Modeling method of industrial data processing model based on natural language interaction and declarative language binding

Abstract

The invention discloses a modeling method of an industrial data processing model based on natural language interaction and declarative language binding. The modeling method comprises the following steps: firstly, acquiring industrial data, then pre-training an interactive model, converting all step parameters and all specific configuration information of modeling required by the industrial data into declarative language commands through continuous interaction between the pre-trained interactive model and a user, and performing modeling on the industrial data according to the declarative language commands. The method comprises the steps that firstly, a declarative language command is sent to a modeling platform, then modeling is conducted on the modeling platform according to the declarative language command, then a modeled data model is trained, the trained data model is deployed according to the declarative language command, and the data model can be used after deployment is completed. The method has high automation and flexibility, the modeling process can be greatly simplified, the modeling efficiency is improved, and the sense of participation and operability of the user in the modeling process are enhanced.

Description

Technical Field

[0001] The present invention relates to the fields of natural language processing and declarative programming, and particularly relates to a modeling method for an industrial data processing model based on natural language interaction and declarative language binding. Background Art

[0002] With the continuous growth of industrial data, how to extract effective information from huge data sets and use it for actual decision-making has become an important challenge faced by the modern industrial and commercial fields. And with the rapid development of artificial intelligence technology, especially in the fields of machine learning and deep learning, the complexity of data modeling and model training has been increasing. Existing modeling platforms usually rely on specific programming languages and underlying technologies, and achieving cross-platform data processing and model training requires a high technical threshold. Users need to master the operation syntax and interfaces of each platform, increasing the complexity of technical operations and development costs.

[0003] At the same time, although existing natural language interaction systems can provide convenient user interfaces, the operations between multiple platforms still lack unified standards. When users use different modeling platforms, they often face technical obstacles such as the inability to seamlessly migrate models and cross-platform calls, which limits the wide application of technologies and the flexibility of systems. Therefore, how to achieve compatibility and migration between natural language interaction and different platforms through a unified declarative language has become a technical problem to be solved urgently in the field of intelligent modeling. Summary of the Invention

[0004] The present invention proposes a modeling method for an industrial data processing model based on natural language interaction and declarative language binding, aiming to enable users to conveniently perform data processing, model training, evaluation, and deployment operations on multiple modeling platforms through the combination of natural language and declarative language, simplify complex technical implementations, and reduce the technical threshold for users.

[0005] The technical solution of the present invention includes the following steps:

[0006] S1. Obtain industrial data through the operation logs of industrial machinery and equipment and sensors installed on industrial machinery and equipment.

[0007] S2. Use an interactive model as a medium for interaction between users and the data modeling platform, and perform pre-training on the interactive model to transform natural language into a declarative language, obtaining a pre-trained interactive model.

[0008] The pre-training of transforming natural language into a declarative language enables the interactive model to transform natural language into a declarative language. The declarative language is a programming paradigm in a computer that only expresses the logic of a program without describing the program control flow. In specific implementations, the present invention can adopt a declarative language in the JSON data format.

[0009] S3. The user inputs industrialized data and their own modeling requirements into the pre-trained interactive model for analysis and processing. The interactive model interacts with the user based on the results of the analysis and processing to obtain a declarative language command for the modeling steps.

[0010] S4. In the computer, select the required data model for modeling according to the user's own modeling requirements, perform modeling processing on the selected data model on the data modeling platform according to the declarative language command for the modeling steps to obtain a completed data model, and input the industrialized data into the completed data model for training to obtain a trained data model.

[0011] S5. Deploy the trained data model on the corresponding platform according to the declarative language command for the modeling steps obtained in step S3, obtain new industrialized data in real time, input the new industrialized data into the data model on the platform for processing, and obtain the results of the data model processing in real time.

[0012] The results of the data model processing obtained in real time are provided to the user in the form of a structured report. The user makes corresponding adjustments and decisions on the industrial equipment according to the results output by the data model.

[0013] The interactive model in step S2 is a GPT model or a DeepSeek model; the data modeling platform in step S2 includes a BML platform and a PAI platform.

[0014] The specific content of step S3 is as follows:

[0015] S31. After the user inputs industrialized data and their own modeling requirements into the pre-trained interactive model for analysis and processing, step parameters are obtained, a problem about the modeling step parameters is generated, and it is returned to the user.

[0016] S32. The user answers the returned problem about the modeling step parameters and inputs it into the interactive model. The user and the interactive model interact continuously until the interactive model extracts all the specific configuration information of the step parameters and then stops interacting with the user. The interactive model converts the obtained step parameters and all the specific configuration information into declarative language to obtain a declarative language command for the modeling steps.

[0017] S33. Perform verification processing on the obtained declarative language command for the modeling steps to obtain a verified declarative language command for the modeling steps.

[0018] The specific content of step S31 is as follows:

[0019] S311. The user inputs industrialized data and their own modeling requirements into the pre-trained interactive model.

[0020] S312. The interactive model extracts the step parameters required for modeling from the input data.

[0021] The step parameters include the category of the data model for modeling, the filling method, normalization, and target variable selection.

[0022] S313. According to the extracted step parameters required for modeling, the interactive model generates modeling step parameter questions and returns the modeling step parameter questions to the user.

