusted closed polygon.

[0066] S44, the adjusted closed polygon is taken as the target boundary of the area to be determined, wherein the coordinate sequence of the boundary vertices of the target boundary is arranged in clockwise or counterclockwise order.

[0067] One approach uses either the Graham scan method or the Andrew algorithm to sort the points in the valid point set by polar angle and calculate the convex hull, obtaining a vertex sequence, and then deriving the minimum convex hull polygon, suitable for representing the approximate outer contour of a park. Another approach involves finding the minimum and maximum longitude and latitude values ​​in the valid point set, constructing a rectangle based on these values, and obtaining the minimum outer rectangle, suitable for scenarios with relatively regular point distribution or where high precision is not required. The choice between the minimum convex hull polygon and the minimum outer rectangle can be made according to actual needs.

[0068] For each edge of the minimum convex hull polygon or the minimum outer rectangle, translate it outward along the normal direction by a preset fine-tuning distance, and connect adjacent edges with arcs to form a new closed polygon, which serves as the target boundary of the proposed park. This accommodates situations where the point coordinates deviate slightly from their actual positions due to surveying errors, GPS drift, etc., and makes the boundary slightly loose to prevent edge enterprises from being misjudged as outside in the future.

[0069] The specific value of the preset fine-tuning distance can be set by the implementer according to the actual situation. For example, in this embodiment, the preset fine-tuning distance is 20 meters.

[0070] The above-mentioned method calculates the minimum convex hull polygon or minimum outer rectangle of the enterprise coordinate points that have been determined to be located inside the initial boundary or within the boundary buffer area, and finely adjusts the buffer outward to generate a closed polygon that closely matches the actual distribution of enterprises and is smooth and reasonable. This improves the smoothness and inclusiveness of the boundary and provides a reliable and efficient geometric basis for the subsequent spatial matching of enterprises and the park.

[0071] S5: Based on the target boundaries of all pending parks and the coordinate data of the enterprises to be matched, establish a relationship between each enterprise to be matched and at least one pending park, and generate a relationship confidence score.

[0072] Among them, the companies to be matched are those whose affiliation to the park has not yet been determined.

[0073] This embodiment adopts a multi-level progressive spatial matching strategy. Based on the spatial relationship between the coordinate data of the enterprise to be matched and the target boundaries of each park, it prioritizes high-precision point-to-surface inclusion matching, followed by buffer proximity matching, and finally nearest neighbor fallback matching. At the same time, it integrates the enterprise's own confidence, park confidence, and spatial matching confidence to output a comprehensive association confidence, which improves the rationality and accuracy of the matching.

[0074] In one specific embodiment, S5 includes the following steps: S51. Obtain the coordinate data and overall confidence score of each enterprise to be matched, as well as the overall confidence score of each park to be determined. The overall confidence score of the enterprise is calculated by weighted fusion based on the multi-source address data of the enterprise to be matched, and the overall confidence score of the park is calculated by weighted fusion based on the initial confidence score of the data source type corresponding to the park to be determined.

[0075] S52, for each enterprise to be matched, based on the spatial relationship between the coordinate data of the enterprise to be matched and the target boundaries of each target park, a multi-level spatial matching strategy is adopted to determine the target park associated with the enterprise to be matched and the corresponding spatial matching confidence. The multi-level spatial matching strategy includes, in sequence: point-to-surface inclusion determination, buffer proximity determination, and nearest neighbor determination.

[0076] S53. Based on the spatial matching confidence, the overall enterprise confidence of the enterprise to be matched, and the overall park confidence of the associated undetermined park, the association confidence between the enterprise to be matched and the associated undetermined park is calculated.

[0077] Among them, the enterprise comprehensive confidence score is used to reflect the overall reliability of the geographic information of the enterprise to be matched. In this embodiment, it is calculated by weighting and integrating factors such as data source credibility, geocoding accuracy, and data timeliness based on the enterprise's business registration address, annual report operating address, map POI, and other multi-source address data. The value range is from 0 to 1, reflecting the credibility of the enterprise data itself.

[0078] The overall confidence level of a park is used to reflect the overall reliability of a park under consideration. It is calculated by weighted fusion of confidence levels based on the initial confidence level of the data source types associated with the park. The value ranges from 0 to 1, reflecting the authority and accuracy of the park's data.

[0079] This embodiment performs three spatial matching strategies sequentially. Once a matching is successful, the process terminates and the corresponding spatial matching confidence level is recorded, thus avoiding unnecessary calculations.

