Sea multi-parameter information extraction method based on term semantics

Abstract

The invention relates to the technical field of ocean multi-parameter information extraction, in particular to an ocean multi-parameter information extraction method based on term semantics, which introduces an advanced clustering algorithm and a classification algorithm to perform deep learning and optimization on ocean terms; using term expression ability and Chinese phrase structure rules to optimize a K-proximity algorithm, and identifying marine entities and attributes thereof in the text; the method comprises the following steps: performing entity and attribute extraction, and performing standardization processing on extracted entities and attributes; designing a maximum correlation algorithm, and evaluating the importance of different parameters in a specific ocean event or process; through multi-dimensional correlation analysis, the extracted parameter information is matched with ocean structural data, multi-dimensional correlation of the information is achieved, the optimized extraction content is output according to the analysis result, in this way, the parameters which have the largest influence on the water quality condition are screened out by combining the ecological system characteristics and seasonal changes of the sea area, and the water quality condition of the sea area is determined. Visual reports and suggestions are generated, and decision basis is provided for managers.

Description

Technical Field

[0001] The present invention relates to the technical field of ocean multi-parameter information extraction, and in particular to an ocean multi-parameter information extraction method based on terminology semantics. Background Art

[0002] With the advancement of ocean observation technology and the advent of the big data era, the amount of ocean data collected has exploded. These data cover multiple dimensions such as physics, chemistry, and biology, and are of great significance for understanding marine ecosystems, predicting marine disasters, and optimizing the development of marine resources.

[0003] However, the complexity, heterogeneity and specialized terminology of ocean data pose great challenges to the effective extraction and analysis of information.

[0004] When dealing with complex scenarios such as ocean data, current multi-parameter information extraction methods often face the defects of scarce and unbalanced data sources, insufficient algorithm generalization ability, and low entity and attribute recognition accuracy, which affects the extraction speed and accuracy of ocean multi-parameter information and limits its use in marine scientific research, environmental monitoring and resource management. Summary of the invention

[0005] The purpose of the present invention is to provide a method for extracting ocean multi-parameter information based on terminology semantics, which solves the problem that the multi-parameter information extraction method often faces the defects of scarce and unbalanced data sources, insufficient algorithm generalization ability, and low entity and attribute recognition accuracy when processing complex scenes such as ocean data, thereby affecting the extraction speed and accuracy of ocean multi-parameter information.

[0006] To achieve the above object, the present invention provides a method for extracting ocean multi-parameter information based on terminology semantics, comprising the following steps:

[0007] Based on the existing marine terminology semantic knowledge system, advanced clustering and classification algorithms are introduced to conduct in-depth learning and optimization of marine terminology;

[0008] By using the expressive power of terms and the rules of Chinese phrase structure, the K-nearest neighbor algorithm is optimized to identify marine entities and their attributes in the text;

[0009] Entity and attribute extraction is performed by combining domain-specific dictionaries and contextual information, and the extracted entities and attributes are standardized;

[0010] Combine scene semantic information to design maximum relevance algorithms to assess the importance of different parameters in specific marine events or processes;

[0011] Through multi-dimensional correlation analysis, the extracted parameter information is matched with the ocean structured data to achieve multi-dimensional correlation of information.

[0012] Based on the analysis results, the optimized extraction content is output, including key parameters, trend prediction, and anomaly detection.

[0013] Among them, in "on the basis of the existing marine terminology semantic knowledge system, introducing advanced clustering algorithms and classification algorithms to conduct deep learning and optimization of marine terminology", the method includes the following steps:

[0014] Collect marine terms and their related descriptions, integrate the data into the database, clean and pre-process the data to remove duplicate, erroneous or irrelevant information;

[0015] Based on the existing semantic knowledge system of marine terminology, clustering and classification algorithms are used, and the algorithms are preliminarily applied to the preprocessed data, and the performance of the algorithms is evaluated through cross-validation;

[0016] Based on the results of the initial implementation of the algorithm, a deep learning model was constructed to perform deep feature extraction and classification of marine terms, and the model was trained using the preprocessed data;

[0017] Combine the output of the deep learning model with the existing semantic knowledge system of marine terminology to update and improve the terminology library and explore new relationships between terms.

[0018] Among them, in “using terminology expression ability and Chinese phrase structure rules to optimize K-nearest neighbor algorithm and identify marine entities and their attributes in text”, the method includes the following steps:

[0019] Analyze the phrase structure of Chinese, extract common phrase structure rules based on a large corpus, and form a rule base as a grammatical basis for identifying marine entities and their attributes;

[0020] Combining the expressive power of terms and the structural rules of Chinese phrases, we designed a feature selection and representation method to convert text data into feature vectors that can be processed by the algorithm. At the same time, we adjusted the parameters of the KNN algorithm and introduced a weight mechanism.

