Spatial geographic data processing method and device

Abstract

The invention relates to a spatial geographic data processing method and device. The method comprises the following steps: acquiring spatial geographic data corresponding to a dynamic geographic entity, wherein the spatial geographic data comprises a position change sequence, an attribute tag sequence and an external event record; performing cross-domain nested coding processing on the spatial geographic data to obtain a multi-domain entity relationship coding set with multi-dimensional non-structural tense features and heterogeneous semantic relationship dimensions; in the multi-domain entity relationship coding set, performing residual graph generation processing on the geographic entity time sequence nodes with label drift and state hopping to obtain a residual graph set of a cross-temporal period; and performing semantic transition compression processing on the non-structurally determined region according to the residual map set to obtain a semantic chain set which is used for representing a space-time semantic reconstruction expression model of the dynamic geographic entity in the multi-stage evolution process. By means of the method, compression expression and semantic modeling can be conducted on the state mutation rule of the dynamic geographic entity in the non-structure-determined area.

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and in particular to a method and device for processing spatial geographic data. Background Art

[0002] In the field of data processing technology, it involves processing the spatial geographic data related to geographic entities to achieve temporal and spatial evolution modeling of the geographic entity status.

[0003] Related data processing methods describe the evolution of geographic entities by constructing temporal location trajectories. However, this method has difficulty effectively capturing the intrinsic semantic connections between discontinuous states when faced with dynamic geographic entities with frequent state jumps and obvious semantic label mutations. This results in a single model expression structure, weak semantic continuity, and insufficient modeling capabilities for abnormal evolutionary stages, making it difficult to adapt to the expression needs of highly dynamic geographic phenomena. Summary of the Invention

[0004] Based on this, it is necessary to provide a spatial geographic data processing method, device, computer equipment and computer-readable storage medium to address the above technical problems, which can be used to compress and express the state mutation laws of dynamic geographic entities in non-structurally determined areas and perform semantic modeling, thereby improving the structural extraction and interpretable expression capabilities of complex geographic evolution processes.

[0005] In a first aspect, the present application provides a spatial geographic data processing method, comprising: Acquire spatial geographic data corresponding to a dynamic geographic entity, wherein the spatial geographic data includes a position change sequence, an attribute label sequence, and an external event record; Performing cross-domain nested coding processing on the spatial geographic data to obtain a multi-domain entity relationship coding set with multi-dimensional unstructured temporal features and heterogeneous semantic relationship dimensions; In the multi-domain entity relationship encoding set, residual graph generation processing is performed on geographic entity time series nodes with label drift and state jump to obtain a residual graph set across temporal periods, each residual graph in the residual graph set representing a residual semantic relationship pattern of the dynamic geographic entity under a discontinuous semantic path; According to the residual graph set, semantic transition compression processing is performed on the non-structural determination area corresponding to the dynamic geographic entity to obtain a semantic chain set for representing the spatiotemporal semantic reconstruction expression model of the dynamic geographic entity in a multi-stage evolution process.

[0006] In a second aspect, the present application further provides a spatial geographic data processing device, comprising: An acquisition module is used to acquire spatial geographic data corresponding to a dynamic geographic entity, wherein the spatial geographic data includes a position change sequence, an attribute label sequence, and an external event record; An encoding module is used to perform cross-domain nested encoding processing on the spatial geographic data to obtain a multi-domain entity relationship encoding set with multi-dimensional unstructured temporal features and heterogeneous semantic relationship dimensions; A residual module is configured to generate residual graphs for geographic entity temporal nodes with label drift and state jump in the multi-domain entity relationship encoding set, thereby obtaining a residual graph set across temporal periods, wherein each residual graph in the residual graph set represents a residual semantic relationship pattern of the dynamic geographic entity under a discontinuous semantic path; A compression module is used to perform semantic transition compression processing on the non-structural determination area corresponding to the dynamic geographic entity according to the residual map set to obtain a semantic chain set for representing the spatiotemporal semantic reconstruction expression model of the dynamic geographic entity in a multi-stage evolution process.

[0007] In a third aspect, the present application further provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the above steps when executing the computer program.

[0008] In a fourth aspect, the present application further provides a computer-readable storage medium on which a computer program is stored, and the computer program implements the above steps when executed by a processor.

[0009] The above-mentioned spatial geographic data processing method, apparatus, computer device and computer-readable storage medium firstly realize the unified acquisition and temporal organization of multi-dimensional evolution information of dynamic geographic entities by constructing spatial geographic data containing position change sequences, attribute label sequences and external event records; secondly, based on cross-domain nested encoding processing of spatial geographic data, a multi-domain entity relationship code set is obtained, thereby establishing an entity expression form covering the multi-dimensional structural relationship between spatial position, semantic label and event effect; thirdly, based on identifying and extracting jump nodes in the multi-domain entity relationship code set to construct a residual graph set, thereby capturing the discontinuous state mutation in the evolution process of dynamic geographic entities and forming quantifiable semantic deviation information; thirdly, based on the residual graph set, semantic transition compression processing is performed on non-structurally determined areas, thereby extracting a semantic chain set with stage characteristics and trend consistency; based on this, the state mutation law of dynamic geographic entities in non-structurally determined areas can be compressed and semantically modeled, thereby improving the structure extraction and interpretable expression capabilities of complex geographic evolution processes. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] In order to more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the following briefly introduces the drawings required for use in the embodiments or related technical descriptions. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0011] Figure 1 1 is a flow chart of a method for processing spatial geographic data in one embodiment; Figure 2 FIG. 4 is a structural block diagram of a spatial geographic data processing device in one embodiment. DETAILED DESCRIPTION

[0012] In order to make the purpose, technical solutions and advantages of this application more clear, the following further describes this application in detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain this application and are not intended to limit this application.

[0013] In one embodiment, Figure 1 As shown, a method for processing spatial geographic data is provided. This embodiment uses the method applied to a server as an example. It is understood that the method can also be applied to a terminal, or to a system including a terminal and a server, and implemented through interaction between the terminal and the server. In this embodiment, the method includes the following steps S101 to S104.

[0014] Step S101 : Acquire spatial geographic data corresponding to a dynamic geographic entity. The spatial geographic data includes a position change sequence, an attribute tag sequence, and an external event record.

[0015] Among them, dynamic geographic entities represent geographic objects with spatial locations and attribute characteristics that change over time, such as shared bicycles in the city, moving typhoon systems, seasonally changing river boundaries, or construction areas.

[0016] Among them, spatial geographic data refers to a data set used to describe the relationship between the spatial location, attribute characteristics and environment of dynamic geographic entities, such as a set of time-series data segments containing coordinate trajectories, land use categories, and traffic event information.

[0017] Among them, the position change sequence represents a set of time series data related to the spatial position changes of dynamic geographic entities in a preset geographic coordinate system, that is, it is used to describe the movement trajectory and spatial range evolution of dynamic geographic entities in the time dimension, such as the GPS positioning coordinate sequence continuously recorded from 7 am to 9 am for a bus.

[0018] Among them, the attribute label sequence represents a time series data set of attribute categories or state identifiers of a dynamic geographic entity at different time points, that is, it is used to describe the changes in the characteristics of a dynamic geographic entity in time evolution. For example, the label of a river changes from "frozen state" to "meltwater state" and then to "flood state" at different periods.

[0019] Among them, external event records represent a set of event information associated with the spatial area of ​​dynamic geographic entities and caused by changes in the external environment, that is, they are used to explain or assist in judging the external driving reasons for changes in the state of dynamic geographic entities, such as "the rainfall intensity reached 100mm / h on July 15, 2021" as an important event record affecting changes in surface runoff.

