Natural resource change pattern spot purification method, device and equipment based on knowledge graph

Abstract

The invention relates to the technical field of remote sensing data and artificial intelligence, in particular to a natural resource change pattern spot purification method, device and equipment based on a knowledge graph, and the method comprises the steps: obtaining an original change pattern spot; inputting the original change pattern spot into a pre-constructed remote sensing space-time knowledge graph model to obtain a model prediction pattern spot, the pre-constructed remote sensing space-time knowledge graph model being constructed based on a remote sensing space-time knowledge graph ontology mode; and based on a preset inference rule, performing inference purification on the model prediction pattern spots to obtain purified change pattern spots. Therefore, the problem of high false alarm rate of change pattern spot detection by a deep learning model in the field of natural resource monitoring in related technologies is solved, and the working efficiency and accuracy of natural resource monitoring work are improved.

Description

Technical Field

[0001] The present invention relates to the field of remote sensing data and artificial intelligence technology, and in particular to a method, device and equipment for purifying natural resource change spots based on a knowledge graph. Background Art

[0002] Natural resource monitoring is an important part of the national natural resource management business. With the development of remote sensing and artificial intelligence technology, using satellite remote sensing images as data sources and using deep learning change detection models to timely detect change patterns has become an important means of natural resource monitoring.

[0003] In related technologies, deep learning change detection models usually only support before and after phase remote sensing images as input, and extract change patterns based on the visual features of the input dual-phase images, which has the characteristics of what you see is what you get.

[0004] However, among the change patches detected by current deep learning models, a high proportion are those not required for business purposes, increasing the workload of subsequent manual verification steps and reducing the overall efficiency of natural resource monitoring. Furthermore, natural resource monitoring urgently requires systematic, formalized, and standardized geographic spatiotemporal knowledge support. On the one hand, a large amount of new knowledge hidden in structured or unstructured geographic spatiotemporal big data awaits discovery; on the other hand, natural resource monitoring requires interpretable, reusable, and reasonable knowledge related to geographic spatiotemporal distribution as an infrastructure to assist artificial intelligence in understanding the laws governing natural resources. Currently, the field of natural resource monitoring is plagued by the phenomenon of "massive data, information explosion, and difficulty in obtaining knowledge." The knowledge base of basic geographic data has not been fully mined, and conventional change monitoring methods are unable to meet the needs of efficient natural resource monitoring, which urgently needs to be addressed. Summary of the Invention

[0005] The present invention provides a method, device and equipment for purifying natural resource change spots based on knowledge graphs to solve the problem of high false alarm rate of change spots detected by deep learning models in the field of natural resource monitoring in related technologies, and to improve the operational efficiency and accuracy of natural resource monitoring work.

[0006] The first aspect of the present invention provides a method for purifying natural resource change spots based on a knowledge graph, comprising the following steps: obtaining original change spots; inputting the original change spots into a pre-constructed remote sensing spatiotemporal knowledge graph model to obtain model prediction spots, wherein the pre-constructed remote sensing spatiotemporal knowledge graph model is constructed based on the remote sensing spatiotemporal knowledge graph ontology model; based on preset inference rules, performing inference purification on the model prediction spots to obtain purified change spots.

[0007] Furthermore, in some embodiments, before inputting the original change map into a pre-built remote sensing spatiotemporal knowledge graph model to obtain the model prediction map, it also includes: determining the application scenario in the natural resources field and the remote sensing spatiotemporal knowledge graph ontology model, wherein the remote sensing spatiotemporal knowledge graph ontology model includes a remote sensing ontology layer, a time series layer, a time series image layer, a semantic association layer, an event evolution layer and a common sense law layer; obtaining remote sensing image data corresponding to the application scenario in the natural resources field, and pre-processing the remote sensing image data, and performing knowledge extraction based on the pre-processing result to obtain a knowledge extraction result; based on the remote sensing ontology layer, the time series layer, the time series image layer, the semantic association layer, the event evolution layer and the common sense law layer, obtaining an initial remote sensing spatiotemporal knowledge graph model according to the knowledge extraction result; based on the preset graph database parallel database acceleration strategy, the initial remote sensing spatiotemporal knowledge graph model is divided into blocks and containerized to obtain the pre-built remote sensing spatiotemporal knowledge graph model.

[0008] Furthermore, in some embodiments, before performing reasoning purification on the model prediction map based on the pre-constructed inference rules to obtain the purified change map, it also includes: extracting key information from the business demand text of natural resource monitoring to form a knowledge triple set, wherein the knowledge triple set includes multiple knowledge triples; constructing a logical expression according to the knowledge triple set, and obtaining the pre-constructed inference rule according to the logical expression, wherein the logical expression is used to describe the demand logic of the natural resource monitoring business.

