Cooperative presentation method of knowledge graph and ocean process model

Abstract

The invention relates to the technical field of ocean data processing, in particular to a collaborative presentation method of a knowledge graph and an ocean process model, which comprises the steps of extracting ocean application scene knowledge for knowledge fusion and representation, obtaining a knowledge representation model, constructing the knowledge graph by combining a graph calculation algorithm, and embedding the knowledge representation model into the knowledge graph. Inputting the ocean data into an ocean process model for initialization and labeling, and performing simulation evolution on the ocean process model; the embedded knowledge graph and the evolved ocean process model are subjected to heterogeneous data fusion and association, query and feedback are conducted, the knowledge graph storage mode of an attribute graph is adopted, the internal structure of the graph is directly reflected, in combination with a graph calculation algorithm, deep mining and reasoning of knowledge are conducted, and through heterogeneous data fusion and association, the deep mining and reasoning of the knowledge graph are achieved. And entity-attribute identification is carried out, so that the precision and the calculation efficiency of the ocean process model are improved, the ocean process is simulated more accurately, and the calculation time is shortened.

Description

Technical Field

[0001] The present invention relates to the field of ocean data processing technology, and in particular to a collaborative presentation method of a knowledge graph and an ocean process model. Background Art

[0002] Ocean numerical models are one of the important tools in the field of ocean research. However, there are still large systematic biases in almost all advanced ocean models, and one of the main reasons for this phenomenon is closely related to ocean processes, which often have spatial scales smaller than the model grid resolution. These sub-grid ocean processes that have not been fully resolved must be parameterized in the ocean model to represent the role they play in the simulation process.

[0003] However, among the current traditional ocean models, most use parameterized models with great uncertainty in turbulence characteristics, such as the K-profile parameterized model (KPP). This uncertainty will cause errors in the simulation and prediction of the ocean models, making their prediction results inaccurate and unreliable.

[0004] In order to solve the problems existing in the existing ocean models, based on the existing Beibu Gulf ocean observations, coastal geology, coastal ecosystems and remote sensing data, this paper takes the coordinated planning, development and industrial services of land and sea natural resources as the guide, and takes the structured and intelligent processing of natural resource data as the core. In combination with the actual needs of coastal land space planning, environmental ecological restoration and regional industrial development, a collaborative presentation method of knowledge graphs and data models is proposed through the structuring of natural resource data, knowledge systematization, modeling and process reconstruction, so as to improve the data service capabilities of ocean models. Summary of the invention

[0005] The purpose of the present invention is to provide a collaborative presentation method of a knowledge graph and an ocean process model, aiming to solve the problem of inaccurate and unreliable simulation and prediction results of ocean models.

[0006] To achieve the above object, the present invention provides a collaborative presentation method of a knowledge graph and an ocean process model, comprising the following steps:

[0007] Extracting marine application scenario knowledge for knowledge fusion and representation, obtaining a knowledge representation model, building a knowledge graph in combination with a graph computing algorithm, and embedding the knowledge representation model into the knowledge graph;

[0008] Inputting ocean data into an ocean process model for initialization and annotation, and simulating the evolution of the ocean process model;

[0009] The embedded knowledge graph and the evolved ocean process model are used to fuse and associate heterogeneous data, and perform query and feedback.

[0010] The specific method of extracting marine application scenario knowledge for knowledge fusion and representation, obtaining a knowledge representation model, building a knowledge graph in combination with a graph computing algorithm, and embedding the knowledge representation model into the knowledge graph is as follows:

[0011] Extract marine application scenario knowledge to form an initial knowledge base;

[0012] Acquire knowledge from different sources of the knowledge base, fuse and represent it, and obtain a knowledge representation model;

[0013] Construct a knowledge graph based on the knowledge base combined with a graph computing algorithm;

[0014] Embed the knowledge representation model into the knowledge graph.

[0015] The marine application scenario knowledge includes entities and relationships related to marine processes in marine scientific literature and databases, such as marine phenomena, physical parameters and biological species.

