Archive knowledge graph construction method and system based on semantic understanding and information depth mining

Abstract

The application provides a file knowledge graph construction method and system based on semantic understanding and information deep mining, relates to the technical field of computer data processing and urban and rural planning file management, and is used for solving the problems of insufficient planning conflict automatic identification and resolution suggestion generation. The method comprises the following steps: associating planning attributes to grid three-dimensional geographic information data to generate three-dimensional grid data with planning attributes; merging adjacent grid units into planning plot units and generating plot spatial relationship data; comparing planning labels in semantics, identifying conflict label combinations and determining conflict types; generating conflict resolution path data based on conflict types and label association data; and constructing a file knowledge graph. The application is used for planning file management and review.

Description

Technical Field

[0001] This application relates to the fields of computer data processing and urban and rural planning archives management technology, specifically to a method and system for constructing an archives knowledge graph based on semantic understanding and deep information mining. Background Technology

[0002] As urbanization continues, planning management departments need to comprehensively coordinate planning schemes in multiple thematic areas such as land use, transportation, and environmental protection. These data often come from different business systems and may not have been fully coordinated during the formulation process, resulting in the same plot of land being given contradictory functional positioning in different themes. For example, the same plot of land may be marked as commercial development land and ecological protection area at the same time, which brings great difficulties to planning approval and file management.

[0003] In existing technologies, some studies combine geographic information systems with planning databases to initially identify land conflicts through spatial overlay analysis; however, these methods mainly rely on manual rule setting and lack a deep understanding of the semantic level of planning labels, making it difficult to automatically distinguish between genuine conflicts and normally compatible combinations of planning attributes.

[0004] Meanwhile, existing methods do not make sufficient use of three-dimensional spatial information and cannot effectively handle conflicts between three-dimensional planning elements such as building height and underground space. After a conflict is discovered, there is also a lack of a systematic conflict resolution suggestion generation mechanism. Furthermore, the plots, labels, spatial relationships and conflict handling records in the planning archives are difficult to form a unified structured expression, making it difficult to reuse planning review experience and conflict handling results. They still rely heavily on the planners' manual experience judgment, which greatly limits work efficiency. Summary of the Invention

[0005] This application provides a method and system for constructing an archival knowledge graph based on semantic understanding and deep information mining, in order to solve the shortcomings of existing technologies such as reliance on manual verification for planning conflict identification, lack of deep semantic understanding, and insufficient conflict resolution suggestions.

[0006] To address the aforementioned technical problems, in a first aspect, this application provides a method for constructing an archival knowledge graph based on semantic understanding and deep information mining, comprising: The planning attributes in the planning thematic data of the target area are associated with the three-dimensional geographic information data of the target area after gridding, so as to generate three-dimensional grid data of the target area with planning attributes. From 3D grid data with planning attributes, adjacent grid cells with similar spatial geometric features and the same planning label are extracted. The extracted adjacent grid cells are merged into planning plot cells of the target area. Based on the spatial adjacency and spatial inclusion relationships between adjacent planning plot cells, plot spatial relationship data is generated. Obtain multiple planning tags corresponding to different planning thematic data for each planning plot unit, perform semantic comparison on multiple planning tags to identify conflicting planning tag combinations on the same planning plot unit, determine the corresponding conflict type based on the planning tag combination, and identify planning plot units with planning tag combinations as conflicting plot units. Based on the spatial relationship data of land parcels, determine the label association data between conflicting land parcel units and adjacent planned land parcel units, and extract conflict resolution path data for resolving conflicting land parcel units from the label association data based on the conflict type; Based on planning plot units, planning tags, plot spatial relationship data, conflict types, and conflict resolution path data, an archival knowledge graph of the target area is constructed.

[0007] Secondly, this application provides an archival knowledge graph construction system, including: The association module is used to associate the planning attributes in the planning thematic data of the target area with the three-dimensional geographic information data of the target area after gridding, so as to generate three-dimensional grid data of the target area with planning attributes. The extraction module is used to extract adjacent grid cells with similar spatial geometric features and the same planning label from 3D grid data with planning attributes. The extracted adjacent grid cells are merged into planning plot cells of the target area. Based on the spatial adjacency and spatial inclusion relationships between adjacent planning plot cells, plot spatial relationship data is generated. The planning plot cells and plot spatial relationship data are then output to the acquisition module, determination module, and construction module. The acquisition module is used to acquire multiple planning tags corresponding to different planning thematic data for each planning plot unit, perform semantic comparison on multiple planning tags to identify conflicting planning tag combinations on the same planning plot unit, determine the corresponding conflict type based on the planning tag combination, and identify planning plot units with planning tag combinations as conflicting plot units. The determination module is used to determine the label association data between conflicting land parcel units and adjacent planned land parcel units based on the land parcel spatial relationship data, extract conflict resolution path data for resolving conflicting land parcel units from the label association data based on the conflict type, and output the conflict resolution path data to the construction module; The construction module is used to build an archival knowledge graph of the target area based on planning plot units, planning labels, plot spatial relationship data, conflict types, and conflict resolution path data.

[0008] Thirdly, this application provides an electronic device, comprising: Memory, used to store computer programs; A processor, used to execute computer programs, implements the steps of the archival knowledge graph construction method based on semantic understanding and deep information mining as described in the first aspect above.

[0009] Fourthly, this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, can implement the steps of the archival knowledge graph construction method based on semantic understanding and deep information mining as described in the first aspect above.

[0010] The technical solution provided in this application has the following beneficial effects: First, by associating 3D geographic information data with multi-thematic planning attributes under the same spatial index, 3D grid data with planning attributes is generated. Based on the consistency of spatial geometric features and planning labels, these data are merged to form planning plot units, enabling planning data from different sources to be expressed in a unified 3D spatial framework, thus reducing conflict identification errors caused by inconsistent spatial scales. Second, by using a label mapping dictionary and a predefined conflict label combination library to perform semantic comparison of multiple planning labels and simultaneously determine the conflict type, the conflict identification process is transformed from manual experience-based judgment to a verifiable rule-based process. Furthermore, by combining conflict types, adjacent plot label sets, and spatial distribution characteristics into label association data, and sorting multiple adjustment directions based on spatial integration values, spatially based conflict resolution path data can be generated. Finally, the constructed archival knowledge graph unifies plot nodes, adjacency relationships, inclusion relationships, conflict types, and conflict resolution paths, providing a structured data foundation for planning archive management, planning review, and the reuse of conflict handling experience. Attached Figure Description

[0011] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0012] Figure 1 Construct a system architecture diagram for the archival knowledge graph of this application; Figure 2 This is a flowchart illustrating the overall process of constructing an archival knowledge graph based on semantic understanding and deep information mining, as described in this application. Figure 3 This application generates an interactive diagram of the attribute association system for three-dimensional mesh data generation and planning. Figure 4 This is a schematic diagram of the three-dimensional mesh data generation process of this application; Figure 5 This is a schematic diagram illustrating the planning attribute association process in this application; Figure 6 This is a schematic diagram illustrating the spatial connectivity merging and parcel identifier allocation in this application; Figure 7 This is a flowchart of the semantic comparison and identification process for conflicting land parcel units in this application; Figure 8 Generate a flowchart for the conflict resolution path data in this application; Figure 9 The node and edge topology network diagram after the archival knowledge graph of this application has been constructed. Detailed Implementation

[0013] The technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are some embodiments of this application, but not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0014] See Figure 1 As shown in the figure, this application discloses a system architecture for an archival knowledge graph construction system. The system architecture includes four layers: a data acquisition layer 101, a spatial processing layer 102, a conflict analysis layer 103, and a knowledge graph layer 104.