[0009] Furthermore, in some embodiments, extracting key information from the natural resource monitoring business requirement text to form a knowledge triple set includes: performing natural language processing on the business requirement text to identify entities, relationships and attributes in the business requirement text to obtain recognition results; generating the multiple knowledge triples based on the recognition results, and forming the knowledge triple set based on the multiple knowledge triples.

[0010] Furthermore, in some embodiments, constructing a logical expression based on the knowledge triple set includes: based on the semantic relationship of each knowledge triple in the knowledge triple set, using a preset logical operator to combine the multiple knowledge triples into a logical expression.

[0011] Furthermore, in some embodiments, the model prediction spots are subjected to reasoning purification based on pre-constructed inference rules to obtain purified change spots, including: based on the pre-constructed inference rules, determining the spots to be purified that do not meet preset business requirements from the model prediction spots; filtering the spots to be purified to obtain the purified change spots.

[0012] The knowledge graph-based natural resource change spot purification method provided by an embodiment of the present invention constructs a remote sensing spatiotemporal knowledge graph for natural resources based on the remote sensing spatiotemporal knowledge graph ontology model. This method leverages a parallel database partitioning acceleration strategy for graph databases to achieve efficient management and application. It also utilizes multi-source prior knowledge and knowledge-based rule reasoning methods to purify raw natural resource change spots from the constructed graph. This method addresses the high false alarm rate of change spot detection using deep learning models in the natural resource monitoring field, improving the efficiency and accuracy of natural resource monitoring.

[0013] The second aspect of the present invention provides a natural resource change spot purification device based on a knowledge graph, wherein the device includes: an acquisition module for obtaining original change spots; a mathematical modeling module for inputting the original change spots into a pre-constructed remote sensing spatiotemporal knowledge graph model to obtain model prediction spots, wherein the pre-constructed remote sensing spatiotemporal knowledge graph model is constructed based on the remote sensing spatiotemporal knowledge graph ontology model; a control module for performing reasoning purification on the model prediction spots based on preset inference rules to obtain purified change spots.

[0014] Furthermore, in some embodiments, before inputting the original change map into a pre-built remote sensing spatiotemporal knowledge graph model to obtain the model prediction map, the mathematical modeling module is also used to: determine the application scenario in the natural resources field and the remote sensing spatiotemporal knowledge graph ontology model, wherein the remote sensing spatiotemporal knowledge graph ontology model includes a remote sensing ontology layer, a time series layer, a time series image layer, a semantic association layer, an event evolution layer and a common sense law layer; obtain remote sensing image data corresponding to the application scenario in the natural resources field, and pre-process the remote sensing image data, and extract knowledge based on the pre-processing result to obtain a knowledge extraction result; based on the remote sensing ontology layer, the time series layer, the time series image layer, the semantic association layer, the event evolution layer and the common sense law layer, obtain an initial remote sensing spatiotemporal knowledge graph model according to the knowledge extraction result; based on the preset graph database parallel database acceleration strategy, the initial remote sensing spatiotemporal knowledge graph model is divided into blocks and containerized to obtain the pre-built remote sensing spatiotemporal knowledge graph model.

[0015] Furthermore, in some embodiments, before performing reasoning purification on the model prediction map based on the pre-constructed inference rules to obtain the purified change map, the mathematical modeling module is also used to: extract key information from the business demand text of natural resource monitoring to form a knowledge triple set, wherein the knowledge triple set includes multiple knowledge triples; construct a logical expression based on the knowledge triple set, and obtain the pre-constructed inference rule based on the logical expression, wherein the logical expression is used to describe the demand logic of the natural resource monitoring business.

[0016] Furthermore, in some embodiments, the mathematical modeling module is also used to: perform natural language processing on the business requirement text, identify entities, relationships and attributes in the business requirement text, and obtain recognition results; generate the multiple knowledge triples based on the recognition results, and form the knowledge triple set based on the multiple knowledge triples.

[0017] Furthermore, in some embodiments, the mathematical modeling module is further used to: combine the multiple knowledge triples into a logical expression using a preset logical operator based on the semantic relationship of each knowledge triple in the knowledge triple set.

[0018] Furthermore, in some embodiments, the control module is specifically used to: determine the spots to be purified that do not meet preset business requirements from the model prediction spots based on the pre-built inference rules; filter the spots to be purified to obtain the purified changed spots.

