Land surface change prediction method based on spatio-temporal rule guided knowledge graph reasoning and application thereof

Abstract

The application relates to the technical field of remote sensing image processing and knowledge graph, in particular to a land surface change prediction method based on spatiotemporal rule guided knowledge graph reasoning and application thereof. The method can maximize the utilization of the characteristics of each land surface change by obtaining the change information of the land surface in time and space in the mode of excavating the knowledge graph rules, can infer the future land surface change by calculating the time sequence constraint and the space constraint on the sampling entity, and can comprehensively consider the complexity of the land surface change by jointly optimizing the embedding vector with the loss, so that the spatiotemporal rules are considered in the land surface change knowledge graph reasoning research, and the inherent characteristics and spatiotemporal rule constraints of the triples in the knowledge graph are effectively utilized. The application aims to solve the problem of how to accurately predict the future land surface change by using the knowledge graph reasoning.

Description

A Land Change Prediction Method Based on Spatiotemporal Rule-Guided Knowledge Graph Reasoning and Its Application Technical Field

[0001] This application relates to the fields of remote sensing image processing and knowledge graph technology, and in particular to a method for predicting land surface changes based on spatiotemporal rules-guided knowledge graph reasoning and its application. Background Technology

[0002] With the rapid development of remote sensing technology, monitoring land surface changes using multi-source remote sensing image data has become an important research direction in the fields of geographic information science and environmental science. Traditional methods for monitoring land surface changes mainly rely on manual interpretation or simple threshold segmentation based on a single remote sensing data source. These methods are not only inefficient, but also difficult to accurately identify and analyze land surface change information in complex geographical environments and variable land surface conditions.

[0003] Existing technical solutions apply deep learning to extract surface features from remote sensing images, effectively extracting information such as land use type and vegetation cover through semantic segmentation algorithms. However, current methods are mainly applied to processing remote sensing images from a single time period or a single image, making it difficult to accurately extract features from remote sensing images spanning longer periods.

[0004] For the application scenario of land surface change monitoring, it is precisely necessary to collect remote sensing images over a relatively long period and perform long-term feature extraction to achieve accurate land surface change monitoring. Existing technologies are mostly based on data-driven deep learning models, such as traditional translation models like TransE, which learn the embedded features of entities and relationships to achieve knowledge graph completion and reasoning. However, in the land surface change monitoring scenario, due to the lack of effective integration of spatiotemporal factors, traditional models struggle to fully utilize the spatiotemporal correlation of land surface changes during the reasoning process, resulting in poor accuracy of the inference results.

[0005] In view of this, this application proposes a knowledge graph-based method for predicting land surface changes that considers the spatiotemporal correlation of land surface changes, thereby enabling accurate prediction of future land surface changes using knowledge graph reasoning. Summary of the Invention

[0006] The main purpose of this application is to provide a method for predicting land surface changes based on knowledge graph reasoning guided by spatiotemporal rules, aiming to solve the problem of how to accurately predict future land surface changes using knowledge graph reasoning.

[0007] To achieve the above objectives, this application provides a method for predicting land surface changes based on knowledge graph reasoning guided by spatiotemporal rules, the method comprising:

[0008] S10, determine the graph loss between positive triples and negative triples in the input land surface change knowledge graph, wherein the positive triples are triples with known relationships in the land surface change knowledge graph, and the negative triples are triples with no known relationships in the land surface change knowledge graph;

[0009] S20, sample spatial entity pairs that satisfy preset spatial propagation rules from the knowledge graph of land surface change, and calculate the spatial constraints of the spatial entity pairs; and sample causal entity pairs that satisfy preset temporal causal rules from the knowledge graph of land surface change, and calculate the temporal constraints of the causal entity pairs.

[0010] S30, the graph loss, the spatial constraint, and the temporal constraint are weighted and summed to obtain the joint loss;

[0011] S40, the embedding vector is optimized based on the joint loss, so as to predict the surface change based on the optimized embedding vector.

[0012] Optionally, the map loss The calculation expression is:

[0013]

[0014] In the formula, They represent The corresponding embedding vector has a size of d is a positive integer; Each represents a different entity, and E represents the entity set; R represents the relation in a positive triple; G represents the relation in a negative triplet; G represents the knowledge graph of land surface change. For Lp norm, This is the interval hyperparameter.

