A cadastral data retrieval method and system based on a knowledge graph

Abstract

The embodiment of the application provides a cadastral data retrieval method and system based on a knowledge graph, and belongs to the technical field of cadastral data processing. The historical land semantics in a retrieval request is structurally analyzed, so that the request content such as historical number, historical right holder, historical location and historical adjacent relationship can be converted into data constraints that can be directly called in subsequent graph retrieval. The land state data, land evolution event data and land adjacency relationship data in the same cadastral business database are used to construct a time-series cadastral knowledge graph. An evolution corridor is generated around a seed state node, and conditional association propagation and candidate scoring processing are performed inside the evolution corridor, so that the retrieval process is no longer limited to static field matching, but can gradually locate the present status inheritance land corresponding to the historical semantics along the land evolution relationship, so that the output result not only has high accuracy, but also has clear inheritance basis and interpretability.

Description

Technical Field

[0001] This invention relates to the field of cadastral data processing technology, and more specifically to a cadastral data retrieval method and system based on knowledge graphs. Background Technology

[0002] Cadastral data retrieval is a fundamental supporting link in natural resource management, real estate registration, land consolidation, planning verification, and access to historical archives. In practice, the retrieval objects typically include land parcels, rights holders, location descriptions, boundary relationships, and various registration status information. Existing technologies commonly employ cadastral data retrieval methods, primarily including field-based retrieval based on parcel numbers, fuzzy matching based on rights holders or location descriptions, and general map-based retrieval based on association relationships.

[0003] In general scenarios, when the search request matches the current status of land parcels in the cadastral database, the above method can return the target result relatively directly. However, in the cadastral management process, land parcels are not always in a static state. Affected by business events such as division, merger, renumbering, boundary adjustments, and ownership changes, the same piece of land may correspond to different parcel numbers, different boundary forms, and different rights statuses at different times. In this case, when business personnel initiate a search, they may still use historical parcel numbers, historical rights holders, or historical adjacent relationships for description, while the target result exists in the current parcel status, thus causing a significant mismatch between the search conditions and the current cadastral data.

[0004] Existing field-based retrieval methods typically only match currently stored fields. When the search criteria pertain to historical semantics while the results pertain to current semantics, this can easily lead to missed or false matches. While conventional map retrieval methods can express the relationships between parcels, many implementations still emphasize static connections between entities, lacking a continuous representation of the evolution of parcel states over time. In particular, they lack the ability to constrain and judge inheritance relationships in complex evolutionary processes such as segmentation, merging, and renumbering. Consequently, when a historical parcel has evolved into multiple current parcels, the system often struggles to accurately identify the true successor parcel from multiple candidate objects.

[0005] Therefore, how to enable the retrieval system to accurately retrieve the corresponding current successor parcels based on historical semantic inputs such as historical parcel numbers, historical rights holder descriptions, or historical adjacency descriptions, while providing traceable evidence of succession, under the condition of using only core data within the same cadastral business database, remains a technical problem that urgently needs to be solved in the field of cadastral data retrieval. Summary of the Invention

[0006] The purpose of this invention is to provide a cadastral data retrieval method based on knowledge graphs, so as to at least solve the technical problem that existing cadastral data retrieval methods cannot accurately locate the current inheriting land parcels based on historical semantics after land parcels have undergone division, merging, renumbering or boundary adjustment.

[0007] To achieve the above objectives, the first aspect of the present invention provides a cadastral data retrieval method based on a knowledge graph, comprising:

[0008] The system acquires retrieval request data and parcel status data, parcel evolution event data, and parcel adjacency relationship data from the same cadastral business database. It then performs structured parsing on the retrieval request data to obtain an entity seed set, a time constraint set, and a relation constraint set.

[0009] A temporal cadastral knowledge graph is constructed based on the parcel status data, the parcel evolution event data, and the parcel adjacency relationship data, and the historical parcel status corresponding to the entity seed set is determined as the seed status node;

[0010] Based on the entity seed set, the time constraint set, and the relation constraint set, an evolutionary corridor is generated along the parcel evolution relationship in the temporal cadastral knowledge graph, and a candidate parcel state set is extracted based on the evolutionary corridor;

[0011] Based on the time constraint set and the relation constraint set, conditional association propagation and relevance calculation are performed on each candidate parcel state in the candidate parcel state set to obtain the candidate ranking result;

[0012] Based on the candidate ranking results, the historical succession chain is traced back, and consistency determination is performed in combination with the time constraint set and the relation constraint set. When the determination meets the preset conditions, the target cadastral retrieval result and the corresponding evidence chain are output.

[0013] Optionally, the retrieval request data is subjected to structured parsing to obtain an entity seed set, a time constraint set, and a relation constraint set, including:

[0014] Extract at least one of the historical parcel number, historical right holder identifier, historical location description, and historical adjacent relationship description from the search request data to form a request element set;

[0015] Based on the set of request elements, identify historical entity identifiers to form an entity seed set, identify time description information to form a time constraint set, and identify relationship description information to form a relationship constraint set;

[0016] The entity seed set, the time constraint set, and the relation constraint set are encoded to form a query representation, and the query representation is stored in association with the entity seed set, the time constraint set, and the relation constraint set.

[0017] Optionally, a temporal cadastral knowledge graph is constructed based on the parcel status data, the parcel evolution event data, and the parcel adjacency relationship data, and the historical parcel status corresponding to the entity seed set is determined as the seed status node, including:

[0018] Based on the occurrence time of the land parcel evolution events in the land parcel status data, the status of each basic entity of the land parcel in the land parcel status data is expanded to form multiple land parcel status nodes, and each land parcel status node corresponds to a valid time interval.

[0019] Configure status number, parcel number, boundary geometry, area, rights status, and location description attributes for each parcel status node;

[0020] Based on the parcel evolution event data, state continuity edges, split edges, merge edges, renumbering edges, and boundary adjustment edges are established between each parcel state node;

[0021] Based on the land parcel adjacency relationship data, adjacency edges are established between land parcel status nodes that correspond in time to form the temporal cadastral knowledge graph.

[0022] The historical entity identifiers in the entity seed set are mapped to the corresponding historical parcel status nodes in the temporal cadastral knowledge graph, and the mapped parcel status nodes are determined as the seed status nodes.

[0023] Optionally, based on the entity seed set, the time constraint set, and the relation constraint set, an evolutionary corridor is generated along the parcel evolution relationship in the temporal cadastral knowledge graph, including:

[0024] Starting from the seed state node, a constrained expansion is performed along the state continuity edge, split edge, merge edge, renumbering edge and boundary adjustment edge in the temporal cadastral knowledge graph;

[0025] For each evolutionary edge pointing from the parent state node to the child state node, calculate the inheritance mapping strength;

[0026] The time constraint set is used to remove evolution edges that conflict with the requested time semantics, and the relation constraint set is used to suppress expansion directions that do not meet the relation conditions;

[0027] Nodes and edges whose inheritance mapping strength meets the preset conditions are retained to form the evolution corridor.