[0019] The knowledge graph-based natural resource change spot purification device provided by an embodiment of the present invention constructs a remote sensing spatiotemporal knowledge graph for natural resources based on the remote sensing spatiotemporal knowledge graph ontology model. It leverages a parallel database partitioning acceleration strategy for graph databases to achieve efficient management and application. It utilizes multi-source prior knowledge and knowledge-based rule reasoning methods to purify raw natural resource change spots from the constructed graph. This device addresses the high false alarm rate of change spot detection using deep learning models in the natural resource monitoring field, improving the efficiency and accuracy of natural resource monitoring.

[0020] The third aspect of the present invention provides an electronic device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the natural resource change spot purification method based on the knowledge graph as described above.

[0021] The fourth aspect of the present invention provides a computer-readable storage medium having a computer program stored thereon, which is executed by a processor to implement the natural resource change spot purification method based on the knowledge graph as described above.

[0022] A fifth aspect of the present invention provides an embodiment including a computer program, which, when executed by a processor, implements the natural resource change spot purification method based on the knowledge graph as described above.

[0023] Therefore, the natural resource change spot purification method based on the knowledge graph in the embodiment of the present invention has the following beneficial effects:

[0024] (1) Based on the need to fully express the spatiotemporal characteristics of multi-source knowledge, the embodiment of the present invention designs a model layer of remote sensing spatiotemporal knowledge graph, which can fully integrate multi-source knowledge such as text, images, and vectors.

[0025] (2) The embodiment of the present invention utilizes multi-source prior knowledge in the remote sensing spatiotemporal knowledge graph and uses a knowledge-based rule reasoning method to purify the original natural resource change map with a high false alarm rate. This can significantly alleviate the problem of high false alarm rate in the change map detection of deep learning models in the current natural resource monitoring field, and is conducive to improving the operational efficiency and accuracy of natural resource monitoring work. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] The above and / or additional aspects and advantages of the present invention will become apparent and readily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which:

[0027] Figure 1 A flowchart of a method for purifying natural resource change spots based on a knowledge graph provided by an embodiment of the present invention;

[0028] Figure 2 A schematic diagram of an example model of a remote sensing spatiotemporal knowledge graph provided according to a specific embodiment of the present invention;

[0029] Figure 3 A schematic diagram of a remote sensing spatiotemporal knowledge graph ontology model according to a specific embodiment of the present invention;

[0030] Figure 4 This is a diagram showing the overall architecture of a natural resource change spot purification system driven by a remote sensing spatiotemporal knowledge graph according to a specific embodiment of the present invention;

[0031] Figure 5 A flow chart of a method for purifying natural resource change spots based on a knowledge graph according to a specific embodiment of the present invention;

[0032] Figure 6 A block diagram of a natural resource change spot purification device based on a knowledge graph according to an embodiment of the present invention;

[0033] Figure 7 A schematic structural diagram of an electronic device provided according to an embodiment of the present invention. DETAILED DESCRIPTION

[0034] The following describes embodiments of the present invention in detail, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present invention, and are not to be construed as limiting the present invention.

[0035] The following describes the natural resource change spot purification method, device and equipment based on the knowledge graph according to the embodiment of the present invention with reference to the accompanying drawings. In response to the problem of high false alarm rate of change spots detected by deep learning models in the field of natural resource monitoring in the related art mentioned in the above background technology, the present invention provides a natural resource change spot purification method based on the knowledge graph, which designs and constructs a remote sensing spatiotemporal knowledge graph in the field of natural resources based on the remote sensing spatiotemporal knowledge graph ontology model, realizes its efficient management and application with the help of the parallel database acceleration strategy of the graph database, and utilizes multi-source prior knowledge and knowledge-based rule reasoning methods to purify the original natural resource change spots in the constructed graph. It solves the problem of high false alarm rate of change spots detected by deep learning models in the field of natural resource monitoring, and improves the operating efficiency and accuracy of natural resource monitoring work.

[0036] Specifically, Figure 1 This is a flowchart of the natural resource change spot purification method based on knowledge graph provided in an embodiment of the present invention.

[0037] like Figure 1 As shown in FIG, the natural resource change map purification method based on the knowledge graph includes the following steps:

[0038] In step S101, the original change pattern is obtained.

[0039] Among them, the original change map is the basic unit reflecting the changes in natural resources and geographical environment, and is an important basic data for subsequent data analysis, processing and decision support.

[0040] Optionally, embodiments of the present invention may collect existing geographic information system data and, through deep learning methods, enable a neural network to automatically select appropriate original change patches from natural resource remote sensing image data.

[0041] In step S102, the original change pattern is input into a pre-built remote sensing spatiotemporal knowledge graph model to obtain a model prediction pattern, wherein the pre-built remote sensing spatiotemporal knowledge graph model is constructed based on the remote sensing spatiotemporal knowledge graph ontology model.