[0080] The spatial matching confidence score, the overall enterprise confidence score, and the overall park confidence score are weighted and summed according to preset weights to obtain the corresponding association confidence score. The preset weight sum of the three confidence scores is 1, and the specific value can be set by the implementer according to the actual situation.

[0081] In one specific embodiment, S52 includes the following steps: S521, for each enterprise to be matched, traverse the target boundaries of all pending parks and perform the first-level matching: determine whether the coordinate data of the enterprise to be matched successfully falls inside any target boundary. If so, determine the pending park corresponding to the target boundary as the pending park associated with the enterprise to be matched, and use the first preset value as the spatial matching confidence level, and stop the subsequent matching steps for the enterprise to be matched.

[0082] S522, If the first-level matching fails, the second-level matching is executed: it is determined whether the coordinate data of the enterprise to be matched successfully falls within the extended buffer of any target boundary. If so, the undetermined park corresponding to the target boundary is determined as the undetermined park associated with the enterprise to be matched, and the spatial matching confidence is calculated based on the shortest distance from the coordinate data of the enterprise to be matched to the target boundary. The subsequent matching steps for the enterprise to be matched are stopped. The extended buffer is the area formed by extending the target boundary outward by a second preset distance.

[0083] S523, if the second-level matching fails, the third-level matching is executed: calculate the shortest distance from the coordinate data of the enterprise to be matched to the target boundary of each undetermined park, take the undetermined park corresponding to the smallest shortest distance as the undetermined park associated with the enterprise to be matched, and calculate the spatial matching confidence based on the smallest shortest distance.

[0084] The specific value of the first preset value can be set by the implementer according to the actual situation. For example, in this embodiment, the first preset value is 1.0 to represent the highest confidence level of the match.

[0085] During the second-level matching process, the spatial matching confidence score is calculated using a linear decay function based on the shortest distance from the coordinates of the company to be matched to the target boundary of the associated industrial park: Conf = 1 - d / D, where d is the shortest distance from the coordinates of the company to be matched to the target boundary, and D is a second preset distance, for example, D = 50 meters, which can be set by the implementer according to the actual situation. Correspondingly, coordinates falling within the extended buffer indicates that the company is adjacent to the industrial park, but with a certain offset, and the closer the distance, the higher the confidence score. Linear decay is used to simply and intuitively characterize the reliability of the match.

[0086] When performing the third-level matching, the spatial matching confidence is calculated using an exponential decay function based on the minimum shortest distance: Conf=e^(-λ×d) min ), where d min λ is the minimum shortest distance from the coordinates of the company to be matched to the target boundary of each potential industrial park, where λ is the decay coefficient, for example, 0.005. Correspondingly, the third-level matching uses a fallback matching method to ensure that each company to be matched can be associated with an industrial park, but the confidence level decreases as the distance increases. Exponential decay can quickly reduce the confidence level of long-distance matching.

[0087] The above-described spatial matching strategy employs a three-tiered progressive approach: point-to-surface inclusion, buffer proximity, and nearest neighbor. In the first tier, point-to-surface inclusion is directly used with the highest spatial confidence, ensuring that companies located within the park receive the most reliable association results—aligning with business intuition and computationally efficient. In the second tier, an extended buffer is introduced, and a linear decay function is used to calculate spatial confidence. This allows neighboring companies located outside the park due to surveying errors, GPS drift, or boundary conditions to still be successfully associated, with their confidence decreasing reasonably with distance, avoiding rigid binary decisions of either inside or outside. In the third tier, nearest neighbor determination is used in conjunction with an exponential decay function, ensuring that all companies not successfully matched in the first two tiers are associated with at least the nearest park location. Simultaneously, the confidence of associations at extremely long distances is rapidly reduced to negligible levels, guaranteeing full matching coverage while preventing weak associations from polluting data quality, thus improving the accuracy and robustness of park-company associations.

[0088] This embodiment identifies and merges the same park from different data sources based on park name data and spatial reference data, effectively integrating multi-source heterogeneous data and solving the problems of incomplete coverage and duplicate data conflicts from a single data source, thus improving the accuracy of park entity identification. By determining the target generation strategy and generating the initial boundary from the preset boundary generation strategy based on the data source type and data completeness, the generation of park boundaries can adaptively adapt to data sources of different types and richness, improving the robustness of park boundary generation. By calculating the minimum convex hull polygon or minimum outer rectangle of the enterprise coordinate points that have been determined to be located inside the initial boundary or within the boundary buffer area and finely adjusting the buffer outward, the smoothness and inclusiveness of the target boundary are improved. Finally, by establishing an association relationship between the target boundary and the coordinate data of the enterprises to be matched and generating an association confidence score, the accurate association between enterprises and parks is achieved.