[0020] For example, in order to support the modeling of the evolution process of dynamic geographic entities, it is necessary to first construct a collection of spatial geographic data related to them. This collection can truly and comprehensively reflect the location status, attribute status, and external influences of the specified dynamic geographic entity at different time stages. To this end, it is necessary to systematically collect the data sources related to the dynamic geographic entity and organize the position change sequence according to the time sequence. The position change sequence reflects the spatial position coordinates of the dynamic geographic entity at different time points, so as to express the trajectory change process experienced by the dynamic geographic entity through the correspondence between timestamps and coordinates. In addition, it is necessary to simultaneously construct an attribute label sequence. This attribute label sequence is a labeled description of the characteristics of the dynamic geographic entity itself (such as category, function, status, etc.) in each time period. The labels are comparable and support time series analysis, aiming to reveal the change path of the dynamic geographic entity attributes over time. Furthermore, records of external events within the activity range of dynamic geographic entities should also be included in the data system. This part of the data may contain event information that may have an impact on the evolution process of dynamic geographic entities. By setting elements such as the time of event occurrence, affected area and event type, the temporal and spatial correspondence between events and dynamic geographic entities can be clarified.

[0021] Step S102 : performing cross-domain nested coding processing on the spatial geographic data to obtain a multi-domain entity relationship coding set having multi-dimensional unstructured temporal features and heterogeneous semantic relationship dimensions.

[0022] Among them, the multi-domain entity relationship coding set represents a coded data structure that is represented in a unified structural organization form after the fusion of multiple types of spatial geographic data. That is, it is used to describe the semantic and temporal structural relationships of dynamic geographic entities at various time nodes under the interaction of multi-source information. For example, a coded data structure containing information such as location, traffic flow, and accident level assigned to a main traffic artery every hour can reflect the dynamic evolution behavior of the road in the traffic and event dimensions.

[0023] Among them, multidimensional unstructured temporal features represent data features composed of time series data from multiple different sources or different types, which lack regular grid arrangement and have time sequence association; heterogeneous semantic relationship dimensions represent dimensional features in which multiple semantic interpretation sources that constitute the dynamic geographic entity encoding data structure have differences in attribute categories, event types, spatial scales, etc.

[0024] For example, to extract information related to multidimensional unstructured temporal features and heterogeneous semantic relationship dimensions from collected spatial geographic data, a cross-domain nested encoding process is required for the spatial geographic data. In this process, the position change sequence, attribute label sequence, and external event records are first unfolded along the same timeline according to chronological order, constructing a time node set for dynamic geographic entities at a unified moment granularity. Subsequently, the three types of data corresponding to each time node are combined into a data unit. By setting a nested structure, the geographic location state is used as the primary dimension, and the attribute label state and event response information within the corresponding time period are embedded within it, forming a nested representation with position change as the main line and attributes and events as branches. On this basis, the data units of these nested representations are then integrated according to the temporal evolution order, and a multi-domain entity relationship encoding set covering the entire evolution process is formed through temporal connection. This multi-domain entity relationship encoding set not only maintains temporal continuity but also preserves the cross-dependency structure between location, attributes, and events in the semantic dimension.

[0025] In step S103, residual graph generation processing is performed on the geographic entity time series nodes with label drift and state jump in the multi-domain entity relationship coding set to obtain a residual graph spectrum set across the temporal period. Each residual graph in the residual graph spectrum set represents a residual semantic relationship pattern of the dynamic geographic entity under the discontinuous semantic path.

[0026] Among them, geographic entity time series nodes with label drift and state jumps indicate that during the temporal evolution of dynamic geographic entities, the entity state corresponding to a specific time point has experienced significant changes in attribute labels or discontinuous transitions in the overall state. This is used to identify mutation behaviors or abnormal change signals in the evolution of dynamic geographic entities. For example, if the state label of a region is "farmland" for many consecutive days, but suddenly changes to "construction land" one day without a matching external event description, the node data of this mutation is considered to have label drift and state jumps.

[0027] Among them, the residual map set across temporal periods represents a set of residual pattern maps covering multiple time intervals, constructed based on the identified mutation node data. Each residual map reflects the residual structure of the corresponding dynamic geographic entity due to semantic changes in different time periods, and is used to represent a residual semantic relationship pattern of the corresponding dynamic geographic entity under a discontinuous semantic path.

[0028] Among them, a residual semantic relationship pattern of dynamic geographic entities under discontinuous semantic paths indicates that after a dynamic geographic entity undergoes a state mutation, some of its semantic attributes still maintain continuity or form a new dependency structure. For example, an industrial area quickly changes from an "active enterprise zone" to "idle land" under policy adjustments, but some semantic elements related to enterprise names and infrastructure categories remain stable. These semantic fragments that are not completely broken constitute the residual semantic relationship pattern.

[0029] For example, to identify the state discontinuities experienced by dynamic geographic entities during their evolution, it is necessary to locate time series nodes experiencing label drift and state transitions within a multi-domain entity relationship encoding set. Based on these, a set of residual graphs reflecting semantic change anomalies is constructed. This process first calculates the degree of semantic deviation between each time series node and its preceding and following nodes based on the temporal variation trends of attribute labels and event response characteristics. This deviation not only reflects changes in label content but also considers the magnitude of spatial position changes and event types to determine whether a nonlinear transition trend exists. When a time series node's semantic expression differs significantly from that of its historical nodes, and this difference cannot be rationally explained by event triggering, the node is marked as a potential transition node. Next, a segment of encoded data containing several time steps preceding and following the transition node is extracted, and the semantic composition of each time series node in this segment is compared to determine the residual semantic differences across dimensions. Finally, this residual semantic difference is constructed as a residual graph reflecting the transition characteristics. Based on this, the residual graph collection constructed in this way gradually constitutes a structured dataset that expresses different residual semantic relationship patterns of dynamic geographic entities in the multi-stage evolution process.

[0030] Step S104 , performing semantic transition compression processing on the non-structural determination area corresponding to the dynamic geographic entity according to the residual graph set, and obtaining a semantic chain set for representing the spatiotemporal semantic reconstruction expression model of the dynamic geographic entity in the multi-stage evolution process.

[0031] Among them, the unstructured region refers to the spatial or temporal segment in which the evolution path of a dynamic geographic entity cannot be described by stable rules due to factors such as label discontinuity, event interference or semantic jump during the spatiotemporal evolution of the entity.

[0032] Among them, the semantic chain set represents a set of semantic chain segments formed by connecting the key semantics of dynamic geographic entities at different time nodes according to time sequence and evolutionary logic. It is used to extract a coherent and deducible semantic expression structure from complex jumps. For example, the evolution chain of a river can sequentially include a series of causal semantic nodes such as "source enhancement, water volume increase, increased flood risk, governance project intervention, and water body shrinkage". A collection of multiple such semantic chain segments constitutes a semantic chain set.

[0033] Among them, the spatiotemporal semantic reconstruction expression model represents a comprehensive model that can characterize the evolution process of dynamic geographic entities, which is established on the basis of the semantic chain set and utilizes the temporal sequence and spatial change relationship therein. That is, it is used to express and restore the evolution process of dynamic geographic entities in the context of complex and nonlinear state migration.

[0034] For example, in order to extract a semantic chain set with structured evolutionary significance from a set of residual graphs with significant jump characteristics, the residual graph set must first be analyzed and processed centrally, especially for the residual nodes located in non-structurally determined areas, and semantic trajectory extraction must be carried out; since these residual nodes are often located in sections with label mutations or state transitions, they themselves lack continuous structural support and are difficult to be directly incorporated into a unified model expression. Therefore, starting from the residual nodes marked as unstable in each residual graph, the time dimension is expanded forward and backward node by node, and the jump path consistent with the direction of change of its attribute label is retrieved, thereby restoring the possible trends of the dynamic geographic entity in the local evolution.