[0021] The preprocessed text data and the annotated marine entities and their attribute information are used to train the optimized KNN model. The model is used to identify and extract marine entities and their attributes from new text data to form structured information, which provides a basis for subsequent analysis and application.

[0022] The recognition results are manually verified, and the algorithms and models are iteratively optimized based on the verification results.

[0023] Among them, in “extracting entities and attributes by combining domain-specific dictionaries and context information, and standardizing the extracted entities and attributes”, the steps include:

[0024] Based on the target domain data, a domain-specific dictionary containing key entities and attributes is constructed. The dictionary will serve as a reference for entity and attribute extraction.

[0025] Preprocess the input text, including removing punctuation and special character noise, word segmentation, and part-of-speech tagging;

[0026] Utilize the constructed domain-specific dictionary and the contextual information of the text to extract entities and attributes. By matching keywords and phrases in the dictionary and analyzing the grammatical structure and semantic relationships in the text, target entities and attributes are identified.

[0027] The extracted entities and attributes are standardized, including unified naming conventions, removal of synonyms and near-synonyms, and determination of attribute types and value ranges, to ensure that the extracted entities and attributes are consistent and comparable in subsequent analysis and applications.

[0028] Among them, in “combining scene semantic information, designing a maximum correlation algorithm, and evaluating the importance of different parameters in specific marine events or processes”, the method also includes the following steps:

[0029] Clarify the scene semantic information of the specific marine event or process that needs to be evaluated, including the type of event, the geographical environment in which it occurs, the temporal background, related elements and their interrelationships, as the basis for subsequent algorithm design and parameter evaluation;

[0030] According to the defined scene semantic information, relevant ocean observation data, historical records, and experimental data are collected, and preprocessed and analyzed to obtain reliable data sets for algorithm training and parameter evaluation;

[0031] Based on the scene semantic information and the collected data, a maximum correlation algorithm is designed to quantify the correlation of different parameters in a specific marine event or process and identify the key parameters that have the greatest impact on the event or process;

[0032] The maximum correlation algorithm was trained using the preprocessed data set, and the accuracy and reliability of the algorithm were verified through cross-validation and model evaluation methods;

[0033] The trained maximum correlation algorithm is used to evaluate the importance of different parameters in a specific marine event or process. By calculating the correlation score and contribution index of the parameters, the key parameters that have the greatest impact on the event or process are identified.

[0034] Among them, in “matching the extracted parameter information with the ocean structured data through multi-dimensional correlation analysis to achieve multi-dimensional correlation of information”, the method includes the following steps:

[0035] Clarify the source of the extracted parameter information, and at the same time, clarify the source of the marine structured data, including marine environmental databases, historical records, and professional research reports;

[0036] Extract relevant parameter information and perform preprocessing, including data cleaning, data format conversion and data standardization. The preprocessed parameter information is matched with the ocean structured data.

[0037] Build a framework for ocean structured data. The framework contains all relevant ocean data fields, including time, location, temperature, salinity, and current speed. Each field has a clear data type and value range.

[0038] After extracting and preprocessing the parameter information and constructing the marine structured data framework, multi-dimensional correlation analysis is performed, including identifying the common features between the parameter information and the marine structured data, establishing association rules, and calculating the degree of correlation. Through multi-dimensional correlation analysis, the intrinsic connection between the parameter information and the marine structured data is revealed;

[0039] According to the results of multi-dimensional association analysis, the extracted parameter information is matched with the ocean structured data to realize the multi-dimensional association of information, including mapping the parameter information to the corresponding fields in the ocean structured data framework and establishing an association table or database.

[0040] Among them, in “outputting optimized extraction content according to the analysis results, including key parameters, trend prediction, and anomaly detection”, the method includes the following steps:

[0041] Organize the results of all previous analysis steps, including the outputs of data preprocessing, multi-dimensional correlation analysis, parameter evaluation, etc.

[0042] From the collated analysis results, extract the key parameters that have the greatest impact on the target ocean event or process;

[0043] Utilize the extracted key parameters, combined with time series analysis and machine learning, to build a trend prediction model;

[0044] Based on trend prediction, through real-time monitoring of key parameters and comparative analysis with historical data, possible abnormal values ​​or abnormal situations can be identified;

[0045] Combine extracted key parameters, trend prediction results, and anomaly detection results into a complete report or dataset.