[0015] The data acquisition layer 101 includes a three-dimensional geographic information database 1011 and a planning thematic database 1012. The three-dimensional geographic information database 1011 is used to store three-dimensional spatial data such as the surface elevation, building outlines and underground space of the target area, while the planning thematic database 1012 is used to store various planning thematic data such as land use, transportation, environmental protection and municipal facilities.

[0016] The spatial processing layer 102 includes a gridding processing module 1021, an R-tree spatial indexing module 1022, and a planning attribute association module 1023. The gridding processing module 1021 acquires three-dimensional geographic information data from the data acquisition layer 101 and generates three-dimensional grid cells at a fixed granularity. The R-tree spatial indexing module 1022 establishes a spatial index for all three-dimensional grid cells. The planning attribute association module 1023 associates the planning attributes in the planning thematic database 1012 with each three-dimensional grid cell according to spatial query rules and outputs three-dimensional grid data with planning attributes. The spatial processing layer 102 also includes a connected domain analysis module 1024 and a land parcel relationship extraction module 1025. The connected domain analysis module 1024 performs spatial connected domain analysis on the three-dimensional grid data with planning attributes, merging homogeneous adjacent grid cells into planned land parcel cells. The land parcel relationship extraction module 1025 extracts the spatial adjacency and spatial inclusion relationships between planned land parcel cells, outputs land parcel spatial relationship data, and transmits the planned land parcel cells and land parcel spatial relationship data to the conflict analysis layer 103 and the knowledge graph layer 104.

[0017] The conflict analysis layer 103 includes a tag extraction module 1031, a semantic comparison module 1032, and a conflict resolution path generation module 1033. The tag extraction module 1031 extracts planning tags for each planning plot unit from various planning thematic data. The semantic comparison module 1032 compares the planning tags with a predefined conflict tag combination library to identify conflicting plot units. The conflict resolution path generation module 1033 calculates the spatial fusion degree value based on plot spatial relationship data and tag correlation data and generates conflict resolution path data. The conflicting plot units, conflict types, and conflict resolution path data are then transmitted to the knowledge graph layer 104.

[0018] The knowledge graph layer 104 includes a graph construction module 1041 and a graph storage and query module 1042. The graph construction module 1041 receives the output results of the spatial processing layer 102 and the conflict analysis layer 103, creates nodes and edges and writes attributes, and summarizes and constructs the archival knowledge graph. The graph storage and query module 1042 persistently stores the constructed archival knowledge graph and provides multiple query interfaces based on parcel identifiers, conflict types and neighborhood traversal.

[0019] Based on the above system architecture, this application provides a method for constructing an archival knowledge graph based on semantic understanding and deep information mining.

[0020] See Figure 2 As shown, the method includes the following steps: Step S1: Associate the planning attributes in the planning thematic data of the target area with the three-dimensional geographic information data of the target area after gridding, so as to generate three-dimensional grid data of the target area with planning attributes.

[0021] The purpose of this step is to establish a unified three-dimensional spatial data foundation for subsequent land parcel delineation and conflict identification, and to integrate planning thematic data from different business systems within the same spatial framework.

[0022] See Figure 3 As shown, the specific implementation process of this step is as follows: First, three-dimensional geographic information data and planning thematic data of the target area are collected from the data acquisition layer 101. The three-dimensional geographic information data refers to a spatial dataset containing three-dimensional elements such as the surface elevation, building outline and underground space of the target area. The data coordinate system needs to be uniformly converted to the target coordinate system before collection. The planning thematic data refers to planning results data from different business fields, which are divided into land use, transportation, environmental protection and municipal facilities types according to business fields.

[0023] Then, the gridding module 1021 divides the collected 3D geographic information data into regular 3D grid units according to a fixed spatial granularity. Gridding refers to dividing a continuous 3D space into non-overlapping cubic units at fixed 3D dimensions. Each 3D grid unit has a unique spatial coordinate number and a defined 3D spatial range. In this embodiment, the spatial granularity of the gridding is determined based on the planning accuracy requirements of the target area, generally ranging from 10m×10m to 50m×50m horizontally and 3m to 10m vertically. For example, for high-density construction areas such as urban center areas, a fine-grained grid of 10m×10m horizontally and 3m vertically is recommended to ensure accuracy; for suburban areas or ecological control areas, a coarse-grained grid of 50m×50m horizontally and 10m vertically can be used to reduce computational load. This embodiment uses a grid granularity of 20m×20m horizontally and 5m vertically as a typical example.

[0024] Next, the R-tree spatial index module 1022 builds an R-tree spatial index for all three-dimensional mesh cells. The R-tree spatial index is a tree index structure that supports efficient querying of multi-dimensional spatial data. The three-dimensional bounding box of each three-dimensional mesh cell is entered into the R-tree as an index node, which reduces the query complexity for any spatial range from linear traversal to logarithmic level.

[0025] Finally, the planning attribute association module 1023 writes the planning attributes from the planning thematic data into the corresponding three-dimensional grid cells. The planning attributes are attribute fields in the planning thematic data used to describe properties such as land use, functional zoning, and development intensity. Since the spatial representation of planning thematic data is divided into three different data forms: areal data, linear data, and point data, areal data is planning thematic data that expresses the spatial range of a planning thematic data with polygonal areas, such as the land use range of each plot in land use planning. Linear data is planning thematic data that expresses the spatial distribution of a planning thematic data with line segments or polylines, such as the road centerline in transportation planning. Point data is planning thematic data that expresses the spatial location of a planning thematic data with coordinate points, such as the facility locations in municipal facility planning. Therefore, different spatial query rules are needed for different data formats to determine which three-dimensional grid cells should be written into the planning attributes. The core logic of the spatial query rules is to establish the correspondence between planning thematic data and three-dimensional grid cells by judging the overlap or inclusion of spatial locations. Only three-dimensional grid cells that are determined to have spatial relationships by the spatial query rules will be written into the corresponding planning attributes.