[0042] Among them, remote sensing spatiotemporal knowledge graph is a technical tool that combines remote sensing data with spatiotemporal information to represent and organize entities, events and their relationships in geographic space in the form of a knowledge graph. It has multimodal, multi-granular, spatiotemporal, event, and truth characteristics to achieve comprehensive analysis and comprehensive application of multimodal prior data.

[0043] Optionally, an embodiment of the present invention constructs a knowledge graph model by determining application scenarios in the field of natural resources and a remote sensing spatiotemporal knowledge graph ontology model, wherein the remote sensing spatiotemporal knowledge graph ontology model includes a remote sensing ontology layer, a time series layer, a time series image layer, a semantic association layer, an event evolution layer, and a common sense law layer; remote sensing image data corresponding to the application scenarios in the field of natural resources is obtained, and the remote sensing image data is preprocessed, and knowledge is extracted based on the preprocessing results to obtain a knowledge extraction result; based on the remote sensing ontology layer, the time series layer, the time series image layer, the semantic association layer, the event evolution layer, and the common sense law layer, an initial remote sensing spatiotemporal knowledge graph model is obtained according to the knowledge extraction result; based on the preset graph database parallel database acceleration strategy, the initial remote sensing spatiotemporal knowledge graph model is divided into blocks and containerized to obtain a pre-constructed remote sensing spatiotemporal knowledge graph model.

[0044] Specifically, an ontology is a formal description of entities, concepts, and relationships in the real world. It defines the terms, concepts, attributes, and relationships used in a knowledge graph, as well as the hierarchical structure and constraints between them. Ontology design includes ontology schema design, entity design, and relationship design. Ontology schema design is the foundation for constructing a priori knowledge graphs for remote sensing, helping computers understand the meaning and context of knowledge.

[0045] in, Figure 2 This is a schematic diagram of an example model of a remote sensing spatiotemporal knowledge graph provided according to a specific embodiment of the present invention. Figure 2 As shown in the figure, the remote sensing prior knowledge graph ontology model is divided into 6 layers from the inside to the outside, namely the remote sensing ontology layer, time series layer, time series image layer, semantic association layer, event evolution layer, and common sense law layer. It flexibly integrates multimodal, multi-granular, spatiotemporal, event and other characteristics to meet the needs of different application scenarios.

[0046] Furthermore, the multimodal nature of remote sensing prior knowledge graphs refers to the fact that the knowledge sources of the prior knowledge graph include text, images, vectors, and other modalities. Text data refers to sequences of computer characters and is the primary source of knowledge for conventional knowledge graphs. Image data refers to remote sensing images stored in the form of raster pixel matrices. It is the most important data source in the remote sensing field and intuitively reflects the surface state of the earth within a certain time and space range. Vector data refers to geometric shape data such as points, lines, and surfaces stored in the form of coordinate sequences, usually with attribute values such as surface semantics, and is an important source of prior knowledge for geographic scenes.

[0047] Multi-granularity refers to the fact that the remote sensing prior knowledge graph encompasses different scales, from macroscopic to microscopic. For example, it encompasses everything from the climate and environmental differentiation of Guangdong Province to detailed features such as a city's road network and buildings. Multi-granularity encompasses different levels of classification, attributes, and relationships, meeting the needs of a wide range of application scenarios.

[0048] Spatiotemporal characteristics refer to modeling the temporal and spatial variations of remote sensing data using a priori remote sensing knowledge graph. For example, for remote sensing images of the same region from different periods, information such as surface cover and changing trends at different stages can be modeled. This also takes into account the various spatial resolutions of remote sensing images and the spatial relationships between different land feature types.

[0049] Event features refer to the modeling of various event information appearing in remote sensing imagery using a priori remote sensing knowledge graph. Examples include natural disasters, human activities, and other events. To enable event-based analysis and prediction, the constructed ontology includes information such as event type, spatiotemporal scope, and influencing factors.

[0050] The principles of events are a further high-level abstraction of the laws of event evolution. That is, the remote sensing prior knowledge graph models the logical relationships between events, such as sequential relationships, causal relationships, conditional relationships, and hierarchical relationships. By analyzing the internal laws of event evolution, it can better drive the purification of change spots and other natural resource business analysis.

[0051] Furthermore, an entity is an independent thing with distinct characteristics that can be distinguished from other things. In the remote sensing prior knowledge graph, the information used to describe these things is the entity, which can be abstract or concrete. Entities are the basic building blocks of the knowledge graph. In the graph database that serves as the carrier of the knowledge graph, entities are represented as vertices.