[0046] The invention discloses a method for extracting marine multi-parameter information based on term semantics. On the basis of the existing marine term semantic knowledge system, the invention introduces advanced clustering algorithms and classification algorithms to conduct deep learning and optimization of marine terms. The invention utilizes the term expression ability and Chinese phrase structure rules to optimize the K-nearest neighbor algorithm to identify marine entities and their attributes in the text. The invention combines domain-specific dictionaries and context information to extract entities and attributes, and standardizes the extracted entities and attributes. The invention combines scene semantic information to design a maximum correlation algorithm to evaluate the importance of different parameters in specific marine events or processes. The invention matches the extracted parameter information with marine structured data through multi-dimensional correlation analysis to realize the multi-dimensional correlation of information. According to the analysis results, the optimized extraction content is output, including key parameters, trend prediction, and anomaly detection. The optimized clustering algorithm performs preliminary classification on water quality parameters: temperature, salinity, and dissolved oxygen, and then extracts key entities and attributes in the text using the KNN algorithm combined with Chinese phrase structure rules. Finally, the maximum correlation algorithm is used to screen out the parameters that have the greatest impact on water quality conditions in combination with the ecosystem characteristics and seasonal changes of the sea area, and an intuitive report and suggestions are generated to provide decision-making basis for managers. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art are briefly introduced below.

[0048] Figure 1 It is a flow chart of a method for extracting ocean multi-parameter information based on terminology semantics of the present invention.

[0049] Figure 2 This is a flowchart of the present invention, which introduces advanced clustering algorithms and classification algorithms on the basis of the existing marine terminology semantic knowledge system to conduct deep learning and optimization of marine terminology.

[0050] Figure 3 It is a flow chart of the present invention for utilizing the terminology expression ability and Chinese phrase structure rules to optimize the K-nearest neighbor algorithm and identify the marine entities and their attributes in the text.

[0051] Figure 4 It is a flowchart of the present invention for extracting entities and attributes by combining domain-specific dictionaries and context information, and for standardizing the extracted entities and attributes.

[0052] Figure 5 It is a flow chart of the present invention for combining scene semantic information, designing a maximum correlation algorithm, and evaluating the importance of different parameters in a specific ocean event or process.

[0053] Figure 6It is a flowchart of the present invention that matches the extracted parameter information with the ocean structured data through multi-dimensional correlation analysis to achieve multi-dimensional correlation of information.

[0054] Figure 7 The present invention outputs the optimized extraction content according to the analysis results, including a flow chart of key parameters, trend prediction and anomaly detection. DETAILED DESCRIPTION

[0055] Embodiments of the present invention are described in detail below. Examples of the embodiments are shown in the accompanying drawings. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present invention, but should not be construed as limiting the present invention.

[0056] See also Figures 1 to 7 , Figure 1 is a flow chart of a method for extracting ocean multi-parameter information based on terminology semantics of the present invention, Figure 2 This is a flowchart of the present invention that introduces advanced clustering algorithms and classification algorithms on the basis of the existing marine terminology semantic knowledge system to conduct deep learning and optimization of marine terminology. Figure 3 This is a flowchart of the present invention that utilizes the terminology expression ability and Chinese phrase structure rules to optimize the K-nearest neighbor algorithm to identify marine entities and their attributes in texts. Figure 4 It is a flowchart of the present invention for extracting entities and attributes by combining domain-specific dictionaries and context information, and for standardizing the extracted entities and attributes. Figure 5 It is a flowchart of the present invention that combines scene semantic information, designs a maximum correlation algorithm, and evaluates the importance of different parameters in a specific marine event or process. Figure 6 It is a flowchart of the present invention that matches the extracted parameter information with the ocean structured data through multi-dimensional correlation analysis to achieve multi-dimensional correlation of information. Figure 7 The present invention outputs the optimized extraction content according to the analysis results, including a flow chart of key parameters, trend prediction and anomaly detection.

[0057] A method for extracting ocean multi-parameter information based on terminology semantics comprises the following steps:

[0058] S1: Based on the existing semantic knowledge system of marine terminology, advanced clustering and classification algorithms are introduced to conduct deep learning and optimization of marine terminology;

[0059] S11: Collect marine terms and their related descriptions, integrate the data into the database, clean and pre-process the data to remove duplicate, erroneous or irrelevant information;

[0060] S12: Based on the existing semantic knowledge system of marine terminology, clustering and classification algorithms are used, and the algorithms are preliminarily applied to the preprocessed data, and the performance of the algorithms is evaluated through cross-validation;

[0061] S13: Based on the results of the preliminary implementation of the algorithm, a deep learning model was constructed to perform deep feature extraction and classification of marine terms, and the model was trained using the preprocessed data;

[0062] S14: Combine the output of the deep learning model with the existing semantic knowledge system of marine terminology to update and improve the terminology library and explore new relationships between terms.

[0063] In this embodiment:

[0064] Step 1: Data preparation and preprocessing

[0065] Collect and integrate data: Collect marine terms and their related descriptions from various marine scientific literature, databases, professional websites and other channels, including the definition, classification, contextual usage and other information of the terms.

[0066] Data cleaning and standardization: Clean the collected data to remove duplicate, erroneous or irrelevant information, and establish unified data formats and standards, such as spelling rules for terms, definition formats, etc.

[0067] Build an initial terminology database: organize the cleaned data into an initial marine terminology semantic knowledge system, including terms and their attributes, relationships, etc.