[0026] The spatial query rules corresponding to the three data formats are shown in Table 1.

[0027] Table 1 Spatial query rules for different data formats

[0028] See Figure 4 As shown, the process of generating the above three-dimensional grid data can be summarized as follows: First, the three-dimensional geographic information data is processed into a grid; then, an R-tree spatial index is established; then, the planning attributes are associated with grid cells according to three spatial query rules: area, line, and point; finally, the planning attributes are converted into semantic planning tags through a tag mapping dictionary.

[0029] See Figure 5 As shown, taking the central urban area of ​​area A as an example, the process of planning attribute association is further explained: the land use thematic data is treated as surface data, and the boundary of its surface features is the polygon range of each plot. The planning attribute association module 1023 performs spatial intersection judgment on each surface feature. If the boundary polygon of a surface feature overlaps with the horizontal projection range of a certain three-dimensional grid cell, then the planning attribute of the surface feature (such as land use code R1 and maximum floor area ratio 2.5) is written into the three-dimensional grid cell.

[0030] The traffic planning thematic data is treated as linear data, with the road centerline as the linear element. The planning attribute association module 1023 extends 20 meters to both sides of each road centerline to form a buffer surface. Spatial intersection judgment is performed on each buffer surface. If the buffer surface overlaps with the horizontal projection of a certain three-dimensional grid cell, the planning attribute of the linear element (e.g., the road grade is expressway) is written into that three-dimensional grid cell. The environmental protection management thematic data is treated as planar data, and the same planar intersection judgment as the land use category is performed. Planning attributes such as the ecological protection zone category are written into the corresponding three-dimensional grid cell.

[0031] After the above processing, each 3D mesh cell retains its original 3D geometric features while also incorporating planning attribute information from multiple planning thematic data, thus forming 3D mesh data with planning attributes. The above example is merely one implementation method of this application.

[0032] This step, by establishing an R-tree spatial index and categorical spatial query rules, enables precise association of multi-source planning thematic data within a unified three-dimensional spatial framework.

[0033] In step S1, after the planning attribute association module 1023 completes the planning attribute association, it is necessary to convert the planning attributes into planning tags through a tag mapping dictionary. The construction process of the tag mapping dictionary is as follows: First, the planning management department extracts all planning attribute field values ​​and their corresponding functional semantics based on national and local land use classification standards, environmental protection regulations, transportation planning specifications, and other documents. Then, planning attribute field values ​​with similar semantics are grouped into the same planning tag. For example, land use codes R1 (Class I residential land) and R2 (Class II residential land) both belong to residential use in terms of functional semantics, so they are uniformly mapped to the residential land tag. Finally, all mapping relationships are entered into a structured correspondence table to form the tag mapping dictionary. The fields of the tag mapping dictionary include tag code, source planning topic, original attribute field value, standard tag name, semantic category, applicable area, and version number. The tag code is used to maintain a unique reference to the same tag in the subsequent conflict tag combination library and archive knowledge graph node attributes. When the same planning attribute field value involves multiple functional semantics (for example, a land use code B1R1 contains both commercial and residential functions), it is mapped to multiple planning tags. Example content of the tag mapping dictionary is shown in Table 2.

[0034] Table 2. Tag Mapping Dictionary (Partial Examples)

[0035] This step, by establishing an R-tree spatial index and categorical spatial query rules, enables precise association of multi-source planning thematic data within a unified three-dimensional spatial framework.

[0036] Step S2: Extract adjacent grid cells with similar spatial geometric features and the same planning label from the 3D grid data with planning attributes. Merge the extracted adjacent grid cells into planning plot cells of the target area. Generate plot spatial relationship data based on the spatial adjacency and spatial inclusion relationships between adjacent planning plot cells.

[0037] After generating 3D grid data with planning attributes in step S1, it is necessary to integrate a large number of fine-grained 3D grid cells into planning plot cells with clear semantics. This is because directly performing conflict identification at the level of a single 3D grid cell would result in excessive computation and fragmented results. Merging homogeneous adjacent grid cells into planning plot cells before performing conflict analysis can reduce the computational scale while ensuring the semantic integrity of conflict identification.

[0038] The specific implementation process for this step is as follows: First, the connected component analysis module 1024 traverses all 3D grid cells in the 3D grid data with planning attributes, calculating the spatial geometric feature differences between each 3D grid cell and its spatially adjacent grid cells. These spatial geometric features include the following three quantitative indicators, as shown in Table 3.

[0039] Table 3 Definition of Spatial Geometric Feature Indicators

[0040] Two adjacent 3D grid cells are considered to have similar spatial geometric characteristics when the difference in shape index ΔSI does not exceed 0.15, the difference in aspect ratio ΔAR does not exceed 0.3, and the relative volume difference ΔV / V_avg does not exceed 20%. These thresholds are default values ​​determined based on commonly used homogeneity criteria in urban and rural planning gridding practices. In practical applications, they can be adjusted according to regional characteristics and planning accuracy requirements.

[0041] Then, based on the similarity of spatial geometric features, it is further determined whether the planning label sets carried by two adjacent three-dimensional grid cells are completely consistent. That is, the same planning labels mean that the sets of all planning labels owned by the two grid cells are completely equal. For each pair of adjacent three-dimensional grid cells that meet the two conditions of similar spatial geometric features and the same planning labels, it is recorded as a candidate merging pair. All candidate merging pairs constitute the grid cell group to be merged.

[0042] Next, the connected component analysis module 1024 performs spatial connected component analysis on the grid cell group to be merged. The spatial connected component analysis treats the grid cell group to be merged as an undirected graph structure (with three-dimensional grid cells as nodes and candidate merging pairs as edges). It uses a breadth-first traversal algorithm to start from any unvisited node and classify all nodes that are directly or indirectly connected to the starting node into the same spatial connected component. The above process is repeated until all nodes are visited, resulting in multiple spatial connected components.

[0043] The specific execution steps of the breadth-first traversal algorithm are as follows: (1) Initialize an empty visited set V and an empty queue Q; (2) Select any grid cell from the grid cell group to be merged that is not yet in V. ,Will Add V and put it into Q, while creating a new spatial connected component. (3) When Q is not empty, take out the first element n of the queue, traverse all candidate merge pairs of n and their neighbor m. If m is not in V, add m to V and Q and assign m to the queue. (4) When Q is empty, Construction complete; if there are still grid cells not yet in V, return to step (2) to create new connected components; (5) after all grid cells have been visited, the complete spatial connected components are obtained { , ,..., }

[0044] See Figure 6 As shown, all three-dimensional mesh cells within each spatial connected domain are merged into a single planning plot cell. The three-dimensional spatial extent of the planning plot cell is obtained by taking the union of the spatial extents of all mesh cells within its spatial connected domain. In other words, the three-dimensional boundaries of each mesh cell in the same spatial connected domain are merged to form a continuous three-dimensional polygonal outline as the spatial boundary of the planning plot cell. A unique plot identifier is assigned to each planning plot cell, and the plot identifier is generated using sequentially increasing integer numbers.