[0028] Optionally, a set of candidate parcel states is extracted based on the evolutionary corridor, including:

[0029] For each candidate path from the seed state node to any reachable parcel state node, calculate the path succession strength;

[0030] The evolutionary inheritance value of the parcel state node is determined based on the product of the inheritance mapping strength of each candidate path.

[0031] Land parcel state nodes with evolution inheritance values ​​not lower than a preset threshold are identified as candidate land parcel state nodes, and the corresponding node set is identified as the candidate land parcel state set.

[0032] The evolution inheritance value corresponding to each candidate parcel state node is associated with and stored with the evolution corridor.

[0033] Optionally, based on the time constraint set and the relation constraint set, conditional association propagation and relevance calculation are performed on each candidate parcel state in the candidate parcel state set to obtain candidate ranking results, including:

[0034] Based on the query representation, time constraint set, and relation constraint set, conditional association propagation is performed on the nodes and their associated edges in the candidate parcel state set to obtain the contextual representation of each candidate parcel state under the current retrieval request conditions.

[0035] Calculate the semantic matching value, evolutionary inheritance value, adjacency satisfaction degree, and right status continuity satisfaction degree for each candidate parcel status.

[0036] The candidate parcel status relevance score is calculated based on the semantic matching value, the evolutionary inheritance value, the adjacency relationship satisfaction degree, and the continuous satisfaction degree of the right status;

[0037] The candidate land parcel status set is sorted from high to low according to the relevance score to form the candidate ranking result.

[0038] Optionally, tracing back the historical succession chain based on the candidate ranking results includes:

[0039] Select a predetermined number of high-scoring candidate parcel states from the candidate ranking results to form a high-scoring candidate set;

[0040] For each candidate parcel state in the high-scoring candidate set, backtrack all feasible paths between it and the seed state node from the evolution corridor;

[0041] For each feasible path, calculate the evidence chain score, determine the feasible path with the highest evidence chain score as the historical succession chain of the corresponding candidate land parcel status, and output the historical succession chain in association with the candidate land parcel status.

[0042] Optionally, performing a consistency determination by combining the set of time constraints and the set of relational constraints includes:

[0043] For the candidate parcel with the highest score in the candidate ranking results, extract its upstream state set, area information and adjacency relationship information from its corresponding historical succession chain;

[0044] The area inheritance residual is calculated based on the upstream state set and the area information, and the adjacency consistency is calculated based on the relationship constraint set and the actual adjacency relationship of the candidate parcel state.

[0045] If the area inheritance residual is not greater than a preset area threshold and the adjacency relationship consistency is not less than a preset relationship threshold, the candidate land parcel status is determined to meet the consistency judgment condition.

[0046] Optionally, when a preset condition is met, the target cadastral retrieval result and the corresponding evidence chain are output, specifically including:

[0047] When the highest-scoring candidate parcel status in the candidate ranking results meets the consistency judgment condition, the current parcel identifier, historical parcel identifier, historical succession chain, key evolution event sequence, relationship matching result, and result confidence level corresponding to the candidate parcel status are output.

[0048] When the highest-scoring candidate parcel status does not meet the consistency determination criteria, the next candidate parcel status is selected according to the candidate ranking results and the consistency determination is repeated until a candidate parcel status that meets the consistency determination criteria is obtained.

[0049] If none of the candidate parcel statuses meet the consistency criteria, the output will show that no target parcel satisfying the historical inheritance constraint was found.

[0050] A second aspect of the present invention provides a knowledge graph-based cadastral data retrieval system, the system being used to execute the aforementioned knowledge graph-based cadastral data retrieval method, the system comprising:

[0051] The parsing unit is used to acquire retrieval request data and parcel status data, parcel evolution event data and parcel adjacency relationship data from the same cadastral business database, and to perform structured parsing on the retrieval request data to obtain entity seed set, time constraint set and relation constraint set;

[0052] The construction unit is used to construct a temporal cadastral knowledge graph based on the parcel status data, the parcel evolution event data and the parcel adjacency relationship data, and to determine the historical parcel status corresponding to the entity seed set as seed status nodes;

[0053] The generation unit is used to generate an evolutionary corridor along the parcel evolution relationship in the temporal cadastral knowledge graph based on the entity seed set, the time constraint set, and the relation constraint set, and to extract a candidate parcel state set based on the evolutionary corridor;

[0054] The sorting unit is used to perform conditional association propagation and relevance calculation on each candidate parcel state in the candidate parcel state set based on the time constraint set and the relation constraint set, so as to obtain the candidate sorting result;

[0055] The output unit is used to trace back the historical succession chain based on the candidate ranking result, and perform consistency determination in combination with the time constraint set and the relation constraint set. When the determination meets the preset conditions, the target cadastral retrieval result and the corresponding evidence chain are output.

[0056] Through the above technical solutions, this invention proposes a cadastral data retrieval method and system based on a knowledge graph. By performing structured parcel semantic analysis on the historical parcels in the retrieval request, the request content, such as historical numbers, historical rights holders, historical locations, and historical adjacency relationships, can be transformed into data constraints that can be directly invoked in subsequent graph retrieval. Based on this, a temporal cadastral knowledge graph is constructed using parcel status data, parcel evolution event data, and parcel adjacency relationship data from the same cadastral business database, allowing the historical, transitional, and current states of parcels to be continuously expressed within the same graph structure. By generating evolutionary corridors around seed state nodes and performing conditional association propagation and candidate scoring processing within these corridors, the retrieval process is no longer limited to static field matching but can gradually locate the current successor parcels corresponding to the historical semantics along the parcel evolution relationship. By tracing back the historical succession chain of high-scoring candidate parcel states and performing consistency judgment in conjunction with area succession relationships and adjacency relationships, the output results not only have high accuracy but also clear succession basis and interpretability.

[0057] Other features and advantages of the embodiments of the present invention will be described in detail in the following detailed description section. Attached Figure Description

[0058] The accompanying drawings are provided to further illustrate embodiments of the present invention and form part of the specification. They are used together with the following detailed description to explain the embodiments of the present invention, but do not constitute a limitation thereof. In the drawings:

[0059] Figure 1 This is a flowchart illustrating the steps of a knowledge graph-based cadastral data retrieval method according to one embodiment of the present invention.

