A time and space-based water environment knowledge graph construction method and device

Abstract

The application relates to a time and space-based water environment knowledge graph construction method and device, and relates to the technical field of water environment data processing.The method comprises the following steps: processing and integrating multi-source data related to a water environment, taking the key attributes of each data as a benchmark, analyzing and connecting entities, generating entities, and constructing an entity data model; the spatial relationship and time sequence relationship are determined through spatial analysis and time sequence analysis between the entities; finally, the spatial correlation of each entity is optimized according to the spatial relationship, the time correlation of each entity is optimized according to the time sequence relationship, and a multi-dimensional water environment knowledge graph is established by further combining the time-space correlation optimization. It can be seen that the embodiment establishes a knowledge graph that comprehensively considers the time and space characteristics according to the complexity of water environment information, and can be used for realizing comprehensive monitoring and management of the water environment.

Description

Technical Field

[0001] This application relates to the field of water environment data processing technology, and in particular to a method and apparatus for constructing a water environment knowledge graph based on time and space. Background Technology

[0002] Existing technologies for constructing water environment knowledge graphs mostly focus on the representation and querying of static data. For example, they utilize graph theory to build hydrological data models or ontology-based water quality monitoring data models. While these methods can handle certain data relationships, they often neglect the spatiotemporal characteristics of water environment data, such as seasonal variations in river water levels and the spatial distribution of pollutant concentrations. They lack the ability to integrate and analyze data across spatiotemporal dimensions, thus limiting the practical application of existing water environment knowledge graphs. Therefore, current methods for constructing water environment knowledge graphs suffer from insufficient data integration capabilities, lack of dynamism, and difficulty in capturing spatiotemporal features, leading to data silos when dealing with massive amounts of spatiotemporal data. Summary of the Invention

[0003] This application provides a method and apparatus for constructing a water environment knowledge graph based on time and space. It utilizes acquired multi-source data related to the water environment to perform time series and spatial relationship analysis, and further optimizes the correlation between time and space to establish a multi-dimensional water environment knowledge graph. This enables the water environment knowledge graph to have multi-dimensional data integration and analysis capabilities, solves the data silo problem in the construction of water environment knowledge graphs in the prior art, and improves the ability to process massive spatiotemporal data.

[0004] Firstly, this application provides a method for constructing a water environment knowledge graph based on time and space, including:

[0005] Preprocessing is performed on the acquired water quality monitoring data, meteorological data, and geographical data to obtain integrated data, in which each data point carries key attributes.

[0006] Based on the aforementioned key attributes, the integrated data is processed through entity analysis and entity concatenation to construct an entity data model. Each entity set in the entity data model consists of entity type, entity name, and core attributes.

[0007] Based on the entity set, spatial and temporal analysis are performed between entities to obtain spatial and temporal relationships.

[0008] A knowledge graph architecture is constructed, spatial association optimization is performed on each entity in the entity set according to the spatial relationship, and temporal association optimization is performed on each entity according to the temporal sequence relationship, and the knowledge graph architecture is updated to obtain a multi-dimensional water environment knowledge graph.

[0009] The water environment knowledge graph is updated by optimizing spatiotemporal correlation.

[0010] Optionally, preprocessing is performed on the acquired water quality monitoring data, meteorological data, and geographic data to obtain integrated data, including:

[0011] Water quality monitoring data, meteorological data, and geographical data are obtained from the data source set as raw data. The data source set includes the data source corresponding to each of the raw data.

[0012] The original data is standardized and deduplicated, and the data attributes of the data source and each original data are used as the key attributes of the original data to establish integrated data.

[0013] Optionally, based on the aforementioned key attributes, an entity data model is constructed by processing the integrated data through entity analysis and entity concatenation, including:

[0014] Entity analysis is performed based on the integrated data to establish entities, and the core attributes of the entities are determined by combining the key attributes with the data sources of each entity.

[0015] Based on the relationships between the entities and the core attributes, an entity triple set is established, and the entities in the entity triple set are linked together to obtain an entity set, thus establishing an entity data model.

[0016] Optionally, based on the relationships between the entities and the core attributes, an entity triple set is established, and multiple entity concatenation is performed on each entity in the entity triple set to obtain an entity set, including:

[0017] Based on the core attribute analysis, the establishment basis of the triplet is determined, and based on the entity analysis relationship type, the entity relationship type is obtained;

[0018] Based on the establishment basis and the entity relationship type, a set of triples is established for each entity;

[0019] Analyze the core entities from the set of triples;

[0020] Based on the core entity, the entities in the triplet set are linked together to form an entity relationship chain, thus obtaining an entity set.

[0021] Optionally, spatial and temporal analysis are performed on the entity set to obtain spatial and temporal relationships, including:

[0022] Coordinate information is extracted from each entity in the entity set, and the coordinate information is determined based on the latitude and longitude coordinates in the core attributes of each entity;

[0023] Based on the coordinate information, the Haversian algorithm is used to analyze the proximity relationships between entities to obtain spatial relationships;

[0024] Determine the time data for each entity, wherein the time data includes timestamps and time sequence information, and the time sequence information is determined based on the collection frequency of the data corresponding to each entity.

[0025] Based on the timestamp and the time sequence information, a dynamic time warping algorithm is used for consistency alignment to obtain the time sequence relationship between entities.

[0026] Optionally, based on the coordinate information, the Haversian algorithm is used to analyze the proximity relationships between entities to obtain spatial relationships, including:

[0027] Based on the coordinate information, entity coordinates are established for the two entities respectively, resulting in the first entity coordinates and the second entity coordinates represented in decimal.

[0028] The first entity coordinates are transformed to obtain the first entity radian, and the second entity coordinates are transformed to obtain the second entity radian;

[0029] Based on the first entity radian and the second entity radian, the Haversian algorithm is used to calculate the spatial distance;

[0030] Based on the spatial distance, a comparison is made with a preset spatial threshold to determine the proximity relationship between entities and obtain the spatial relationship.

[0031] Optionally, a knowledge graph architecture is constructed, and spatial association optimization is performed on each entity in the entity set based on the spatial relationships, and temporal association optimization is performed on each entity based on the temporal series relationships, including:

[0032] A knowledge graph architecture is used to construct a water environment knowledge graph, and each entity in the entity set is added to the water environment knowledge graph.

[0033] In the water environment knowledge graph, the R-tree algorithm is used to optimize the spatial index of entities related to the water environment, and spatial association is performed in combination with spatial relationships to update the water environment knowledge graph;

[0034] Based on the aforementioned time series relationship, a B+ tree index is used for timestamp conversion and sorting to optimize the time index.

[0035] In the water environment knowledge graph, knowledge graph mapping is optimized based on the time series relationship and the time index, time axis nodes are established for each entity, the water environment knowledge graph is updated, and the water environment knowledge graph is visualized.