[0045] Next, the land parcel relationship extraction module 1025 obtains the spatial boundary information of each planned land parcel unit. The spatial boundary information is the three-dimensional polygon boundary calculated by merging the outer contours of all three-dimensional mesh units contained in the planned land parcel unit. The spatial adjacency relationship is extracted based on the spatial boundary information of adjacent planned land parcel units. That is, all pairs of planned land parcel units are traversed. If the outer contours of two planned land parcel units have a common line segment and the length of the common line segment is not less than the side length of one mesh unit, then it is determined that the two have a spatial adjacency relationship.

[0046] Meanwhile, the land parcel relationship extraction module 1025 extracts spatial inclusion relationships based on the spatial boundary information of each planned land parcel unit. That is, if the spatial boundary of one planned land parcel unit is completely located inside the spatial boundary of another planned land parcel unit, it is determined that there is a spatial inclusion relationship. The determination of spatial inclusion relationship takes into account tolerance processing. When the maximum distance deviation between the boundaries of two planned land parcel units does not exceed 1 grid unit size, it is still considered as complete inclusion.

[0047] Finally, adjacency edges are generated for each pair of planned land parcels that have spatial adjacency, and containment edges (direction from the outer layer to the inner layer) are generated for each pair of planned land parcels that have spatial containment. All adjacency edges and containment edges are aggregated into a unified data structure to generate land parcel spatial relationship data.

[0048] This step integrates a large number of fine-grained grid cells into semantically clear planning plot cells through breadth-first traversal spatial connectivity analysis, and systematically records the topological relationships between plots, laying a structured spatial relationship foundation for subsequent conflict identification and resolution analysis.

[0049] Step S3: Obtain multiple planning tags corresponding to different planning thematic data for each planning plot unit, perform semantic comparison on multiple planning tags to identify conflicting planning tag combinations on the same planning plot unit, determine the corresponding conflict type based on the planning tag combination, and identify the planning plot unit with the planning tag combination as the conflicting plot unit.

[0050] After generating the planning plot units and plot spatial relationship data in step S2, it is necessary to detect whether there are semantic conflicts between planning tags from different themes for each planning plot unit. This is because the same planning plot unit may simultaneously carry planning tags from multiple themes such as land use, transportation, and environmental protection, and the tags from different themes may be semantically mutually exclusive.

[0051] See Figure 7 As shown, the specific implementation process of this step is as follows: First, the tag extraction module 1031 uses the spatial boundary information of each planning plot unit as the query condition to perform spatial queries on various types of planning thematic data. The query rules are consistent with step S1. The planning attributes that fall within or intersect within the spatial range of the planning plot unit are extracted and then converted into planning tags through the tag mapping dictionary established in step S1.

[0052] Then, all the planning tags extracted from all types of planning thematic data for a planning plot unit are aggregated into a set, namely the tag set of the planning plot unit. The tag set is a finite set of all planning tags extracted from various types of planning thematic data for a planning plot unit, and each planning tag records the planning thematic data type from which it originates.

[0053] Next, the semantic comparison module 1032 performs a semantic comparison between the tag set of each planned plot unit and each conflict tag record in the preset conflict tag combination library. The construction process of the conflict tag combination library is as follows: First, experts in the field of planning management sort out the land use function combinations, construction intensity control restrictions, and spatial boundary constraints that are clearly prohibited from being compatible in national and local planning regulations, and extract them as the initial conflict rule set; Second, based on typical conflict scenarios found in planning practice cases, supplement the conflict combinations that are not explicitly covered by regulations but frequently occur in practice; Third, perform logical deduplication and consistency checks on the initial conflict rule set to ensure that there are no contradictory rules; Fourth, enter the conflict rules that pass the checks into a structured data table in a unified format to form the conflict tag combination library; The fields of the conflict tag combination library include rule number, conflict planning tag A, conflict planning tag B, conflict type, applicable conditions, rule source, priority, and version number. The conflict type is determined by the rule number that matches the tag combination. The conflict label combination library supports incremental updates. When regulations are revised or new planning topics are added, new conflict rule records are added according to the above process, and the existence of label codes, non-empty conflict types, and priority conflict are checked for the new rules.

[0054] Conflict types are divided into three categories: use conflict refers to the same planning plot unit being designated for mutually exclusive uses; intensity conflict refers to the construction intensity planning label exceeding the upper limit label of the construction intensity control of the planning plot unit; and spatial conflict refers to the construction boundary planning label exceeding the spatial control range label of the planning plot unit.

[0055] The specific process of semantic comparison involves set inclusion judgment. For a conflict record in the conflict tag combination library, if the tag set of a certain planned land parcel unit contains all the planning tags in that record, then the planned land parcel unit is determined to have a corresponding conflict tag combination, identified as a conflicting land parcel unit, and the conflict type field in that conflict record is recorded as the corresponding conflict type identifier. If the tag set of a certain planned land parcel unit matches multiple conflict records, then multiple conflict types are recorded. An example of the predefined conflict tag combination library is shown in Table 4.

[0056] Table 4. Predefined conflict tag combination library (partial examples)

[0057] In practical engineering applications, taking the central urban area of ​​Area A as an example, and using the planned plot unit with plot identifier 201 as an example, the semantic comparison process is explained: the tag extraction module 1031 extracts a set of tags for the planned plot unit, which includes three planning tags: residential land, ecological protection zone and transportation corridor. The semantic comparison module 1032 performs a set inclusion judgment on the tag set and the first record in the conflict tag combination library. The first record requires both residential land and ecological protection zone to be included. The tag set contains both residential land and ecological protection zone, so it is determined that the tag set includes all the planning tags in the first record, and the planned land parcel unit is identified as a conflicting land parcel unit, with the corresponding conflict type identifier being land use conflict. The module continues to perform a set inclusion judgment on the tag set and records 2 to 6 one by one. Record 2 requires both commercial development land and basic farmland protection zone to be included. The tag set does not contain commercial development land, so there is no match. Records 3 to 6 are similarly not matched. Finally, it is determined that the conflict tag combination of the planned land parcel unit with parcel identifier 201 is residential land and ecological protection zone, with the conflict type identifier being land use conflict.

[0058] Taking another planned plot unit identified as 305 as an example, the tag set of this planned plot unit includes the high-density construction tag and the intensity control upper limit tag, which matches the 3rd record in the conflict tag combination library, and is identified as a conflict plot unit, with the conflict type identified as intensity conflict.

[0059] After performing the above set inclusion comparison on each of the 1200 planned land parcels, a total of 47 conflicting land parcels were identified, including 28 with conflicting land use, 13 with conflicting intensity, and 6 with conflicting spatial characteristics. The above example is only one implementation method of this application.