[0060] Figure 2This is a system architecture diagram of a knowledge graph-based cadastral data retrieval system provided in one embodiment of the present invention. Detailed Implementation

[0061] The specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0062] like Figure 1 As shown, this invention provides a cadastral data retrieval method based on a knowledge graph. This method is primarily applicable to scenarios where cadastral parcels have undergone splitting, merging, renumbering, boundary adjustments, or changes in rights status, but the retrieval request still uses historical parcel numbers, historical rights holder descriptions, historical location descriptions, or historical adjacency descriptions as input conditions. In this scenario, traditional retrieval methods that directly compare current parcel fields are prone to problems such as missing results, too many candidate results, or candidate results lacking genuine succession relationships. This invention, by uniformly organizing parcel status data, parcel evolution event data, and parcel adjacency relationship data within the same cadastral business database, constructs a temporal cadastral knowledge graph capable of expressing the succession relationships between historical, transitional, and current states. Furthermore, it organizes subsequent data processing flows around the historical semantics in the retrieval request, thereby achieving accurate retrieval of current successor parcels.

[0063] In this embodiment of the invention, core cadastral data from the same cadastral business database is preferably used for processing. The core cadastral data includes at least parcel status data, parcel evolution event data, and parcel adjacency relationship data. Specifically, the parcel status data records information such as parcel number, boundary geometry, area, rights status, and location description corresponding to parcels at different effective periods; the parcel evolution event data records the event type and effective time when parcels are divided, merged, renumbered, have boundaries adjusted, or undergo ownership changes; and the parcel adjacency relationship data describes the spatial contact, adjacency, or inclusion relationships between parcels under corresponding time conditions.

[0064] The method provided by the embodiments of the present invention includes the following steps:

[0065] S10: Obtain the retrieval request data and the parcel status data, parcel evolution event data and parcel adjacency relationship data from the same cadastral business database, and perform structured parsing on the retrieval request data to obtain the entity seed set, time constraint set and relation constraint set.

[0066] Specifically, the process involves acquiring retrieval request data along with parcel status data, parcel evolution event data, and parcel adjacency relationship data from the same cadastral business database. At least one of the following is extracted from the retrieval request data: historical parcel number, historical right holder identifier, historical location description, and historical adjacency relationship description, forming a request element set. Based on the request element set, historical entity identifiers are identified to form an entity seed set; time description information is identified to form a time constraint set; and relationship description information is identified to form a relationship constraint set. The entity seed set, the time constraint set, and the relationship constraint set are encoded to form a query representation, which is then associated and stored with the entity seed set, the time constraint set, and the relationship constraint set.

[0067] In this embodiment of the invention, step S10 is used to complete the input preparation for the entire cadastral retrieval process. It should be noted that the retrieval request is no longer simply regarded as a string of text content to be matched, but is further parsed into entity constraints, time constraints, and relational constraints that can participate in subsequent map calculations, so that the retrieval input is transformed from static string conditions into a computable retrieval intent representation.

[0068] In practical applications, retrieval request data can originate from historical parcel numbers, historical right holder names, original location descriptions, or descriptions of historical relationships such as being adjacent to or derived from a parcel, entered by business personnel in the cadastral retrieval interface. Alternatively, it can originate from structured requests automatically transmitted during historical archive retrieval, registration review, or real estate verification processes. Specifically, the system first identifies explicit entity content in the retrieval request, organizing data that can be directly mapped to parcels, right holders, or related historical identifiers into a set of entity seeds. Simultaneously, it identifies whether the request contains time-limited information, such as statements like "before a certain year," "before the change," "at the original registration time," or "in a historical state," and organizes these into a set of time constraints. Further identify the relational constraints involved in the request, such as expressions like adjacency, origin, inheritance, and derived from, and organize them into a set of relational constraints. Through the above processing, the subsequent retrieval process not only knows what to search for, but also knows the historical time range and the semantic relationships to search for.

[0069] Furthermore, in one executable implementation, to make the entity seed set Time constraint set Set of relational constraints This invention also encodes the data into a unified query representation, enabling it to be directly invoked in subsequent steps within the same data structure. It can be in the following form:

[0070] ;

[0071] Where s represents the entity seed feature vector, t represents the time constraint feature vector, and r represents the relation constraint feature vector. , and ...

[0072] In one specific implementation, suppose the search request entered by the business user is "to query the current status of the land parcel after the subdivision of the original land parcel A-102, requiring it to still be adjacent to the original land parcel B-204 and correspond to the land parcel under the name of the original rights holder A". In this implementation, the system can identify "original land parcel A-102" and "original rights holder A" as entity seed content, forming an entity seed set. Identify the historical to current temporal semantics corresponding to the "divided current land parcels" to form a set of temporal constraints. Identify the relational constraints corresponding to "still adjacent to the original parcel B-204", and form a set of relational constraints. After the above data is generated, it is then uniformly encoded to form a query representation. This serves as the data entry point for subsequent steps.

[0073] S20: Construct a temporal cadastral knowledge graph based on the parcel status data, the parcel evolution event data, and the parcel adjacency relationship data, and determine the historical parcel status corresponding to the entity seed set as seed status nodes.

[0074] Specifically, based on the occurrence time of parcel evolution events in the parcel status data, the basic entities of each parcel in the parcel status data are expanded to form multiple parcel status nodes, each corresponding to a valid time interval; each parcel status node is configured with a status number, parcel number, boundary geometry, area, rights status, and location description attributes; based on the parcel evolution event data, continuous state edges, split edges, merge edges, renumbered edges, and boundary adjustment edges are established between each parcel status node; based on the parcel adjacency relationship data, adjacency edges are established between temporally corresponding parcel status nodes to form the temporal cadastral knowledge graph; historical entity identifiers in the entity seed set are mapped to corresponding historical parcel status nodes in the temporal cadastral knowledge graph, and the mapped parcel status nodes are determined as the seed status nodes.

[0075] In this embodiment of the invention, step S20 is used to organize the original cadastral business data into a temporal cadastral knowledge graph that can express the historical state change process of land parcels. This allows the retrieval process to move beyond the static land parcel entity level and instead identify the land parcel inheritance relationship along the historical state chain. It should be noted that traditional cadastral retrieval systems typically treat a land parcel as a fixed object. However, in real cadastral business, a land parcel may correspond to different land parcel numbers, different boundary geometries, and different rights states at different times. If a single static entity is still used to represent it, the continuous process of a land parcel evolving from its historical state to its current state cannot be accurately described. Therefore, this invention introduces a method for constructing land parcel state nodes in this step, expanding and organizing the state of a land parcel within different effective time intervals into independent nodes in the graph.

[0076] In practical applications, the system first reads parcel status data, parcel evolution event data, and parcel adjacency relationship data. Then, it unfolds the status of each parcel's basic entity according to the chronological order of the parcel evolution events. Specifically, the states of the same parcel at different effective periods can be constructed into multiple parcel status nodes. Each parcel status node carries attribute information such as status number, parcel number, boundary geometry, area, rights status, and location description. Through this method, historical information originally compressed into a single static parcel record is decomposed into a chronologically ordered state chain, thus providing a foundation for subsequent historical succession relationship analysis.