[0036] Optionally, the water environment knowledge graph can be optimized and updated through spatiotemporal correlation, including:

[0037] Spatial two-dimensional data and temporal data corresponding to entities are selected from the water environment knowledge graph. The spatial two-dimensional data is represented in the form of coordinate data, and the temporal data is represented in the form of one-dimensional time.

[0038] Using the aforementioned two-dimensional spatial data as input, the spatial distance is analyzed and spatial weights are determined based on the Haversian algorithm and a negative exponential decay model.

[0039] Using the aforementioned time data as input, the time difference is analyzed, and a linear decay model is used to analyze the time distance and determine the time weight.

[0040] Based on the spatial weight and the temporal weight, the spatiotemporal correlation strength between entities is quantified;

[0041] Based on the fusion of the spatial two-dimensional data and the temporal data, spatiotemporal three-dimensional data is obtained, and spatiotemporal correlation optimization is performed using a spatial data indexing algorithm to establish a spatiotemporal index storage.

[0042] The water environment knowledge graph is updated based on the spatiotemporal correlation strength and the spatiotemporal index storage.

[0043] Optionally, after optimizing and updating the water environment knowledge graph through spatiotemporal correlation, the method further includes:

[0044] Based on the time series of each entity in the water environment knowledge graph, key events are analyzed and the corresponding water environment regions are determined. The key events include at least extreme climate events and sudden changes in water quality.

[0045] Develop a predictive model;

[0046] For the key events, using the time series as input, the prediction model predicts the future state of the water environment in the water environment area, and outputs the key event alert data through the prediction model. The alert data includes intelligent early warning information and decision information.

[0047] The water environment knowledge graph is updated and dynamically evolved in real time using the updated data from the integrated data.

[0048] Secondly, this application provides a time- and space-based water environment knowledge graph construction device, comprising:

[0049] The multi-source data acquisition module is used to preprocess the acquired water quality monitoring data, meteorological data and geographical data to obtain integrated data, in which each data carries key attributes;

[0050] The entity construction module is used to construct an entity data model based on the key attributes, through entity analysis and entity concatenation processing of the integrated data. Each entity set in the entity data model consists of entity type, entity name and core attributes.

[0051] The spatiotemporal analysis module is used to perform spatial and temporal series analysis between entities based on the entity set, and to obtain spatial and temporal series relationships.

[0052] The knowledge graph construction module is used to construct a knowledge graph architecture, optimize the spatial association of each entity in the entity set according to the spatial relationship, optimize the temporal association of each entity according to the temporal sequence relationship, update the knowledge graph architecture, and obtain a multi-dimensional water environment knowledge graph.

[0053] The spatiotemporal correlation module is used to optimize and update the water environment knowledge graph through spatiotemporal correlation.

[0054] In summary, this embodiment first collects water quality monitoring data, meteorological data, and geographical data from data sources for preprocessing, integrates the data, and constructs key attributes for each data point based on the data sources. Then, based on these key attributes, entities are established through entity analysis; further, the core attributes of each entity are clarified based on these key attributes. Entity relationships are analyzed through entities and core attributes, and entity sets are established through entity chaining, thereby constructing an entity data model. Subsequently, spatial and temporal series analyses are performed on each entity to clarify the spatial and temporal relationships between entities. A water environment knowledge graph is established using each entity, with spatial association optimization based on spatial relationships and temporal association optimization based on temporal series relationships. Furthermore, the water environment knowledge graph is optimized through spatiotemporal association to establish a multi-dimensional water environment knowledge graph. It is evident that this embodiment establishes a knowledge graph that comprehensively considers temporal and spatial characteristics to address the complexity of water environment information. This can be used to achieve comprehensive monitoring and management of the water environment, solving the data silo problem in existing water environment knowledge graph construction technologies. This embodiment can effectively improve the processing capability of massive spatiotemporal data during the water environment knowledge graph construction process. Attached Figure Description

[0055] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0056] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0057] Figure 1 A flowchart illustrating a time- and space-based method for constructing a water environment knowledge graph, as provided in an embodiment of this application;

[0058] Figure 2 This is a flowchart illustrating the steps of a time- and space-based water environment knowledge graph construction method provided in an optional embodiment of this application.

[0059] Figure 3 This is a structural block diagram of a time- and space-based water environment knowledge graph construction device provided in an embodiment of this application. Detailed Implementation

[0060] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0061] To facilitate understanding of the embodiments of this application, further explanations and descriptions will be provided below in conjunction with the accompanying drawings and specific embodiments. These embodiments do not constitute a limitation on the embodiments of this application.

[0062] Figure 1 The flowchart illustrates a method for constructing a water environment knowledge graph based on time and space, as provided in this application embodiment. The method may specifically include the following steps:

[0063] Step 110: Based on the acquired water quality monitoring data, meteorological data, and geographical data, preprocessing is performed to obtain integrated data.

[0064] Each piece of data in the integrated data carries key attributes.

[0065] In this embodiment, raw data related to the water environment can be obtained from corresponding data sources to form multi-source data. Among the multi-source data: water quality monitoring data is mainly obtained from data sources such as automatic monitoring stations, and mainly includes multiple relevant indicators and monitoring metadata; meteorological data is mainly obtained from automatic meteorological monitoring stations in the monitoring area, including relevant meteorological data; geographic data refers to Geographic Information System (GIS) data, which has a wide range of sources, including GIS vector databases, satellite remote sensing images, and official administrative division websites, and mainly includes various spatial data.

[0066] In actual implementation, all acquired data can carry key attributes. Key attributes can be determined based on the data source. For example, in water quality monitoring data, key attributes can include at least the monitoring station ID, and the same applies to other data. Key attributes can be used as the core attributes of each entity subsequently established, and can also be used to achieve spatial and temporal associations, etc. This embodiment will not elaborate on these aspects.

[0067] In existing technologies, water environment knowledge graphs are typically composed of single or simple data, mainly because: ① Existing technologies have relatively singular monitoring needs for water environment and usually do not consider complex water environment issues; ② Existing technologies have limited data integration capabilities. Traditional knowledge graphs primarily process structured data, while unstructured or semi-structured data, such as raw environmental monitoring data (including water quality monitoring data and meteorological data) and satellite remote sensing images (which are part of geographic data), are difficult to integrate from multiple sources.

[0068] To address the technical problems of the prior art, in this embodiment, multi-source data is integrated, and a series of preprocessing operations are performed on the acquired data to integrate the preprocessed data into integrated data.

[0069] Step 120: Based on the key attributes, construct an entity data model by performing entity analysis and entity concatenation processing on the integrated data.

[0070] In the entity data model, each entity set consists of entity type, entity name, and core attributes.

[0071] In this embodiment, corresponding entities can be created for each type of data in the integrated data. At least one type of entity can be created for each type of data, and a corresponding entity name is set to distinguish the entities. Then, for each entity, core attributes are created based on the key attributes corresponding to the original data. The basis for creating the core attributes is the data source corresponding to each original data, thereby obtaining the entity set.