[0016] The specific method of inputting ocean data into the ocean process model for initialization and annotation, and simulating the evolution of the ocean process model is as follows:

[0017] Select appropriate ocean process models;

[0018] Inputting ocean data into the ocean process model for initialization and annotation;

[0019] The parameters and boundary conditions of the ocean process model are adjusted to simulate the evolution of the ocean process model.

[0020] The specific method of inputting the ocean data into the ocean process model for initialization and annotation is as follows:

[0021] Obtain ocean and coastal observation data, historical data, and acquisition data;

[0022] Standardizing the collected data, cleaning redundancy, errors and outliers, and obtaining ocean data;

[0023] The ocean data is input into the ocean process model for initialization and annotation.

[0024] The specific method of fusing and associating heterogeneous data of the embedded knowledge graph and the evolved ocean process model is as follows:

[0025] Perform image preprocessing on the embedded knowledge graph to obtain a processed graph;

[0026] Preprocessing the input text of the evolved ocean process model to obtain a processing model;

[0027] The processing graph and the processing model are subjected to feature extraction, fusion and association.

[0028] The specific method of performing feature extraction, fusion and association on the processing graph and the processing model is as follows:

[0029] Performing feature extraction and fusion on the processing graph and the processing model based on a deep learning model;

[0030] The processing graph and the processing model are associated based on a machine learning algorithm.

[0031] The present invention provides a collaborative presentation method of a knowledge graph and an ocean process model, extracts ocean application scenario knowledge for knowledge fusion and representation, obtains a knowledge representation model, constructs a knowledge graph in combination with a graph computing algorithm, and embeds the knowledge representation model into the knowledge graph; inputs ocean data into the ocean process model for initialization and annotation, and simulates the evolution of the ocean process model; performs heterogeneous data fusion and association on the embedded knowledge graph and the evolved ocean process model, and performs query and feedback. This method adopts a knowledge graph storage method of an attribute graph, directly reflects the internal structure of the graph, and combines a graph computing algorithm to perform knowledge fusion and representation. Deep mining and reasoning use GIS data knowledge annotation, including point, line, and surface model annotation, close to the scenes in the natural resources field, and entity-attribute recognition through heterogeneous data fusion and association to improve the accuracy and computational efficiency of ocean process models, so as to more accurately simulate ocean processes and reduce computing time. It can monitor changes in the ocean environment in real time, such as the dynamic changes of parameters such as temperature, salinity, and flow rate, and quickly and directly provide users with full-dimensional search results of the ocean within the observation range, improve the knowledge and data service capabilities required for ocean scene applications, and solve the problem of inaccurate and unreliable simulation and prediction results of ocean models. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0033] Figure 1 It is a flow chart of a collaborative presentation method of a knowledge graph and an ocean process model provided by the present invention.

[0034] Figure 2It is a flowchart of the specific method of extracting marine application scenario knowledge for knowledge fusion and representation, obtaining a knowledge representation model, building a knowledge graph in combination with a graph computing algorithm, and embedding the knowledge representation model into the knowledge graph.

[0035] Figure 3 It is a flowchart of a specific method for inputting ocean data into an ocean process model for initialization and annotation, and simulating the evolution of the ocean process model.

[0036] Figure 4 It is a flowchart of a specific method for inputting ocean data into the ocean process model for initialization and annotation.

[0037] Figure 5 It is a flowchart of the specific method of fusing and associating heterogeneous data between the embedded knowledge graph and the evolved ocean process model.

[0038] Figure 6 It is a flowchart of a specific method for performing feature extraction, fusion and association on the processing graph and the processing model. DETAILED DESCRIPTION

[0039] Embodiments of the present invention are described in detail below, 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 should not be construed as limiting the present invention.

[0040] See also Figures 1 to 6 The present invention provides a collaborative presentation method of a knowledge graph and an ocean process model, comprising the following steps:

[0041] S1 extracts marine application scenario knowledge for knowledge fusion and representation, obtains a knowledge representation model, constructs a knowledge graph in combination with a graph computing algorithm, and embeds the knowledge representation model into the knowledge graph;

[0042] Specific method:

[0043] S11 extracts marine application scenario knowledge to form an initial knowledge base;

[0044] In the embodiment of the present invention, the marine application scenario knowledge includes entities and relationships related to marine processes in marine scientific literature and databases, such as marine phenomena, physical parameters, and biological species.