[0077] Furthermore, after completing node construction, the system establishes evolutionary edge relationships between parcel state nodes based on parcel evolution event data. Specifically, it establishes state continuity edges between state nodes belonging to the same parcel and changing continuously over time; it establishes splitting edges when one parcel state evolves into multiple parcel states; it establishes merging edges when multiple parcel states merge into one parcel state; it establishes renumbering edges when parcel numbers change but the parcel entity continues to exist; and it establishes boundary adjustment edges when boundaries are corrected or partially adjusted but the inheritance relationship of the parcel entity is not interrupted. Simultaneously, the system also establishes adjacency edges between parcel state nodes that have spatial adjacency under the same valid time conditions based on parcel adjacency relationship data. Through the above processing, a temporal cadastral knowledge graph G that takes into account parcel states, evolutionary relationships, and spatial relationships can be formed.

[0078] After forming the map, in this embodiment of the invention, the entity seed set obtained in step S10 is further... This is mapped to the corresponding historical parcel status nodes in the temporal cadastral knowledge graph. Specifically, if the entity seed set contains historical parcel numbers, the system first searches the graph for a set of parcel numbers that match and conform to the time constraints. The land parcel state nodes are determined; if the entity seed set also contains historical rights holder or location information, this information is further used to constrain the candidate historical state nodes to improve mapping accuracy. Finally, the successfully mapped historical land parcel state nodes are determined as seed state nodes and serve as the starting nodes for subsequent evolutionary relationship expansion in the graph.

[0079] It should be noted that the construction of the temporal cadastral knowledge graph in this step is not merely for the purpose of storing graph data, but rather to make the business fact of how historical parcels evolved into current parcels computable in terms of data structure. In other words, the graph G constructed in this step is the direct basis for generating evolutionary corridors in the subsequent step S30, and also the source of the original graph data for calculating candidate relevance in step S40.

[0080] In one specific implementation, suppose the land parcel status data records a historical land parcel A-102 in time... to The interior is a complete parcel of land, in time A partitioning event occurred, dividing the land into two new parcel states, A-102-1 and A-102-2. Subsequently, A-102-2 was... A renumbering occurred, changing the current parcel number to C-501. In this step, the system will sequentially construct the historical parcel status node, the segmented sub-parcel status nodes, and the renumbered current parcel status node, connecting them via segmentation edges and renumbering edges. If the search request contains historical parcel number A-102, the system can map the historical status node corresponding to A-102 as a seed status node for subsequent steps to continue expanding along the graph.

[0081] S30: Based on the entity seed set, the time constraint set, and the relation constraint set, generate an evolution corridor along the parcel evolution relationship in the temporal cadastral knowledge graph, and extract a candidate parcel state set based on the evolution corridor.

[0082] Specifically, starting from the seed state node, constrained expansion is performed along the state continuity edges, split edges, merge edges, renumbering edges, and boundary adjustment edges in the temporal cadastral knowledge graph; for each evolutionary edge pointing from the parent state node to the child state node, the inheritance mapping strength is calculated; the time constraint set is used to remove evolutionary edges that conflict with the requested time semantics, and the relation constraint set is used to suppress expansion directions that do not meet the relation conditions; nodes and edges whose inheritance mapping strength meets the preset conditions are retained to form the evolutionary corridor.

[0083] Furthermore, for each candidate path from the seed state node to any reachable parcel state node, the path inheritance strength is calculated; based on the product of the inheritance mapping strengths of each candidate path, the evolutionary inheritance value of the parcel state node is determined; parcel state nodes with evolutionary inheritance values ​​not lower than a preset threshold are determined as candidate parcel state nodes, and the corresponding node set is determined as the candidate parcel state set; the evolutionary inheritance value corresponding to each candidate parcel state node is associated and stored with the evolutionary corridor.

[0084] In this embodiment of the invention, step S30 is used to extract the candidate subgraph most relevant to the current retrieval request from the complete temporal cadastral knowledge graph, avoiding indiscriminate expansion of the entire graph in subsequent searches, thereby reducing interference from irrelevant nodes. It should be noted that when historical parcels have undergone multiple splits, mergers, renumberings, and boundary adjustments, a direct breadth-first search of the complete graph easily yields a large number of reachable but not truly inherited candidate parcel states. Therefore, in this step, the invention generates evolutionary corridors through inheritance mapping strength constraints. This allows subsequent processing to focus on nodes that are truly likely to inherit the target historical land parcel.

[0085] In practical applications, the system first starts with the seed state node determined in step S20 and performs constrained expansion along the state continuity edges, splitting edges, merging edges, renumbering edges, and boundary adjustment edges in the temporal cadastral knowledge graph. During the expansion process, the system does not treat every edge equally; instead, it calculates the inheritance mapping strength for each evolutionary edge pointing from parent state node u to child state node v. In one executable implementation, the inheritance mapping strength satisfies:

[0086] ;

[0087] in, This represents the area inherited by the parent state node u from the child state node v. This represents the total area of ​​the parent state node u. This represents the boundary length inherited from parent state node u to child state node v. I(u,v) represents the total boundary length of the parent state node u, and I(u,v) represents the continuity indicator value of the right state between the parent state node u and the child state node v. , and, , where is the weighting coefficient. It should be noted that this formula is used to comprehensively measure the credibility of the inheritance of an evolutionary edge from three aspects: area inheritance, boundary inheritance, and continuity of rights status, thereby avoiding one-sided judgments based solely on area overlap or changes in parcel number.

[0088] In this embodiment of the invention, the time constraint set Set of relational constraints This process also involves expansion control in this step. Specifically, for evolutionary edges that clearly conflict with the time constraint set, such as future state edges that occur after the time limit specified in the retrieval request or that are inconsistent with historical semantics, the system directly eliminates them. For expansion directions that are inconsistent with the relation constraint set, such as when the retrieval request emphasizes that the extension is still adjacent to a certain historical parcel, but a candidate expansion direction no longer maintains that adjacency under the corresponding time conditions, the system suppresses or lowers the retention priority of that expansion direction. Through the above methods, the generation process of the evolutionary corridor is simultaneously constrained by both historical time semantics and historical relation semantics, thus better reflecting real cadastral retrieval scenarios.