[0015] Optionally, the known relations in the triples include spatial relations. Event-related relationships Relationship with time ;

[0016] The preset spatial propagation rules Represented as containing spatial class relations The logical rules are expressed as follows:

[0017] ;

[0018] The preset temporal causal rules Represented as a relationship containing event classes The logical rules are expressed as follows:

[0019] ;

[0020] In the formula, X, Y, and Z represent entity variables in different triples.

[0021] Optionally, the spatial constraint The mathematical expression is:

[0022]

[0023] In the formula, Preset spatial propagation rules confidence level It is the sigmoid activation function. For vector concatenation, Let be the weight matrix of the spatial propagation rule, with size . , and For spatial entity pairs and The embedding vector;

[0024] in:

[0025]

[0026] In the formula, The support is represented by the statistically derived support from the land surface change knowledge graph, which simultaneously satisfies the rule body. And the rule header H A It is also obtained from the existing quantity; Represents the objects that satisfy the rules in the knowledge graph of land surface change. Quantity; This represents n triples in the spatial propagation rule.

[0027] Optionally, the timing constraints The mathematical expression is:

[0028]

[0029] In the formula, To pre-determine the causal rules of time sequence confidence level The weight matrix for time-series causal rules has a size of [value missing]. , and For causal entity pairs and The embedding vector;

[0030] in:

[0031]

[0032] In the formula, The support is represented by the statistical expression obtained from the knowledge graph of land surface change, which simultaneously satisfies n rules. And the rule header H B Also obtained from the existing quantity, where the rule header H B It includes results of future land surface changes; Represents the objects that satisfy the rules in the knowledge graph of land surface change. Quantity; This represents the n triples in a temporal causal rule.

[0033] Optionally, the mathematical expression for the joint loss is:

[0034]

[0035] In the formula, For joint losses, For weight hyperparameters, For map loss, Due to spatial constraints, For timing constraints.

[0036] Optionally, the prediction of land surface changes based on the optimized embedding vector includes:

[0037] S41, Calculate the score of the optimized embedding vector:

[0038]

[0039] In the formula, , , They are respectively , and Embedded vector, , and These represent the head entity, candidate entity, and relation in the triplet to be predicted, respectively. It is an Lp norm;

[0040] S42, Select the entity with the highest score. As a result of inferences about surface changes:

[0041]

[0042] In the formula, E represents the entity set.

[0043] In addition, to achieve the above objectives, this application also provides a land surface change prediction method based on spatiotemporal rule-guided knowledge graph reasoning as described in any of the preceding claims, and its application in land surface change monitoring.

[0044] In addition, to achieve the above objectives, this application also provides a computer system, the computer system comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the computer program is executed by the processor, it implements the steps of the land surface change prediction method based on spatiotemporal rule-guided knowledge graph reasoning as described in any of the preceding claims.

[0045] In addition, to achieve the above objectives, this application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the land surface change prediction method based on spatiotemporal rule-guided knowledge graph reasoning as described in any of the preceding claims.

[0046] This application has at least the following beneficial effects:

[0047] 1. By mining knowledge graph rules to obtain information on temporal and spatial changes in the Earth's surface, the characteristics of each change in the Earth's surface can be maximized.

[0048] 2. By sampling entities to compute temporal and spatial constraints, the embedding vector is optimized using joint loss to infer future surface changes;

[0049] 3. Taking into account the complexity of surface changes, spatiotemporal rules are considered in the reasoning research of knowledge graphs on surface changes, and the inherent characteristics of triples in the knowledge graph and spatiotemporal rule constraints are effectively utilized. Attached Figure Description

[0050] Figure 1 is a flowchart illustrating the land surface change prediction method based on spatiotemporal rule-guided knowledge graph reasoning according to an embodiment of this application.

[0051] Figure 2 is an example diagram of land surface change detection involved in the embodiments of this application;

[0052] Figure 3 is an example diagram of the knowledge graph of land surface change involved in the embodiments of this application;

[0053] Figure 4 is a schematic diagram of the hardware operating environment of the computer system involved in the embodiments of this application.

[0054] The realization of the purpose, functional features and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0055] To better understand the above technical solutions, exemplary embodiments of this disclosure will be described in more detail below with reference to the accompanying drawings. While exemplary embodiments of this disclosure are shown in the drawings, it should be understood that this disclosure can be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of this disclosure to those skilled in the art.