[0035] Next, to further reduce the complexity of the transition paths, it is necessary to aggregate the parts of all transition paths with similar semantic trends. Specifically, the transition paths with similar semantic span characteristics are merged, that is, the residual nodes with consistent label migration directions and similar state transition rhythms are merged into a single semantic segment, and then connected and spliced ​​in chronological order. Finally, by integrating all the semantic segments, a semantic chain set covering the entire unstructured area can be constructed. This semantic chain set is used to replace the scattered transition data in the original residual map that lacks structural expression capabilities, and becomes the key structural unit for characterizing dynamic geographic entities in the irregular evolution stage.

[0036] In the above-mentioned spatial geographic data processing method, first, spatial geographic data containing position change sequences, attribute label sequences, and external event records are constructed to achieve unified acquisition and temporal organization of the multidimensional evolution information of dynamic geographic entities. Second, a multi-domain entity relationship code set is obtained by cross-domain nested coding of the spatial geographic data, thereby establishing an entity expression form covering the multi-dimensional structural relationship between spatial location, semantic labels, and event effects. Third, a residual graph set is constructed by identifying and extracting jump nodes in the multi-domain entity relationship code set to capture the discontinuous state mutations in the evolution process of dynamic geographic entities and form quantifiable semantic deviation information. Third, semantic transition compression is performed on non-structurally determined areas based on the residual graph set, thereby extracting a semantic chain set with stage characteristics and trend consistency. Based on this, the state mutation laws of dynamic geographic entities in non-structurally determined areas can be compressed and semantically modeled, thereby improving the structure extraction and interpretable expression capabilities of complex geographic evolution processes.

[0037] In an exemplary embodiment, cross-domain nested coding is performed on spatial geographic data to obtain a multi-domain entity relationship coding set with multi-dimensional non-structural temporal features and heterogeneous semantic relationship dimensions, including steps S201 to S203.

[0038] Step S201 : performing time alignment processing on the position change sequence, the attribute label sequence, and the external event record in the spatial geographic data to obtain a temporally aligned sequence set.

[0039] Among them, the temporally aligned sequence set represents a multidimensional synchronization sequence set composed of position change sequences, attribute label sequences, and external event records aligned at the same time granularity after unified processing in the time dimension. It is used to ensure that the content of data from different sources at each time node is comparable and synchronized.

[0040] For example, based on the evolution timeline of dynamic geographic entities, the position change sequence, attribute label sequence and external event record of the same time granularity are mapped to a unified time axis to achieve time alignment of the above three types of data. Specifically, due to differences in data collection frequency, start time and time interval of different data sources, each data sequence needs to be timestamp standardized before performing time alignment, that is, all original time labels are converted into a unified format and relocated to a common reference moment set; then, the original data is interpolated, cropped or repeatedly filled according to the common reference moment set to ensure that each time node can obtain complete values ​​in the three data dimensions. Based on this, the set of temporal sequences that are finally aligned has synchronous time series, which can ensure that at any time node, the spatial position, attribute label and event status are in a comparable unified framework.

[0041] Step S202 : performing association modeling processing on the spatial positions, attribute labels, and event states corresponding to different time nodes according to the temporal alignment sequence set, and obtaining a composite entity node set with a time index, a spatial label state, and an event response scalar.

[0042] The composite entity node set represents a set of structured composite expression units (i.e., composite entity nodes) composed of the spatial position, attribute label, and event status of the corresponding time node at each aligned time node, which is used to describe the complete status of the dynamic geographic entity at each composite entity node, including the time index, spatial label status, and event response scalar.

[0043] Among them, the time index represents the time tag used to uniquely identify each composite entity node, and is used as a reference axis in constructing sequence structures or performing operations such as sorting and searching; the spatial label state represents the attribute category or state label of the dynamic geographic entity on a certain composite entity node, and is used to express the semantic meaning of the entity at the corresponding time node; the event response scalar represents the response result of an external event corresponding to a certain composite entity node to the entity state, and is used to quantify the intensity or scope of influence of the event.

[0044] For example, to achieve the joint expression of semantic information and spatial behavior based on a set of aligned temporal sequences, the spatial location, attribute label, and event state corresponding to each time node are structurally combined to form a composite entity node set with a time index. Specifically, at each time node, the projected area of ​​the dynamic geographic entity in geographic space is determined based on its spatial location and serves as the georeference field of the time node. Simultaneously, the attribute label of the corresponding time node is extracted as a semantic state field and structured into the node content. Combined with the event state information recorded at that time point, such as the impact type, scope, or event level, this is expressed as an event response scalar reflecting changes in the external environment. This is combined with other fields to form a complete node representation, namely, the complete state of a composite entity node. Furthermore, all composite entity nodes are arranged using the time index as the primary key to ensure the structural continuity of the dynamic geographic entity during its evolution. Based on this, the resulting composite entity node set possesses both spatial distribution properties and embeds the semantic state and event impact structure, capable of expressing the complete state of the dynamic geographic entity at a specific moment.

[0045] Step S203 : By extracting the semantic connection relationship, temporal adjacency relationship and event causal relationship between the composite entity nodes, a nested structure construction process is performed on the composite entity node set to obtain a multi-domain entity relationship coding set.

[0046] Among them, the semantic connection relationship represents the semantic-level association relationship formed by the attribute label change trend between adjacent composite entity nodes, such as the label change relationship of "agricultural land" gradually becoming "urban construction land".

[0047] The temporal adjacency relationship represents the temporal relationship between consecutive composite entity nodes connected in time sequence. For example, the time index of a node is "2023-06-01", and the time of its next adjacent node is "2023-06-02".

[0048] Among them, event causality represents the causal relationship between the state changes caused by a certain event between different composite entity nodes. For example, if a "flood event" corresponding to a certain node is triggered, and the label of another entity changes from "cultivated land" to "bare land", then there is an event causal relationship between the two nodes.

[0049] For example, in order to enable the composite entity node set to express the cross-domain relationship structure of dynamic geographic entities in the entire temporal process, it is necessary to introduce multiple structural connection mechanisms on the basis of the composite entity node set to construct nested relationships. First, by analyzing the similarity and trend changes of the attribute label values ​​between adjacent composite entity nodes, a semantic connection relationship is established to reflect the semantic evolution logic of dynamic geographic entities in the time dimension. Secondly, by calculating the time interval and position displacement information between temporally continuous composite entity nodes, a temporal adjacency relationship is established to reflect the spatiotemporal evolution logic of dynamic geographic entities in the spatial trajectory and time advancement process. Furthermore, through the correlation between the common causal characteristics and response results of the event states of the composite entity nodes, an event causal relationship is constructed to reflect the state transition logic caused by the influence of events between different composite entity nodes.

[0050] After the above-mentioned relationship structure is constructed, the three relationship structures are embedded in the topological description of each composite entity node, and the relationship between the composite entity nodes is encapsulated as a computable data unit in the form of a nested structure, thereby constructing a multi-domain entity relationship coding set; this multi-domain entity relationship coding set retains the information characteristics of each composite entity node in terms of space, semantics, and events, and at the same time clarifies the cross-dimensional connection method between composite entity nodes.

[0051] In this embodiment, first, a unified time alignment process is performed on the position change sequence, attribute label sequence, and external event record to obtain a temporal alignment sequence set, thereby ensuring the synchronous expression structure of various types of data at the same time node and improving the fusion basis of multi-source information; secondly, based on the temporal alignment sequence set, the spatial positions, attribute labels, and event states corresponding to different time nodes are associated and modeled to obtain a composite entity node set, thereby achieving a structured representation of the complete state characteristics of the dynamic geographic entity at each specific moment on the time axis; thirdly, based on the semantic connection relationship, temporal adjacency relationship, and event causal relationship between the composite entity nodes, the composite entity node set is nested and constructed, thereby establishing a multi-domain entity relationship coding set that cross-dimensionally expresses the information characteristics of each node and the connection relationship between nodes. Based on this, unified modeling and association organization of multi-source and multi-dimensional temporal information of dynamic geographic entities can be achieved.