[0089] Example 2 Embodiment 2 of the present invention provides a non-transitory computer-readable storage medium, which can be disposed in an electronic device to store at least one instruction or at least one program related to implementing a method in the method embodiment. The at least one instruction or at least one program is loaded and executed by the processor to implement the method for constructing the relationship between parks and enterprises provided in the above embodiment.

[0090] Example 3 Embodiment 3 of the present invention provides an electronic device, which includes a processor and the non-transitory computer-readable storage medium of Embodiment 2 of the present invention.

[0091] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A method for constructing the relationship between industrial parks and enterprises, characterized in that, The method includes the following steps: S1. Based on the park name data and park spatial reference data within the target geographical area, identify and merge the same park from different data sources to obtain at least one undetermined park. S2, for each pending zone, based on the data source type and data completeness corresponding to the pending zone, determine the target generation strategy from the preset boundary generation strategy, and generate the initial boundary of the pending zone according to the target generation strategy. S3, for each enterprise coordinate point in the coordinate data of the park-related enterprises corresponding to the undetermined park, calculate the spatial distance from the enterprise coordinate point to the initial boundary, and determine whether the enterprise coordinate point is located inside the initial boundary or within the boundary buffer area based on the spatial distance. S3 includes the following steps: S31, For any enterprise coordinate point in the coordinate data of the park-related enterprises corresponding to the undetermined park, calculate the shortest spatial distance from the enterprise coordinate point to the initial boundary, wherein the shortest spatial distance is the minimum value of the vertical distance from the enterprise coordinate point to each side of the initial boundary. S32, when the shortest spatial distance is equal to 0, it is determined that the enterprise coordinate point is located inside the initial boundary; S33, when the shortest spatial distance is greater than 0 and less than or equal to a preset buffer distance threshold, it is determined that the enterprise coordinate point is located within the boundary buffer area of ​​the initial boundary; S34, when the shortest spatial distance is greater than the buffer distance threshold, it is determined that the enterprise coordinate point is outside the boundary buffer area of ​​the initial boundary; S4, based on the coordinates of all enterprises located inside or within the boundary buffer zone of the initial boundary, spatially adjust the initial boundary to obtain the target boundary of the proposed park. The enterprise coordinates include those located inside the initial boundary and those located within the boundary buffer zone. The target boundary is a closed polygon, composed of an ordered sequence of boundary vertex coordinates, used to characterize the actual geographical extent of the proposed park. S4 includes the following steps: S41, collect all enterprise coordinate points that are determined to be located inside the initial boundary or within the boundary buffer area, and form a valid point set; S42, Calculate the minimum convex hull polygon or minimum bounding rectangle of the effective point set; S43, buffer the minimum convex hull polygon or the minimum outer rectangle outward by a preset fine-tuning distance to obtain the adjusted closed polygon; S44, the adjusted closed polygon is used as the target boundary of the area to be determined, wherein the coordinate sequence of the boundary vertices of the target boundary is arranged in clockwise or counterclockwise order; S5. Based on the target boundaries of all pending parks and the coordinate data of the enterprises to be matched, establish an association between each enterprise to be matched and at least one pending park, and generate an association confidence score. The association confidence score is obtained by weighted summation of spatial matching confidence score, enterprise comprehensive confidence score and park comprehensive confidence score.

2. The method for constructing the relationship between industrial parks and enterprises according to claim 1, characterized in that, The park name data includes the name text of several initial parks, and the park spatial reference data includes the coordinates of spatial reference points of several initial parks. S1 includes the following steps: S11, construct a set of parks based on all the initial parks; S12, For any two parks in the park set, calculate the text similarity between the name texts of the two parks, and the spatial distance between the spatial reference point coordinates of the two parks; S13, when the text similarity is greater than the first preset threshold and the spatial distance is less than the second preset threshold, the two parks are merged into one undetermined park, and the two parks in the park set are replaced with the undetermined park to obtain the updated park set; S14, For the updated set of parks, repeat steps S12 to S13 until no two parks in the updated set of parks simultaneously satisfy the condition that the corresponding text similarity is greater than the first preset threshold and the corresponding spatial distance is less than the second preset threshold. When the repeated execution stops, each park in the set of parks is determined as a pending park.