[0072] Subsequently, based on the core attributes, the relationships between entities are analyzed and linked together. Specifically, entities from three types of data with related relationships are usually linked together. For example, entities with related relationships can be selected from water quality monitoring data, meteorological data, and geographic data, and linked together to form a complete relationship chain.

[0073] Finally, an entity data model is constructed based on the established entity set and the interconnection of multiple entities.

[0074] Step 130: Perform spatial analysis and time series analysis between entities based on the entity set to obtain spatial relationships and time series relationships.

[0075] Step 140: Construct a knowledge graph architecture, optimize the spatial association of each entity in the entity set according to the spatial relationship, optimize the temporal association of each entity according to the temporal sequence relationship, update the knowledge graph architecture, and obtain a multi-dimensional water environment knowledge graph.

[0076] Step 150: Optimize and update the water environment knowledge graph through spatiotemporal correlation.

[0077] Steps 130-150 are described uniformly as follows:

[0078] In constructing a water environment knowledge graph based on space and time, a water environment knowledge graph can be built using entities within a pre-defined knowledge graph architecture. Subsequently, spatial and temporal relationships are established for each entity in the entity set, and further spatiotemporal optimization is performed by combining spatiotemporal elements to update the water environment knowledge graph. This allows the knowledge graph to possess characteristics of different dimensions, including time and space. The following explains how to build a knowledge graph that comprehensively considers both temporal and spatial characteristics:

[0079] When establishing spatial relationships between entities, the spatial relationships between entities are first analyzed. Specifically, this process involves traversing the entity set, selecting two entities of different types, and analyzing their spatial relationships, such as proximity, based on core attributes. This determines whether the two entities are spatially adjacent. If a proximity relationship exists between the two entities, their spatial proximity is confirmed, and this spatial relationship indicates that an association can be established. Furthermore, to avoid traversing all entities for insertion operations in practical knowledge graph applications, this embodiment further introduces spatial index optimization to establish efficient query operations for all entities within a region, thereby optimizing spatial relationships.

[0080] When establishing time relationships for various entities, the first step is to analyze the time series relationships between them. Specifically, this involves traversing each entity, extracting its time data from its original data, and analyzing whether the timestamps of each data point are consistent and whether there is time series consistency. For entities that satisfy both timestamp consistency and time series consistency, valid time relationships are established to obtain the time series relationships. Then, based on these time series relationships, time relationship optimization is performed through time indexing to achieve efficient time-based query operations.

[0081] Based on the established spatial and temporal associations, a spatiotemporal joint index is built using temporal and spatial relationships, and the water environment knowledge graph is further updated through spatiotemporal association optimization.

[0082] In practical applications, the constructed water environment knowledge graph can enable comprehensive monitoring and management of the water environment. For example, prediction models can be used to predict key events and propose early warnings and decisions based on the water environment knowledge graph.

[0083] As can be seen, this embodiment processes and integrates multi-source data related to the water environment. Based on the key attributes of each data point, it analyzes and connects entities to generate entities and constructs entity data models. Through spatial and temporal series analysis between entities, spatial and temporal relationships are determined. Finally, spatial association optimization is performed on each entity based on spatial relationships, and temporal association optimization is performed on each entity based on temporal series relationships. Furthermore, by combining spatiotemporal association optimization, a multi-dimensional water environment knowledge graph is established. Therefore, this embodiment addresses the complexity of water environment information by establishing a knowledge graph that comprehensively considers both temporal and spatial characteristics, which can be used to achieve comprehensive monitoring and management of the water environment.

[0084] Reference Figure 2 This illustration shows a flowchart of a time- and space-based water environment knowledge graph construction method according to an optional embodiment of this application. The method specifically includes the following steps:

[0085] Step 210: Based on the acquired water quality monitoring data, meteorological data, and geographical data, preprocessing is performed to obtain integrated data.

[0086] Each piece of data in the integrated data carries key attributes.

[0087] Among related technologies, existing water environment knowledge graphs have the following main drawbacks and shortcomings:

[0088] 1. Limited data integration capabilities: Traditional knowledge graphs primarily process structured data, making it difficult to integrate unstructured or semi-structured data, such as raw environmental monitoring data and satellite remote sensing images. 2. Lack of dynamism: Knowledge graphs are typically used as static data storage and query tools, failing to effectively represent and analyze the trends of water environment data over time. 3. Difficulty in capturing spatiotemporal features: Existing technologies struggle to effectively capture and represent the spatiotemporal characteristics of water environment data, such as the impact of pollutant emissions and seasonal variations on water quality. 4. Insufficient analytical capabilities: When dealing with complex water environment issues, existing knowledge graphs lack the ability to analyze complex relationships and patterns, limiting their capacity for in-depth understanding and decision support. 5. Limited applicability: Current water environment knowledge graphs are mostly used for specific water bodies or water environment problems, lacking universality and scalability, making it difficult to meet the data integration and analysis needs of different regions and water bodies. 6. Low level of intelligence: The construction and application of knowledge graphs suffer from low levels of intelligence, relying heavily on manual intervention and expert experience, resulting in low efficiency and limited accuracy in data processing and analysis.

[0089] To address this, this application implements a knowledge graph that comprehensively considers temporal and spatial characteristics to realize comprehensive monitoring and management of the water environment, taking into account the complexity of water environment information. Addressing the technical problems of existing technologies, this knowledge graph can be used to achieve: 1. Cross-platform integration: It can integrate water environment data from different sources, including water quality monitoring, meteorological data, geospatial information, etc.; 2. Time series analysis: It can process the temporal characteristics of water environment data, such as seasonal changes and long-term trends; 3. Spatial relationship analysis: It can identify and analyze the spatial relationships of various elements in the water environment, such as the diffusion paths of water pollution; 4. Intelligent early warning and decision support: It uses data analysis conclusions to provide decision support for water environment management and pollution control; 5. Real-time updates and dynamic evolution: It can update the graph in real time with new data input and dynamically evolve over time. The following will describe the specific implementation process in detail with reference to each step and optional embodiments.

[0090] In one optional embodiment, this embodiment preprocesses the acquired water quality monitoring data, meteorological data, and geographical data to obtain integrated data. Specifically, this may include: acquiring water quality monitoring data, meteorological data, and geographical data from a data source set as raw data, wherein the data source set includes the data source corresponding to each of the raw data; performing standardization and deduplication preprocessing on the raw data, and using the data attributes of the data source and each of the raw data as the key attributes of the raw data to establish integrated data.