[0052] In an exemplary embodiment, by extracting the semantic connection relationship, temporal adjacency relationship and event causal relationship between composite entity nodes, a nested structure construction process is performed on the composite entity node set to obtain a multi-domain entity relationship coding set, including steps S301 to S302.

[0053] Step S301: Based on the semantic connection relationship, temporal adjacency relationship and event causal relationship between the composite entity nodes, the composite entity node set is subjected to relationship triple extraction processing to obtain a multi-source relationship graph including a semantic connection edge set, a temporal adjacency edge set and an event causal edge set.

[0054] Among them, the multi-source relationship graph represents a set of graph structures composed of multiple composite entity nodes and different types of semantic and temporal association relationships between them. It is used to uniformly organize the interaction structure between multi-source information such as spatial location, attribute labels and event impacts. For example, a node in the multi-source relationship graph represents the status of an urban area every hour, and the edges represent the relationship connections between attribute evolution, time advancement and event-driven.

[0055] Among them, the semantic connection edge set represents the edge set in the multi-source relationship graph where the semantic labels between two composite entity nodes have an evolutionary trend or associated change. For example, the semantic connection edge between the "agricultural land" node and the "construction land" node represents the trend of land function change at the semantic level.

[0056] Among them, the temporal adjacent edge set represents the edge set established in time order between adjacent composite entity nodes in the multi-source relationship graph. For example, the temporal adjacent edge between the two nodes on June 1, 2023 and June 2, 2023 represents the diurnal evolution direction of the dynamic geographic entity state.

[0057] Among them, the event causal edge set represents the edge set in which an event occurring at a certain composite entity node in the multi-source relationship graph drives the attribute change of another composite entity node. For example, the event causal edge between the "heavy rainfall event" node and the subsequent "flooded surface" node represents the evolutionary path triggered by the event.

[0058] For example, first, for each pair of composite entity nodes, the similarity of their semantic labels is calculated. If the semantic labels are related or have an evolutionary trend at the semantic level, a semantic connection edge is established between the two composite entity nodes. This semantic connection edge is used to represent the change path of the attribute label. Then, the composite entity nodes that are adjacent in time are connected. That is, each group of composite entity nodes that are continuous in time index is connected through a temporal adjacency edge to ensure that the constructed graph structure has temporal traceability. Then, in the event dimension, the mutual influence of the event response scalars in the two composite entity nodes is analyzed to determine whether the state of a node is triggered or affected by the event information carried by the previous and next nodes. If specific causal conditions are met, an event causal edge is established to describe the state jump caused by the event. After completing the generation of the above three types of edges, all composite entity nodes and the three types of edges are uniformly constructed into a multi-source relationship graph containing a semantic connection edge set, a temporal adjacency edge set, and an event causal edge set. This multi-source relationship graph not only retains the temporal order and spatial semantic information of the original nodes, but also establishes a cross-dimensional association path at the structural level.

[0059] Step S302: In the multi-source relationship graph, a double-layer association clustering process is performed on the semantic connection edge set and the event causal edge set to construct a labeled event nested subgraph, and each labeled event nested subgraph is subjected to a nested graph fusion process according to the topological skeleton constructed according to the temporal adjacency edge set to obtain a multi-domain entity relationship encoding set.

[0060] Among them, the label event nested subgraph represents the local graph structure obtained by double clustering of semantic connection edges and event causal edges in the multi-source relationship graph. It is used to express the co-evolutionary relationship between a group of nodes in the label state change and event impact. For example, a group of nodes changes from "forestland to wasteland and then to developed land" and are all related to the "land acquisition event", which constitutes a label event nested subgraph.

[0061] Among them, the topological skeleton constructed by the temporal adjacent edge set represents a trunk structure with a time sequence formed in the entire multi-source relationship graph based on the temporal adjacent edges, which is used to carry the temporal organization framework of each nested subgraph of the label event. For example, the nested subgraphs of the label event representing different evolutionary stages are embedded in sequence on the time main line graph constructed by consecutive days to form a complete evolutionary process.

[0062] For example, in a multi-source relationship graph, we first perform a preliminary clustering of nodes with highly similar semantic trends based on the attribute label change paths covered by semantic connection edges. We then further subdivide these preliminary clustering results based on the presence of event causal edges between these nodes, ensuring that the nodes within the cluster units are related in both semantic and event dimensions. Based on this, the result of this two-layer clustering process is a set of nested label-event subgraphs. Each nested label-event subgraph aggregates a subset of nodes that are interconnected in terms of semantic evolution and event-drivenness, forming a nested unit that expresses the local evolutionary structure of dynamic geographic entities.

[0063] Then, at the overall topological level of the multi-source relationship graph, based on temporal adjacency edges, a topological skeleton is extracted along the complete timeline. Each labeled event nested subgraph is sequentially superimposed on this topological skeleton, ensuring that each labeled event nested subgraph maintains its own internal structural independence while maintaining consistency with the global temporal progression. Finally, through nested graph fusion processing, each labeled event nested subgraph is integrated into a unified topological framework, forming a multi-domain entity relationship encoding set with multi-dimensional structural hierarchy and semantic evolution expression capabilities.

[0064] In this embodiment, first, the semantic connection relationship, temporal adjacency relationship and event causal relationship between composite entity nodes are subjected to triple extraction processing, thereby constructing a multi-source relationship map that simultaneously reflects semantic evolution, time advancement and event-driven mechanism, thereby enhancing the structural expression ability between nodes; secondly, the semantic connection edges and event causal edges are subjected to double-layer clustering and integrated into the topological skeleton constructed by temporal adjacency edges, thereby realizing the nested integration of multi-dimensional evolutionary relationships and forming a multi-domain entity relationship coding set with temporal organization and semantic nesting. Based on this, it is possible to achieve unified modeling of the multi-dimensional structural relationships of dynamic geographic entities in a complex evolutionary background.

[0065] In an exemplary embodiment, in a multi-domain entity relationship coding set, residual graph generation processing is performed on geographic entity time series nodes with label drift and state jump to obtain a residual graph set across temporal periods, including steps S401 to S403.

[0066] Step S401 : performing difference measurement processing on the attribute label state and event response state of each geographic entity time series node under the continuous time index in the multi-domain entity relationship coding set to obtain a label drift intensity matrix and a state jump probability matrix.

[0067] Among them, the label drift intensity matrix represents a two-dimensional numerical matrix used to measure the degree of change in the attribute label status of dynamic geographic entities during the time evolution process. For example, if the label of a certain area changes from "farmland" to "industrial land" between "2020-2021", a higher label drift intensity value will be recorded at the matrix position corresponding to this time interval.

[0068] Among them, the state transition probability matrix represents a two-dimensional numerical matrix used to reflect the probability of state transition caused by external events. For example, when a "flood disaster" event occurs in a certain area at a certain time point, followed by a "bare land" state, a higher state transition probability value is recorded at the matrix position corresponding to the time interval.

[0069] For example, based on the time axis, the attribute label state changes of dynamic geographic entities at adjacent time nodes are compared item by item, and combined with their corresponding event response states, the joint index of label change amplitude and state transition significance, namely the label drift intensity matrix and the state jump probability matrix, is calculated. Specifically, it is first necessary to extract the semantic distance between the current attribute label and the attribute label of the previous time node for each time series node, and use this to construct the original label drift intensity matrix that describes the label drift intensity. Subsequently, the event response state changes of the current node are combined with the event response state of its historical period for correlation comparison, and the original state jump probability matrix that describes the jump probability is constructed.

[0070] To ensure comparability between label differences and event transitions within a unified evaluation framework, the original label drift intensity matrix and the original state transition probability matrix must be normalized and weighted. This allows for a clear characterization of state changes between any pair of time nodes, both in the context of semantic mutations and event inductions. The resulting label drift intensity matrix and state transition probability matrix are used to characterize the severity of dynamic geographic entity evolution at the label level and the likelihood of transitions at the event level, respectively, providing a quantitative basis for subsequent anomalous trajectory identification.

