Industrial cluster analysis method and system based on space-time knowledge graph
By constructing an industry domain ontology model and geographic grid system based on spatiotemporal knowledge graphs, the problem of low efficiency in big data analysis in existing technologies is solved, enabling multi-dimensional analysis and prediction of industrial clusters and revealing the collaborative connections and forms of industrial clusters.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- GUANGDONG URBAN & RURAL PLANNING & DESIGN INST
- Filing Date
- 2025-12-24
- Publication Date
- 2026-05-19
AI Technical Summary
Existing technologies suffer from high computational overhead, low query efficiency, inability to deeply explore the intrinsic mechanisms of industrial clusters, and inability to support multi-scale, hierarchical spatial retrieval and analysis when processing complex industrial spatiotemporal big data.
We employ a spatiotemporal knowledge graph-based approach to construct an ontology model for the industry sector. We discretize the data using a geographic grid system to generate triplet data, build a knowledge graph, and perform statistical, spatial, network, and evolutionary analyses. Finally, we combine this with graph neural networks for prediction.
It integrates multi-scale entities and diverse relationships, identifies industrial clusters, reveals the collaborative connections and complex cluster forms of industrial clusters, and provides a multi-dimensional and quantifiable analytical framework for the evaluation and planning of regional industrial clusters.
Smart Images
Figure CN122066464A_ABST
Abstract
Description
Technical Field
[0001] This application belongs to the field of industrial cluster data analysis technology, and proposes an industrial cluster analysis method and system based on spatiotemporal knowledge graph. Background Technology
[0002] Industrial cluster analysis is a key tool for studying the structure, form, synergy, and evolution of regional industries, aiming to provide data-driven decision support for regional industrial layout planning. This analysis involves various elements such as enterprises, personnel, capital, and resources, which together constitute massive, multi-source, and dynamically evolving spatiotemporal data. However, due to the significant heterogeneity of these elements in terms of geometric shape, temporal characteristics, and business semantics, how to efficiently organize, manage, integrate, and deeply understand this industrial data has become a core challenge in achieving accurate and intelligent industrial cluster modeling and analysis in the current industrial economics field.
[0003] Currently, this field often relies on Geographic Information Systems (GIS) as the primary carrier for data processing, storage, and visualization analysis. Although semantic network technologies, represented by knowledge graphs, have provided new approaches to knowledge organization, existing GIS-based technical systems still face a series of inherent bottlenecks when processing complex industrial spatiotemporal big data. At the computational level, coordinate-based spatial relationship calculations incur enormous overhead, leading to low efficiency in spatial querying and analysis under large-scale data. Simultaneously, existing methods lack depth and hierarchy in knowledge structures, hindering in-depth exploration of the intrinsic mechanisms of industrial clusters and the deduction of development patterns. They also cannot naturally support multi-scale, hierarchical spatial retrieval and analysis. Especially when facing large-scale queries that require simultaneous processing of complex semantics and spatiotemporal conditions, the system index structure often cannot accommodate these demands, easily resulting in severe performance bottlenecks in large-scale graphs. Summary of the Invention
[0004] To overcome the shortcomings of the prior art in terms of low analysis efficiency under large-scale data, this invention proposes an industrial cluster analysis method and system based on spatiotemporal knowledge graphs.
[0005] To achieve the above-mentioned technical effects, the technical solution of the present invention is as follows: A method for analyzing industrial clusters based on spatiotemporal knowledge graphs. The knowledge graph's schema layer is constructed based on a pre-defined industry domain ontology model, which includes spatial inclusion relationships, temporal process relationships, and property rights relationships of industry clusters. Data of the target industry cluster is obtained based on the data required by the ontology model of the industry sector; the data of the industry cluster is spatially discretized based on a preset geographic grid system; Based on the pattern layer, entities, attributes, and relationships between entities in the industrial cluster data are extracted. Based on the entities, attributes, and relationships between entities, triple data between space, relationships, and industries are generated. A knowledge graph is constructed based on the triple data. The target industry cluster data is analyzed based on the knowledge graph to obtain industry cluster analysis results; wherein the analysis includes at least one of statistical analysis, spatial analysis, network analysis, connection analysis and evolutionary analysis.