[0045] S12 obtains knowledge from different sources of the knowledge base, fuses and represents it, and obtains a knowledge representation model;

[0046] In the embodiment of the present invention, knowledge from different sources is integrated, redundancy and conflict are removed, and a suitable knowledge representation model (such as ontology language) is used to represent the ocean process knowledge in a structured manner.

[0047] S13 constructs a knowledge graph based on the knowledge base combined with a graph computing algorithm;

[0048] S14 embeds the knowledge representation model into the knowledge graph.

[0049] In an embodiment of the present invention, the entities and relationships in the knowledge graph are embedded into a vector space to form a low-dimensional vector representation, which can be used to measure the similarity between entities and the strength of relationships.

[0050] S2 inputs the ocean data into the ocean process model for initialization and annotation, and simulates the evolution of the ocean process model;

[0051] Specific method:

[0052] S21Selection of appropriate ocean process models;

[0053] In an embodiment of the present invention, the ocean process model includes a physical ocean model, an ecological ocean model, etc.

[0054] S22 inputs the ocean data into the ocean process model for initialization and annotation;

[0055] Specific method:

[0056] S221 obtains ocean and coastal observation data and historical data to obtain collected data;

[0057] S222 performs standardization processing on the collected data, cleans up redundancy, errors and abnormal values, and obtains ocean data;

[0058] In the embodiment of the present invention, the standardization process cleans redundancy, errors and outliers to ensure the consistency and comparability of data formats and improve data quality.

[0059] S223 inputs the ocean data into the ocean process model for initialization and annotation.

[0060] S23 adjusts the ocean process model parameters and boundary conditions, and simulates the evolution of the ocean process model.

[0061] S3 fuses and associates the embedded knowledge graph and the evolved ocean process model with heterogeneous data, and performs query and feedback.

[0062] In the embodiment of the present invention, the knowledge graph and the ocean process model heterogeneous data are fused and associated to perform query and feedback, and the content of the knowledge graph can be dynamically updated according to the simulation results of the ocean process model and new knowledge input. At the same time, the display results can be corrected and improved through the feedback mechanism to improve the accuracy and reliability of the system, improve the accuracy and calculation efficiency of the ocean process model, so as to more accurately simulate the ocean process and reduce the calculation time, and monitor the changes in the ocean environment in real time, such as the dynamic changes of parameters such as temperature, salinity, and flow rate, and quickly and directly provide users with full-dimensional search results of the ocean within the observation range, so as to improve the knowledge and data service capabilities required for ocean scene applications.

[0063] Specific method:

[0064] S31 performs image preprocessing on the embedded knowledge graph to obtain a processed graph;

[0065] In an embodiment of the present invention, the knowledge graph is adjusted to a suitable size and ratio, and a deep learning model such as a convolutional neural network (CNN) is used to extract feature vectors of the image.

[0066] S32 performs input text preprocessing on the evolved ocean process model to obtain a processing model;

[0067] In an embodiment of the present invention, the input text is cleaned to remove irrelevant characters, stop words and noise in the text, the text is segmented into words or phrases, and word embedding technology (such as Word2Vec, BERT, etc.) is used to convert the words or phrases into vector representations.

[0068] S33 performs feature extraction, fusion and association on the processing graph and the processing model.

[0069] Specific method:

[0070] S311 performs feature extraction and fusion on the processing graph and the processing model based on a deep learning model;

[0071] In an embodiment of the present invention, a deep learning model (such as a convolutional neural network, a recurrent neural network, a Transformer, etc.) is used to extract features and fuse the processing graph and the processing model. For example, CNN can be used to extract image features, RNN or Transformer can be used to extract text features, and then the two can be fused at a certain intermediate layer. In addition to deep learning model fusion, two methods, a matrix decomposition-based method and a graph-based method, can also be used.