[0071] In step S402 , abnormal trajectory identification processing is performed on each geographic entity time series node according to the label drift intensity matrix and the state jump probability matrix to obtain a set of abnormal trajectory nodes marked with semantic offset nodes.

[0072] Among them, the set of abnormal trajectory nodes marked with semantic offset nodes represents the set of time series nodes with significant semantic variation characteristics identified based on the label drift intensity matrix and the state jump probability matrix in the time trajectory analysis of dynamic geographic entities. It is used to calibrate the key time points where jump, abnormal or mutation behaviors occur in the evolution path of dynamic geographic entities.

[0073] For example, first, each time series node is used as an evaluation unit, and its note drift strength value in the label drift strength matrix and its state jump probability value in the state jump probability matrix are retrieved respectively. If at least one of these two indicators of a time series node is significantly higher than the overall change benchmark of the time series, it can be judged that the time series node has potential semantic drift behavior. Furthermore, in order to enhance the stability of the recognition results, a sliding window mechanism should be introduced to jointly analyze multiple continuous time periods to identify areas where labels are continuously unstable or events are frequently triggered, so as to avoid the interference of isolated noise points. Then, after obtaining the preliminary recognition results, the time series nodes with high overlap and overlapping label change paths are merged and reduced to finally generate an abnormal trajectory node set, and the key metadata such as the time position, label change direction, and event excitation degree of each semantic drift node are retained in the abnormal trajectory node set; the semantic drift node represents the key moment point in the evolution path of the dynamic geographic entity where there is a semantic jump risk or an abnormal driving mechanism.

[0074] Step S403 , performing a residual graph generation process across time periods on the abnormal trajectory node set to obtain a residual graph set across temporal periods.

[0075] For example, first, with each semantic offset node as the core point, construct the adjacent trajectories in several time periods before and after it, extract its continuous behavior information in the state evolution process from the adjacent trajectories, and compare it with the ideal evolution trend or the historical average evolution trend to generate residual differences in the time series. Then, by mapping the residual differences in this time series into the graph structure, a residual subgraph centered on the semantic offset node can be obtained. Furthermore, in order to make these residual subgraphs have cross-period expression capabilities, it is necessary to aggregate the subgraph segments with semantic similarity or jump features that appear repeatedly in different time periods, so as to construct a residual graph set across temporal periods.

[0076] In this embodiment, first, a label drift intensity matrix and a state jump probability matrix are constructed based on the difference measurement processing of the attribute label status and event response status on continuous time series nodes, thereby realizing the quantitative expression of the semantic changes and event-driven characteristics of dynamic geographic entities; secondly, each time series node is jointly analyzed and processed based on the above two matrices, thereby identifying a set of abnormal trajectory nodes with potential semantic mutation characteristics, thereby enhancing the ability to locate discontinuous evolutionary behaviors; thirdly, a cross-cycle residual modeling processing is performed on the set of abnormal trajectory nodes, thereby generating a set of residual graphs reflecting the semantic offset structure, thereby realizing the graphical expression of the jump path and abnormal structure in the evolution process of dynamic geographic entities.

[0077] In an exemplary embodiment, abnormal trajectory identification processing is performed on each geographic entity time series node according to the label drift intensity matrix and the state jump probability matrix to obtain a set of abnormal trajectory nodes marked with semantic offset nodes, including steps S501 to S503.

[0078] Step S501 : performing joint anomaly measurement processing on the label drift intensity matrix and the state jump probability matrix to obtain the semantic mutation gradient sequence of each geographic entity time series node under the continuous time index.

[0079] Among them, the semantic mutation gradient sequence represents a numerical sequence composed of the joint measurement results of the label drift intensity and state jump probability of the physical geographic entity at continuous time series nodes. It is used to describe the change trend and mutation degree of the semantic state of the dynamic geographic entity during the evolution process.

[0080] For example, the label drift strength matrix is ​​first extracted from each time series node in sequence as the basic measure of semantic variation. At the same time, the state transition probability matrix of the time series node in response to an event is numerically sampled as another measure. Subsequently, these two types of measures are combined with a unified time index, and a fused semantic mutation index is generated through linear weighting, distance transformation, or other standardized methods. This semantic mutation index not only reflects the intensity of label changes at a single time point, but also considers the interference effect of event responses on the state evolution path. Based on this, the sequence of semantic mutation indicators formed at all time series nodes constitutes a semantic mutation gradient sequence, which is used to describe the semantic change density, mutation tendency, and fluctuation trajectory of dynamic geographic entities throughout the entire evolution cycle. In this way, the evolution trends at the label level and the event level can be merged into the same measurement system, providing a unified structural input for the subsequent clustering analysis of abnormal nodes.

[0081] Step S502 : clustering the time index positions with abnormal jump amplitudes in the semantic mutation gradient sequence according to the state transition direction and the mutation time difference to obtain a mutation node set.

[0082] Among them, the state transition direction represents the semantic change trend direction of the attribute label state between adjacent time series nodes, which is used to determine whether the entity evolution path presents a consistent evolution logic. For example, the attribute label of a certain area changes from "arable land" to "construction land" and then to "commercial land". The state transition direction of this state evolution path can be abstractly expressed as the direction of "advancing from natural land to urban functional areas".

[0083] Among them, the mutation time difference represents the interval size of the corresponding time index between adjacent time series nodes, which is used to measure the distribution density and concentration of mutation events on the time axis. For example, if two nodes occur in the third and fourth quarters of 2020 respectively, their mutation time difference is one quarter. Furthermore, if the mutation time differences of multiple nodes are relatively concentrated, it indicates that the dynamic geographic entity has concentrated semantic change behavior within a certain time period.

[0084] Among them, the abnormal jump amplitude indicates the significant degree of numerical jump of a time series node relative to its previous and next nodes in the semantic mutation gradient sequence, which is used to identify the severity of the state mutation. For example, if the gradient value in the continuous time series jumps from 0.2 to 0.95, the abnormal jump amplitude of the node is 0.75, indicating that the node has undergone a drastic label mutation or semantic shift.

[0085] Among them, the mutation node set represents a group of time series nodes that show abnormal behaviors in terms of jump amplitude, mutation direction and time density, which are clustered and identified from the semantic mutation gradient sequence. It is used to centrally represent the state mutation concentration area of ​​dynamic geographic entities during the evolution process.

[0086] For example, taking the time index as the basic unit and combining the numerical change trend of each time series node in the semantic mutation gradient sequence, the time index position where the jump amplitude increases abnormally is identified, that is, the mutation node, and the state transition direction and mutation time difference in the time period before and after it are calculated. On this basis, the nodes with similar state transition directions, relatively concentrated mutation time differences and mutation amplitudes of the same order of magnitude between mutation nodes are classified into the same clustering unit, thereby avoiding the interference of isolated mutation nodes on the overall evolution trend, and identifying time periods showing continuous jumps or high-frequency mutations by clustering. Based on this, the set of mutation nodes generated thereby is a node cluster that exhibits relatively concentrated and consistent jump characteristics in the time series, which constitutes the core mutation fragment in the spatial semantic evolution structure.

[0087] Step S503 : performing context consistency verification and labeling on the mutation node set according to a preset context threshold, and obtaining an abnormal trajectory node set labeled with semantically offset nodes.

[0088] Among them, the context threshold represents the judgment standard for the tolerance of context state trend differences set when verifying semantic consistency. It is used to distinguish real semantic deviation nodes from normal fluctuation nodes. For example, setting the context threshold to 30% means that if the state similarity between a time series node and its previous and next nodes is less than 30%, the time series node is considered to be a context-inconsistent semantic deviation node.