[0052] Furthermore, a relationship refers to a directed, semantic representation between entities. In a knowledge graph, the associations and connections between entities are represented as relationships, and in a graph database, which serves as the carrier of the knowledge graph, relationships are represented as edges.

[0053] Furthermore, a knowledge graph is a network composed of entities and their relationships. Each entity and its associated attribute key-value pairs describe a piece of knowledge, while each relationship and its attributes represent the relationships between knowledge points. The basic unit of a knowledge graph is a triple of the form <head entity, relationship, tail entity>, which describes directed relationships between entities.

[0054] Among them, in some embodiments, the remote sensing prior knowledge graph instance model designs 7 types of entities, including text, image, vector, time, event, individual or organization, and concept, and establishes multiple semantic relationships within the same entity and between multiple types of entities, realizing comprehensive analysis and comprehensive application of multimodal prior knowledge.

[0055] Figure 3 This is a schematic diagram of the remote sensing spatiotemporal knowledge graph ontology model provided according to a specific embodiment of the present invention. Figure 3 As shown in the figure, vector entities include change survey vectors, geographic conditions monitoring vectors, normalization results patch vectors, model prediction patch vectors, image interpretation result vectors, rule applicable geographic scope and its index, and feature entities. Image entities include change survey images, geographic conditions monitoring images, monthly images, and other image entities. Text entities include root and subclass entities of the natural resource classification, geographic conditions monitoring classification, patch new code classification, patch old code classification, and image interpretation classification system.

[0056] Furthermore, in some embodiments, a pre-defined graph database parallel sharding acceleration strategy divides the graph database into multiple independent blocks and implements Docker containerization, improving read and write efficiency when processing large amounts of data and optimizing the overall performance of the knowledge graph. Each database block runs in an independent graph database instance, enhancing parallel processing capabilities.

[0057] Specifically, the database partitioning strategy is based on the division of administrative boundaries, that is, the data is separated according to the geographical boundaries of the administrative divisions. Taking a single prefecture-level city as an example, the city is treated as an independent data block, and its corresponding data is stored in an independent graph database. The above division method can better organize and manage data at the scale of prefecture-level cities and improve the cohesion of data in geographic space. Through this strategy, the data of each prefecture-level city will run in a relatively independent graph database instance, thereby reducing cross-prefecture-level city data access during queries and operations, and can improve the system's read and write efficiency and the efficiency of data analysis and queries within a specific prefecture-level city. The above-mentioned partitioning strategy based on administrative divisions conforms to the geographical characteristics of the data, is conducive to the maintenance and expansion of the system, and can improve the management and query efficiency of the knowledge graph.

[0058] Thus, the construction of the remote sensing spatiotemporal knowledge graph model is completed, and the model prediction pattern is obtained by inputting the original change pattern into the constructed remote sensing spatiotemporal knowledge graph model.

[0059] In step S103, based on preset inference rules, the model prediction pattern is subjected to inference purification to obtain purified change patterns.

[0060] Furthermore, in some embodiments, before performing reasoning purification on the model prediction map based on pre-constructed inference rules to obtain the purified change map, it also includes: extracting key information from the business demand text of natural resource monitoring to form a knowledge triple set, wherein the knowledge triple set includes multiple knowledge triples; constructing a logical expression according to the knowledge triple set, and obtaining the pre-constructed inference rule according to the logical expression, wherein the logical expression is used to describe the demand logic of the natural resource monitoring business.

[0061] Furthermore, in some embodiments, key information is extracted from the natural resource monitoring business requirement text to form a knowledge triple set, including: performing natural language processing on the business requirement text, identifying entities, relationships and attributes in the business requirement text, and obtaining recognition results; based on the recognition results, generating multiple knowledge triples, and forming a knowledge triple set based on the multiple knowledge triples.

[0062] The entities in the recognition results are taken as the subjects of the knowledge triples, the relationships in the recognition results are taken as the relationships of the knowledge triples, and the attributes in the recognition results are taken as the objects of the knowledge triples to form knowledge triples.

[0063] Furthermore, in some embodiments, constructing a logical expression based on a set of knowledge triples includes: combining multiple knowledge triples into a logical expression using a preset logical operator based on the semantic relationship of each knowledge triple in the set of knowledge triples.