[0060] This step achieves automated identification of planning label conflicts by using a set inclusion judgment method based on a predefined conflict rule base, transforming traditional manual verification into structured rule matching, thus improving the efficiency and consistency of conflict resolution.

[0061] Step S4: Determine the label association data between conflicting land parcel units and adjacent planned land parcel units based on the land parcel spatial relationship data, and extract conflict resolution path data for resolving conflicting land parcel units from the label association data based on the conflict type.

[0062] After identifying the conflicting land parcel units in step S3, in order to provide planners with reasonable conflict resolution suggestions, it is also necessary to analyze the surrounding spatial environment and label distribution of the conflicting land parcel units, determine which direction the conflict labels should be adjusted to and to what target label, and generate objective and operable conflict resolution path data.

[0063] See Figure 8 As shown, the specific implementation process of this step is as follows: In the first step, the path generation module 1033 takes the plot identifier of the conflicting plot unit as input and performs a first-order neighborhood query in the plot spatial relationship data generated in step S2. That is, it extracts all nodes directly connected to the conflicting plot unit from the directed graph of the plot spatial relationship data and forms the neighboring plot set of the conflicting plot unit by the query results. The neighboring plot set is a set of all adjacent planned plot units that have an adjacency relationship edge or an inclusion relationship edge with the conflicting plot unit.

[0064] The second step is to extract all the planning tags of the conflicting land parcel units from the tag set constructed in step S3, and use them as the first planning tag set; for each adjacent planned land parcel unit in the adjacent land parcel set, extract all its planning tags as the second planning tag set of the adjacent planned land parcel unit; wherein the first planning tag set contains at least one set of conflicting tag combinations, and the second planning tag set reflects the existing planning intention of the adjacent planned land parcel unit.

[0065] The third step involves calculating the following spatial distribution characteristic indicators based on the spatial boundary information of the conflicting land parcel units and the spatial boundary information of each adjacent planned land parcel unit: the spatial distance is calculated using the Euclidean distance between the centroids of the two planned land parcel units; the contact boundary length is obtained by calculating the total length of the common line segment after finding the intersection of the spatial boundaries of the two planned land parcel units; the shared boundary ratio is obtained by dividing the contact boundary length by the total perimeter of the conflicting land parcel units; and the relative orientation is obtained by mapping the azimuth angle of the centroid of the adjacent planned land parcel unit relative to the centroid of the conflicting land parcel unit to one of eight discrete orientations.

[0066] The specific indicators of spatial distribution characteristics are shown in Table 5.

[0067] Table 5 Explanation of Spatial Distribution Characteristic Indicators

[0068] The fourth step involves combining the first planning label set, the second planning label set for each adjacent planning plot unit, and the corresponding spatial distribution feature vector into a structured data record. This record represents the label association data between the conflicting plot unit and its adjacent planning plot unit. The fields of this label association data include the conflicting plot unit identifier, the adjacent planning plot unit identifier, the first planning label set, the second planning label set, and the spatial distribution feature vector. A record is generated for each adjacent planning plot unit in the adjacent plot set, and all records are aggregated into a label association dataset.

[0069] Fifth, extract the resolution direction information from the label association dataset according to the conflict type and generate multiple adjustment directions based on the following mapping rules: For conflicts of use, adjustment target labels are generated based on the main functional use labels in the second planning label set of adjacent planning plot units. The main functional use labels are determined according to the priority of the label source topic, the priority of the rule, and the frequency of occurrence of adjacent plot labels. For intensity conflicts, the planned intensity value in the conflicting plot unit is compared with the corresponding upper limit of construction intensity control. If the planned intensity value is greater than the upper limit of construction intensity control, the adjustment target label is determined to be an intensity label that matches the upper limit of construction intensity control. For spatial conflicts, the construction boundary of the conflicting plot unit is compared with the control boundary corresponding to the boundary constraint label of the adjacent planning plot unit. If the construction boundary exceeds the control boundary, the adjustment target label is determined by referring to the boundary constraint label of the adjacent planning plot unit. Each adjacent planning plot unit corresponds to an adjustment direction, and each adjustment direction contains an adjustment target label.

[0070] Step 6: Perform maximum and minimum value normalization on the contact boundary length and spatial distance values ​​of all adjacent planned land parcels, denoted as Lnorm and Dnorm respectively. The calculation method for maximum and minimum value normalization is as follows: For the contact boundary length value, Lnorm = (Li - Lmin) / (Lmax - Lmin), where Li is the contact boundary length of the current adjacent land parcel, and Lmin and Lmax are the minimum and maximum contact boundary lengths among all adjacent land parcels, respectively; for the spatial distance value, Dnorm = (Di - Dmin) / (Dmax - Dmin), where Di is the spatial distance value of the current adjacent land parcel, and Dmin and Dmax are the minimum and maximum spatial distance values ​​among all adjacent land parcels, respectively; when Lmax = Lmin or Dmax = Dmin, the corresponding normalization value is set to 0.5 to avoid division by zero and maintain the neutral contribution of each adjacent land parcel to this indicator. The purpose of normalization is to map indicators of different dimensions uniformly to the [0, 1] interval, allowing for weighted summation. Since a smaller spatial distance and a longer contact boundary indicate a higher degree of fusion, the formula for calculating the spatial fusion degree S is shown in equation (1).

[0071]

[0072] Where equation (1) is The contact boundary length weight indicates the degree of importance attached to the tightness of the boundary contact. This represents the spatial distance weight, indicating the degree of importance attached to spatial distance; and The sum of these values ​​equals 1, serving as a constraint to ensure that the fusion degree value S ranges from [0, 1]. The default value is... = =0.5, meaning that the closeness of boundary contact and spatial distance are given equal weight. In practical applications, if the planning management department believes that the closeness of boundary contact has a greater impact on conflict resolution decisions, the weight can be appropriately increased. The value of (e.g.) =0.6, =0.4); conversely, if the effect of spatial distance is more significant, it can be improved. The value of is recommended. The weight adjustment range is suggested to be between [0.3, 0.7] to avoid any factor being overly ignored.

[0073] The spatial integration degree is calculated for each adjacent planned land parcel unit in the adjacent land parcel set. Then, based on the spatial fusion value corresponding to each adjustment direction, the multiple adjustment directions are sorted, that is, all adjustment directions are sorted from largest to smallest according to S, and the adjustment direction with the largest spatial fusion value is determined as the priority adjustment direction; when multiple adjustment directions have the same spatial fusion value, the shared boundary ratio is used as the secondary sorting index.