[0068] Step 2: Select and optimize the algorithm

[0069] Research advanced algorithms: Research current advanced clustering algorithms (such as DBSCAN, spectral clustering, hierarchical clustering, etc.) and classification algorithms (such as support vector machines, random forests, deep learning models, etc.) to understand their principles, advantages, and applicable scenarios.

[0070] Algorithm selection and preliminary testing: According to the characteristics and requirements of marine terminology, select appropriate clustering algorithms and classification algorithms for preliminary testing. During testing, indicators such as cross-validation, accuracy, and recall can be used to evaluate the performance of the algorithm.

[0071] Algorithm optimization: Based on the results of preliminary tests, perform optimization operations such as parameter adjustment and feature selection on the algorithm to improve the performance and accuracy of the algorithm.

[0072] Step 3: Deep learning model training

[0073] Build deep learning models: Based on the optimized clustering algorithm and classification algorithm, build deep learning models such as convolutional neural network (CNN), recurrent neural network (RNN) or Transformer for deep learning and feature extraction of marine terms.

[0074] Model training: Use the preprocessed ocean terminology data to train the deep learning model so that it can learn the semantic relationship and features between terms. During the training process, it is necessary to continuously adjust the model parameters and learning rate to improve the generalization ability of the model.

[0075] Model evaluation and adjustment: Use the test data set to evaluate the trained model and measure the performance of the model through indicators such as accuracy, recall, and F1 score. Based on the evaluation results, further adjust and optimize the model.

[0076] Step 4: Optimize the terminology knowledge system

[0077] Update the terminology database: Integrate the features and relationships extracted by the deep learning model into the existing marine terminology semantic knowledge system to update and improve the terminology database.

[0078] Terminology relationship mining: Utilize the output of deep learning models to mine new relationships between terms, such as similarity, correlation, causality, etc., to further enrich the terminology knowledge system.

[0079] Knowledge system verification: The optimized terminology knowledge system is verified through expert review, field testing, etc. to ensure its accuracy and reliability.

[0080] Step 5: Continuous Update and Maintenance

[0081] Real-time monitoring and updating: Establish a real-time monitoring system to track emerging marine terminology and semantic relationships, and update the new information to the terminology knowledge system in a timely manner.

[0082] Model iteration and optimization: With the development of marine science and the continuous increase of data, deep learning models are regularly iterated and optimized to adapt to new data and needs.

[0083] User feedback and interaction: Encourage users to participate in the process of improving the terminology knowledge system and optimizing the model, and further improve the accuracy and practicality of the terminology semantic knowledge system through user feedback.

[0084] S2: Using the terminology expression ability and Chinese phrase structure rules, we optimize the K-nearest neighbor algorithm to identify marine entities and their attributes in the text:

[0085] S21: Analyze the phrase structure of Chinese, extract common phrase structure rules based on a large corpus, and form a rule base as a grammatical basis for identifying marine entities and their attributes;

[0086] S22: Combining the expression ability of terms and the structural rules of Chinese phrases, design feature selection and representation methods to convert text data into feature vectors that can be processed by the algorithm. At the same time, adjust the parameters of the KNN algorithm and introduce a weight mechanism;

[0087] S23: Use the preprocessed text data and the annotated marine entities and their attribute information to train the optimized KNN model, and use the model to identify and extract marine entities and their attributes from new text data to form structured information, providing a basis for subsequent analysis and application;

[0088] S24: Manually verify the recognition results, and iteratively optimize the algorithms and models based on the verification results.

[0089] In this embodiment:

[0090] Step 1: Termbase construction and preprocessing

[0091] Collect marine terminology: Collect marine-related terminology from authoritative marine science literature, databases, professional websites and other channels, including marine life, marine phenomena, marine resources, marine environment and other aspects.

[0092] Terminology standardization: Standardize the collected terms, including unified spelling, definition, classification, etc., to ensure the accuracy and consistency of the terms.

[0093] Improve the expressiveness of terms: Evaluate and improve the expressiveness of terms based on the frequency of use of terms, context, and other information, and provide rich features for subsequent algorithm applications.

[0094] Step 2: Extraction of Chinese phrase structure rules

[0095] Analyze Chinese phrase structure: conduct in-depth research on Chinese phrase structure, including subject-predicate structure, verb-object structure, attributive-predicate structure, etc., and understand the expression forms and rules of these structures in the text.

[0096] Extract structural rules: Based on a large amount of Chinese text corpus, common Chinese phrase structure rules are extracted to form a rule base, which provides a grammatical basis for the subsequent identification of marine entities and their attributes.

[0097] Step 3: K-nearest neighbor algorithm optimization

[0098] Feature selection and representation: Combining the expressive power of terms and the structural rules of Chinese phrases, we design appropriate feature selection and representation methods to convert text data into feature vectors that can be processed by the algorithm.