[0064] For example, for the problem "neither newly built greenhouses nor demolished greenhouses on the surface of ponds are extracted", we construct inference rules, such as:

[0065]

[0066] For the question "whether the greenhouse on the cultivated garden grass is removed or the cultivated garden grass is not extracted", an inference rule is constructed, for example:

[0067]

[0068] For "buildings less than 200 square meters outside the scope of cultivated land can be excluded", an inference rule is constructed, for example:

[0069]

[0070] Furthermore, in some embodiments, based on pre-constructed inference rules, the model prediction spots are subjected to inference purification to obtain purified change spots, including: based on pre-constructed inference rules, determining the spots to be purified that do not meet preset business requirements from the model prediction spots; filtering the spots to be purified to obtain purified change spots.

[0071] In this way, the construction of the remote sensing spatiotemporal knowledge graph model was completed, and the knowledge extraction speed was improved through the parallel acceleration strategy of the graph database, and the purification of the map spots was achieved based on inference rules.

[0072] In order to enable relevant technical personnel in this field to better understand the natural resource change map purification method based on knowledge graph in an embodiment of the present invention, it will be explained below with reference to specific embodiments.

[0073] Figure 4 This is a diagram showing the overall architecture of a natural resource change map purification system driven by a remote sensing spatiotemporal knowledge graph according to a specific embodiment of the present invention. Figure 5 This is a flow chart of a method for purifying natural resource change spots based on a knowledge graph according to a specific embodiment of the present invention.

[0074] like Figure 4 As shown in Figure 1, the system consists of a knowledge service layer, a resource management layer, and an infrastructure layer. The knowledge service layer includes the construction, visualization, and application of remote sensing spatiotemporal knowledge graphs; the resource management layer includes the data source for knowledge extraction, the data structure of the spatiotemporal knowledge graph, and user information management; the infrastructure layer includes development environments such as Vue.js and NeoVis.js, and storage environments such as Docker and MySQL. This allows the construction of a natural resource change map purification system driven by remote sensing spatiotemporal knowledge graphs.

[0075] Furthermore, if Figure 5 As shown in the figure, the natural resource change map purification method based on knowledge graph first instantiates the remote sensing spatiotemporal knowledge graph ontology model as the remote sensing spatiotemporal knowledge graph. After parallel partitioning of the remote sensing spatiotemporal knowledge graph, the original change map obtained is purified according to the inference rules, and finally the change map purification result is obtained.

[0076] The knowledge graph-based natural resource change spot purification method provided by an embodiment of the present invention constructs a remote sensing spatiotemporal knowledge graph for natural resources based on the remote sensing spatiotemporal knowledge graph ontology model. This method leverages a parallel database partitioning acceleration strategy for graph databases to achieve efficient management and application. It also utilizes multi-source prior knowledge and knowledge-based rule reasoning methods to purify raw natural resource change spots from the constructed graph. This method addresses the high false alarm rate of change spot detection using deep learning models in the natural resource monitoring field, improving the efficiency and accuracy of natural resource monitoring.

[0077] Next, a natural resource change spot purification device based on a knowledge graph proposed according to an embodiment of the present invention will be described with reference to the accompanying drawings.

[0078] Figure 6A block diagram of a natural resource change spot purification device based on a knowledge graph provided according to an embodiment of the present invention.

[0079] like Figure 6 As shown, the natural resource change spot purification device 10 based on knowledge graph includes: an acquisition module 100, a mathematical modeling module 200 and a control module 300.

[0080] Among them, the acquisition module 100 is used to obtain the original change pattern; the mathematical modeling module 200 is used to input the original change pattern into a pre-constructed remote sensing spatiotemporal knowledge graph model to obtain the model prediction pattern, wherein the pre-constructed remote sensing spatiotemporal knowledge graph model is constructed based on the remote sensing spatiotemporal knowledge graph ontology model; the control module 300 is used to perform reasoning purification on the model prediction pattern based on preset inference rules to obtain the purified change pattern.

[0081] Furthermore, in some embodiments, before inputting the original change map into a pre-built remote sensing spatiotemporal knowledge graph model to obtain the model prediction map, the mathematical modeling module 200 is also used to: determine the application scenario of the natural resources field and the remote sensing spatiotemporal knowledge graph ontology model, wherein the remote sensing spatiotemporal knowledge graph ontology model includes a remote sensing ontology layer, a time series layer, a time series image layer, a semantic association layer, an event evolution layer and a common sense law layer; obtain remote sensing image data corresponding to the application scenario of the natural resources field, and pre-process the remote sensing image data, and extract knowledge based on the pre-processing result to obtain a knowledge extraction result; based on the remote sensing ontology layer, the time series layer, the time series image layer, the semantic association layer, the event evolution layer and the common sense law layer, obtain an initial remote sensing spatiotemporal knowledge graph model according to the knowledge extraction result; based on the preset graph database parallel database acceleration strategy, the initial remote sensing spatiotemporal knowledge graph model is divided into blocks and containerized to obtain the pre-built remote sensing spatiotemporal knowledge graph model.