[0074] The seventh step is to generate conflict resolution path data based on the priority adjustment direction. The conflict resolution path data is a structured record containing the unique identifier of the conflicting land parcel unit, the conflict type, the adjustment target label, and the spatial integration degree value. The above seven steps are performed on all conflicting land parcel units to generate a complete set of conflict resolution path data.

[0075] In practical engineering applications, taking conflicting land parcel unit P1 as an example, the conflict type of P1 is land use conflict, the first planning label set includes residential land and ecological protection area, and the adjacent land parcel set includes 4 adjacent planning land parcel units: Q1, Q2, Q3 and Q4.

[0076] The spatial distribution characteristics of P1 and its adjacent planned plot units are calculated as follows: the centroid Euclidean distance between P1 and Q1 is 80 meters, and the shared boundary length is 60 meters; the centroid Euclidean distance between P1 and Q2 is 50 meters, and the shared boundary length is 120 meters; the centroid Euclidean distance between P1 and Q3 is 110 meters, and the shared boundary length is 40 meters; the centroid Euclidean distance between P1 and Q4 is 70 meters, and the shared boundary length is 80 meters; the total boundary perimeter of the conflicting plot unit P1 is 320 meters.

[0077] The contact boundary length values ​​from Q1 to Q4 are normalized to a maximum and minimum value, with a maximum of 120 meters and a minimum of 40 meters. The corresponding values ​​for each adjacent planned land parcel are also normalized. They are respectively: Q1 corresponds to =(60-40)÷(120-40)=0.25; Q2 corresponds to =(120-40)÷(120-40)=1.00; Q3 corresponds to =(40-40)÷(120-40)=0.00; Q4 corresponds to =(80-40)÷(120-40)=0.50; The spatial distance values ​​were normalized to a maximum of 110 meters and a minimum of 50 meters. The corresponding values ​​for each adjacent planned land parcel unit were also normalized. They are respectively: Q1 corresponds to =(80-50)÷(110-50)=0.50; Q2 corresponds to =(50-50)÷(110-50)=0.00; Q3 corresponds to =(110-50)÷(110-50)=1.00; Q4 corresponds to =(70-50)÷(110-50)=0.33.

[0078] Based on equation (1), calculate the spatial integration degree value corresponding to each adjacent planning plot unit: Q1 corresponds to S = 0.5 × 0.25 + 0.5 × (1 - 0.50) = 0.125 + 0.250 = 0.375; Q2 corresponds to S = 0.5 × 1.00 + 0.5 × (1 - 0.00) = 0.500 + 0.500 = 1.000; Q3 corresponds to S = 0.5 × 0.00 + 0.5 × (1 - 1.00) = 0.000 + 0.000 = 0.000; Q4 corresponds to S = 0.5 × 0.50 + 0.5 × (1 - 0.33) = 0.250 + 0.335 = 0.585; Q2 has the highest spatial integration value, so the Q2 direction is determined as the priority adjustment direction; Q2 only contains residential land labels, so the target label for adjustment is residential land; finally, conflict resolution path data for P1 is generated: unique identifier P1, conflict type is land use conflict, target label for adjustment is residential land, and spatial integration value is 1.000. The above example is only one implementation method of this application.

[0079] This step generates quantitative conflict resolution path data by performing rule-based correlation analysis on the comprehensive spatial distribution characteristics and label semantic information, providing planners with objective and operable conflict resolution references.

[0080] Step S5: Based on the planning plot units, planning tags, plot spatial relationship data, conflict types, and conflict resolution path data, construct an archival knowledge graph for the target area.

[0081] After generating all conflict resolution path data in step S4, all the structured data generated in the first four steps need to be organized into a queryable and traversable archival knowledge graph. The archival knowledge graph is a graph data structure with planning plot units as nodes, plot spatial relationships as edges, and planning tags and conflict resolution information as node attributes.

[0082] See Figure 9 As shown, the specific implementation process of this step is as follows: In the first step, the map construction module 1041 traverses all the planned land parcel units generated in step S2, creates a node for each planned land parcel unit, and the node identifier is consistent with the land parcel identifier of the corresponding planned land parcel unit. The set of tags constructed for the planned land parcel unit in step S3 is written into the planning tag attribute of the node.

[0083] In the second step, the graph construction module 1041 traverses the spatial relationship data of the land parcels generated in step S2, records each adjacency relationship edge, and creates an undirected adjacency edge connecting the corresponding two nodes in the archive knowledge graph; records each containment relationship edge, and creates a directed containment edge from the outer node to the inner node.

[0084] The third step is to traverse all conflicting parcel unit identifiers output in step S3, find the corresponding node in the archive knowledge graph, and write a conflict type attribute for it. The attribute value should be consistent with the conflict type recorded in step S3. If a conflicting parcel unit matches multiple conflict records, the conflict type attribute value should be a list containing all conflict types.

[0085] The fourth step is to traverse all the conflict resolution path data generated in step S4, find the corresponding conflict parcel unit node in the archive knowledge graph, and write the resolution path attributes to it. The content of the resolution path attributes includes a unique identifier, conflict type, adjustment target label, and spatial integration degree value.

[0086] In the fifth step, the graph construction module 1041 summarizes all nodes, adjacent edges, containing edges, and node attributes created in the above four steps into the graph database of the graph storage and query module 1042. The summarization process is to write each node in the order of first node and then edge. After writing, a full verification is performed on the graph database. The specific rules of the full verification include: (1) Edge endpoint integrity check - confirm that the two endpoint nodes of each adjacent edge and containing edge exist in the graph database. If the endpoint is found to be missing, record the abnormal log and reconstruct the missing node from the output data of the corresponding step and then retry; (2) Conflict attribute completeness check - check whether the conflict type attribute and resolution path attribute of all nodes identified as conflict parcel units in step S3 have been correctly written. If the attribute is missing, read it again from the output data of step S3 and step S4 and rewrite it; (3) Tag attribute non-empty check - confirm that the planning tag attribute of each node is not an empty set. After all the verifications pass, the complete target area archive knowledge graph is output.

[0087] The knowledge graph supports three query methods: querying all attributes of a single node based on the parcel identifier; querying all conflicting parcel unit nodes in batches based on the conflict type; and performing a neighborhood traversal query starting from any node and following adjacent or containing edges.

[0088] The elements of the archival knowledge graph are shown in Table 6.