[0099] Algorithm parameter adjustment: According to the characteristics of marine entity recognition, the parameters of the KNN algorithm are adjusted, such as the selection of K value, distance measurement method, etc., to improve the recognition accuracy and efficiency of the algorithm.

[0100] Introducing a weight mechanism: Considering the impact of different features on the recognition results, a weight mechanism is introduced to weight important features to further improve the recognition ability of the algorithm.

[0101] Step 4: Identification of ocean entities and their attributes

[0102] Training model: Use the preprocessed text data and annotated marine entities and their attribute information to train the optimized KNN model.

[0103] Model evaluation and optimization: Evaluate the performance of the model through cross-validation, accuracy, recall and other indicators, and further optimize the model based on the evaluation results.

[0104] Identification and extraction: Use the trained model to identify and extract ocean entities and their attributes from new text data to form structured information.

[0105] Step 5: Result verification and feedback

[0106] Manual verification: Manual verification of the recognition results to ensure their accuracy and reliability.

[0107] Feedback and iteration: Based on the verification results, the algorithm and model are iteratively optimized to improve recognition accuracy and generalization ability.

[0108] Continuous updates: With the development of marine science and the continuous increase of text data, the terminology library and phrase structure rule library are continuously updated to ensure the timeliness and accuracy of the algorithm.

[0109] S3: Entity and attribute extraction is performed by combining domain-specific dictionaries and context information, and the extracted entities and attributes are standardized;

[0110] S31: Based on the target domain data, a domain-specific dictionary containing key entities and attributes is constructed. The dictionary will serve as a reference for entity and attribute extraction;

[0111] S32: preprocessing the input text, including removing noise from punctuation marks and special characters, word segmentation, and part-of-speech tagging;

[0112] S33: Use the constructed domain-specific dictionary and the contextual information of the text to extract entities and attributes. By matching keywords and phrases in the dictionary and analyzing the grammatical structure and semantic relationship in the text, target entities and attributes are identified.

[0113] S34: Standardize the extracted entities and attributes, including unifying naming conventions, removing synonyms and near-synonyms, and determining the type and value range of attributes to ensure that the extracted entities and attributes are consistent and comparable in subsequent analysis and application.

[0114] In this embodiment:

[0115] Step 1: Build a domain-specific dictionary

[0116] First, a domain-specific dictionary containing key entities and attributes is constructed based on the expertise of the target domain. This dictionary will serve as a reference for entity and attribute extraction to ensure the accuracy and professionalism of the extraction.

[0117] Step 2: Text preprocessing and word segmentation

[0118] Preprocess the input text, including removing noise (such as punctuation, special characters, etc.), word segmentation and part-of-speech tagging. This step is the basis for subsequent entity and attribute extraction, ensuring the cleanliness and readability of the text data.

[0119] Step 3: Extraction based on dictionary and context information

[0120] Entities and attributes are extracted by using the constructed domain-specific dictionary and combining the contextual information of the text. The target entities and attributes are identified by matching the keywords and phrases in the dictionary and analyzing the grammatical structure and semantic relationships in the text.

[0121] Step 4: Entity and attribute standardization

[0122] Standardize the extracted entities and attributes, including unifying naming conventions, removing synonyms and near-synonyms, determining the type and value range of attributes, etc. The purpose of this step is to ensure that the extracted entities and attributes are consistent and comparable in subsequent analysis and application.

[0123] Step 5: Verification and Correction

[0124] Finally, verify and correct the standardized entities and attributes. Ensure the accuracy and completeness of extraction and standardization through manual review or comparison with other reliable data sources. Make timely corrections and supplements to any errors or omissions found.

[0125] S4: Combine scene semantic information to design a maximum correlation algorithm to evaluate the importance of different parameters in specific marine events or processes;

[0126] S41: Clarify the scenario semantic information of the specific marine event or process to be evaluated, including the type of event, the geographical environment in which it occurred, the temporal context, the relevant elements and their interrelationships, as the basis for subsequent algorithm design and parameter evaluation;

[0127] S42: Based on the defined scene semantic information, collect relevant ocean observation data, historical records, and experimental data, and perform preprocessing and analysis to obtain reliable data sets for algorithm training and parameter evaluation;

[0128] S43: Based on the scene semantic information and the collected data, design a maximum correlation algorithm that quantifies the correlation of different parameters in a specific marine event or process and identifies the key parameters that have the greatest impact on the event or process;

[0129] S44: Use the preprocessed data set to train the maximum correlation algorithm, and verify the accuracy and reliability of the algorithm through cross-validation and model evaluation methods;

[0130] S45: Use the trained maximum correlation algorithm to evaluate the importance of different parameters in a specific marine event or process, and identify the key parameters that have the greatest impact on the event or process by calculating the correlation score and contribution index of the parameters.