[0056] First Embodiment

[0057] Referring to Figure 1, this embodiment provides a method for predicting land surface changes based on knowledge graph reasoning guided by spatiotemporal rules, including the following steps:

[0058] S10, determine the graph loss between positive triples and negative triples in the input land surface change knowledge graph, wherein the positive triples are triples with known relationships in the land surface change knowledge graph, and the negative triples are triples with no known relationships in the land surface change knowledge graph;

[0059] In this embodiment, the map loss is first calculated from the positive and negative triples set in the knowledge graph of land surface change.

[0060] The ternary unit is the basic unit in the knowledge graph of land surface change. Specifically, the knowledge graph of land surface change is defined as: , where E represents the entity set and R represents the relation set.

[0061] Positive triples are triples with known relationships in the land surface change knowledge graph, while negative triples are triples with no known relationships in the land surface change knowledge graph. Specifically, a positive triple is represented as... ,in Representing different entities, represent and The relationships between them. From the set of relationships. Select except Other than the relationship, it is denoted as Replace the relations in the positive triples to obtain the negative triples, denoted as: .

[0062] Optionally, map loss The calculation expression is:

[0063]

[0064] In the formula, They represent The corresponding embedding vector has a size of d is a positive integer; Each represents a different entity, and E represents the entity set; R represents the relation in a positive triple; G represents the relation in a negative triplet; G represents the knowledge graph of land surface change. For Lp norm, This is the interval hyperparameter.

[0065] For example, in some optional implementations, multi-temporal remote sensing images are acquired through satellite data receiving systems, data sharing platforms, etc. Then, the acquired multi-temporal remote sensing images are used to extract and number change areas using a change detection model, as shown in Figure 2. Based on prior knowledge, the land cover types and change types of the change areas are determined as land cover entities and change entities.

[0066] The extracted land feature entities (such as plot 1, plot 2, etc.) and change entities (such as reduced vegetation cover, conversion of arable land to construction land) are used as nodes in the land surface change knowledge graph, and spatial relationships are added. (e.g., adjacent within 1km, etc.), event-based relationships (e.g., changes in type, impact, consequences, etc.), time-related relationships Using edges such as 2022-2024 and 2020-2022 as the knowledge graph, the land surface change knowledge graph is shown in Figure 3. The land surface change knowledge graph G is represented using graph theory and defined as follows: In this context, E represents the set of entities, and R represents the set of relations. The triplet is the basic unit in the land surface change knowledge graph. Examples of triples in the land surface change knowledge graph include: Change Type (Plot 6, Farmland to Construction Land), Adjacent 1km (Plot 2, Plot 3), 2020-2022 (Plot 2, Reduced Vegetation Cover).

[0067] The existing triples in the knowledge graph of land surface change are used as positive triples. For example: 2020-2022 (Plot 1, deforestation), impact (deforestation, conversion of forest land to farmland), adjacent 1km (Plot 1, Plot 2), 2020-2022 (Plot 2, deforestation), adjacent 1km (Plot 2, Plot 3), change type (Plot 3, conversion of forest land to farmland), etc. From the set of relationships. Select except Replace the relation in a positive triple with the relation other than the one in the positive triple to obtain a negative triple. Examples include: adjacent 1km (plot 1, forest land logging), adjacent 1km (forest land logging, forest land converted to farmland), change type (plot 1, plot 2), adjacent 1km (plot 2, forest land logging), 2020-2022 (plot 2, plot 3), adjacent 10km (plot 3, forest land converted to farmland), etc.

[0068] S20, sample spatial entity pairs that satisfy preset spatial propagation rules from the knowledge graph of land surface change, and calculate the spatial constraints of the spatial entity pairs; and sample causal entity pairs that satisfy preset temporal causal rules from the knowledge graph of land surface change, and calculate the temporal constraints of the causal entity pairs.

[0069] In this step, spatial and temporal constraints in the knowledge graph of land surface change are also calculated.

[0070] It should be noted that, based on the spatial propagation and temporal causal characteristics of land surface change, this embodiment uses the AMIE rule mining algorithm to analyze the triples in the land surface change knowledge graph and extract logical rules. ,in, Represents the rule body. It represents n triples, including spatiotemporal attributes and surface feature attributes. This represents the rule header, which contains the results of future surface changes.

[0071] The pre-defined logical rules include spatial propagation rules and temporal causal rules, where X, Y, and Z represent different entity variables.