[0089] For example, first, based on the preset context threshold, for each mutation node, the state trajectory, label trend and event changes in the previous and next time windows are extracted, and the similarity and continuity in semantic trends between the mutation node and its context are analyzed. If the mutation node lacks a predictable connection with the corresponding upper and lower nodes in the semantic trend, and there is a significant causal disconnection between it and the event information, it is marked as a semantic offset node. Based on this, the judgment process not only considers the single-point mutation characteristics, but also emphasizes the discontinuity of local semantic trends to enhance the rigor and stability of the annotation; after the annotation is completed, in the final set of abnormal trajectory nodes, all nodes that meet the context mutation characteristics and whose structures exhibit semantic breakpoint behavior can be classified as semantic offset nodes.

[0090] In this embodiment, first, a semantic mutation gradient sequence is constructed based on the joint anomaly measurement processing of label drift intensity and state jump probability, thereby achieving a quantitative expression of the semantic change trend of dynamic geographic entities; secondly, cluster analysis is performed in the semantic mutation gradient sequence according to the state transition direction and mutation time difference, thereby extracting a set of mutation nodes, enhancing the ability to identify centralized jump behaviors in the evolution process; thirdly, consistency verification and annotation processing are performed on the mutation node set based on the context threshold, thereby screening out a set of abnormal trajectory nodes that truly exhibit semantic drift characteristics, thereby achieving accurate identification and structural calibration of the jump nodes of dynamic geographic entities.

[0091] In an exemplary embodiment, semantic transition compression processing is performed on the non-structural determination area corresponding to the dynamic geographic entity according to the residual atlas set to obtain a semantic chain set, including steps S601 to S602.

[0092] Step S601: traverse the residual nodes located in the non-structural determination area in the residual graph set through a preset semantic traversal algorithm, and extract jump paths that are consistent with the attribute label variation directions of each residual node.

[0093] Among them, the semantic traversal algorithm represents a path extraction method based on the semantic relationship and structural connection relationship between residual nodes in the residual graph, so as to identify node sequences with coherence in the direction of attribute label variation, namely, jump paths, from non-structural determination areas.

[0094] For example, a preset semantic traversal algorithm is used to gradually search for a sequence of nodes in the residual graph that are continuous or convergent in the direction of attribute label variation. Specifically, starting from each residual node, the algorithm searches forward or backward along the direction of change of its attribute label for its successor nodes in the adjacent graph nodes that have a semantically coherent relationship. For example, a continuous transition from "agriculture" to "construction" and then to "commerce" is considered to be a jump path with consistent direction. Furthermore, when determining the jump path, contextual information such as the event response status must be combined to determine whether the path is affected by external driving factors, ensuring that each path segment in the jump path has a clear semantic transfer interpretation basis. Based on this, when a group of residual nodes have consistency in the direction of attribute label variation and present a traversable connection form in the graph structure, the link corresponding to the group of residual nodes can be finally identified as a jump path. Therefore, a series of jump paths with clear jump directions and traceable node structures are extracted from the irregular and fragmented residual graph.

[0095] Step S602 : Based on the semantic span tensor of each transition path, the residual nodes with the same label migration direction and matching state transition period in all transition paths are compressed into the same semantic segment, and multiple semantic segments are integrated to obtain a semantic chain set.

[0096] Among them, the semantic span tensor representation of the jump path is used to quantify the multi-dimensional structured expression of the label change range, state transition amplitude and time span in a jump path, so as to compare the overall degree of change of different jump paths in the process of semantic evolution.

[0097] For example, a semantic span tensor is used to characterize factors such as the overall magnitude of label transitions within a transition path, the number of change dimensions, and the duration of state transitions. By constructing a semantic span tensor for each transition path, we can determine the total amount of change that transition path undergoes within the semantic classification system and its duration over time. These tensors are then compared and analyzed to identify transition paths with consistent migration directions, similar tensor magnitudes, and matching time spans. These paths are then merged and compressed into unified semantic segments. Furthermore, to ensure that the compressed semantic segments retain the semantic trend characteristics of the original transition paths, the node order, state start and end points, and event triggering conditions within the different transition paths are regularized, ensuring that the semantic segments have a unified starting and end logic. Based on this, after generating multiple semantic segments, they are then integrated according to time index order and structural connection rules. Redundant nodes are removed, overlapping portions are merged, and the sequence of semantic segments is adjusted to ultimately construct a complete semantic chain set.

[0098] In this embodiment, first, the residual graph nodes are traversed based on the consistency of the attribute label variation direction, thereby extracting the jump path, thereby realizing the quantitative extraction of the semantic evolution trend in the non-structural determination area; secondly, the semantic span tensor of the jump path is compressed and merged, and the jump paths with consistent label migration direction and similar state transition characteristics are integrated into semantic chain segments, thereby improving the aggregation degree and chain organization ability of the residual structure expression; based on this, the mutation behavior of dynamic geographic entities in semantically discontinuous areas can be compressed modeled and chained, enhancing the structural clarity and analysis controllability of spatiotemporal semantic modeling.

[0099] In an exemplary embodiment, after performing semantic transition compression processing on the non-structural determination area corresponding to the dynamic geographic entity according to the residual map set to obtain the semantic chain set, the method further includes steps S701 to S703.

[0100] Step S701 : performing stage-by-stage division processing on each semantic chain segment in the semantic chain set to obtain a staged semantic chain set including each semantic state stage.

[0101] Among them, the staged semantic chain set containing each semantic state stage represents a chain set formed after each semantic chain segment is divided into multiple semantic state stages with internal consistency and structural independence based on the state change law, event response characteristics and label transition trend on the basis of the original semantic chain set. It is used to refine the staged behavioral characteristics in the evolution process of dynamic geographic entities; for example, a semantic chain segment expressing "from cultivated land to industrial land and then to commercial land" is divided into three semantic state stages of "agricultural development stage", "industrial transformation stage" and "commercial construction stage" to form a staged structural sequence.

[0102] Among them, the semantic state stage represents a local state unit in a semantic chain segment, which is composed of several semantic state nodes, is continuous in time, and has a stable evolution direction in semantic label variation.

[0103] For example, first, it is necessary to analyze the attribute label evolution trend of each node in each semantic segment in the semantic chain set, identify the node position where the semantic state has a significant turning point, and use it as a candidate node for the stage demarcation point. Then, combined with the time index change rate of the node, it is judged whether the state change has a stable trend. If the label change direction is consistent within a certain time range and there is no drastic fluctuation, it is classified as a continuous stage, namely the semantic state stage. In addition, the event response state can be introduced as an auxiliary judgment basis. When the event triggering frequency in a certain semantic segment is concentrated and the corresponding label state changes are consistent, the semantic segment can also be classified as a separate semantic state stage. Based on this, after the division is completed, each semantic segment is organized into multiple semantic state stages. Each semantic state stage records its start and end time, core semantic label and change type, thereby forming a staged semantic chain set containing multiple semantic state stages.

[0104] Step S702 , based on the spatial position distribution and state change trend of each semantic state stage in the staged semantic chain set, the evolutionary relationship between each semantic state stage is subjected to spatiotemporal causal coding processing to obtain a multi-stage evolutionary relationship set.

[0105] Among them, the spatial position distribution and state change trend of the semantic state stage represent the coverage, spatial extension direction and change speed of a certain semantic state stage in the geographic space, as well as its structural change path in the semantic label evolution system.

[0106] Among them, the multi-stage evolutionary relationship set represents a structured association set formed by the causal, sequential, and spatial conduction paths constructed between multiple semantic state stages, which is used to describe the state transition logic and propagation mechanism of dynamic geographic entities between multiple semantic state stages.

[0107] For example, first, for each semantic state stage, its location coverage and expansion direction in geographic space are extracted. The state change trends and evolution rates within each semantic state stage are modeled to clarify the spatial distribution characteristics and change trend characteristics of each semantic state stage. Next, based on the temporal order, the continuity relationship between the previous and subsequent semantic state stages is structured, and causal relationships are determined based on whether they exhibit continuous evolution. Based on this, if two semantic state stages partially overlap spatially, have a temporal evolutionary sequence, and their label change paths exhibit causal logical coherence, they are considered to constitute a spatiotemporal causal unit. Furthermore, when determining each spatiotemporal causal unit, not only the main evolutionary path is considered, but also branch paths, concurrent paths, and jump paths formed by event interruptions are recorded. This results in a multi-stage evolutionary relationship set containing multiple types of causal connections. This multi-stage evolutionary relationship set structurally reflects the dependencies and influence paths between different semantic state stages, providing a causal logical basis for dynamic geographic entity modeling.