[0131] In this embodiment,

[0132] Step 1: Define scene semantic information

[0133] First, the scene semantic information of the specific marine event or process to be evaluated is clarified, including the type of event, the geographical environment in which it occurred, the time background, related elements (such as ocean temperature, salinity, current velocity, etc.) and their interrelationships. This information will serve as the basis for subsequent algorithm design and parameter evaluation.

[0134] Step 2: Collect and analyze data

[0135] According to the defined scene semantic information, relevant ocean observation data, historical records, experimental data, etc. are collected, preprocessed and analyzed. This step aims to obtain a reliable data set for algorithm training and parameter evaluation to ensure the accuracy and effectiveness of subsequent evaluations.

[0136] Step 3: Design the maximum correlation algorithm

[0137] Based on the scene semantic information and the collected data, a maximum correlation algorithm is designed. The algorithm should be able to quantify the correlation of different parameters in a specific marine event or process and identify the key parameters that have the greatest impact on the event or process. The design of the algorithm should take into account the weights of the parameters, their interrelationships, and the dynamic changes of their impact on the event or process.

[0138] Step 4: Algorithm training and verification

[0139] The maximum correlation algorithm is trained using the preprocessed data set, and the accuracy and reliability of the algorithm are verified through cross-validation, model evaluation, etc. This step is to ensure the applicability and robustness of the algorithm in different scenarios and provide a solid foundation for subsequent parameter evaluation.

[0140] Step 5: Evaluate parameter importance

[0141] Finally, the trained maximum correlation algorithm is used to evaluate the importance of different parameters in a specific marine event or process. By calculating the correlation score, contribution and other indicators of the parameters, the key parameters that have the greatest impact on the event or process are identified, providing decision support for subsequent marine scientific research, disaster warning, resource management, etc. At the same time, according to the evaluation results, the algorithm is adjusted and optimized as necessary to improve the accuracy and practicality of the evaluation.

[0142] S5: Through multi-dimensional correlation analysis, the extracted parameter information is matched with the ocean structured data to achieve multi-dimensional correlation of information.

[0143] S51: Clarify the source of the extracted parameter information and the source of the marine structured data, including marine environment databases, historical records, and professional research reports;

[0144] S52: extract relevant parameter information and perform preprocessing, including data cleaning, data format conversion and data standardization, and match the preprocessed parameter information with the ocean structured data;

[0145] S53: Build a framework for ocean structured data. The framework contains all relevant ocean data fields, including time, location, temperature, salinity, and current velocity. Each field has a clear data type and value range.

[0146] S54: After extracting and preprocessing the parameter information and constructing the marine structured data framework, multi-dimensional correlation analysis is performed, including identifying the common features between the parameter information and the marine structured data, establishing association rules, and calculating the degree of correlation. Through multi-dimensional correlation analysis, the intrinsic connection between the parameter information and the marine structured data is revealed;

[0147] S55: According to the results of the multi-dimensional association analysis, the extracted parameter information is matched with the ocean structured data to achieve multi-dimensional association of the information, including mapping the parameter information to the corresponding fields in the ocean structured data framework and establishing an association table or database.

[0148] In this embodiment,

[0149] Step 1: Identify the source of parameter information and ocean structured data

[0150] First, it is necessary to clarify the source of the extracted parameter information, which may come from sensor data, remote sensing observations, laboratory analysis results, etc. At the same time, it is also necessary to clarify the source of marine structured data, which usually includes marine environmental databases, historical records, professional research reports, etc. Clarifying the data source is the basis for subsequent matching and association analysis.

[0151] Step 2: Extract and preprocess parameter information

[0152] Next, relevant parameter information is extracted from various sources and preprocessed. This includes steps such as data cleaning (removing outliers, filling missing values, etc.), data format conversion (ensuring the consistency of data format), and data standardization (converting data into comparable forms). The preprocessed parameter information will be easier to match with ocean structured data.

[0153] Step 3: Build an ocean structured data framework

[0154] In order to match the parameter information, a framework of ocean structured data needs to be constructed. This framework should contain all relevant ocean data fields, such as time, location, temperature, salinity, current speed, etc., and ensure that each field has a clear data type and value range. The constructed data framework will serve as the basis for subsequent matching and association analysis.

[0155] Step 4: Multi-dimensional correlation analysis

[0156] After extracting and preprocessing the parameter information and constructing the marine structured data framework, multi-dimensional association analysis is performed. This includes identifying the common features between the parameter information and the marine structured data, establishing association rules, and calculating the degree of association. Through multi-dimensional association analysis, the intrinsic connection between the parameter information and the marine structured data can be revealed, providing guidance for subsequent information matching.