[0089] Table 6. Explanation of the Composition of Archival Knowledge Graph Elements

[0090] See Figure 9 As shown, taking the completed knowledge graph of the urban center area of ​​City A as an example, its topological structure contains 1200 nodes and 3645 edges. Ordinary planning plot unit nodes only carry planning label attributes; for example, the planning label attribute of the node with plot identification 100 is residential land. Conflict plot unit nodes simultaneously carry planning label attributes, conflict type attributes, and resolution path attributes. For example, the conflict plot unit node P1 with plot identification 201 has planning label attributes of residential land and ecological protection zone, conflict type attribute of land use conflict, and the adjustment target label in the resolution path attribute is residential land with a spatial integration degree of 1.000. Conflict plot unit node P1 is connected to adjacent planning plot unit nodes Q1, Q2, Q3, and Q4 through adjacency edges. The ordinary node with plot identification 100 points to node R1 with plot identification 501, which is spatially contained within it, through an inclusion edge. The above topological relationships are... Figure 9 The network is represented as a network of nodes and edges. Nodes are represented by circles and labeled with node identifiers and attribute summaries. Adjacent edges are represented by solid lines, and containing edges are represented by dashed lines with arrows. The above example is only one implementation of this application.

[0091] This step organizes the planning plot units, spatial topological relationships, conflict information, and resolution paths into a unified graph data structure, realizing the knowledge-based representation of planning archives and facilitating the planning management department to quickly locate conflicting plots and obtain resolution suggestions.

[0092] In summary, this application, through the organic collaboration of steps S1 to S5, achieves end-to-end processing from raw planning data to structured knowledge graph, improves the accuracy of planning conflict identification, reduces the workload of manual verification, and provides an objective and operable reference for resolving planning conflicts.

[0093] This application also provides an archival knowledge graph construction system, the system comprising: The association module is used to associate the planning attributes in the planning thematic data of the target area with the three-dimensional geographic information data of the target area after gridding, so as to generate three-dimensional grid data of the target area with planning attributes.

[0094] The extraction module is used to extract adjacent grid cells with similar spatial geometric features and the same planning label from 3D grid data with planning attributes. The extracted adjacent grid cells are merged into planning plot cells of the target area, and plot spatial relationship data is generated based on the spatial adjacency and spatial inclusion relationships between adjacent planning plot cells.

[0095] The acquisition module is used to acquire multiple planning tags corresponding to different planning thematic data for each planning plot unit, perform semantic comparison on multiple planning tags to identify conflicting planning tag combinations on the same planning plot unit, determine the corresponding conflict type based on the planning tag combination, and identify planning plot units with planning tag combinations as conflicting plot units.

[0096] The determination module is used to determine the label association data between conflicting land parcel units and adjacent planned land parcel units based on the spatial relationship data of land parcels, and to extract conflict resolution path data for resolving conflicting land parcel units from the label association data based on the conflict type.

[0097] The construction module is used to build an archival knowledge graph of the target area based on planning plot units, planning labels, plot spatial relationship data, conflict types, and conflict resolution path data.

[0098] The archival knowledge graph construction system of this application embodiment is used to implement the aforementioned archival knowledge graph construction method based on semantic understanding and deep information mining. Therefore, the specific implementation of the archival knowledge graph construction system can be found in the embodiment section of the archival knowledge graph construction method based on semantic understanding and deep information mining above. The specific implementation can be referred to the description of the corresponding embodiment, and will not be repeated here.

[0099] This application also provides an electronic device, including: a memory for storing a computer program; and a processor for implementing the steps of the above-described method for constructing an archival knowledge graph based on semantic understanding and deep information mining when executing the computer program.

[0100] This application also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of any of the above-described methods for constructing an archival knowledge graph based on semantic understanding and deep information mining.

[0101] In one exemplary embodiment, the aforementioned computer-readable storage medium may include, but is not limited to, various media capable of storing computer programs, such as USB flash drives, read-only memory, random access memory, portable hard drives, magnetic disks, or optical disks.

[0102] The embodiments of this application also provide a computer program product, which includes a computer program that, when executed by a processor, implements the steps in any of the above embodiments of the archival knowledge graph construction method based on semantic understanding and deep information mining.

[0103] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0104] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in one or more embodiments of this specification are all information and data authorized by the user or fully authorized by all parties. Furthermore, the collection, use and processing of related data must comply with relevant laws, regulations and standards, and corresponding operation entry points are provided for users to choose to authorize or refuse.

[0105] The foregoing has provided a detailed description of the method and system for constructing an archival knowledge graph based on semantic understanding and deep information mining, as provided in this application. Specific examples have been used to illustrate the principles and implementation methods of this application. The descriptions of the embodiments above are merely for the purpose of helping to understand the method and its core ideas. It should be noted that those skilled in the art can make various improvements and modifications to this application without departing from its principles, and these improvements and modifications also fall within the protection scope of this application.

Claims

1. A method for constructing an archival knowledge graph based on semantic understanding and deep information mining, characterized in that, include: The planning attributes in the planning thematic data of the target area are associated with the three-dimensional geographic information data of the target area after gridding, so as to generate three-dimensional grid data of the target area with planning attributes. From the three-dimensional grid data with planning attributes, adjacent grid cells with similar spatial geometric features and the same planning label are extracted. The extracted adjacent grid cells are merged into planning plot cells of the target area. Based on the spatial adjacency and spatial inclusion relationships between adjacent planning plot cells, plot spatial relationship data is generated. Obtain multiple planning tags corresponding to different planning thematic data for each planning plot unit, perform semantic comparison on multiple planning tags to identify conflicting planning tag combinations on the same planning plot unit, determine the corresponding conflict type based on the planning tag combinations, and identify the planning plot unit with the planning tag combinations as conflicting plot units. Based on the spatial relationship data of the land parcels, determine the label association data between the conflicting land parcel units and the adjacent planned land parcel units, and extract the conflict resolution path data for resolving the conflicting land parcel units from the label association data based on the conflict type; Based on the planned land parcel units, the planning tags, the land parcel spatial relationship data, the conflict types, and the conflict resolution path data, an archival knowledge graph of the target area is constructed.

2. The method according to claim 1, characterized in that, The step of determining the label association data between conflicting land parcel units and adjacent planned land parcel units based on the land parcel spatial relationship data, and extracting conflict resolution path data for resolving the conflicting land parcel units from the label association data based on the conflict type, includes: Based on the spatial relationship data of the land parcels, adjacent planned land parcels that have spatial adjacency or spatial inclusion relationships with the conflicting land parcels are obtained, thus obtaining the set of adjacent land parcels of the conflicting land parcels. Obtain the first set of planning tags for conflicting land parcel units and the second set of planning tags for each adjacent planned land parcel unit in the adjacent land parcel set; Based on the spatial boundary information of the conflicting land parcel unit and the spatial boundary information of each adjacent planned land parcel unit, calculate the spatial distribution characteristics between the conflicting land parcel unit and each adjacent planned land parcel unit; The first planning tag set, the second planning tag set of each adjacent planning plot unit, and the spatial distribution features are combined to construct tag association data between conflicting plot units and each adjacent planning plot unit; Based on a preset mapping relationship between conflict types and resolution directions, resolution direction information matching the conflict type is extracted from the tag association data, and conflict resolution path data is generated according to the resolution direction information.