[0072] Specifically, the preset spatial propagation rules Represented as containing spatial class relations The logical rules are expressed as follows:

[0073] ;

[0074] The preset temporal causal rules Represented as a relationship containing event classes The logical rules are expressed as follows:

[0075] ;

[0076] In the formula, X, Y, and Z represent entity variables in different triples.

[0077] In some alternative implementations, among the mined logical rules, spatial propagation rules and temporal causal rules with a confidence level greater than the threshold θ are selected as the corresponding spatial entity pairs and causal entity pairs.

[0078] Specifically, and optionally, for the calculation of spatial constraints, samples satisfying spatial propagation rules are taken from the knowledge graph of surface changes. For entities, select the entity pair in the rule header, and denot it as a spatial entity pair. Computational space constraints :

[0079]

[0080] In the formula, Preset spatial propagation rules confidence level It is the sigmoid activation function. For vector concatenation, Let be the weight matrix of the spatial propagation rule, with size . , and For spatial entity pairs and The embedding vector;

[0081] in:

[0082]

[0083] In the formula, The support is represented by the statistically derived support from the land surface change knowledge graph, which simultaneously satisfies the rule body. And the rule header H A It is also obtained from the existing quantity; Represents the objects that satisfy the rules in the knowledge graph of land surface change. Quantity; This represents n triples in the spatial propagation rule.

[0084] Specifically and optionally, for the calculation of temporal constraints, samples satisfying temporal causal rules are sampled from the land surface change knowledge graph. Entity pairs selected from the rule header are denoted as causal entity pairs. Calculate timing constraints :

[0085]

[0086] In the formula, To pre-determine the causal rules of time sequence confidence level The weight matrix for time-series causal rules has a size of [value missing]. , and For causal entity pairs and The embedding vector;

[0087] in:

[0088]

[0089] In the formula, The support is represented by the statistical expression obtained from the knowledge graph of land surface change, which simultaneously satisfies n rules. And the rule header H B Also obtained from the existing quantity, where the rule header H B It includes results of future land surface changes; Represents the objects that satisfy the rules in the knowledge graph of land surface change. Quantity; This represents the n triples in a temporal causal rule.

[0090] For example, also based on the content of the previous example, let's assume the spatial propagation rules. satisfy:

[0091] “Adjacent 1km (Plot 2, Plot 3) ∧ Change type (Plot 3, Forest land converted to farmland) → Type change (Plot 2, Forest land converted to farmland) β1=0.83”.

[0092] Temporal causality rules satisfy:

[0093] “2020-2022 (Plot 1, forest land logging) ∧ 2022-2024 (Plot 1, forest land converted to farmland) → Impact (forest land logging, forest land converted to farmland) β2=0.7”.

[0094] The spatial entity pairs are obtained as follows: N1 = {(Plot 2, forest land converted to farmland), (Plot 5, farmland converted to construction land)};

[0095] Causal entity pair: N2 = {(forest land logging, forest land conversion to farmland)}.

[0096] Note that the above example is only a logical illustration and not an actual data structure.

[0097] S30, the graph loss, the spatial constraint, and the temporal constraint are weighted and summed to obtain the joint loss;

[0098] In this embodiment, the mathematical expression for the joint loss is:

[0099]

[0100] In the formula, For joint losses, For weight hyperparameters, For map loss, Due to spatial constraints, For timing constraints.

[0101] S40, the embedding vector is optimized based on the joint loss, so as to predict the surface change based on the optimized embedding vector.

[0102] In this embodiment, backpropagation is performed based on the joint loss to update the embedding vectors of entities and relations. A batch number is set, and iterative training is performed until the joint loss converges, at which point the optimized embedding vectors are saved.

[0103] Optionally, the prediction of surface changes based on the optimized embedding vector is as follows:

[0104] S41, Calculate the score of the optimized embedding vector:

[0105]

[0106] In the formula, , , They are respectively , and Embedded vector, , and These represent the head entity, candidate entity, and relation in the triplet to be predicted, respectively. It is an Lp norm;

[0107] S42, Select the entity with the highest score. As a result of inferences about surface changes:

[0108]

[0109] In the formula, E represents the entity set.