[0091] In practical implementation, considering the complex water environment issues, this embodiment collects corresponding data from different data sources. Each raw data item may include data type, subdivided data items, data source (including data source and / or collection method), and key attributes. For example, the data collection and composition can be referred to in Table 1 below:

[0092] Data types Subdivided data items Data source / collection method Key attributes (used for entity association) Water quality monitoring data 1. Standard indicators: pH value, dissolved oxygen (DO), chemical oxygen demand (COD), ammonia nitrogen (NH3-N), total phosphorus (TP), total nitrogen (TN), turbidity; 2. Characteristic indicators: heavy metals (e.g., Cu, Pb, Cr), volatile organic compounds (VOCs); 3. Monitoring metadata: monitoring station number, monitoring time (accurate to the minute), monitoring equipment model, data collection frequency (e.g., once / hour), data validity indicator (normal / exceeding standard), preliminary determination of the cause of exceeding the standard (e.g., industrial discharge / domestic sewage). Automatic monitoring stations, manual sampling and laboratory analysis, and unmanned surface vessel (USV) mobile monitoring. Monitoring station ID, monitoring time, and coordinates of monitoring point Meteorological data 1. Meteorological elements: daily average temperature, daily precipitation, hourly wind speed / direction, relative humidity, sunshine duration, air pressure; 2. Extreme weather: heavy rain (24-hour precipitation ≥ 50 mm), high temperature (daily maximum temperature ≥ 35℃); 3. Meteorological metadata: meteorological station number, observation time (accurate to the hour), meteorological station altitude, data transmission delay (e.g., ≤ 15 minutes). Regional automatic weather monitoring station Weather station ID, observation time, weather station coordinates Geographic Information System (GIS) data 1. Spatial data of water bodies: River / lake vector boundaries (latitude and longitude range), drainage area, average water depth, shoreline length, water type (natural / artificial), main water supply sources (e.g., precipitation / groundwater / reservoir replenishment); 2. Spatial data of monitoring stations: Latitude and longitude coordinates (WGS84 coordinate system) of water quality monitoring stations / meteorological stations, administrative region to which the station belongs; 3. Spatial data of pollution source enterprises: Latitude and longitude coordinates of enterprises, industry type, and associated monitoring data of pollution source enterprises; 4. Auxiliary spatial data: Administrative division boundaries (province / city / county), topographic elevation data (DEM), land use type (e.g., cultivated land / industrial land / residential area), main traffic arteries (e.g., bridges / waterways, affecting pollution diffusion), and location of sewage outlets. GIS vector database, satellite remote sensing imagery, official website of administrative divisions Spatial entity ID, latitude and longitude range / coordinates, region ID

[0093] Table 1

[0094] Referring to Table 1 above, in this embodiment, the collected raw data carries key attributes, which are mainly composed of the data source and data attributes, including: the data source represented by ID, time (which can be used for subsequent time series analysis), and coordinate data (which can be used for subsequent spatial analysis).

[0095] To address the technical problem of limited cross-platform data integration capabilities in existing technologies, this embodiment further processes the collected raw data by performing standardization, deduplication, and other operations to obtain integrated data.

[0096] Step 220: Perform entity analysis based on the integrated data, establish entities, and determine the core attributes of the entities based on the key attributes and the data sources of each entity.

[0097] In practical implementation, different types of data are selected from the integrated data to create corresponding entities. Each entity mainly consists of entity type, entity name, and core attributes. Each type of data can create at least one type of entity, and each entity type (i.e., entity name) can include at least one entity (referring to entity name). For example, referring to Table 1 above, geographic GIS information includes enterprises, regions, water bodies, sewage outlets, etc. Therefore, an entity type of "GIS geographic entity" can be created, including entity names such as "water body," "pollution source enterprise," "administrative division," and "sewage outlet."

[0098] When creating entities, for each type of entity, detailed key attributes are extracted as core attributes. For some entities, detailed coordinates (such as specific latitude and longitude coordinates) and time-related information (such as specific collection frequency and time) can be further extracted.

[0099] For example, entities can be created with reference to Table 2 below:

[0100] Entity type Entity Name Core attributes (corresponding data sources) Water quality monitoring entity Water quality monitoring station Station ID (WS-001), longitude, latitude, river basin (Yangtze River Basin), monitoring indicators (DO, COD), equipment model, data update frequency (once per hour). Meteorological entities Weather station Station ID (MS-001), longitude, latitude, altitude (23m), observation elements (temperature, precipitation), observation frequency (1 time / hour) GIS geographic entities water body Water body ID (WT-001), name, water body type, latitude and longitude range, length, average water depth, main pollutants GIS geographic entities Pollution source enterprises Pollution source enterprise ID (PE-001), name, industry type, longitude, latitude, and unique ID of associated pollution discharge monitoring data. GIS geographic entities Administrative divisions Region ID (AD-001), Name, Level (Prefecture-level City), Boundary Latitude and Longitude Range (113.41°E-115.05°E, 23.58°N-26.22°N), Belonging River Basin GIS geographic entities sewage outlet Discharge outlet ID (DP-001), longitude, latitude, company, discharge type, daily discharge volume, and main pollutants.

[0101] Table 2

[0102] Step 230: Analyze the relationships between entities based on each entity and the core attribute, establish an entity triple set, and perform multi-entity concatenation on each entity in the entity triple set to obtain an entity set and establish an entity data model.

[0103] In the entity data model, each entity set consists of entity type, entity name, and core attributes.

[0104] In practical implementation, the initially established entities are discrete, but they actually possess certain chain / serial relationships. To facilitate water environment monitoring in practical applications, the relationships between entities can be further analyzed based on the initially established entities to determine the types of entity relationships. Then, using relationship type analysis, multiple entities can be established as triples, linking multiple entities of different types together to form entity triples. Each entity triple can consist of three types of entities: entities representing water quality, meteorological, and GIS data.

[0105] Therefore, this embodiment converts data of different types and structures into entities, and further realizes cross-platform integration by combining serial entity processing, thereby effectively improving data integration capabilities.

[0106] Optionally, based on the relationships between the entities and the core attributes, an entity triple set is established, and multiple entities in the entity triple set are linked together to obtain an entity set. This may include: analyzing the basis for establishing the triples based on the core attributes, and analyzing the relationship types of the entities to obtain entity relationship types; establishing a triple set based on the basis for establishing the triples and the entity relationship types; analyzing the core entities from the triple set; and linking the entities in the triple set together based on the core entities to form an entity relationship chain to obtain an entity set.

[0107] When linking entities, the first step is to analyze the basis for establishing triples based on core attributes. This involves determining which data entities can be linked together to form triples, analyzing the relationship types between entities (i.e., whether there are certain relationships between entities), and identifying the regions corresponding to the entities, such as using entity coordinates to analyze the region where the entity is located. A region can be selected, and the core entities within that region can be identified. Based on the basis for establishing triples and the entity relationship types, the three types of entities within the region are linked together to obtain entity triples. For example, in water environment monitoring, the selected region could be a river, and within the river region, three types of entities belonging to that region are selected. Through linking multiple entities, an entity relationship chain is formed. Combined with existing entities, an entity data model can be established.