[0089] After forming the extended path, this embodiment of the invention further processes the path from the seed state node to any reachable parcel state node. The path inheritance strength is calculated, and the evolution inheritance value of the state node of the parcel is determined accordingly. In one executable implementation, the evolved inheritance value satisfies:

[0090] ;

[0091] in, Represents the set of seed state nodes To the parcel status node The set of all feasible paths, Indicates the path The strength of the inheritance mapping of the corresponding evolutionary edges is considered. The technical principle behind this approach is that the reliability of a path depends not only on whether the endpoint and the starting point are reachable, but also on whether each evolutionary relationship within the path has sufficient inheritance basis. If the inheritance mapping strength of a certain edge in the path is very low, that edge will significantly reduce the overall reliability of the entire path, thereby suppressing erroneous candidates that are locally similar but have discontinuous overall inheritance relationships.

[0092] In this embodiment of the invention, the system will evolve and inherit values. Land parcel status nodes that are not lower than a preset threshold are retained as candidate land parcel status nodes, and these nodes are combined into a candidate land parcel status set. At the same time, the system will also retain these nodes and their interconnections, forming an evolutionary corridor around the current retrieval request. It should be noted that the evolutionary corridor It is not the final result, but rather a range-controlled and semantically continuous candidate space for subsequent candidate scores; a set of candidate parcel states. This is the direct input for step S40 to perform the candidate state correlation calculation.

[0093] In one specific implementation, assuming the seed state node corresponds to historical parcel A-102, there are three reachable paths in the graph: the first path is to obtain current parcel C-501 by dividing and renumbering A-102; the second path is to form current parcel C-503 by boundary adjustment of A-102; the third path, although reachable on the graph, only connects to current parcel D-117 through a single low-inheritance-strength merging event. In this implementation, after calculating the path inheritance strength for each of the above paths, the system may find that C-501 and C-503 have higher evolutionary inheritance values, while D-117 has a significantly lower evolutionary inheritance value. Therefore, the system retains the parcel state nodes corresponding to C-501 and C-503 in the candidate parcel state set, while removing the node corresponding to D-117 from the evolutionary corridor or downgrading it to a low-priority candidate, thereby reducing noise interference for subsequent steps.

[0094] S40: Based on the time constraint set and the relation constraint set, perform conditional association propagation and correlation calculation on each candidate parcel state in the candidate parcel state set to obtain the candidate ranking result.

[0095] Specifically, based on the query representation, time constraint set, and relation constraint set, conditional association propagation is performed on the nodes and their associated edges in the candidate parcel state set to obtain the contextual representation of each candidate parcel state under the current retrieval request conditions; the semantic matching value, evolutionary inheritance value, adjacency satisfaction degree, and right state continuity satisfaction degree of each candidate parcel state are calculated respectively; the relevance score of the candidate parcel state is calculated according to the semantic matching value, the evolutionary inheritance value, the adjacency satisfaction degree, and the right state continuity satisfaction degree; the candidate parcel state set is sorted from high to low according to the relevance score to form the candidate ranking result.

[0096] In this embodiment of the invention, the step is used to obtain the candidate parcel state set in step S30. and evolutionary corridor Based on this, the matching degree between each candidate land parcel status and the current search request is comprehensively calculated, and a candidate ranking result with a sequential order is output. It should be noted that although the candidate land parcel status set obtained after step S30 has eliminated most significantly irrelevant nodes, it may still contain multiple candidate land parcel statuses with certain inheritance relationships. For example, a historical land parcel may have been divided into multiple current land parcels, or multiple candidate land parcels may have certain correspondences with historical semantics in terms of area and location. Therefore, this step requires further fine-grained ranking of the candidate results from multiple dimensions such as semantics, time, adjacency relationships, and continuity of right status.

[0097] In practical applications, the system first processes the query representation obtained in step S10. Time constraint set Set of relational constraints and the evolutionary inheritance value obtained in step S30 This information is then input into the candidate subgraph processing. Specifically, within the subgraph corresponding to the candidate parcel state set, conditional association propagation is performed on the candidate parcel state nodes and their surrounding edges. This conditional association propagation is not unconstrained propagation throughout the entire graph, but rather within the evolution corridor. Within the defined subgraph, each candidate parcel state node retains its own attribute information while also absorbing contextual information reflected by its upstream inheritance relationships, adjacency relationships, and corresponding time constraints. In this way, the final representation of a candidate parcel state is no longer determined solely by its current parcel number, area, or location fields, but rather reflects the inheritance chain in which the parcel state exists, whether it satisfies the historical relationship constraints in the retrieval request, and whether its surrounding adjacency relationships are continuous. After obtaining the contextual representation of each candidate parcel state, the system calculates the semantic matching value, evolutionary inheritance value, adjacency satisfaction degree, and right status continuity satisfaction degree for each candidate parcel state. Furthermore, in one executable implementation, the relevance score of the i-th candidate parcel state... It can be represented as: ;

[0098] in, Represents the status and query representation of the i-th candidate land parcel. semantic matching value, This represents the evolutionary inheritance value of the i-th candidate parcel state. Let represent the adjacency satisfaction degree of the i-th candidate parcel state. This represents the degree of continuous satisfaction of the rights status of the i-th candidate land parcel. , , and These are the weighting coefficients.

[0099] In this embodiment of the invention, semantic matching value This primarily reflects the degree of consistency between the candidate land parcel status and the semantics of the search request in terms of parcel number, location, rights holder, and related descriptions; evolutionary inheritance value. This primarily reflects whether the candidate land parcel's state truly originates from historical seed state nodes along the inheritance path; and the degree of adjacency satisfaction. This primarily reflects whether the candidate land parcel status maintains the adjacency or spatial relationship characteristics required in the search request; and the degree of continuity in the satisfaction of rights status. This primarily reflects whether the status of candidate land parcels has continuity or a reasonable inheritance relationship with historical claims in the rights chain. By comprehensively weighting the above four dimensions, erroneous conclusions can be effectively avoided by relying solely on the similarity of a single field or the single spatial proximity.

[0100] After calculating the relevance score, the system calculates the relevance score based on the relevance score. Set of candidate parcel states The candidates are sorted from highest to lowest to obtain the ranking results. It should be noted that the output of this step is not the final retrieval conclusion, but rather provides a priority order for performing evidence chain backtracking and consistency determination in step S50.

[0101] In one specific implementation, if step S30 yields two highly reliable candidate parcel states C-501 and C-503, where C-501 has a high evolutionary inheritance value with historical parcel A-102 and still maintains its adjacency with the original parcel B-204, while C-503, although also originating from historical parcel A-102, has undergone significant changes in its current adjacency relationship, then in this step, the system will, based on... , , and Based on the comprehensive calculation results, C-501 is assigned a higher relevance score, thus placing it in a higher position in the candidate ranking results.

[0102] S50: Based on the candidate ranking results, trace back the historical succession chain, and perform consistency determination in combination with the time constraint set and the relation constraint set. When the determination meets the preset conditions, output the target cadastral retrieval result and the corresponding evidence chain.