[0006] As a preferred embodiment, the entities in the industrial cluster data include spatial unit entities, industrial carrier entities, and functional resource entities. The spatial unit entities include administrative region entities and grid entities, the industrial carrier entities include industrial cluster entities and enterprise entities, and the functional resource entities include at least one of property rights chains, value chains, supply chains, or innovation chains.
[0007] As a preferred embodiment, the spatial discretization process includes the following steps: The preset geographic grid system is used as a spatial reference frame, and grid identifiers are generated according to the preset spatial accuracy. Based on the spatial inclusion relationship, a mapping relationship is established between industrial clusters and their corresponding grid identifiers.
[0008] As a preferred embodiment, the steps for extracting entities, attributes, and relationships between entities from the industrial cluster data based on the schema layer include: Based on the pattern layer, define data rule configurations, and extract entities, attributes, and relationships between entities based on the data rule configurations. The data rule configuration includes: Entity rules include the entity's name and the allowed attribute constraints for the entity; Attribute rules, including the allowed node types for edges; Relationship rules include the relationship type, the allowed source node types, and the target node types.
[0009] As a preferred embodiment, the statistical analysis includes: Based on entities within the knowledge graph, aggregate calculations are performed on at least one of the following for the target industrial cluster: number of enterprises, proportion of enterprises, and industrial economic indicators.
[0010] As a preferred embodiment, the spatial analysis includes: Entities within the knowledge graph are used as spatial analysis units to calculate the number of enterprises or spatial density of the target industrial cluster in the grid; spatial clustering is then performed based on the density to obtain the spatial clustering area of the industrial cluster.
[0011] As a preferred embodiment, the steps of the connection analysis include: Based on the structure of the knowledge graph, the clustering coefficient of the node clusters in the knowledge graph is calculated to obtain the comprehensive connection density within the industrial cluster, and its expression is as follows:
[0012] Where C is the internal comprehensive connection density coefficient, Ci represents the connection density between industry type i and other industry types, ki represents the total number of connections between industry type i and other industry types, Ei represents the maximum number of connections between industry types that are connected to industry type i, and N represents the number of industry networks within the industry cluster. Based on the structure of the knowledge graph, the edge betweenness number of the edges in the knowledge graph is calculated to obtain the connection strength between various links within the industrial cluster. The expression is as follows:
[0013] in, Represents the connection between a pair of links in an industrial cluster. The number of edge betweennesses; This represents the total number of shortest paths from node s in the industrial cluster to node t in the industrial cluster. This indicates that these shortest paths pass through edges. The quantity.
[0014] As a preferred embodiment, the evolutionary analysis steps include: Construct a weighted graph for any target industry cluster within the knowledge graph; The weighted graph is divided into time steps using a time-series cyclic unit; The graph neural network aggregates the graph structure information at each time step and predicts the link relationships between nodes and the weight changes of edges in future time steps. The expression for the weighted graph is as follows:
[0015] in, This represents the total weighted graph of the industrial cluster. This represents the set of industry cluster nodes at time node t; It is the set of supply chain edges between industrial clusters, representing the strength of the supply chain connection between cluster Es and cluster Ec at time t. The weight matrix representing the edges; This indicates the characteristics of node indicators.
[0016] This invention also proposes an industrial cluster analysis system based on spatiotemporal knowledge graphs, which applies the aforementioned industrial cluster analysis method based on spatiotemporal knowledge graphs, including: Data acquisition module: used to acquire target industry cluster data based on the data required by the industry domain ontology model; and to perform spatial discretization processing on the industry cluster data based on a preset geographic grid system; The knowledge graph construction module is used to extract entities, attributes, and relationships between entities from the industrial cluster data based on the pattern layer; generate triple data between space, relationships, and industries based on the entities, attributes, and relationships between entities; and construct a knowledge graph based on the triple data. The graph analysis module is used to perform industrial cluster analysis based on the spatiotemporal knowledge graph.
[0017] As a preferred embodiment, the system further includes a query optimization module, which is used to construct and maintain a hybrid index structure within the knowledge graph. The hybrid index structure includes a grid-spatiotemporal multidimensional index constructed based on grid ID Z-order encoding or Hilbert encoding and time attributes, as well as node type, relation type, and attribute indexes based on the graph database.