[0108] Step S703 : performing state recursive modeling processing on the multi-stage evolution relationship set to obtain a spatiotemporal semantic reconstruction expression model representing the dynamic geographic entity in the multi-stage evolution process.

[0109] For example, within a multi-stage evolutionary relationship set, each semantic state stage and the causal, temporal, and spatial relationships between them are transformed into a structural system that can be used for state evolution deduction, thereby achieving holistic modeling of the multi-stage evolution of dynamic geographic entities. Specifically, a corresponding state representation structure is established for each semantic state stage, specifying its basic attributes such as spatial location, semantic label, duration, and event associations. This structure serves as the initial node corresponding to that semantic state stage in the state transition diagram. Next, paths related to the multi-type causal connections between semantic state stages, as recorded in the multi-stage evolutionary relationship set, are used as directed edges in the state transition diagram. Furthermore, recursive deduction is performed based on the connected nodes in the constructed state transition diagram. Starting from the initial state, the semantic state change process of the dynamic geographic entity is gradually advanced according to the state connection relationships, thereby deducing the complete spatiotemporal evolution path. Furthermore, the recursive deduction process not only considers the sequential connections between nodes but also handles nonlinear evolutionary behaviors such as node mergers, splits, or interruptions to express the multi-directional evolutionary trajectory of the dynamic geographic entity under complex changing conditions. Finally, by iteratively expanding the initial nodes and dynamically introducing external environmental influencing parameters in this process, a spatiotemporal semantic reconstruction expression model is finally formed to characterize the entire multi-stage semantic evolution process of dynamic geographic entities.

[0110] In this embodiment, first, the semantic segments are divided into stages, so that the continuous semantic segments containing multiple semantic evolution structures are refined into semantic state stages with internal consistency and external divisibility, thereby improving the structural clarity of the semantic evolution expression; secondly, the spatiotemporal causal relationship between the semantic state stages is established according to the spatial position distribution and state change trend of each semantic state stage, thereby constructing a multi-stage evolutionary relationship set that can reveal the semantic evolution logic and event-driven logic; thirdly, state recursive modeling is performed based on the multi-stage evolutionary relationship set, thereby generating a spatiotemporal semantic reconstruction expression model with logical traceability and process continuity; based on this, the structured, staged and logical expression of the complex semantic evolution process of dynamic geographic entities can be achieved, thereby enhancing the expression and reasoning capabilities of geographic semantic modeling.

[0111] It should be understood that, although the various steps in the flowcharts involved in the various embodiments described above are displayed in sequence according to the instructions of the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and these steps can be executed in other orders. Moreover, at least a portion of the steps in the flowcharts involved in the various embodiments described above can include multiple steps or multiple stages, and these steps or stages are not necessarily executed and completed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a portion of steps or stages in other steps.

[0112] Based on the same inventive concept, embodiments of the present application also provide a spatial geographic data processing device for implementing the aforementioned spatial geographic data processing method. The implementation solution provided by this device is similar to the implementation solution described in the aforementioned method. Therefore, the specific limitations in one or more of the following embodiments of the spatial geographic data processing device can be found in the above-mentioned limitations on the spatial geographic data processing method and will not be further elaborated here.

[0113] In an exemplary embodiment, Figure 2 As shown, a spatial geographic data processing device is provided, comprising: an acquisition module 201, an encoding module 202, a residual module 203 and a compression module 204, wherein: An acquisition module 201 is used to acquire spatial geographic data corresponding to a dynamic geographic entity. The spatial geographic data includes a position change sequence, an attribute tag sequence, and an external event record. The encoding module 202 is used to perform cross-domain nested encoding processing on spatial geographic data to obtain a multi-domain entity relationship encoding set with multi-dimensional unstructured temporal features and heterogeneous semantic relationship dimensions; The residual module 203 is used to generate residual graphs for geographic entity temporal nodes with label drift and state jump in the multi-domain entity relationship encoding set, thereby obtaining a residual graph set across temporal periods. Each residual graph in the residual graph set represents a residual semantic relationship pattern of a dynamic geographic entity under a discontinuous semantic path. The compression module 204 is used to perform semantic transition compression processing on the non-structural determination area corresponding to the dynamic geographic entity according to the residual map set to obtain a semantic chain set for representing the spatiotemporal semantic reconstruction expression model of the dynamic geographic entity in the multi-stage evolution process.

[0114] In an exemplary embodiment, the encoding module 202 is also used to: perform time alignment processing on the position change sequence, attribute label sequence and external event record in spatial geographic data to obtain a temporal alignment sequence set; perform association modeling processing on the spatial positions, attribute labels and event states corresponding to different time nodes according to the temporal alignment sequence set to obtain a composite entity node set with time index, spatial label state and event response scalar; and perform nested structure construction processing on the composite entity node set by extracting the semantic connection relationship, temporal adjacency relationship and event causal relationship between the composite entity nodes to obtain a multi-domain entity relationship encoding set.

[0115] In an exemplary embodiment, the encoding module 202 is also used to: perform relationship triple extraction processing on the composite entity node set according to the semantic connection relationship, temporal adjacency relationship and event causal relationship between the composite entity nodes, and obtain a multi-source relationship graph containing a semantic connection edge set, a temporal adjacency edge set and an event causal edge set; in the multi-source relationship graph, perform double-layer association clustering processing on the semantic connection edge set and the event causal edge set to construct a labeled event nested subgraph, and perform nested graph fusion processing on each labeled event nested subgraph according to the topological skeleton constructed by the temporal adjacency edge set to obtain a multi-domain entity relationship encoding set.

[0116] In an exemplary embodiment, the residual module 203 is also used to: perform difference measurement processing on the attribute label state and event response state of each geographic entity time series node under the continuous time index in the multi-domain entity relationship coding set to obtain a label drift intensity matrix and a state jump probability matrix; perform abnormal trajectory identification processing on each geographic entity time series node according to the label drift intensity matrix and the state jump probability matrix to obtain an abnormal trajectory node set marked with semantic offset nodes; perform residual graph generation processing on the abnormal trajectory node set across time periods to obtain a residual graph set across temporal periods.

[0117] In an exemplary embodiment, the residual module 203 is also used to: perform joint anomaly measurement processing on the label drift intensity matrix and the state jump probability matrix to obtain a semantic mutation gradient sequence of each geographic entity time series node under a continuous time index; cluster the time index positions with abnormal jump amplitudes in the semantic mutation gradient sequence according to the state transfer direction and the mutation time difference to obtain a mutation node set; perform context consistency verification and labeling on the mutation node set according to a preset context threshold to obtain an abnormal trajectory node set labeled with semantic offset nodes.

[0118] In an exemplary embodiment, the compression module 204 is also used to: traverse the residual nodes located in the non-structural determination area in the residual graph set through a preset semantic traversal algorithm, and extract jump paths that are consistent with the attribute label variation direction of each residual node; according to the semantic span tensor of each jump path, compress the residual nodes with consistent label migration directions and matching state transition cycles in all jump paths into the same semantic chain segment, and integrate multiple semantic chain segments to obtain a semantic chain set.

[0119] In an exemplary embodiment, the device also includes a modeling module, which is used to: perform stage-by-stage division processing on each semantic chain segment in the semantic chain set to obtain a staged semantic chain set containing each semantic state stage; perform spatiotemporal causal coding processing on the evolutionary relationship between each semantic state stage based on the spatial position distribution and state change trend of each semantic state stage in the staged semantic chain set to obtain a multi-stage evolutionary relationship set; perform state recursive modeling processing on the multi-stage evolutionary relationship set to obtain a spatiotemporal semantic reconstruction expression model that characterizes dynamic geographic entities in the multi-stage evolution process.