[0103] Specifically, a predetermined number of high-scoring candidate land parcel states are selected from the candidate ranking results to form a high-scoring candidate set; for each candidate land parcel state in the high-scoring candidate set, all feasible paths between it and the seed state node are traced back from the evolution corridor; an evidence chain score is calculated for each feasible path, and the feasible path with the largest evidence chain score is determined as the historical succession chain of the corresponding candidate land parcel state, and the historical succession chain is associated with the candidate land parcel state and output.

[0104] Furthermore, for the candidate parcel state with the highest score in the candidate ranking results, the upstream state set, area information, and adjacency relationship information in its corresponding historical succession chain are extracted; the area succession residual is calculated based on the upstream state set and the area information, and the adjacency relationship consistency is calculated based on the relationship constraint set and the actual adjacency relationship of the candidate parcel state; if the area succession residual is not greater than a preset area threshold and the adjacency relationship consistency is not less than a preset relationship threshold, the candidate parcel state is determined to meet the consistency judgment condition.

[0105] Furthermore, when the highest-scoring candidate parcel status in the candidate ranking results meets the consistency determination condition, the current parcel identifier, historical parcel identifier, historical succession chain, key evolution event sequence, relationship matching result, and result confidence level corresponding to the candidate parcel status are output; when the highest-scoring candidate parcel status does not meet the consistency determination condition, the next candidate parcel status is selected according to the order of the candidate ranking results and the consistency determination is repeated until a candidate parcel status that meets the consistency determination condition is obtained; when none of the candidate parcel statuses meet the consistency determination condition, the result that no target parcel satisfying the historical succession constraint was found is output.

[0106] In this embodiment of the invention, step S50 is used to further determine the target parcel, output the inheritance basis, and verify the business consistency of the results based on the candidate ranking results. It should be noted that in cadastral business scenarios, simply providing a current parcel number is usually insufficient to support business verification, especially when historical parcels have undergone multiple evolutions. Business personnel also need to know why the parcel was identified as a search result, its inheritance path, and whether it is consistent with the area, time, and adjacency relationships involved in historical requests. Therefore, this invention introduces an evidence chain backtracking and consistency determination mechanism in this step.

[0107] In practical applications, the system first selects a preset number of high-scoring candidate parcel states from the candidate ranking results obtained in step S40, forming a high-scoring candidate set. For each candidate parcel state in the high-scoring candidate set, the system uses the evolution corridor obtained in step S30. The process involves backtracking all feasible paths between the current state node and the seed state node, and calculating an evidence chain score for each feasible path. In one executable implementation, the evidence chain score can be expressed as:

[0108] ;

[0109] in, Representing a path The strength of the inheritance mapping of the evolutionary edge. Represents a small constant. Indicates path length. Representing a path For time constraint set The degree of violation, and This is the penalty coefficient. The principle behind this formula is that if each edge in a path has a high succession mapping strength, its evidence chain score will be high; if the path is too long, has too many detours, or has a significant conflict with the retrieval request in terms of time sequence, its evidence chain score will be lowered. Therefore, the system can prioritize succession chains with more direct succession relationships, more reasonable time logic, and overall simplicity as the target evidence chain.

[0110] In this embodiment of the invention, the system determines the feasible path with the highest evidence chain score as the historical succession chain of the corresponding candidate land parcel state, and records this historical succession chain in association with the candidate land parcel state. In this way, each high-scoring candidate land parcel state no longer corresponds to only one score, but also to a historical succession chain that can be used to explain its origin. Through this processing, the system can then clearly show business personnel: from which historical land parcel state did the target land parcel originate, through which splitting, merging, renumbering, or boundary adjustment events were performed, ultimately forming the current candidate land parcel state.

[0111] After completing the evidence chain backtracking, this embodiment of the invention further performs a consistency determination on the highest-scoring candidate parcel status in the candidate ranking results. Specifically, the system extracts the upstream status set, area information, and adjacency relationship information from the historical succession chain corresponding to the candidate parcel status, and calculates the area succession residual accordingly. Consistency of Adjacency In one executable implementation, the above two metrics can be expressed as:

[0112] ;

[0113] ;

[0114] in, This indicates the current area of ​​the candidate land parcel. This indicates the inherited area obtained by tracing back from the historical succession chain. This represents the set of target adjacency relations extracted based on the set of relational constraints. This represents the set of actual adjacency relationships for candidate parcels. This represents a small constant. It should be noted that the area inheritance residual is mainly used to verify whether the area of ​​a candidate land parcel can be reasonably explained by historical inheritance relationships, while the adjacency consistency is mainly used to verify whether the candidate land parcel status still maintains the spatial relationship characteristics emphasized in the request. If a candidate land parcel status has a high score, but its area is significantly inconsistent with the historical inheritance chain, or its current adjacency relationship deviates significantly from the request semantics, it indicates that the candidate result only appears similar and is not the target land parcel that truly meets the business retrieval requirements.

[0115] In this embodiment of the invention, when the area inheritance residual is not greater than a preset area threshold and the adjacency consistency is not less than a preset relationship threshold, the system determines that the current candidate parcel status meets the consistency judgment condition and outputs it as the final target cadastral retrieval result. Simultaneously, the system also outputs the current parcel identifier, historical parcel identifier, historical inheritance chain, key evolutionary event sequence, relationship matching result, and result confidence level corresponding to the candidate parcel status. If the current highest-scoring candidate parcel status does not meet the consistency judgment condition, the system does not terminate immediately but selects the next candidate parcel status according to the candidate ranking result and repeats the consistency judgment until a candidate result that meets the consistency judgment condition is found. If all candidate parcel statuses fail the consistency judgment, the system outputs that no target parcel satisfying the historical inheritance constraint was found.

[0116] In one specific implementation, assume that the highest-scoring candidate parcel status output in step S40 is C-501, and the second highest-scoring candidate parcel status is C-503. The system first backtracks C-501 to obtain its historical succession chain, which originates from A-102 through one segmentation and one renumbering. It then calculates the area succession residual and adjacency consistency, finding that C-501 can not only be explained by the historical succession chain in terms of area but also maintains an adjacency relationship with the original parcel B-204. Therefore, the system determines it as the final retrieval result. If C-501 has the highest relevance score but its current adjacency relationship no longer meets the retrieval request, while C-503 passes the consistency determination, then the system outputs C-503 as the final result. Through this method, the present invention outputs not only the result object itself but also a complete retrieval conclusion including the result object, the succession basis, and the consistency conclusion, thereby making the results more consistent with the verification requirements of cadastral business.