[0082] Furthermore, in some embodiments, before performing reasoning purification on the model prediction map based on the pre-constructed inference rules to obtain the purified change map, the mathematical modeling module 200 is also used to: extract key information from the business demand text of natural resource monitoring to form a knowledge triple set, wherein the knowledge triple set includes multiple knowledge triples; construct a logical expression based on the knowledge triple set, and obtain the pre-constructed inference rule based on the logical expression, wherein the logical expression is used to describe the demand logic of the natural resource monitoring business.

[0083] Furthermore, in some embodiments, the mathematical modeling module 200 is also used to: perform natural language processing on the business requirement text, identify entities, relationships and attributes in the business requirement text, and obtain recognition results; generate the multiple knowledge triples based on the recognition results, and form the knowledge triple set based on the multiple knowledge triples.

[0084] Furthermore, in some embodiments, the mathematical modeling module 200 is further configured to: combine the plurality of knowledge triples into a logical expression using a preset logical operator based on the semantic relationship of each knowledge triple in the knowledge triple set.

[0085] Furthermore, in some embodiments, the control module 300 is specifically used to: determine the spots to be purified that do not meet preset business requirements from the model prediction spots based on the pre-built inference rules; filter the spots to be purified to obtain the purified changed spots.

[0086] It should be noted that the above explanation of the embodiment of the natural resource change spot purification method based on the knowledge graph is also applicable to the natural resource change spot purification device based on the knowledge graph of this embodiment, and will not be repeated here.

[0087] The knowledge graph-based natural resource change spot purification device provided by an embodiment of the present invention constructs a remote sensing spatiotemporal knowledge graph for natural resources based on the remote sensing spatiotemporal knowledge graph ontology model. It leverages a parallel database partitioning acceleration strategy for graph databases to achieve efficient management and application. It utilizes multi-source prior knowledge and knowledge-based rule reasoning methods to purify raw natural resource change spots from the constructed graph. This device addresses the high false alarm rate of change spot detection using deep learning models in the natural resource monitoring field, improving the efficiency and accuracy of natural resource monitoring.

[0088] Figure 7 This is a schematic diagram of the structure of an electronic device provided according to an embodiment of the present invention. The electronic device may include:

[0089] Memory 701 , processor 702 , and computer programs stored in the memory 701 and executable on the processor 702 .

[0090] When the processor 702 executes the program, the natural resource change spot purification method based on the knowledge graph provided in the above embodiment is implemented.

[0091] Furthermore, the electronic device further includes:

[0092] The communication interface 703 is used for communication between the memory 701 and the processor 702 .

[0093] The memory 701 is used to store computer programs that can be run on the processor 702 .

[0094] The memory 701 may include a high-speed RAM memory, and may also include a non-volatile memory (non-volatile memory), such as at least one disk memory.

[0095] If the memory 701, processor 702, and communication interface 703 are implemented independently, the communication interface 703, memory 701, and processor 702 can be interconnected via a bus and communicate with each other. The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus. The bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 7 Only one thick line is used in the diagram, but this does not mean that there is only one bus or one type of bus.

[0096] Optionally, in a specific implementation, if the memory 701, the processor 702 and the communication interface 703 are integrated on a chip, the memory 701, the processor 702 and the communication interface 703 can communicate with each other through an internal interface.

[0097] The processor 702 may be a central processing unit (CPU), an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of the present invention.

[0098] In addition, an embodiment of the present invention also provides a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, it implements the natural resource change spot purification method based on the knowledge graph as described above.

[0099] In addition, an embodiment of the present invention also provides a computer program product, including a computer program, which is executed to implement the natural resource change spot purification method based on the knowledge graph as described above.

[0100] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic expressions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or N embodiments or examples in a suitable manner. In addition, those skilled in the art can combine and combine different embodiments or examples described in this specification and the features of different embodiments or examples without contradiction.

[0101] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be understood to indicate or imply relative importance or implicitly specify the number of technical features indicated. Thus, a feature specified as "first" or "second" may explicitly or implicitly include at least one such feature. In the description of the present invention, "N" means at least two, such as two, three, etc., unless otherwise specifically defined.

[0102] Any process or method description in a flowchart or otherwise described herein may be understood to represent a module, segment or portion of code comprising one or more executable instructions for implementing a custom logical function or step of a process, and the scope of the preferred embodiments of the invention includes alternative implementations in which functions may be performed out of the order shown or discussed, including performing functions in a substantially simultaneous manner or in reverse order depending on the functions involved, as should be understood by those skilled in the art to which embodiments of the invention pertain.