[0018] As a preferred embodiment, the query optimization module includes the following query steps: Parse the query request and extract its semantic, spatial, and temporal conditions; Based on the extracted condition types, an index selection strategy is formulated, and a retrieval plan is generated based on the index selection strategy. The index selection strategy includes: For queries that include spatial or temporal conditions, the grid-spatiotemporal index is selected for retrieval; For queries that include semantic conditions, select the appropriate graph index for retrieval; For queries that include multiple conditions, the order in which different indexes are used is evaluated, an index combination strategy that minimizes the intermediate result set is generated, and retrieval is performed based on the index combination strategy.
[0019] Compared with the prior art, the beneficial effects of the present invention are as follows: This invention integrates geospatial analysis and knowledge graph technology, combining multi-scale entities and multi-relationships within the knowledge graph to identify industrial clusters based on gridded geospatial data. Through the knowledge graph, the collaborative connections and complex cluster forms of industrial clusters can be revealed at both the internal structure and cross-cluster association levels, thus providing a multi-dimensional and quantifiable analytical framework for the evaluation and planning of regional industrial clusters. Attached Figure Description
[0020] Figure 1 This is a flowchart illustrating the implementation of an industrial cluster analysis method based on spatiotemporal knowledge graphs proposed in Example 1. Figure 2 This is a structural diagram of the knowledge graph in Example 1; Figure 3 This is a schematic diagram of an industrial cluster analysis framework based on spatiotemporal knowledge graphs, as shown in Example 1. Figure 4 This is an architecture diagram of an industrial cluster analysis system based on spatiotemporal knowledge graphs proposed in Example 2. Detailed Implementation
[0021] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.
[0022] Example 1 This embodiment proposes an industrial cluster analysis method based on spatiotemporal knowledge graphs, and the implementation flowchart is as follows: Figure 1 As shown.
[0023] This embodiment proposes an industrial cluster analysis method based on spatiotemporal knowledge graphs, which includes the following steps: S1. Construct a pattern layer of the knowledge graph based on a preset industry domain ontology model, wherein the industry domain ontology model includes the spatial inclusion relationship, temporal process relationship and property rights relationship of industry clusters; S2. Obtain target industry cluster data based on the data required by the industry domain ontology model; perform spatial discretization processing on the industry cluster data based on a preset geographic grid system; S3. Extract entities, attributes, and relationships between entities from the industrial cluster data based on the pattern layer; generate triple data between space, relationships, and industries based on the entities, attributes, and relationships between entities; and construct a knowledge graph based on the triple data. S4. Analyze the target industry cluster data based on the knowledge graph to obtain industry cluster analysis results; wherein, the analysis includes at least one of statistical analysis, spatial analysis, network analysis, connection analysis and evolutionary analysis.
[0024] In this embodiment, a spatiotemporal knowledge graph data model that integrates geographic grid coding and can uniformly express industrial entities, spatial locations, and temporal evolution is created. Geographic grid coding is used as the core attribute of the entity, which can extract semantic association information related to the industrial cluster and deeply integrate it with the temporal attributes generated by each chain to form a three-in-one industrial cluster data representation of "semantics-space-time".
[0025] In one optional embodiment, the entities of the industrial cluster data include spatial unit entities, industrial carrier entities, and functional resource entities, wherein the spatial unit entities include administrative region entities and grid entities, the industrial carrier entities include industrial cluster entities and enterprise entities, and the functional resource entities include at least one of property rights chain, value chain, supply chain, or innovation chain.
[0026] In this embodiment, entities include spatial units, industrial carriers, and functions or resources that are attached to or derived from the operation of the industry. They are represented as industrial cluster knowledge concepts, and are instantiated objects mapped according to the business needs of regional industrial cluster spatial analysis and network analysis.
[0027] Furthermore, the knowledge relationships within industrial clusters include various relationships such as spatial inclusion, temporal process, property rights, and inheritance.
[0028] Spatial inclusion relationships: including spatial relationships between administrative regions at all levels, spatial relationships between administrative regions and grids, and spatial relationships between grids and enterprises, etc.
[0029] Ownership relationships: including headquarters-branch relationships, and investment-investee relationships. Sequential process relationships: reflecting the process of an enterprise's creation, changes (changes in name, address, and other information), and demise.
[0030] like Figure 2 The diagram shown is a structural diagram of a knowledge graph.