[0117] To more clearly illustrate the operation of this invention, a specific embodiment is described below. In this embodiment, a set of land parcel data from a certain cadastral business database is selected as a sample. The data includes land parcel status data, land parcel evolution event data, and land parcel adjacency relationship data. Assume that historical land parcel A-102 was a complete land parcel in its early state, with its boundary adjacent to land parcel B-204, and the registered right holder was A. In subsequent business processes, A-102 underwent one division and one numbering adjustment, forming two current land parcel states, C-501 and C-503. Only C-501 maintains its adjacency relationship with B-204 in its current state.

[0118] In this embodiment, the business user inputs a search request: "Query the current status of the land parcel after the original land parcel A-102 has been divided, requiring it to still be adjacent to the original land parcel B-204 and correspond to the land parcel under the name of the original rights holder A." The system first performs structured parsing on the search request in step S10 to obtain a set of entity seeds. Time constraint set Set of relational constraints and query representation Subsequently, in step S20, the system constructs a temporal cadastral knowledge graph based on parcel status data, parcel evolution event data, and parcel adjacency relationship data in the cadastral business database, and determines the historical parcel status node corresponding to historical parcel A-102 as the seed status node.

[0119] In step S30, the system uses the historical state node corresponding to A-102 as the starting point and generates an evolutionary corridor along the dividing edge, renumbering edge, and state continuity edge in the graph. It then calculates the inheritance mapping strength and evolutionary inheritance value from A-102 to C-501 and from A-102 to C-503, respectively. The calculations show that both C-501 and C-503 are located within the evolutionary corridor, but C-501 has a higher evolutionary inheritance value, and its path is more stable in terms of boundary inheritance and right state continuity. Subsequently, in step S40, the system calculates the relevance by comprehensively considering the semantic matching value, evolutionary inheritance value, adjacency satisfaction, and right state continuity satisfaction of the candidate parcel states. The results show that C-501 has a higher relevance score than C-503.

[0120] In step S50, the system further traces back the historical succession chain of C-501 and calculates its evidence chain score, area succession residual, and adjacency consistency. Finally, it outputs C-501 as the target cadastral retrieval result, along with its historical succession chain formed by segmenting and renumbering A-102, and corresponding key event descriptions. This invention can accurately locate the current successor parcel even when there is a significant misalignment between the historical semantic input and the current parcel status, and output succession evidence for business review.

[0121] like Figure 2 As shown, this invention provides a knowledge graph-based cadastral data retrieval system. The system is used to execute the aforementioned knowledge graph-based cadastral data retrieval method. The system includes:

[0122] The parsing unit is used to acquire retrieval request data and parcel status data, parcel evolution event data and parcel adjacency relationship data from the same cadastral business database, and to perform structured parsing on the retrieval request data to obtain entity seed set, time constraint set and relation constraint set;

[0123] The construction unit is used to construct a temporal cadastral knowledge graph based on the parcel status data, the parcel evolution event data and the parcel adjacency relationship data, and to determine the historical parcel status corresponding to the entity seed set as seed status nodes;

[0124] The generation unit is used to generate an evolutionary corridor along the parcel evolution relationship in the temporal cadastral knowledge graph based on the entity seed set, the time constraint set, and the relation constraint set, and to extract a candidate parcel state set based on the evolutionary corridor;

[0125] The sorting unit is used to perform conditional association propagation and relevance calculation on each candidate parcel state in the candidate parcel state set based on the time constraint set and the relation constraint set, so as to obtain the candidate sorting result;

[0126] The output unit is used to trace back the historical succession chain based on the candidate ranking result, and perform consistency determination in combination with the time constraint set and the relation constraint set. When the determination meets the preset conditions, the target cadastral retrieval result and the corresponding evidence chain are output.

[0127] Those skilled in the art will understand that all or part of the steps in the methods of the above embodiments can be implemented by a program instructing related hardware. This program is stored in a storage medium and includes several instructions to cause a microcontroller, chip, or processor to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as a USB flash drive, a portable hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0128] The optional embodiments of the present invention have been described in detail above with reference to the accompanying drawings. However, the embodiments of the present invention are not limited to the specific details described above. Within the scope of the technical concept of the embodiments of the present invention, various simple modifications can be made to the technical solutions of the embodiments of the present invention, and these simple modifications all fall within the protection scope of the embodiments of the present invention. It should also be noted that the various specific technical features described in the above specific embodiments can be combined in any suitable manner without contradiction. To avoid unnecessary repetition, the embodiments of the present invention will not further describe the various possible combinations.

[0129] Furthermore, various different embodiments of the present invention can be combined in any way, as long as they do not violate the spirit of the embodiments of the present invention, they should also be regarded as the content disclosed by the embodiments of the present invention.

Claims

1. A cadastral data retrieval method based on knowledge graphs, characterized in that, include: The system acquires retrieval request data and parcel status data, parcel evolution event data, and parcel adjacency relationship data from the same cadastral business database. It then performs structured parsing on the retrieval request data to obtain an entity seed set, a time constraint set, and a relation constraint set. A temporal cadastral knowledge graph is constructed based on the parcel status data, the parcel evolution event data, and the parcel adjacency relationship data, and the historical parcel status corresponding to the entity seed set is determined as the seed status node; Specifically, this includes: expanding the state of each basic entity in the land parcel status data according to the occurrence time of land parcel evolution events in the land parcel status data to form multiple land parcel status nodes, each corresponding to a valid time interval; configuring status number, land parcel number, boundary geometry, area, rights status, and location description attributes for each land parcel status node; establishing state continuity edges, split edges, merge edges, renumbering edges, and boundary adjustment edges between each land parcel status node based on the land parcel evolution event data; establishing adjacency edges between temporally corresponding land parcel status nodes based on the land parcel adjacency relationship data to form the temporal cadastral knowledge graph; mapping historical entity identifiers in the entity seed set to corresponding historical land parcel status nodes in the temporal cadastral knowledge graph, and determining the mapped land parcel status nodes as the seed status nodes; Based on the entity seed set, the time constraint set, and the relation constraint set, an evolutionary corridor is generated along the parcel evolution relationship in the temporal cadastral knowledge graph, and a candidate parcel state set is extracted based on the evolutionary corridor; Based on the time constraint set and the relation constraint set, conditional association propagation and relevance calculation are performed on each candidate parcel state in the candidate parcel state set to obtain the candidate ranking result; Based on the candidate ranking results, the historical succession chain is traced back, and consistency determination is performed in combination with the time constraint set and the relation constraint set. When the determination meets the preset conditions, the target cadastral retrieval result and the corresponding evidence chain are output.