[0108] As an example, combining Tables 1 and 2 above, with "X River section (WT-001)" as the core, the entities of water quality monitoring data (as the core entity), meteorological data, and GIS data are linked to form a complete relationship chain: sewage outlet of a chemical enterprise in Y town (DP-001, GIS data) → discharged to → X River section (WT-001, GIS data) → monitored → X River section water quality monitoring station (WS-001, water quality data) → affected by → Y meteorological station (MS-001, meteorological data) → belongs to → Y town (AD-001, GIS data).

[0109] A further example, based on the entity data in Tables 1 and 2, shows the relationships between entities as illustrated in Table 3 below:

[0110] Relationship type Triple group Relationship attributes Establish basis Monitoring relationship (water quality - water body) X River Section Water Quality Monitoring Station (WS-001) - Monitoring - X River Section (WT-001) Monitoring period (January 1, 2025 to May 31, 2025), core monitoring indicators (DO, COD) The latitude and longitude of WS-001 are within the same range as WT-001, and the monitoring data is labeled "Monitoring object is the X River section". Influencing Relationship (Meteorology - Water Quality) Y Meteorological Station (MS-001) - Impact - X River Section Water Quality Monitoring Station (WS-001) Impact factor (precipitation), impact time (2025-05-01 08:00), impact result (DO value of WS-001 increased from 8.2 mg / L to 8.8 mg / L) MS-001 and WS-001 are within a spatial distance of less than 10 km (neighborhood), and the DO value of WS-001 changes significantly 1 hour after precipitation. Subordination relationship (water body - administration) Section X of the river (WT-001) belongs to Town Y (AD-001) Jurisdiction The latitude and longitude range of WT-001 is completely contained within the administrative boundaries of AD-001. Discharge relationship (sewage outlet - water body) A chemical plant in Town Y discharges wastewater (DP-001) into the X River section (WT-001). Discharge time (08:00-18:00 daily), pollutant contribution rate (COD accounts for 5% of the total load of the X River section). DP-001's latitude and longitude are within 500m of the bank of WT-001, and its discharge permit states "discharged into section X of the river".

[0111] Table 3

[0112] Step 240: Perform spatial analysis and time series analysis between entities based on the entity set to obtain spatial relationships and time series relationships.

[0113] Optionally, the above-mentioned spatial and temporal series analysis of entities based on the entity set to obtain spatial and temporal series relationships may include the following sub-steps:

[0114] Sub-step 2401: Extract coordinate information from each entity in the entity set. The coordinate information is determined based on the latitude and longitude coordinates in the core attributes of each entity.

[0115] Sub-step 2402: Based on the coordinate information, the Haversian algorithm is used to analyze the proximity relationship between entities to obtain the spatial relationship.

[0116] A unified explanation is provided for sub-steps 2401 and 2402:

[0117] In practice, spatial association between entities is mainly achieved through spatial relationship judgment (such as "whether the water quality monitoring station is in the water body" or "whether the sewage outlet is adjacent to the water quality monitoring station") and spatial index optimization (quickly querying associated entities within the region), which involves spatial relationship theory and spatial indexing algorithm.

[0118] There are two main types of spatial relationships: spatial proximity and spatial dispersion. This embodiment establishes a relationship between two spatially proximity entities, but not between two spatially discrete entities. Specifically, firstly, the spatial coordinates of each entity are extracted, primarily latitude and longitude coordinates (see Tables 1 and 2 in this embodiment for details). Then, the shortest distance between two entities (such as a water quality monitoring station and a meteorological station) is calculated using these spatial coordinates. In this embodiment, the shortest distance can be quantified as the shortest distance on the Earth's surface based on latitude and longitude coordinates. The proximity relationship is then determined based on this shortest distance. The Haversine algorithm can be used to quantify the shortest distance on the Earth's surface between the latitude and longitude coordinates of two entities when calculating the shortest distance.

[0119] In an optional embodiment, the above-mentioned analysis of the proximity relationship between entities using the Haversian algorithm based on the coordinate information to obtain the spatial relationship may specifically include: establishing entity coordinates for two entities based on the coordinate information to obtain a first entity coordinate and a second entity coordinate represented in decimal; converting the first entity coordinate to obtain a first entity radian, and converting the second entity coordinate to obtain a second entity radian; calculating the spatial distance using the Haversian algorithm based on the first entity radian and the second entity radian; and comparing the spatial distance with a preset spatial threshold to determine the proximity relationship between entities and obtain the spatial relationship.

[0120] In practical implementation, the latitude and longitude coordinates of entities are extracted from the coordinate information and used as entity coordinates. Each entity coordinate can be represented in decimal. For example, suppose there is entity A (such as water quality monitoring station WS-001) and entity B (such as weather station MS-001). The coordinates of entity A are the first entity coordinates, assumed to be: Let the coordinates of entity A be the coordinates of the first entity, for example: ,in, Longitude (in radians) Latitude (radians);

[0121] Then, convert the coordinate units: convert decimal latitude and longitude (e.g., 114.32°E, 30.58°N) to radians (the formula is: ), to obtain the entity radians corresponding to the two entity coordinates;

[0122] Next, substitute the converted radians into the Haversine formula to calculate the distance and obtain the spatial distance;

[0123] Finally, the spatial distance is compared with a preset spatial threshold (preferably 10 km). If the spatial distance is less than or equal to the spatial threshold, the spatial relationship between the two entities is determined to be spatially adjacent, and an association can be established; otherwise, the spatial relationship between the two entities is determined to be spatially discrete, and an association can not be established.

[0124] Furthermore, in the process of establishing spatial relationships, GIS can be used to analyze geospatial data of each entity, classify and encode the data, and establish a geospatial coordinate system, thereby facilitating the calculation of spatial distances.

[0125] Sub-step 2403: Determine the time data of each entity. The time data includes timestamps and time sequence information. The time sequence information is determined based on the collection frequency of the data corresponding to each entity.

[0126] Sub-step 2404: Based on the timestamp and the time sequence information, a dynamic time warping algorithm is used to perform consistency alignment processing to obtain the time sequence relationship between entities.

[0127] A unified explanation is provided for sub-steps 2403 and 2404:

[0128] Time association addresses the question of when entities are associated. It requires time series alignment (including but not limited to handling differences in the frequency of data collection for different entities, such as water quality monitoring stations collecting data once per hour and sewage outlets collecting data once per day) and time index optimization (quickly querying associated entities within a certain time period). It mainly involves time series analysis and time series alignment algorithms.