[0031] As an example, the data layer construction process of the knowledge graph uses a PostgreSQL database to integrate spatial data and graph data.
[0032] In an optional embodiment, the spatial discretization process includes: The preset geographic grid system is used as a spatial reference frame, and grid identifiers are generated according to the preset spatial accuracy. Based on the spatial inclusion relationship, a mapping relationship is established between industrial clusters and their corresponding grid identifiers.
[0033] As an example, the default geographic grid system is the Geohash grid.
[0034] In this embodiment, the Geohash grid is selected, which is highly consistent in terms of lexicographical order and spatial proximity of the encoding. It recursively divides the entire Earth surface (including land and ocean) into multi-level grids. Each level of grid has a unique string code (grid ID), and the sub-level grid code is prefixed with the parent level grid code information.
[0035] In an optional embodiment, the step of extracting entities, attributes, and relationships between entities from the industry cluster data based on the schema layer includes: Based on the pattern layer, define data rule configurations, and extract entities, attributes, and relationships between entities based on the data rule configurations. The data rule configuration includes: Entity rules include the entity's name and the allowed attribute constraints for the entity; Attribute rules, including the allowed node types for edges; Relationship rules include the relationship type, the allowed source node types, and the target node types.
[0036] More specifically, based on the ontology model's conceptual attributes and relationships constructed in the schema layer, along with their constraint rules, entities, attributes, and relationships between entities are extracted from various types of data sources accessed in the data layer. This forms an instantiated representation of ontology knowledge. Triples include constructing triples such as "space-relation-space," "space-relation-industry," "industry-relation-function," and "function-relation-function."
[0037] Data set representation of the unified description framework for spatiotemporal knowledge graphs of industrial clusters:
[0038] Where s represents a spatial unit, e represents an industrial unit, r represents a relationship, and v represents industrial functional resources.
[0039] Example of a "space-relationship-space" triple:
[0040] Where s is the spatial unit scale triplet, sd is the administrative region unit, rs is the spatial inclusion relationship, and sg is the grid unit.
[0041] Example of the "space-relationship-industry" triple:
[0042] Where e represents the enterprise triple within the industrial carrier, sd represents the administrative region unit, re represents the spatial inclusion relationship, and es represents the enterprise entity under the industrial unit.
[0043] Example of the "industry-relationship-function" triple:
[0044] Where v is the industrial functional resource triplet, rc is the incidental or derived relationship, and va is the property rights chain functional resource derived by the enterprise (e.g., information on subordinate branch enterprises derived by the enterprise).
[0045] In an optional embodiment, the statistical analysis includes: Based on entities within the knowledge graph, aggregate calculations are performed on at least one of the following for the target industrial cluster: number of enterprises, proportion of enterprises, and industrial economic indicators.
[0046] Furthermore, commonly used queries include: By using grid spatial relationships, we can statistically analyze the number and proportion of specific industrial clusters and their respective links in the industrial chain (upstream, midstream, and downstream) within the region, as well as the detailed industry categories. By using grid spatial relationships, the economic value (total profit of enterprises) of specific industrial clusters and their respective links in the industrial chain within the region can be statistically analyzed. The calculation formula is as follows: Enterprise percentage = (Number of enterprises in the target grid from a specific industry cluster) / (Total number of enterprises in that industry cluster across the entire region) * 100% Economic value concentration = (Total profit of enterprises in a specific industrial cluster within the target grid / Total profit of that industrial cluster across the entire region) * 100% Balance of industrial chain structure = Economic value of enterprises in a certain link of the industrial chain (upstream / midstream / downstream) / Total economic value of the industrial cluster * 100% In an optional embodiment, the spatial analysis includes: Entities within the knowledge graph are used as spatial analysis units to calculate the number of enterprises or spatial density of the target industrial cluster in the grid; spatial clustering is then performed based on the density to obtain the spatial clustering area of the industrial cluster.
[0047] More specifically, grid cells are used as the basic spatial analysis unit to calculate the number, proportion, or density of enterprises in a specific industry cluster within each grid in the region. Density-based spatial clustering algorithms are then used to cluster enterprises based on the geographical location represented by the grid codes they occupy.