[0120] Each module in the aforementioned spatial geographic data processing device may be implemented in whole or in part through software, hardware, or a combination thereof. Each module may be embedded in or independent of a processor in a computer device in the form of hardware, or may be stored in a memory in the computer device in the form of software, so that the processor can call and execute the corresponding operations of each module.

[0121] In an exemplary embodiment, a computer device is provided, including a memory and a processor. The memory stores a computer program, and the processor implements the steps in any of the above embodiments when executing the computer program.

[0122] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps in any of the above embodiments are implemented.

[0123] Those skilled in the art will appreciate that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the above-mentioned embodiments. In particular, any reference to memory, database, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The databases involved in the various embodiments provided herein may include at least one of a relational database and a non-relational database. Non-relational databases may include, but are not limited to, distributed databases based on blockchains. The processors involved in the various embodiments provided herein may be, but are not limited to, general-purpose processors, central processing units (CPUs), graphics processing units (GPUs), digital signal processors (DSPs), programmable logic devices (PLDs), data processing logic devices based on quantum computing, and the like.

[0124] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0125] The above-described embodiments merely represent several implementation methods of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present application. It should be noted that a person of ordinary skill in the art may make various modifications and improvements without departing from the spirit of the present application, and these modifications and improvements fall within the scope of protection of the present application. Therefore, the scope of protection of the present application shall be determined by the appended claims.

Claims

1. A spatial geographic data processing method, characterized in that: The method comprises: Acquire spatial geographic data corresponding to a dynamic geographic entity, wherein the spatial geographic data includes a position change sequence, an attribute label sequence, and an external event record; Performing cross-domain nested coding processing on the spatial geographic data to obtain a multi-domain entity relationship coding set with multi-dimensional unstructured temporal features and heterogeneous semantic relationship dimensions; In the multi-domain entity relationship encoding set, residual graph generation processing is performed on geographic entity time series nodes with label drift and state jump to obtain a residual graph set across temporal periods, each residual graph in the residual graph set representing a residual semantic relationship pattern of the dynamic geographic entity under a discontinuous semantic path; According to the residual graph set, semantic transition compression processing is performed on the non-structural determination area corresponding to the dynamic geographic entity to obtain a semantic chain set for representing the spatiotemporal semantic reconstruction expression model of the dynamic geographic entity in a multi-stage evolution process.

2. The method according to claim 1, characterized in that The cross-domain nested coding process is performed on the spatial geographic data to obtain a multi-domain entity relationship coding set with multi-dimensional unstructured temporal features and heterogeneous semantic relationship dimensions, including: In the spatial geographic data, time alignment processing is performed on the position change sequence, the attribute label sequence, and the external event record to obtain a temporally aligned sequence set; Performing association modeling on the spatial positions, attribute labels, and event states corresponding to different time nodes according to the temporal alignment sequence set to obtain a composite entity node set with a time index, a spatial label state, and an event response scalar; By extracting the semantic connection relationship, temporal adjacency relationship and event causal relationship between composite entity nodes, a nested structure construction process is performed on the composite entity node set to obtain a multi-domain entity relationship coding set.

3. The method according to claim 2, characterized in that The method extracts the semantic connection relationship, temporal adjacency relationship and event causal relationship between the composite entity nodes, performs nested structure construction processing on the composite entity node set, and obtains a multi-domain entity relationship coding set, including: According to the semantic connection relationship, temporal adjacency relationship and event causal relationship between the composite entity nodes, the composite entity node set is subjected to relationship triple extraction processing to obtain a multi-source relationship graph including a semantic connection edge set, a temporal adjacency edge set and an event causal edge set; In the multi-source relationship graph, the semantic connection edge set and the event causal edge set are subjected to double-layer association clustering processing to construct a labeled event nested subgraph, and each labeled event nested subgraph is subjected to nested graph fusion processing according to the topological skeleton constructed according to the temporal adjacency edge set to obtain a multi-domain entity relationship encoding set.

4. The method according to claim 1, wherein In the multi-domain entity relationship encoding set, residual graph generation processing is performed on the geographic entity time series nodes with label drift and state jump to obtain a residual graph set across the temporal period, including: Performing difference measurement processing on the attribute label state and event response state of each geographic entity time series node in the multi-domain entity relationship coding set under the continuous time index to obtain a label drift intensity matrix and a state jump probability matrix; According to the label drift strength matrix and the state jump probability matrix, abnormal trajectory identification processing is performed on each geographic entity time series node to obtain a set of abnormal trajectory nodes marked with semantic offset nodes; A residual graph generation process is performed on the abnormal trajectory node set across time periods to obtain a residual graph set across temporal periods.

5. The method according to claim 4, characterized in that The abnormal trajectory identification process is performed on each geographic entity time series node according to the label drift strength matrix and the state jump probability matrix to obtain an abnormal trajectory node set marked with semantic offset nodes, including: Performing joint anomaly measurement processing on the label drift intensity matrix and the state jump probability matrix to obtain a semantic mutation gradient sequence of each geographic entity time series node under a continuous time index; According to the state transfer direction and the mutation time difference, clustering is performed on the time index positions with abnormal jump amplitudes in the semantic mutation gradient sequence to obtain a mutation node set; According to a preset context threshold, context consistency verification and labeling are performed on the mutation node set to obtain an abnormal trajectory node set labeled with semantic offset nodes.

6. The method according to claim 1, characterized in that The performing semantic transition compression processing on the non-structural determination area corresponding to the dynamic geographic entity according to the residual atlas set to obtain a semantic chain set includes: Using a preset semantic traversal algorithm, traverse the residual nodes located in the non-structural determination area in the residual graph set, and extract jump paths that are consistent with the attribute label variation directions of each residual node; According to the semantic span tensor of each jump path, the residual nodes with consistent label migration direction and matching state transition period in all jump paths are compressed into the same semantic chain segment, and multiple semantic chain segments are integrated to obtain a semantic chain set.

7. The method according to claim 1, characterized in that After performing semantic transition compression processing on the non-structural determination area corresponding to the dynamic geographic entity according to the residual graph set to obtain a semantic chain set, the method further includes: Performing phase division processing on each semantic chain segment in the semantic chain set to obtain a phased semantic chain set including each semantic state stage; According to the spatial position distribution and state change trend of each semantic state stage in the staged semantic chain set, the evolutionary relationship between each semantic state stage is subjected to spatiotemporal causal coding processing to obtain a multi-stage evolutionary relationship set; State recursive modeling is performed on the multi-stage evolution relationship set to obtain a spatiotemporal semantic reconstruction expression model that characterizes the dynamic geographic entity in the multi-stage evolution process.

8. A spatial geographic data processing device, characterized in that: The device comprises: An acquisition module is used to acquire spatial geographic data corresponding to a dynamic geographic entity, wherein the spatial geographic data includes a position change sequence, an attribute label sequence, and an external event record; An encoding module is used to perform cross-domain nested encoding processing on the spatial geographic data to obtain a multi-domain entity relationship encoding set with multi-dimensional unstructured temporal features and heterogeneous semantic relationship dimensions; A residual module is configured to generate residual graphs for geographic entity temporal nodes with label drift and state jump in the multi-domain entity relationship encoding set, thereby obtaining a residual graph set across temporal periods, wherein each residual graph in the residual graph set represents a residual semantic relationship pattern of the dynamic geographic entity under a discontinuous semantic path; A compression module is used to perform semantic transition compression processing on the non-structural determination area corresponding to the dynamic geographic entity according to the residual map set to obtain a semantic chain set for representing the spatiotemporal semantic reconstruction expression model of the dynamic geographic entity in a multi-stage evolution process.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images