[0110] In the technical solution provided in this embodiment, information on changes in land surface in time and space is obtained by mining knowledge graph rules. This maximizes the use of the characteristics of each land surface change. By sampling entities to calculate temporal and spatial constraints, the embedding vector is optimized by joint loss, and future land surface changes are inferred. This comprehensively considers the complexity of land surface changes, takes into account spatiotemporal rules in the research on knowledge graph reasoning of land surface changes, and effectively utilizes the characteristics of the triples in the knowledge graph and the spatiotemporal rule constraints.

[0111] Furthermore, as an implementation scheme, this application provides a method for predicting land surface changes based on knowledge graph reasoning guided by spatiotemporal rules, as described in any of the preceding claims, and its application in land surface change monitoring.

[0112] Furthermore, as one implementation scheme, Figure 4 is a schematic diagram of the hardware operating environment of the computer system involved in the embodiment of this application.

[0113] As shown in Figure 4, the computer system may include: a processor 1001, such as a CPU; a memory 1005; a user interface 1003; a network interface 1004; and a communication bus 1002. The communication bus 1002 is used to enable communication between these components. The user interface 1003 may include a display screen or an input unit such as a keyboard. Optionally, the user interface 1003 may also include a standard wired interface or a wireless interface. The network interface 1004 may optionally include a standard wired interface or a wireless interface (such as a Wi-Fi interface). The memory 1005 may be high-speed RAM or non-volatile memory, such as a disk drive. Optionally, the memory 1005 may also be a storage device independent of the aforementioned processor 1001.

[0114] Those skilled in the art will understand that the computer system architecture shown in Figure 4 does not constitute a limitation on the computer system and may include more or fewer components than shown, or combine certain components, or have different component arrangements.

[0115] As shown in Figure 4, the memory 1005, as a storage medium, may include an operating system, a network communication module, a user interface module, and computer programs. The operating system is a program that manages and controls the hardware and software resources of the computer system, as well as the operation of the computer programs and other software or programs.

[0116] In the computer system shown in Figure 4, the user interface 1003 is mainly used to connect to the terminal and communicate with the terminal; the network interface 1004 is mainly used to communicate with the backend server; and the processor 1001 can be used to call the computer program stored in the memory 1005.

[0117] In this embodiment, the computer system includes: a memory 1005, a processor 1001, and a computer program stored in the memory and executable on the processor, wherein:

[0118] When processor 1001 calls a computer program stored in memory 1005, it performs the following operations:

[0119] S10, determine the graph loss between positive triples and negative triples in the input land surface change knowledge graph, wherein the positive triples are triples with known relationships in the land surface change knowledge graph, and the negative triples are triples with no known relationships in the land surface change knowledge graph;

[0120] S20, sample spatial entity pairs that satisfy preset spatial propagation rules from the knowledge graph of land surface change, and calculate the spatial constraints of the spatial entity pairs; and sample causal entity pairs that satisfy preset temporal causal rules from the knowledge graph of land surface change, and calculate the temporal constraints of the causal entity pairs.

[0121] S30, the graph loss, the spatial constraint, and the temporal constraint are weighted and summed to obtain the joint loss;

[0122] S40, the embedding vector is optimized based on the joint loss, so as to predict the surface change based on the optimized embedding vector.

[0123] Furthermore, those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program includes program instructions and can be stored in a storage medium, which is a computer-readable storage medium. The program instructions are executed by at least one processor in a computer system to implement the process steps of the embodiments of the above methods.

[0124] Therefore, this application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the various steps of the land surface change prediction method based on spatiotemporal rule-guided knowledge graph reasoning as described in the above embodiments.

[0125] The computer-readable storage medium can be any computer-readable storage medium capable of storing program code, such as a USB flash drive, portable hard drive, read-only memory (ROM), magnetic disk, or optical disk.

[0126] It should be noted that, since the storage medium provided in the embodiments of this application is the storage medium used to implement the methods of the embodiments of this application, those skilled in the art can understand the specific structure and variations of the storage medium based on the methods described in the embodiments of this application, and therefore will not be repeated here. All storage media used in the methods of the embodiments of this application fall within the scope of protection of this application.

[0127] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0128] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, as well as combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in one or more blocks of the flowchart illustrations and / or one or more blocks of the block diagrams.

[0129] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means that implement the functions specified in one or more flowcharts and / or one or more block diagrams.

[0130] These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process, such that the instructions, which execute on the computer or other programmable apparatus, provide steps for implementing the functions specified in one or more flowcharts and / or one or more block diagrams.