[0048] The formula for calculating spatial density is as follows: Enterprise spatial density = Number of enterprises in a specific industrial cluster within the grid / Grid area In an optional embodiment, the steps of the connection analysis include: Based on the structure of the knowledge graph, the clustering coefficient of the node clusters in the knowledge graph is calculated to obtain the comprehensive connection density within the industrial cluster, and its expression is as follows:
[0049] Where C is the internal comprehensive connection density coefficient, Ci represents an industry type i with degree ki in the industry network, ki represents the total number of connections between industry type i and other industry types, and Ei represents the maximum number of connections between the ki connected industry types of industry type i. The range of Ci is [0,1]. When Ki=0 or Ki=1, Ei=0. In this case, Ci=0.
[0050] Based on the structure of the knowledge graph, the edge betweenness number of the edges in the knowledge graph is calculated to obtain the connection strength between various links within the industrial cluster. The expression is as follows:
[0051] in, Representing an edge The number of edge betweennesses; This represents the total number of shortest paths from node s to node t; This indicates that these shortest paths pass through edges. The quantity.
[0052] Furthermore, the steps for analyzing network connections between various industry clusters are as follows: 1) Analysis of the Importance of Industrial Clusters Output strength: Measures the importance of a target industrial cluster as an upstream supplier in a region. That is, the strength of the ownership linkages between this industrial cluster and other industrial clusters, expressed as follows:
[0053] in, This refers to the weight (connection strength) of the edge pointing from the target node i to other nodes j.
[0054] Input strength: Measures the importance of a target industrial cluster as a downstream demand side in a region. That is, the strength of ownership ties between other industrial clusters and the target industrial cluster, expressed as follows:
[0055] in, This refers to the weight (connection strength) of the edge pointing from other node j to target node i.
[0056] 2) Stability analysis of industrial cluster centers Betweenness centrality: Using ownership chains (headquarters branches, external investment institutions) as the basic element for connectivity analysis, it retrieves enterprise ownership chain information through grid spatial relationships. It measures the degree to which a cluster acts as an "intermediary" or "hub" in the network. Damage to a cluster with high betweenness centrality will have a significant impact on the connectivity of the entire regional industrial cluster network.
[0057] Formula for calculating betweenness centrality:
[0058] in, Representing nodes (industrial clusters) Betweenness centrality; This represents the total number of shortest paths from node s to node t; This indicates the nodes passed through in these shortest paths. The number of terms; the range of summation covers all terms that satisfy the condition. The node pairs s and t.
[0059] In an optional embodiment, the evolutionary analysis steps include: Construct a weighted graph for any target industry cluster within the knowledge graph; The weighted graph is divided into time steps using a time-series cyclic unit; The graph neural network aggregates the graph structure information at each time step and predicts the link relationships between nodes and the weight changes of edges in future time steps. The expression for the weighted graph is as follows:
[0060] in, This represents the total weighted graph of the industrial cluster. This represents the set of industry cluster nodes at time node t; It is the set of supply chain edges between industrial clusters, representing the strength of the supply chain connection between cluster Es and cluster Ec at time t. The weight matrix representing the edges; This indicates the characteristics of node indicators.
[0061] As an example, an LSTM recurrent unit is used in the time dimension, and a GNN graph neural network is used within each time step to aggregate graph information. Given a graph sequence of T historical time steps... Learn a mapping function f to predict the graph state τ time steps ahead. Its expression is as follows:
[0062] Among them, the prediction target Including future node indicator characteristics and weight matrix
[0063] Through future diagram state Predictable: ① Property rights chain link prediction: Predicting which industrial clusters will form new property rights chain relationships at time T+t, or how the strength of existing property rights chain relationships will change. This can identify potential industrial cooperation layouts.
[0064] ② Industrial Cluster Role Prediction: Predict which industrial clusters will become core suppliers (highest output intensity) or key demanders (highest input intensity) in the future, and identify potential bottleneck clusters in the property rights chain (high betweenness, low substitutability).
[0065] ③ Predict the evolution of industrial cluster ecosystems: Predict whether the current industrial clusters will consolidate, split, or new clusters will emerge in the future.
[0066] like Figure 3 The diagram shown is a schematic of an industrial cluster analysis framework based on spatiotemporal knowledge graphs.