2. The cadastral data retrieval method based on knowledge graphs according to claim 1, characterized in that, The retrieval request data is structured and parsed to obtain a set of entity seeds, a set of time constraints, and a set of relation constraints, including: Extract at least one of the historical parcel number, historical right holder identifier, historical location description, and historical adjacent relationship description from the search request data to form a request element set; Based on the set of request elements, identify historical entity identifiers to form an entity seed set, identify time description information to form a time constraint set, and identify relationship description information to form a relationship constraint set; The entity seed set, the time constraint set, and the relation constraint set are encoded to form a query representation, and the query representation is stored in association with the entity seed set, the time constraint set, and the relation constraint set.

3. The cadastral data retrieval method based on knowledge graphs according to claim 1, characterized in that, Based on the entity seed set, the time constraint set, and the relation constraint set, an evolutionary corridor is generated along the parcel evolution relationship in the temporal cadastral knowledge graph, including: Starting from the seed state node, a constrained expansion is performed along the state continuity edge, split edge, merge edge, renumbering edge and boundary adjustment edge in the temporal cadastral knowledge graph; For each evolutionary edge pointing from the parent state node to the child state node, calculate the inheritance mapping strength; The time constraint set is used to remove evolution edges that conflict with the requested time semantics, and the relation constraint set is used to suppress expansion directions that do not meet the relation conditions; Nodes and edges whose inheritance mapping strength meets the preset conditions are retained to form the evolution corridor.

4. The cadastral data retrieval method based on knowledge graphs according to claim 3, characterized in that, Based on the evolutionary corridor, a set of candidate parcel states is extracted, including: For each candidate path from the seed state node to any reachable parcel state node, calculate the path succession strength; The evolutionary inheritance value of the parcel state node is determined based on the product of the inheritance mapping strength of each candidate path. Land parcel state nodes with evolution inheritance values ​​not lower than a preset threshold are identified as candidate land parcel state nodes, and the corresponding node set is identified as the candidate land parcel state set. The evolution inheritance value corresponding to each candidate parcel state node is associated with and stored with the evolution corridor.

5. The cadastral data retrieval method based on knowledge graphs according to claim 2, characterized in that, Based on the time constraint set and the relation constraint set, conditional association propagation and relevance calculation are performed on each candidate parcel state in the candidate parcel state set to obtain the candidate ranking result, including: Based on the query representation, time constraint set, and relation constraint set, conditional association propagation is performed on the nodes and their associated edges in the candidate parcel state set to obtain the contextual representation of each candidate parcel state under the current retrieval request conditions. Calculate the semantic matching value, evolutionary inheritance value, adjacency satisfaction degree, and right status continuity satisfaction degree for each candidate parcel status. The candidate parcel status relevance score is calculated based on the semantic matching value, the evolutionary inheritance value, the adjacency relationship satisfaction degree, and the continuous satisfaction degree of the right status; The candidate land parcel status set is sorted from high to low according to the relevance score to form the candidate ranking result.

6. The cadastral data retrieval method based on knowledge graphs according to claim 1, characterized in that, Based on the candidate ranking results, the historical succession chain is traced back, including: Select a predetermined number of high-scoring candidate parcel states from the candidate ranking results to form a high-scoring candidate set; For each candidate parcel state in the high-scoring candidate set, backtrack all feasible paths between it and the seed state node from the evolution corridor; For each feasible path, calculate the evidence chain score, determine the feasible path with the highest evidence chain score as the historical succession chain of the corresponding candidate land parcel status, and output the historical succession chain in association with the candidate land parcel status.

7. The cadastral data retrieval method based on knowledge graphs according to claim 6, characterized in that, Performing a consistency determination by combining the set of time constraints and the set of relation constraints includes: For the candidate parcel with the highest score in the candidate ranking results, extract its upstream state set, area information and adjacency relationship information from its corresponding historical succession chain; The area inheritance residual is calculated based on the upstream state set and the area information, and the adjacency consistency is calculated based on the relationship constraint set and the actual adjacency relationship of the candidate parcel state. If the area inheritance residual is not greater than a preset area threshold and the adjacency relationship consistency is not less than a preset relationship threshold, the candidate land parcel status is determined to meet the consistency judgment condition.

8. The cadastral data retrieval method based on knowledge graphs according to claim 7, characterized in that, When the preset conditions are met, the target cadastral retrieval results and corresponding evidence chain are output, specifically including: When the highest-scoring candidate parcel status in the candidate ranking results meets the consistency judgment condition, the current parcel identifier, historical parcel identifier, historical succession chain, key evolution event sequence, relationship matching result, and result confidence level corresponding to the candidate parcel status are output. When the highest-scoring candidate parcel status does not meet the consistency determination criteria, the next candidate parcel status is selected according to the candidate ranking results and the consistency determination is repeated until a candidate parcel status that meets the consistency determination criteria is obtained. If none of the candidate parcel statuses meet the consistency criteria, the output will show that no target parcel satisfying the historical inheritance constraint was found.

9. A cadastral data retrieval system based on knowledge graphs, characterized in that, The system is used to execute the knowledge graph-based cadastral data retrieval method according to any one of claims 1-8, and the system comprises: The parsing unit is used to acquire retrieval request data and parcel status data, parcel evolution event data and parcel adjacency relationship data from the same cadastral business database, and to perform structured parsing on the retrieval request data to obtain entity seed set, time constraint set and relation constraint set; A construction unit is used to construct a temporal cadastral knowledge graph based on the parcel status data, the parcel evolution event data, and the parcel adjacency relationship data, and to determine the historical parcel status corresponding to the entity seed set as seed status nodes. Specifically, this includes: expanding the status of each basic entity in the parcel status data according to the occurrence time of the parcel evolution events in the parcel status data to form multiple parcel status nodes, each corresponding to a valid time interval; configuring status number, parcel number, boundary geometry, area, rights status, and location description attributes for each parcel status node; establishing continuous state edges, split edges, merge edges, renumbered edges, and boundary adjustment edges between parcel status nodes based on the parcel evolution event data; establishing adjacency edges between temporally corresponding parcel status nodes based on the parcel adjacency relationship data to form the temporal cadastral knowledge graph; mapping the historical entity identifiers in the entity seed set to the corresponding historical parcel status nodes in the temporal cadastral knowledge graph, and determining the mapped parcel status nodes as seed status nodes. The generation unit is used to generate an evolutionary corridor along the parcel evolution relationship in the temporal cadastral knowledge graph based on the entity seed set, the time constraint set, and the relation constraint set, and to extract a candidate parcel state set based on the evolutionary corridor; The sorting unit is used to perform conditional association propagation and relevance calculation on each candidate parcel state in the candidate parcel state set based on the time constraint set and the relation constraint set, so as to obtain the candidate sorting result; The output unit is used to trace back the historical succession chain based on the candidate ranking result, and perform consistency determination in combination with the time constraint set and the relation constraint set. When the determination meets the preset conditions, the target cadastral retrieval result and the corresponding evidence chain are output.

Images