[0129] This embodiment analyzes time series relationships primarily involving time series alignment. First, time-series-related data for each entity is acquired, including timestamps (typically composed of time zone and format, such as the collection time format "YYYY-MM-DD HH:MM:SS") and time series consistency (referring to collection frequency). To establish time series relationships between entities, this embodiment employs Dynamic Time Warping (DTW) to align timestamps and time series consistency, ensuring uniformity and establishing effective time associations. Specifically, during consistency alignment, time series similarity is calculated by evaluating the time series of different entities (i.e., collection frequency and time series length), achieving time series alignment and establishing a time association between them. This process mainly includes stationarity processing (meaning the time series statistical characteristics do not change over time) and time series similarity processing (measuring the similarity between two time series, such as whether the trends of water quality monitoring data and meteorological data are consistent). Consistency alignment ensures that time series relationships exist between entities.

[0130] Therefore, this embodiment addresses the shortcomings of existing technologies in time series analysis and spatial relationships. First, it fully utilizes the key / core attributes of each entity data (mainly including the coordinates of each entity) to identify the spatial positional relationships of each element in the water environment, thereby establishing spatial relationships between entities. Second, it fully utilizes the time data collected from each entity data to process the temporal characteristics of the water environment data and establish time series relationships between entities.

[0131] Step 250: Construct a water environment knowledge graph using a knowledge graph architecture, and add each entity in the entity set to the water environment knowledge graph.

[0132] In practical implementation, a water environment knowledge graph can be established using a pre-defined knowledge graph architecture. Based on the aforementioned fact that the entities in the known entity set have already established preliminary spatial and temporal relationships, each entity can be updated in the water environment knowledge graph. At this point, the obtained water environment knowledge graph will have preliminary temporal and spatial characteristics.

[0133] Step 260: In the water environment knowledge graph, the R-tree algorithm is used to optimize the spatial index of entities related to the water environment, and spatial association is performed in combination with spatial relationships to update the water environment knowledge graph.

[0134] In practical implementation, based on the aforementioned embodiments, it is known that in addition to establishing spatial relationships, spatial index optimization can also be established to improve the query efficiency of entities within a region. In this embodiment, spatial index optimization uses a minimum bounding rectangle (MBR) combinatorial optimization algorithm for multi-scale entity matching, introduces the R-tree algorithm, and processes the MBR of water environment entities based on this algorithm. For example, storing the MBR of all water environment entities enables efficient operation of querying all entities within a certain region (such as "querying all water quality monitoring stations in the X River section" in Table 3), avoiding traversing all entities.

[0135] After spatial indexing optimization, spatial relationships can be combined to establish spatial associations for each entity in the water environment knowledge graph.

[0136] Step 270: Based on the time series relationship, use a B+ tree index to perform timestamp conversion and sorting, and establish time index optimization.

[0137] Step 280: In the water environment knowledge graph, knowledge graph mapping is optimized based on the time series relationship and the time index, time axis nodes are established for each entity, the water environment knowledge graph is updated, and the water environment knowledge graph is visualized.

[0138] A unified explanation of steps 270-280 is provided below:

[0139] In practical implementation, after analyzing the time series relationships, time index optimization can be performed. A B+ tree index can be used to convert the unified format timestamps into Unix timestamps and store them in sorted order. From the perspective of time series, efficient operations can be established for entities to achieve the query of all data within a certain time period, such as querying water quality monitoring data and meteorological data from 00:00 on 2024-05-01 to 23:59 on 2024-05-01, ultimately achieving fast matching of time-related data.

[0140] In practical processing, time series data can be mapped onto a water environment knowledge graph to form time axis nodes. By visualizing the water environment knowledge graph, a visual interface can be established to intuitively display the relationships between entities, as well as the spatial and temporal relationships between entities. This improves analysis efficiency and applicability, and effectively solves the technical problems of existing technologies, such as difficulty in capturing spatiotemporal features, insufficient analysis capabilities, and limited applicability.

[0141] Step 290: Optimize and update the water environment knowledge graph through spatiotemporal correlation.

[0142] Optionally, this embodiment optimizes and updates the water environment knowledge graph through spatiotemporal correlation, which specifically includes: selecting spatial two-dimensional data and temporal data corresponding to entities from the water environment knowledge graph, wherein the spatial two-dimensional data is represented in the form of coordinate data and the temporal data is represented in the form of one-dimensional time; using the spatial two-dimensional data as input, analyzing spatial distance and determining spatial weights based on the Haversian algorithm and a negative exponential decay model; using the temporal data as input, analyzing time differences and analyzing temporal distances using a linear decay model to determine temporal weights; quantifying the spatiotemporal correlation strength between entities based on the spatial weights and the temporal weights; fusing the spatial two-dimensional data and the temporal data to obtain spatiotemporal three-dimensional data, and using a spatial data indexing algorithm to optimize spatiotemporal correlation and establish a spatiotemporal index storage; and updating the water environment knowledge graph according to the spatiotemporal correlation strength and the spatiotemporal index storage.

[0143] In practical implementation, after establishing a multi-dimensional water environment knowledge graph encompassing time and space, spatiotemporal relationships can be further established for each entity in the knowledge graph. These spatiotemporal relationships involve spatiotemporal weight allocation (i.e., the corresponding weight allocation for time and space) and spatiotemporal index union. Specifically, the spatiotemporal weight allocation is determined to quantify the strength of the spatiotemporal relationships between entities; for example, a sewage outlet within 1 km of the entity has a stronger impact on a water quality monitoring station within the past 3 days. The spatiotemporal index union (including storing three-dimensional data of entities, time, and space) establishes a four-dimensional structure (including entities, time, space, and corresponding weights). The following provides a detailed explanation:

[0144] Spatiotemporal weight calculation: Primarily based on a joint distance decay and time decay algorithm, this method quantifies the strength of spatiotemporal associations between entities, adding weight attributes to the temporal and spatial relationships in the graph. This facilitates the assessment of the strength of spatiotemporal associations. In essence, the closer the weight is to 1, the higher the association. The calculation steps are as follows:

[0145] ① Extract spatial two-dimensional data and temporal data of each entity from the current knowledge graph. The spatial two-dimensional data is also represented in coordinate form; the temporal data can be represented in one-dimensional time form (referring to specific time values, such as 2024-05-01).

[0146] ② Calculate the spatial weight, assuming it to be Ws: Using the Haversine algorithm, analyze the spatial distance d between entities (the specific process can be found in the steps above, and will not be detailed here). Of course, since the previous steps have already calculated this, it can be obtained directly. Using the spatial distance d as input, analyze the spatial distance using a negative exponential decay model to determine Ws. It can be understood that the closer the distance, the larger Ws is.

[0147] ③ Calculate the time weight, assuming it to be Wt: Based on the event data of two entities, calculate the time difference, assuming it to be Δt (the difference between the current time and the associated time). Using the time difference as input, analyze Δt using a linear decay model to determine Wt. It can be understood that the closer the times (the smaller Δt), the larger Wt.