[0067] Example 2 This embodiment proposes an industrial cluster analysis system based on spatiotemporal knowledge graphs, applying an industrial cluster analysis method based on spatiotemporal knowledge graphs proposed in Embodiment 1. For example... Figure 3 The diagram shown is an architecture diagram of an industrial cluster analysis system based on spatiotemporal knowledge graphs in this embodiment.
[0068] This embodiment proposes an industrial cluster analysis system based on spatiotemporal knowledge graphs, including: Data acquisition module: used to acquire target industry cluster data based on the data required by the industry domain ontology model; and to perform spatial discretization processing on the industry cluster data based on a preset geographic grid system; The knowledge graph construction module is used to extract entities, attributes, and relationships between entities from the industrial cluster data based on the pattern layer; generate triple data between space, relationships, and industries based on the entities, attributes, and relationships between entities; and construct a knowledge graph based on the triple data. The graph analysis module is used to perform industrial cluster analysis based on the spatiotemporal knowledge graph.
[0069] Optionally, the system further includes a query optimization module, which is used to construct and maintain a hybrid index structure within the knowledge graph; the query steps of the query optimization module include: Parse the query request and extract its semantic, spatial, and temporal conditions; Based on the extracted condition types, an index selection strategy is formulated, and a retrieval plan is generated based on the index selection strategy. The index selection strategy includes: For queries that include spatial or temporal conditions, the grid-spatiotemporal index is selected for retrieval; For queries that include semantic conditions, select the appropriate graph index for retrieval; For queries that include multiple conditions, the order in which different indexes are used is evaluated, an index combination strategy that minimizes the intermediate result set is generated, and retrieval is performed based on the index combination strategy.
[0070] As an example, the formula for generating grid Z-order encoding is as follows: If the grid ID is a string (such as Geohash), the Z-order encoding can be generated through the following steps:
[0071] The formula for Hilbert curve encoding is as follows:
[0072] in, grid_id A unique identifier representing a grid cell; time_stamp Indicates time attributes (such as company registration time, data collection time); lat and lng Latitude and longitude of the grid center point The terminology used in the accompanying drawings is for illustrative purposes only and should not be construed as limiting the scope of this patent. Obviously, the above embodiments of the present invention are merely examples for clearly illustrating the present invention, and are not intended to limit the implementation of the present invention. Those skilled in the art can make other variations or modifications based on the above description. It is neither necessary nor possible to exhaustively describe all embodiments here. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the claims of the present invention.
Claims
1. A method for analyzing industrial clusters based on spatiotemporal knowledge graphs, characterized in that, Includes the following steps: The knowledge graph's schema layer is constructed based on a pre-defined industry domain ontology model, which includes spatial inclusion relationships, temporal process relationships, and property rights relationships of industry clusters. Data of the target industry cluster is obtained based on the data required by the ontology model of the industry sector; the data of the industry cluster is spatially discretized based on a preset geographic grid system; Based on the pattern layer, entities, attributes, and relationships between entities in the industrial cluster data are extracted, and triple data between space, relationship, and industry are generated based on the entities, attributes, and relationships between entities. A knowledge graph is constructed based on the triplet data; The target industry cluster data is analyzed based on the knowledge graph to obtain industry cluster analysis results; wherein the analysis includes at least one of statistical analysis, spatial analysis, network analysis, connection analysis and evolutionary analysis.
2. The industrial cluster analysis method based on spatiotemporal knowledge graphs according to claim 1, characterized in that, The entities in the industrial cluster data include spatial unit entities, industrial carrier entities, and functional resource entities. The spatial unit entities include administrative region entities and grid entities. The industrial carrier entities include industrial cluster entities and enterprise entities. The functional resource entities include at least one of the following: property rights chain, value chain, supply chain, or innovation chain.
3. The industrial cluster analysis method based on spatiotemporal knowledge graphs according to claim 1, characterized in that, The spatial discretization process includes the following steps: The preset geographic grid system is used as a spatial reference frame, and grid identifiers are generated according to the preset spatial accuracy. Based on the spatial inclusion relationship, a mapping relationship is established between industrial clusters and their corresponding grid identifiers.