3. The method for constructing the relationship between industrial parks and enterprises according to claim 1, characterized in that, S2 includes the following steps: S21, for each pending zone, determine whether the data source type corresponding to the pending zone is an official boundary file; S22, If it is an official boundary file, then the target generation strategy is determined to be a first preset strategy, wherein the first preset strategy is used to directly import the boundary data in the official boundary file as the initial boundary; S23, If it is not an official boundary file, then obtain the coordinate data of the park-related enterprises collected by the park to be determined, and count the number of enterprise coordinate data in the park-related enterprise coordinate data as the data completeness. S24, if the data completeness is greater than a preset quantity threshold, then the target generation strategy is determined to be the second preset strategy, wherein the second preset strategy is used to cluster the coordinate data of the park-related enterprises and generate an initial boundary based on the convex hull of the clustering results. S25, if the data completeness is less than or equal to the preset quantity threshold, then the target generation strategy is determined to be the third preset strategy, wherein the third preset strategy is used to generate an initial boundary by boundary fusion based on remote sensing building outline data and / or map POI data.

4. The method for constructing the relationship between industrial parks and enterprises according to claim 3, characterized in that, The step of generating the initial boundary based on the convex hull of the clustering results includes: S241, Extract the convex hull polygons of each cluster in the clustering results; S242, acquire road network data, buffer the convex hull polygon outward by a first preset distance, and then intersect and trim it with the road network data to obtain the initial boundary.

5. The method for constructing the relationship between industrial parks and enterprises according to claim 1, characterized in that, S5 includes the following steps: S51, obtain the coordinate data and comprehensive confidence score of each enterprise to be matched, as well as the comprehensive confidence score of each park to be determined. The comprehensive confidence score of the enterprise is calculated by weighted fusion of confidence scores based on the multi-source address data of the enterprise to be matched, and the comprehensive confidence score of the park is calculated by weighted fusion of confidence scores based on the initial confidence score of the data source type corresponding to the park to be determined. S52, for each enterprise to be matched, based on the spatial positional relationship between the coordinate data of the enterprise to be matched and the target boundaries of each target park, a multi-level spatial matching strategy is adopted to determine the target park associated with the enterprise to be matched and the corresponding spatial matching confidence. The multi-level spatial matching strategy includes, in sequence: point-to-surface inclusion determination, buffer proximity determination, and nearest neighbor determination. S53. Based on the spatial matching confidence, the overall enterprise confidence of the enterprise to be matched, and the overall park confidence of the associated undetermined park, the association confidence between the enterprise to be matched and the associated undetermined park is calculated.

6. The method for constructing the relationship between industrial parks and enterprises according to claim 5, characterized in that, S52 includes the following steps: S521, for each enterprise to be matched, traverse the target boundaries of all pending parks and perform the first-level matching: determine whether the coordinate data of the enterprise to be matched successfully falls inside any target boundary. If so, determine the pending park corresponding to the target boundary as the pending park associated with the enterprise to be matched, and use the first preset value as the spatial matching confidence level, and stop the subsequent matching steps for the enterprise to be matched. S522, if the first-level matching fails, the second-level matching is performed: it is determined whether the coordinate data of the enterprise to be matched successfully falls within the extended buffer of any target boundary. If so, the undetermined park corresponding to the target boundary is determined as the undetermined park associated with the enterprise to be matched, and the spatial matching confidence is calculated based on the shortest distance from the coordinate data of the enterprise to be matched to the target boundary. The subsequent matching steps for the enterprise to be matched are stopped. The extended buffer is the area formed by extending the target boundary outward by a second preset distance. S523, if the second-level matching fails, the third-level matching is performed: calculate the shortest distance from the coordinate data of the enterprise to be matched to the target boundary of each undetermined park, take the undetermined park corresponding to the smallest shortest distance as the undetermined park associated with the enterprise to be matched, and calculate the spatial matching confidence based on the smallest shortest distance.

7. A non-transitory computer-readable storage medium, wherein the non-transitory computer-readable storage medium stores at least one instruction or at least one program segment, characterized in that, The at least one instruction or the at least one program segment is loaded and executed by the processor to implement the method for constructing the relationship between the park and the enterprise as described in any one of claims 1-6.

8. An electronic device, characterized in that, Includes a processor and the non-transitory computer-readable storage medium as described in claim 7.

Images