[0148] ④ Calculate the total spatiotemporal weight, assuming it is Wst. Determine the total spatiotemporal weight according to Wst=Ws×Wt. Preferably, the value range is [0,1]. The closer the weight is to 1, the stronger the spatiotemporal correlation.

[0149] Next, we will establish a spatiotemporal joint index: First, based on the extracted spatial two-dimensional data and temporal data, we will construct spatiotemporal three-dimensional data, assuming it is as follows. ,in, Spatial coordinates; For time, including the time interval [t_start, t_end] (one-dimensional time).

[0150] Then, the R*-Tree algorithm (i.e., a spatial data indexing algorithm) is used to construct an indexed storage for spatiotemporal three-dimensional data. Specifically, the entity's MBR (spatial two-dimensional) is merged with the time interval to form a spatiotemporal MBR, assuming it is... By storing the spatiotemporal MBR according to the R*-Tree structure, the overlap of node MBRs is reduced and the query efficiency is improved. Thus, this embodiment solves the tedious problem of querying all related entities within a certain spatiotemporal range in complex water environment problems. For example, querying the association between all sewage outlets and water quality monitoring stations in the X River section from 2024-05-01 to 2024-05-03.

[0151] Furthermore, in practical implementation, after establishing a knowledge graph that comprehensively considers temporal and spatial characteristics, this embodiment can use the knowledge graph for comprehensive monitoring and management of the water environment. Specifically, after updating the water environment knowledge graph through spatiotemporal correlation optimization, it may further include: analyzing key events based on the time series of each entity in the water environment knowledge graph, and determining the water environment region corresponding to the key events, wherein the key events include at least extreme climate events and sudden changes in water quality; establishing a prediction model; for the key events, using the time series as input, predicting the future state of the water environment in the water environment region through the prediction model, and outputting prompt data for the key events through the prediction model, wherein the prompt data includes intelligent early warning information and decision-making information; and updating and dynamically evolving the water environment knowledge graph in real time through the updated data of the integrated data.

[0152] In practical implementation, established time-series relationships can be used as time-series data to build a prediction model (such as a time-series model; this implementation does not impose restrictions on this). Key events in the time series are analyzed from the knowledge graph. When a key event occurs, the future state of the water environment within a certain area is predicted using the prediction model, with the time-series data as input. When dealing with complex water environment problems, intelligent early warning and decision support can be achieved, outputting early warning information. Specifically, by combining prediction models and machine learning techniques, intelligent early warnings are provided for events in the knowledge graph, and a decision support system can be introduced to guide actual water resource management activities, thereby achieving intelligent early warning and decision support and solving the technical problem of low intelligence levels in the application of water environment knowledge graphs in existing technologies.

[0153] Furthermore, this embodiment can also be updated and dynamically evolved in real time. Through a real-time data update mechanism, the knowledge graph is updated when data / entities are updated to ensure the real-time nature of the graph information. To adapt to the needs of changes in the water environment, the graph can also be dynamically evolved.

[0154] In summary, this application proposes a method for constructing a water environment knowledge graph based on spatiotemporal information fusion, aiming to solve the data silo problem existing in current water environment monitoring systems and improve the processing capability of massive spatiotemporal data. Specifically, water quality monitoring data, meteorological data, and geographical data are collected from data sources, preprocessed, and integrated. Key attributes of each data source are constructed based on the data sources. Then, entities are established through entity analysis based on the key attributes; the core attributes of each entity are further clarified based on the key attributes. Entity relationships are analyzed through entities and core attributes, and entity sets are established through entity concatenation, thereby constructing an entity data model. Subsequently, spatial and temporal series analyses are performed on each entity to clarify the spatial and temporal relationships between entities. A water environment knowledge graph is constructed using each entity, with spatial association optimization based on spatial relationships and temporal association optimization based on temporal series relationships. Furthermore, the water environment knowledge graph is optimized through spatiotemporal association to establish a multi-dimensional water environment knowledge graph. As can be seen, the beneficial effects of this embodiment include: 1. Improved information integration capabilities: It can process multi-dimensional and large-scale data, avoid the problem of information silos, and improve comprehensive management efficiency; 2. Enhanced decision support: Through accurate early warning and analysis, it enhances the scientific nature and efficiency of decision-making; 3. Enhanced responsiveness: Real-time updates and dynamic evolution capabilities ensure that it can continuously provide feedback and response strategies when facing complex and ever-changing water environment problems; 4. Intelligent environmental management: This embodiment provides a new way to realize intelligent water environment management, which has broad application prospects and promotional value.

[0155] It should be noted that, for the sake of simplicity, the method embodiments are described as a series of actions. However, those skilled in the art should know that the embodiments of this application are not limited to the described order of actions, because according to the embodiments of this application, some steps may be performed in other orders or simultaneously.

[0156] like Figure 3 As shown in the figure, this application embodiment also provides a time- and space-based water environment knowledge graph construction device 300, including:

[0157] The multi-source data acquisition module 310 is used to preprocess the acquired water quality monitoring data, meteorological data and geographical data to obtain integrated data, wherein each data in the integrated data carries key attributes.

[0158] The entity construction module 320 is used to construct an entity data model based on the key attributes, through entity analysis and entity concatenation processing of the integrated data. Each entity set in the entity data model consists of entity type, entity name and core attributes.

[0159] The spatiotemporal analysis module 330 is used to perform spatial analysis and time series analysis between entities based on the entity set to obtain spatial relationships and time series relationships;

[0160] The knowledge graph construction module 340 is used to construct a knowledge graph architecture, optimize the spatial association of each entity in the entity set according to the spatial relationship, optimize the temporal association of each entity according to the temporal sequence relationship, update the knowledge graph architecture, and obtain a multi-dimensional water environment knowledge graph.

[0161] The spatiotemporal correlation module 350 is used to optimize and update the water environment knowledge graph through spatiotemporal correlation.

[0162] It should be noted that the time- and space-based water environment knowledge graph construction device provided in this application embodiment can execute the time- and space-based water environment knowledge graph construction method provided in any embodiment of this application, and has the corresponding functions and beneficial effects of the execution method.

[0163] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0164] The above description is merely a specific embodiment of this application, enabling those skilled in the art to understand or implement this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features claimed herein.