3. The method according to claim 2, characterized in that, The step of generating conflict resolution path data based on the conflict resolution direction information includes: Based on the resolution direction information, multiple adjustment directions corresponding to the conflicting land parcel unit are obtained. Each adjustment direction includes an adjustment target label that adjusts the planning label of the conflicting land parcel unit to the planning label of the adjacent planning land parcel unit. Based on the spatial distribution characteristics, the contact boundary length and spatial distance between the conflicting plot unit and each adjacent planned plot unit are obtained. The contact boundary length value and the spatial distance value are normalized to obtain the spatial integration degree value between each adjacent planned land parcel unit and the conflicting land parcel unit. Based on the spatial fusion degree value corresponding to each adjustment direction, the multiple adjustment directions are sorted, and the adjustment direction with the largest spatial fusion degree value in the sorting results is taken as the priority adjustment direction. Based on the preferred adjustment direction, conflict resolution path data is generated. The conflict resolution path data includes the unique identifier of the conflicting plot unit, the conflict type, the adjustment target label corresponding to the preferred adjustment direction, and the spatial integration degree value corresponding to the preferred adjustment direction.

4. The method according to claim 1, characterized in that, The process of acquiring multiple planning tags corresponding to different planning thematic data for each planning plot unit, performing semantic comparison on the multiple planning tags to identify conflicting planning tag combinations on the same planning plot unit, determining the corresponding conflict type based on the planning tag combinations, and identifying planning plot units containing the planning tag combinations as conflicting plot units includes: Based on the spatial boundary information of each planning plot unit, the planning tags corresponding to each planning plot unit are extracted from various types of planning thematic data, resulting in multiple planning tags for each planning plot unit corresponding to different planning thematic data, and the multiple planning tags corresponding to each planning plot unit are combined to form the tag set of the planning plot unit. The tag set of each planned land parcel unit is compared with each tag combination in the preset conflict tag combination library. Based on the comparison results, planned land parcel units containing all planning tags in the tag combination are selected from the tag set to obtain conflicting land parcel units. The conflict type corresponding to the tag combination is determined as the conflict type of the conflicting land parcel unit.

5. The method according to claim 1, characterized in that, The construction of an archival knowledge graph of the target area based on the planned land parcel units, the planning tags, the spatial relationship data of the land parcels, the conflict types, and the conflict resolution path data includes: Each planned land parcel unit is treated as a node in the archive knowledge graph. Each node is assigned a node identifier, and the planning tag corresponding to each planned land parcel unit is stored as the node attribute of the node. Based on the spatial relationship data of the land parcels, the connection between each pair of planned land parcels with spatial adjacency is taken as an adjacency edge in the archive knowledge graph, and the connection between each pair of planned land parcels with spatial inclusion is taken as an inclusion edge in the archive knowledge graph. Based on the conflict type of each conflicting land parcel unit, a conflict type attribute is added to the node corresponding to the conflicting land parcel unit, and the conflict type attribute is stored as the node attribute of the node. Based on the conflict resolution path data, a resolution path attribute is added to the node corresponding to the conflicting land parcel unit, and the conflict resolution path data is stored as the resolution path attribute of the node. By summarizing all nodes, all adjacent edges, all containing edges, and all node attributes, an archival knowledge graph of the target region is constructed.

6. The method according to claim 1, characterized in that, The step of merging the extracted adjacent grid cells into planned land parcel cells of the target area, and generating land parcel spatial relationship data based on the spatial adjacency and spatial inclusion relationships between adjacent planned land parcel cells, includes: Based on the adjacent grid cells, construct a grid cell group to be merged, perform spatial connectivity analysis on the grid cell group to be merged, divide the spatially connected grid cell group to be merged into a spatial connectivity domain, and obtain multiple spatial connectivity domains; All grid cells within each spatial connectivity domain are merged into a single planning plot cell. A unique plot identifier is assigned to the merged planning plot cell, resulting in a planning plot cell with a plot identifier. Obtain the spatial boundary information of each planned land parcel unit with the land parcel identifier, and extract the spatial adjacency relationship between adjacent planned land parcel units due to shared boundaries based on the spatial boundary information of adjacent planned land parcel units; Based on the spatial boundary information of each planning plot unit and the spatial geometric features of each grid unit, the spatial inclusion relationship is extracted, in which the spatial boundary of one planning plot unit completely contains the spatial boundary of another planning plot unit. For each pair of planned land parcels that have a spatial adjacency relationship, an adjacency relationship edge is generated; for each pair of planned land parcels that have a spatial inclusion relationship, an inclusion relationship edge is generated; the adjacency relationship edges and the inclusion relationship edges are summarized to generate land parcel spatial relationship data.

7. The method according to claim 1, characterized in that, The step of associating the planning attributes in the planning thematic data of the target area with the gridded 3D geographic information data of the target area to generate 3D gridded data of the target area with planning attributes includes: Collect 3D geographic information data and planning thematic data of the target area; The three-dimensional geographic information data is processed into a grid and a spatial index is established. The planning attributes in the planning thematic data are associated with each grid cell of the spatial index to obtain three-dimensional grid data with planning attributes.

8. A knowledge graph construction system for archives, characterized in that, include: The association module is used to associate the planning attributes in the planning thematic data of the target area with the three-dimensional geographic information data of the target area after gridding, so as to generate and output the three-dimensional grid data of the target area with planning attributes; The extraction module is used to extract adjacent grid cells with similar spatial geometric features and the same planning label from the three-dimensional grid data with planning attributes, merge the extracted adjacent grid cells into planning plot cells of the target area, generate plot spatial relationship data according to the spatial adjacency and spatial inclusion relationships between adjacent planning plot cells, and output the planning plot cells and the plot spatial relationship data to the acquisition module, the determination module and the construction module. The acquisition module is used to acquire multiple planning tags corresponding to different planning thematic data for each planning plot unit, perform semantic comparison on multiple planning tags to identify conflicting planning tag combinations on the same planning plot unit, determine the corresponding conflict type based on the planning tag combination, and identify the planning plot unit with the planning tag combination as a conflicting plot unit. The determination module is used to determine the label association data between conflicting land parcel units and adjacent planned land parcel units based on the land parcel spatial relationship data, extract conflict resolution path data for resolving the conflicting land parcel units from the label association data based on the conflict type, and output the conflict resolution path data to the construction module; The construction module is used to construct an archival knowledge graph of the target area based on the planned land parcel units, the planning tags, the land parcel spatial relationship data, the conflict types, and the conflict resolution path data.

9. An electronic device, characterized in that, include: Memory, used to store computer programs; A processor, configured to implement the steps of the method for constructing an archival knowledge graph based on semantic understanding and deep information mining as described in any one of claims 1 to 7 when executing the computer program.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, enables the construction method of an archival knowledge graph based on semantic understanding and deep information mining as described in any one of claims 1 to 7.

Images