[0103] It should be understood that various parts of the present invention can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiment, the N steps or methods can be implemented using software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented using hardware, as in another embodiment, any one of the following technologies known in the art or a combination thereof can be used: a discrete logic circuit having logic gate circuits for implementing logic functions on data signals, an application-specific integrated circuit having suitable combinational logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0104] Those skilled in the art will understand that all or part of the steps in the method of the above embodiment can be completed by instructing related hardware through a program, and the program can be stored in a computer-readable storage medium. When the program is executed, it includes one or a combination of the steps of the method embodiment.

Claims

1. A natural resource change spot purification method based on knowledge graph, characterized in that: The following steps are involved: Get the original change pattern; Inputting the original change pattern into a pre-built remote sensing spatiotemporal knowledge graph model to obtain a model prediction pattern, wherein the pre-built remote sensing spatiotemporal knowledge graph model is constructed based on a remote sensing spatiotemporal knowledge graph ontology model; Based on pre-constructed inference rules, the model prediction pattern is subjected to inference purification to obtain a purified change pattern.

2. The method according to claim 1, characterized in that Before inputting the original change pattern into the pre-built remote sensing spatiotemporal knowledge graph model to obtain the model predicted pattern, the method further includes: Determine application scenarios in the field of natural resources and the remote sensing spatiotemporal knowledge graph ontology model, wherein the remote sensing spatiotemporal knowledge graph ontology model includes a remote sensing ontology layer, a time series layer, a time series image layer, a semantic association layer, an event evolution layer, and a common sense law layer; Acquiring remote sensing image data corresponding to the application scenario in the natural resources field, preprocessing the remote sensing image data, and performing knowledge extraction based on the preprocessing results to obtain a knowledge extraction result; Based on the remote sensing ontology layer, the time series layer, the time series image layer, the semantic association layer, the event evolution layer and the common sense law layer, an initial remote sensing spatiotemporal knowledge graph model is obtained according to the knowledge extraction result; Based on the preset graph database parallel partitioning acceleration strategy, the initial remote sensing spatiotemporal knowledge graph model is divided into blocks and containerized to obtain the pre-built remote sensing spatiotemporal knowledge graph model.

3. The method according to claim 1, characterized in that Before performing reasoning purification on the model prediction pattern based on the pre-built inference rules to obtain the purified change pattern, the method further includes: Extracting key information from a business requirement text of natural resource monitoring to form a knowledge triple set, wherein the knowledge triple set includes a plurality of knowledge triples; A logical expression is constructed according to the knowledge triple set, and the pre-constructed inference rule is obtained according to the logical expression, wherein the logical expression is used to describe the demand logic of the natural resource monitoring business.

4. The method according to claim 3, characterized in that The key information is extracted from the natural resource monitoring business requirement text to form a knowledge triple set, including: Performing natural language processing on the business requirement text to identify entities, relationships, and attributes in the business requirement text to obtain an identification result; Based on the recognition result, the plurality of knowledge triples are generated, and the knowledge triple set is formed according to the plurality of knowledge triples.

5. The method according to claim 3, characterized in that The constructing of a logical expression according to the knowledge triple set includes: Based on the semantic relationship of each knowledge triple in the knowledge triple set, the plurality of knowledge triples are combined into a logical expression using a preset logical operator.

6. The method according to claim 1, characterized in that The method of performing reasoning purification on the model prediction pattern based on the pre-built inference rules to obtain the purified change pattern includes: Based on the pre-built inference rules, determining the spots to be cleaned that do not meet the preset business requirements from the model prediction spots; The to-be-purified image spots are filtered to obtain the purified changed image spots.

7. A natural resource change spot purification device based on knowledge graph, characterized in that: The device comprises: An acquisition module is used to obtain the original change pattern; A mathematical modeling module is used to input the original change pattern into a pre-built remote sensing spatiotemporal knowledge graph model to obtain a model prediction pattern, wherein the pre-built remote sensing spatiotemporal knowledge graph model is constructed based on a remote sensing spatiotemporal knowledge graph ontology model; The control module is used to perform reasoning purification on the model prediction pattern based on pre-built inference rules to obtain purified change pattern.

8. An electronic device, characterized in that: include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the method for purifying natural resource change spots based on a knowledge graph as described in any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: The program is executed by a processor to implement the natural resource change spot purification method based on knowledge graph as described in any one of claims 1 to 6.

10. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the method for purifying natural resource change spots based on a knowledge graph as described in any one of claims 1 to 6 is implemented.

Images