Claims

1. A method for constructing a water environment knowledge graph based on time and space, characterized in that, include: Preprocessing is performed on the acquired water quality monitoring data, meteorological data, and geographic data to obtain integrated data. Each data point in the integrated data carries key attributes, including data source, time, and coordinate data. Based on the aforementioned key attributes, an entity data model is constructed. Each entity set in the entity data model consists of entity type, entity name, and core attributes. The entity types include monitoring entities, meteorological entities, and geographic entities. The core attributes include data source, latitude and longitude data, pollutants, and data collection frequency. Based on the entity set, spatial and temporal analysis are performed between entities to obtain spatial and temporal relationships. A knowledge graph architecture is constructed. Based on the spatial relationships, spatial association optimization is performed on each entity in the entity set. Based on the time series relationships, temporal association optimization is performed on each entity. The knowledge graph architecture is then updated to obtain a multi-dimensional water environment knowledge graph. Specifically: In the water environment knowledge graph, the R-tree algorithm is used to optimize the spatial index of entities related to the water environment, and spatial association is performed in conjunction with spatial relationships to update the water environment knowledge graph; based on the time series relationships, B+ tree indexing is used for timestamp conversion and sorting to establish time index optimization; based on the time series relationships and the time index optimization, knowledge graph mapping is performed, timeline nodes are established for each entity, and the water environment knowledge graph is updated. The water environment knowledge graph is updated through spatiotemporal correlation optimization, wherein: spatial two-dimensional data and temporal data corresponding to entities are selected from the water environment knowledge graph; using the spatial two-dimensional data as input, spatial distance is analyzed using a negative exponential decay model based on the Haversian algorithm to determine spatial weights; using the temporal data as input, time difference is analyzed, and temporal distance is analyzed using a linear decay model to determine temporal weights; based on the spatial weights and temporal weights, the spatiotemporal correlation strength between entities is quantified; the spatial two-dimensional data and the temporal data are fused to obtain spatiotemporal three-dimensional data; and the water environment knowledge graph is updated.

2. The method according to claim 1, characterized in that, Based on the acquired water quality monitoring data, meteorological data, and geographic data, preprocessing is performed to obtain integrated data, including: Water quality monitoring data, meteorological data, and geographical data are obtained from the data source set as raw data. The data source set includes the data source corresponding to each of the raw data. The original data is standardized and deduplicated, and the data attributes of the data source and each original data are used as the key attributes of the original data to establish integrated data.

3. The method according to claim 1, characterized in that, Based on the aforementioned key attributes, an entity data model is constructed by integrating the data through entity analysis and entity concatenation processing, including: Entity analysis is performed based on the integrated data to establish entities, and the core attributes of the entities are determined by combining the key attributes with the data sources of each entity. Based on the relationships between the entities and the core attributes, an entity triple set is established, and the entities in the entity triple set are linked together to obtain an entity set, thus establishing an entity data model.

4. The method according to claim 3, characterized in that, Based on the relationships between the entities and the core attributes, an entity triple set is established, and multiple entity concatenation is performed on each entity in the entity triple set to obtain an entity set, including: Based on the core attribute analysis, the establishment basis of the triplet is analyzed, and based on the entity analysis relationship type, the entity relationship type is obtained; Based on the establishment criteria and the entity relationship type, a set of triples is established for each entity; Analyze the core entities from the set of triples; Based on the core entity, the entities in the triplet set are linked together to form an entity relationship chain, thus obtaining an entity set.

5. The method according to claim 1, characterized in that, Based on the entity set, spatial and temporal series analyses are performed between entities to obtain spatial and temporal relationships, including: Coordinate information is extracted from each entity in the entity set, and the coordinate information is determined based on the latitude and longitude coordinates in the core attributes of each entity; Based on the coordinate information, the Haversian algorithm is used to analyze the proximity relationships between entities to obtain spatial relationships; Determine the time data for each entity, wherein the time data includes timestamps and time sequence information, and the time sequence information is determined based on the collection frequency of the data corresponding to each entity. Based on the timestamp and the time sequence information, a dynamic time warping algorithm is used for consistency alignment to obtain the time sequence relationship between entities.

6. The method according to claim 5, characterized in that, Based on the coordinate information, the Haversian algorithm is used to analyze the proximity relationships between entities to obtain spatial relationships, including: Based on the coordinate information, entity coordinates are established for the two entities respectively, resulting in the first entity coordinates and the second entity coordinates represented in decimal. The first entity coordinates are transformed to obtain the first entity radian, and the second entity coordinates are transformed to obtain the second entity radian; Based on the first entity radian and the second entity radian, the Haversian algorithm is used to calculate the spatial distance; Based on the spatial distance, a comparison is made with a preset spatial threshold to determine the proximity relationship between entities and obtain the spatial relationship.

7. The method according to any one of claims 1-6, characterized in that, After optimizing and updating the water environment knowledge graph through spatiotemporal correlation, it also includes: Based on the time series of each entity in the water environment knowledge graph, key events are analyzed and the corresponding water environment regions are determined. The key events include at least extreme climate events and sudden changes in water quality. Develop a predictive model; For the key events, using the time series as input, the prediction model predicts the future state of the water environment in the water environment area, and outputs the key event alert data through the prediction model. The alert data includes intelligent early warning information and decision information. The water environment knowledge graph is updated and dynamically evolved in real time using the updated data from the integrated data.

8. A time- and space-based water environment knowledge graph construction device, characterized in that, include: The multi-source data acquisition module is used to preprocess the acquired water quality monitoring data, meteorological data, and geographical data to obtain integrated data. Each data in the integrated data carries key attributes, including data source, time, and coordinate data. The entity construction module is used to construct an entity data model based on the key attributes. Each entity set in the entity data model consists of entity type, entity name and core attributes. The entity types include monitoring entities, meteorological entities and geographic entities. The core attributes include data source, latitude and longitude data, pollutants and data collection frequency. The spatiotemporal analysis module is used to perform spatial and temporal series analysis between entities based on the entity set, and to obtain spatial and temporal series relationships. A knowledge graph construction module is used to construct a knowledge graph architecture, optimize spatial associations of entities in the entity set based on spatial relationships, and optimize temporal associations of entities based on temporal series relationships, thereby updating the knowledge graph architecture and obtaining a multi-dimensional water environment knowledge graph. Specifically: in the water environment knowledge graph, the R-tree algorithm is used to optimize the spatial index of entities related to the water environment, and spatial associations are performed in conjunction with spatial relationships to update the water environment knowledge graph; based on the temporal series relationships, B+ tree indexing is used for timestamp conversion and sorting to establish a time index optimization; based on the temporal series relationships and the time index optimization, knowledge graph mapping is performed, timeline nodes are established for each entity, and the water environment knowledge graph is updated. The spatiotemporal correlation module is used to optimize and update the water environment knowledge graph through spatiotemporal correlation, wherein: spatial two-dimensional data and temporal data corresponding to entities are selected from the water environment knowledge graph; using the spatial two-dimensional data as input, spatial distance is analyzed using a negative exponential decay model based on the Haversian algorithm to determine spatial weights; using the temporal data as input, time difference is analyzed, and temporal distance is analyzed using a linear decay model to determine temporal weights; based on the spatial weights and the temporal weights, the spatiotemporal correlation strength between entities is quantified; the spatial two-dimensional data and the temporal data are fused to obtain spatiotemporal three-dimensional data; and the water environment knowledge graph is updated.

Images