4. The industrial cluster analysis method based on spatiotemporal knowledge graphs according to claim 1, characterized in that, The steps for extracting entities, attributes, and relationships between entities from the industrial cluster data based on the schema layer include: Based on the pattern layer, define data rule configurations, and extract entities, attributes, and relationships between entities based on the data rule configurations. The data rule configuration includes: Entity rules include the entity's name and the allowed attribute constraints for the entity; Attribute rules, including the allowed node types for edges; Relationship rules include the relationship type, the allowed source node types, and the target node types.
5. The industrial cluster analysis method based on spatiotemporal knowledge graphs according to any one of claims 1 to 4, characterized in that, The statistical analysis includes: Based on entities within the knowledge graph, aggregate calculations are performed on at least one of the following for the target industrial cluster: number of enterprises, proportion of enterprises, and industrial economic indicators.
6. The industrial cluster analysis method based on spatiotemporal knowledge graphs according to any one of claims 1 to 4, characterized in that, The spatial analysis includes: Entities within the knowledge graph are used as spatial analysis units to calculate the number of enterprises or spatial density of the target industrial cluster in the grid; spatial clustering is then performed based on the density to obtain the spatial clustering area of the industrial cluster.
7. The industrial cluster analysis method based on spatiotemporal knowledge graphs according to any one of claims 1 to 4, characterized in that, The steps of the connection analysis include: Based on the structure of the knowledge graph, the clustering coefficient of the node clusters in the knowledge graph is calculated to obtain the comprehensive connection density within the industrial cluster, and its expression is as follows: Where C is the internal comprehensive connection density coefficient, Ci represents the connection density between industry type i and other industry types, ki represents the total number of connections between industry type i and other industry types, Ei represents the maximum number of connections between industry types that are connected to industry type i, and N represents the number of industry networks within the industry cluster. Based on the structure of the knowledge graph, the edge betweenness number of the edges in the knowledge graph is calculated to obtain the connection strength between various links within the industrial cluster. The expression is as follows: in, Represents the connection between a pair of links in an industrial cluster. The number of edge betweennesses; This represents the total number of shortest paths from node s in the industrial cluster to node t in the industrial cluster. This indicates that these shortest paths pass through edges. The quantity.
8. The industrial cluster analysis method based on spatiotemporal knowledge graphs according to any one of claims 1 to 4, characterized in that, The steps of the evolutionary analysis include: Construct a weighted graph for any target industry cluster within the knowledge graph; The weighted graph is divided into time steps using a time-series cyclic unit; The graph neural network aggregates the graph structure information at each time step and predicts the link relationships between nodes and the weight changes of edges in future time steps. The expression for the weighted graph is as follows: in, This represents the total weighted graph of the industrial cluster. This represents the set of industry cluster nodes at time node t; It is the set of supply chain edges between industrial clusters, representing the strength of the supply chain connection between cluster Es and cluster Ec at time t. The weight matrix representing the edges; This indicates the characteristics of node indicators.
9. An industrial cluster analysis system based on spatiotemporal knowledge graphs, employing the industrial cluster analysis method based on spatiotemporal knowledge graphs as described in any one of claims 1 to 8, characterized in that, The system includes: Data acquisition module: used to acquire target industry cluster data based on the data required by the industry domain ontology model; and to perform spatial discretization processing on the industry cluster data based on a preset geographic grid system; The knowledge graph construction module is used to extract entities, attributes, and relationships between entities from the industrial cluster data based on the pattern layer; generate triple data between space, relationships, and industries based on the entities, attributes, and relationships between entities; and construct a knowledge graph based on the triple data. The graph analysis module is used to perform industrial cluster analysis based on the spatiotemporal knowledge graph.
10. The industrial cluster analysis system based on spatiotemporal knowledge graphs according to claim 9, characterized in that, The system also includes a query optimization module, which is used to construct and maintain a hybrid index structure within the knowledge graph; the query steps of the query optimization module include: Parse the query request and extract its semantic, spatial, and temporal conditions; Based on the extracted condition types, an index selection strategy is formulated, and a retrieval plan is generated based on the index selection strategy. The index selection strategy includes: For queries that include spatial or temporal conditions, select the grid-spatiotemporal index for retrieval; For queries that include semantic conditions, select the appropriate graph index for retrieval; For queries that include multiple conditions, the order in which different indexes are used is evaluated, an index combination strategy that minimizes the intermediate result set is generated, and retrieval is performed based on the index combination strategy.