[0157] Step 5: Realize multi-dimensional association of information

[0158] Finally, according to the results of multi-dimensional association analysis, the extracted parameter information is matched with the marine structured data to achieve multi-dimensional association of information. This includes mapping the parameter information to the corresponding fields in the marine structured data framework, establishing an association table or database, and other steps. By achieving multi-dimensional association of information, it is easier to query, analyze and utilize these data, providing strong support for marine scientific research, environmental protection, disaster warning and other fields.

[0159] S6: Based on the analysis results, output the optimized extraction content, including key parameters, trend prediction, and anomaly detection.

[0160] S61: Organize the results obtained from all previous analysis steps, including the outputs of data preprocessing, multi-dimensional correlation analysis, parameter evaluation, etc.;

[0161] S62: Extract the key parameters that have the greatest impact on the target ocean event or process from the collated analysis results;

[0162] S63: Use the extracted key parameters, combined with time series analysis and machine learning, to build a trend prediction model;

[0163] S64: Based on trend prediction, identify possible abnormal values ​​or abnormal situations through real-time monitoring of key parameters and comparative analysis with historical data;

[0164] S65: Integrate the extracted key parameters, trend prediction results, and anomaly detection results into a complete report or data set.

[0165] In this embodiment:

[0166] Step 1: Organize and analyze the results

[0167] First, the results of all previous analysis steps are collated. This includes but is not limited to the output of data preprocessing, multi-dimensional correlation analysis, parameter evaluation, etc. Ensure the accuracy and completeness of all analysis results, and lay the foundation for the subsequent extraction of optimization content.

[0168] Step 2: Extract key parameters

[0169] From the collated analysis results, the key parameters that have the greatest impact on the target marine event or process are extracted based on methods such as importance assessment and correlation analysis. These parameters should be representative and able to reflect the main characteristics and changing trends of the marine environment.

[0170] Step 3: Build a trend prediction model

[0171] The trend prediction model is constructed by using the extracted key parameters combined with time series analysis, machine learning and other methods. The model should be able to predict the changing trend of key parameters in the future and provide decision support for marine scientific research, disaster warning, etc.

[0172] Step 4: Implement anomaly detection

[0173] Based on trend prediction, anomaly detection is implemented. Through real-time monitoring of key parameters and comparative analysis with historical data, possible abnormal values ​​or abnormal situations can be identified. These abnormal values ​​may indicate sudden changes or potential risks in the marine environment and require special attention.

[0174] Step 5: Output optimized extracted content

[0175] Finally, the extracted key parameters, trend prediction results, and anomaly detection results are integrated into a complete report or data set. Ensure that the content of the report or data set is clear, accurate, and easy to understand. At the same time, according to actual needs, the optimized extracted content is presented in an appropriate form (such as tables, charts, reports, etc.) to facilitate subsequent analysis and application.

[0176] What is disclosed above is only one or more preferred embodiments of the present application, and cannot be used to limit the scope of rights of the present application. Ordinary technicians in this field can understand that all or part of the processes of implementing the above embodiments and equivalent changes made according to the claims of the present application are still within the scope covered by the present application.

Claims

1. A method for extracting ocean multi-parameter information based on terminology semantics, characterized in that: The steps include: Based on the existing marine terminology semantic knowledge system, advanced clustering and classification algorithms are introduced to conduct in-depth learning and optimization of marine terminology; By using the expressive power of terms and the rules of Chinese phrase structure, the K-nearest neighbor algorithm is optimized to identify marine entities and their attributes in the text; Entity and attribute extraction is performed by combining domain-specific dictionaries and contextual information, and the extracted entities and attributes are standardized; Combine scene semantic information to design maximum relevance algorithms to assess the importance of different parameters in specific marine events or processes; Through multi-dimensional correlation analysis, the extracted parameter information is matched with the ocean structured data to achieve multi-dimensional correlation of information; Based on the analysis results, the optimized extraction content is output, including key parameters, trend prediction, and anomaly detection.

2. The method for extracting ocean multi-parameter information based on terminology semantics according to claim 1, characterized in that: In "Based on the existing marine terminology semantic knowledge system, advanced clustering algorithms and classification algorithms are introduced to conduct deep learning and optimization of marine terminology", the method includes the following steps: Collect marine terms and their related descriptions, integrate the data into the database, clean and pre-process the data to remove duplicate, erroneous or irrelevant information; Based on the existing semantic knowledge system of marine terminology, clustering and classification algorithms are used, and the algorithms are preliminarily applied to the preprocessed data, and the performance of the algorithms is evaluated through cross-validation; Based on the results of the initial implementation of the algorithm, a deep learning model was constructed to perform deep feature extraction and classification of marine terms, and the model was trained using the preprocessed data; Combine the output of the deep learning model with the existing semantic knowledge system of marine terminology to update and improve the terminology library and explore new relationships between terms.