[0131] It should be noted that any reference signs placed between parentheses in the claims should not be construed as limiting the claims. The word "comprising" does not exclude the presence of components or steps not listed in the claims. The word "a" or "an" preceding a component does not exclude the presence of a plurality of such components. This application can be implemented by means of hardware comprising several different components and by means of a suitably programmed computer. In a unit claim enumerating several means, several of these means may be embodied by the same item of hardware. The use of the words first, second, and third, etc., does not indicate any order. These words can be interpreted as names.

[0132] Although preferred embodiments of this application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of this application.

[0133] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.

Claims

1. A method for predicting land surface changes based on knowledge graph reasoning guided by spatiotemporal rules, characterized in that, Includes the following steps: S10, determine the graph loss between positive and negative triples in the input land surface change knowledge graph, wherein the positive triples are triples with known relationships in the land surface change knowledge graph, and the negative triples are triples without known relationships in the land surface change knowledge graph; S20, sample spatial entity pairs that satisfy preset spatial propagation rules from the land surface change knowledge graph and calculate the spatial constraints of the spatial entity pairs; and sample causal entity pairs that satisfy preset temporal causality rules from the land surface change knowledge graph and calculate the temporal constraints of the causal entity pairs; S30, perform a weighted sum of the graph loss, the spatial constraints, and the temporal constraints to obtain a joint loss; S40, optimize the embedding vector based on the joint loss, and predict land surface changes based on the optimized embedding vector; the graph loss The calculation expression is: In the formula, They represent The corresponding embedding vector has a size of d is a positive integer; Each represents a different entity, and E represents the entity set; R represents the relation in a positive triple; G represents the relation in a negative triplet; G represents the knowledge graph of land surface change. For Lp norm, The interval hyperparameter; the known relations in the triples include spatial class relations. Event-related relationships Relationship with time The preset spatial propagation rules Represented as containing spatial class relations The logical rules are expressed as follows: The preset temporal causal rules Represented as a relationship containing event classes The logical rules are expressed as follows: In the formula, X, Y, and Z represent entity variables in different triples; the prediction of land surface changes based on the optimized embedding vector includes: S41, calculating the score of the optimized embedding vector: In the formula, 、 、 They are respectively 、 and Embedded vector, 、 and These represent the head entity, candidate entity, and relation in the triplet to be predicted, respectively. For Lp norm; S42, select the entity with the highest score. As a result of inferences about surface changes: In the formula, E represents the entity set.

2. The method as described in claim 1, characterized in that, The spatial constraints The mathematical expression is: In the formula, Preset spatial propagation rules confidence level It is the sigmoid activation function. For vector concatenation, Let be the weight matrix of the spatial propagation rule, with size . ， and For spatial entity pairs and The embedding vector; where: In the formula, The support is represented by the statistically derived support from the land surface change knowledge graph, which simultaneously satisfies the rule body. And the rule header H A It is also obtained from the existing quantity; Represents the objects that satisfy the rules in the knowledge graph of land surface change. Quantity; This represents n triples in the spatial propagation rule.

3. The method as described in claim 1, characterized in that, The timing constraints The mathematical expression is: In the formula, To pre-determine the causal rules of time sequence confidence level The weight matrix for time-series causal rules has a size of [value missing]. ， and For causal entity pairs and The embedding vector; where: In the formula, The support is represented by the statistical expression obtained from the knowledge graph of land surface change, which simultaneously satisfies n rules. And the rule header H B Also obtained from the existing quantity, where the rule header H B It includes results of future land surface changes; Represents the objects that satisfy the rules in the knowledge graph of land surface change. Quantity; This represents the n triples in a temporal causal rule.

4. The method according to any one of claims 1 to 3, characterized in that, The mathematical expression for the joint loss is: In the formula, For joint losses, For weight hyperparameters, For map loss, Due to spatial constraints, For timing constraints.

5. The application of a land surface change prediction method based on spatiotemporal rule-guided knowledge graph reasoning as described in any one of claims 1 to 4 in land surface change monitoring.

6. A computer system, characterized in that, The computer system includes: a memory, a processor, and a computer program stored in the memory and executable on the processor. When the computer program is executed by the processor, it implements the steps of the land surface change prediction method based on spatiotemporal rule-guided knowledge graph reasoning as described in any one of claims 1 to 4.

7. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the steps of the land surface change prediction method based on spatiotemporal rule-guided knowledge graph reasoning as described in any one of claims 1 to 4.

Images