[0072] Among them, the matrix decomposition-based method is to represent the feature vectors of text and image in matrix form, and then use matrix decomposition techniques (such as singular value decomposition, non-negative matrix decomposition, etc.) to extract their common features and fuse them.

[0073] There are two graph-based methods: the feature vectors of text and image are represented as nodes in a graph, and then techniques such as graph neural network (GNN) are used to capture the relationship between them and fuse them.

[0074] S312 associates the processing graph and the processing model based on a machine learning algorithm.

[0075] In an embodiment of the present invention, a machine learning algorithm (such as a decision tree, a neural network, etc.) is used to mine the association relationship between data.

[0076] What is disclosed above is only a preferred embodiment of a method for collaborative presentation of a knowledge graph and an ocean process model of the present invention. Of course, this cannot be used to limit the scope of rights of the present invention. Ordinary technicians in this field can understand that all or part of the processes of the above embodiments and equivalent changes made in accordance with the claims of the present invention still fall within the scope of the invention.

Claims

1. A collaborative presentation method of a knowledge graph and an ocean process model, characterized in that: The following steps are involved: Extracting marine application scenario knowledge for knowledge fusion and representation, obtaining a knowledge representation model, building a knowledge graph in combination with a graph computing algorithm, and embedding the knowledge representation model into the knowledge graph; Inputting ocean data into an ocean process model for initialization and annotation, and simulating the evolution of the ocean process model; The embedded knowledge graph and the evolved ocean process model are used to fuse and associate heterogeneous data, and perform query and feedback.

2. The collaborative presentation method of the knowledge graph and the ocean process model as claimed in claim 1, characterized in that ； The specific method of extracting marine application scenario knowledge for knowledge fusion and representation, obtaining a knowledge representation model, combining a graph computing algorithm to construct a knowledge graph, and embedding the knowledge representation model into the knowledge graph is as follows: Extract marine application scenario knowledge to form an initial knowledge base; Acquire knowledge from different sources of the knowledge base, fuse and represent it, and obtain a knowledge representation model; Construct a knowledge graph based on the knowledge base combined with a graph computing algorithm; Embed the knowledge representation model into the knowledge graph.

3. The collaborative presentation method of the knowledge graph and the ocean process model as claimed in claim 1, It is characterized by: The marine application scenario knowledge includes entities and relationships related to marine processes in marine scientific literature and databases, such as marine phenomena, physical parameters and biological species.

4. The collaborative presentation method of the knowledge graph and the ocean process model according to claim 1, characterized in that; The specific method of inputting ocean data into the ocean process model for initialization and annotation, and simulating the evolution of the ocean process model is as follows: Select appropriate ocean process models; Inputting ocean data into the ocean process model for initialization and annotation; The parameters and boundary conditions of the ocean process model are adjusted to simulate the evolution of the ocean process model.

5. The collaborative presentation method of the knowledge graph and the ocean process model as claimed in claim 4, characterized in that; The specific method of inputting the ocean data into the ocean process model for initialization and annotation is: Obtain ocean and coastal observation data, historical data, and acquisition data; Standardizing the collected data, cleaning redundancy, errors and outliers, and obtaining ocean data; The ocean data is input into the ocean process model for initialization and annotation.

6. The collaborative presentation method of the knowledge graph and the ocean process model according to claim 1, characterized in that; The specific method of fusing and associating heterogeneous data of the embedded knowledge graph and the evolved ocean process model is as follows: Perform image preprocessing on the embedded knowledge graph to obtain a processed graph; Preprocessing the input text of the evolved ocean process model to obtain a processing model; The processing graph and the processing model are subjected to feature extraction, fusion and association.

7. The collaborative presentation method of the knowledge graph and the ocean process model as claimed in claim 6, It is characterized by: The specific method of performing feature extraction, fusion and association on the processing graph and the processing model is as follows: Performing feature extraction and fusion on the processing graph and the processing model based on a deep learning model; The processing graph and the processing model are associated based on a machine learning algorithm.

Images