3. The method for extracting ocean multi-parameter information based on terminology semantics according to claim 2, characterized in that: In "Using terminology expression ability and Chinese phrase structure rules to optimize K-nearest neighbor algorithm and identify marine entities and their attributes in text", the method includes the following steps: Analyze the phrase structure of Chinese, extract common phrase structure rules based on a large corpus, and form a rule base as a grammatical basis for identifying marine entities and their attributes; Combining the expressive power of terms and the structural rules of Chinese phrases, we designed a feature selection and representation method to convert text data into feature vectors that can be processed by the algorithm. At the same time, we adjusted the parameters of the KNN algorithm and introduced a weight mechanism. The preprocessed text data and the annotated marine entities and their attribute information are used to train the optimized KNN model. The model is used to identify and extract marine entities and their attributes from new text data to form structured information, which provides a basis for subsequent analysis and application. The recognition results are manually verified, and the algorithms and models are iteratively optimized based on the verification results.

4. The method for extracting ocean multi-parameter information based on terminology semantics according to claim 1, characterized in that: In “extracting entities and attributes by combining domain-specific dictionaries and context information, and standardizing the extracted entities and attributes”, the steps include: Based on the target domain data, a domain-specific dictionary containing key entities and attributes is constructed. The dictionary will serve as a reference for entity and attribute extraction. Preprocess the input text, including removing punctuation and special character noise, word segmentation, and part-of-speech tagging; Utilize the constructed domain-specific dictionary and the contextual information of the text to extract entities and attributes. By matching keywords and phrases in the dictionary and analyzing the grammatical structure and semantic relationships in the text, target entities and attributes are identified. The extracted entities and attributes are standardized, including unified naming conventions, removal of synonyms and near-synonyms, and determination of attribute types and value ranges, to ensure that the extracted entities and attributes are consistent and comparable in subsequent analysis and applications.

5. The method for extracting ocean multi-parameter information based on terminology semantics according to claim 1, characterized in that: In "Combining scene semantic information, designing a maximum correlation algorithm, and evaluating the importance of different parameters in a specific marine event or process", the method also includes the following steps: Clarify the scene semantic information of the specific marine event or process that needs to be evaluated, including the type of event, the geographical environment in which it occurs, the time background, related elements and their interrelationships, as the basis for subsequent algorithm design and parameter evaluation; According to the defined scene semantic information, relevant ocean observation data, historical records, and experimental data are collected, and preprocessed and analyzed to obtain reliable data sets for algorithm training and parameter evaluation; Based on the scene semantic information and the collected data, a maximum correlation algorithm is designed to quantify the correlation of different parameters in a specific marine event or process and identify the key parameters that have the greatest impact on the event or process; The maximum correlation algorithm was trained using the preprocessed data set, and the accuracy and reliability of the algorithm were verified through cross-validation and model evaluation methods; The trained maximum correlation algorithm is used to evaluate the importance of different parameters in a specific marine event or process. By calculating the correlation score and contribution index of the parameters, the key parameters that have the greatest impact on the event or process are identified.

6. The method for extracting ocean multi-parameter information based on terminology semantics according to claim 1, characterized in that: In "matching the extracted parameter information with the ocean structured data through multi-dimensional correlation analysis to achieve multi-dimensional correlation of information", the method includes the following steps: Clarify the source of the extracted parameter information, and at the same time, clarify the source of the marine structured data, including marine environmental databases, historical records, and professional research reports; Extract relevant parameter information and perform preprocessing, including data cleaning, data format conversion and data standardization. The preprocessed parameter information is matched with the ocean structured data. Build a framework for ocean structured data. The framework contains all relevant ocean data fields, including time, location, temperature, salinity, and current speed. Each field has a clear data type and value range. After extracting and preprocessing the parameter information and constructing the marine structured data framework, multi-dimensional correlation analysis is performed, including identifying the common features between the parameter information and the marine structured data, establishing association rules, and calculating the degree of correlation. Through multi-dimensional correlation analysis, the intrinsic connection between the parameter information and the marine structured data is revealed; According to the results of multi-dimensional association analysis, the extracted parameter information is matched with the ocean structured data to realize the multi-dimensional association of information, including mapping the parameter information to the corresponding fields in the ocean structured data framework and establishing an association table or database.

7. The method for extracting ocean multi-parameter information based on terminology semantics according to claim 1, characterized in that: In "outputting optimized extraction content, including key parameters, trend prediction, and anomaly detection based on the analysis results", the method includes the following steps: Organize the results of all previous analysis steps, including the outputs of data preprocessing, multi-dimensional correlation analysis, parameter evaluation, etc. From the collated analysis results, extract the key parameters that have the greatest impact on the target ocean event or process; Utilize the extracted key parameters, combined with time series analysis and machine learning, to build a trend prediction model; Based on trend prediction, through real-time monitoring of key parameters and comparative analysis with historical data, possible abnormal values ​​or abnormal situations can be identified; Combine extracted key parameters, trend prediction results, and anomaly detection results into a complete report or dataset.

Images