[0023] The specific content of step S32 is as follows:

[0024] S321. The user answers according to the returned modeling step parameter questions and inputs them into the interactive model.

[0025] S322. The interactive model further extracts the specific configuration information of the required step parameters from the user's answers. According to the obtained specific configuration information, the interactive model generates new modeling step parameter questions again and returns them to the user.

[0026] S323. Repeat steps S321 - S322 until the interactive model stops interacting with the user after extracting all the specific configuration information of the step parameters.

[0027] S324. The interactive model converts all the obtained step parameters and all the specific configuration information into declarative language, and summarizes all the converted declarative language to obtain the modeling step declarative language command.

[0028] The verification process in step S33 includes consistency check of data types, validity check of parameters, and correctness check of logic flow.

[0029] The specific content of step S4 is as follows:

[0030] S41. According to the user's own modeling requirements, use machine learning algorithms to analyze and obtain the category of the data model required for modeling, and select the data model required for modeling on the data modeling platform.

[0031] S42. According to the modeling step declarative language command after verification processing in the computer, perform modeling processing on the selected data model on the data modeling platform to obtain the modeled data model.

[0032] S43. Input the industrial data into the modeled data model, and train the data model according to the modeling step declarative language command after verification processing to obtain the trained data model.

[0033] The categories of the data models required for modeling selected in step 41 include machine learning models and deep learning models.

[0034] The platforms deployed in step S5 include cloud platforms, enterprise internal servers, and edge devices.

[0035] The beneficial effects of the present invention are as follows:

[0036] 1. The present invention simplifies the modeling process. Users only need to interact with the system through natural language, and the system can automatically identify user requirements and generate corresponding modeling processes. This interaction method reduces the requirements for professional skills, enabling users without a data science background to quickly complete data modeling tasks and improving work efficiency.

[0037] 2. The present invention improves data processing efficiency. Through the automatic generation and execution of declarative languages, data processing processes (such as missing value filling, data normalization, feature selection, etc.) are accurately described and automatically executed. The system can automatically recommend and configure the most suitable data processing pipeline according to the characteristics of the specific dataset and user requirements, thus effectively improving the efficiency and quality of data preprocessing.

[0038] 3. The present invention enhances cross-platform flexibility. The declarative operation language has high generality and can be executed on multiple modeling platforms, including streaming modeling platforms (such as BML, PAI). Users can perform model training, evaluation, and deployment using a unified declarative language interface without being restricted by platform differences, greatly simplifying the complexity of cross-platform operations.

[0039] 4. The present invention improves user experience and business application efficiency. Through standardized and simplified modeling processes, users can quickly achieve data modeling and model deployment. Users can focus more on business decision-making and model application without worrying about underlying technical details, thus improving the overall efficiency of business applications. Description of the Drawings

[0040] Figure 1 It is a schematic diagram of the specific process of the method of the present invention. Detailed Embodiments

[0041] The present invention will be described in more detail below with reference to the drawings and embodiments, but the present invention is not limited thereto. For those of ordinary skill in the art in the technical field, without departing from the principle of the present invention, several improvements and refinements can be made, and these improvements and refinements are also considered within the protection scope of the present invention. The content not detailedly described in this specification belongs to the prior art well-known to those skilled in the art.

[0042] Embodiment 1

[0043] As Figure 1 shown, the specific steps of this embodiment are as follows:

[0044] S1. Obtain industrialization data through the operation logs of industrial machinery and equipment and sensors installed on the industrial machinery and equipment.

[0045] S2. Use an interactive model as the medium for interaction between the user and the data modeling platform, and perform pre-training on the interactive model to transform natural language into a declarative language, obtaining the pre-trained interactive model.

[0046] The interactive model is a GPT model or a DeepSeek model; the data modeling platform includes a BML platform and a PAI platform.

[0047] The pre-training of transforming natural language into a declarative language enables the interactive model to transform natural language into a declarative language.

[0048] The declarative language is a programming paradigm in a computer that only expresses the logic of a program without describing the program control flow. In specific implementation, the declarative language in the JSON data format can be adopted in the present invention.

[0049] S3. The user inputs the industrialization data and their own modeling requirements into the pre-trained interactive model for analysis and processing, and the interactive model conducts interactive processing with the user based on the results of the analysis and processing to obtain a modeling step declarative language command.

[0050] S31. After the user inputs the industrialization data and their own modeling requirements into the pre-trained interactive model for analysis and processing, step parameters are obtained, a modeling step parameter question is generated, and it is returned to the user.

[0051] S311. The user inputs the industrialization data and their own modeling requirements into the pre-trained interactive model.

[0052] S312. The interactive model extracts the step parameters required for modeling from the input data.

[0053] The step parameters include the category of the data model for modeling, the filling method, normalization, and target variable selection.

[0054] In specific implementation, the step parameters also include uploading data and data processing schemes, etc. The data processing scheme is specifically data cleaning methods and feature engineering, etc.

[0055] S313. Based on the extracted step parameters required for modeling, the interactive model generates a modeling step parameter question and returns the modeling step parameter question to the user to more accurately collect the user's requirements.

[0056] S32. The user answers according to the returned modeling step parameter questions and inputs them into the interactive model. The user and the interactive model interact continuously until the interactive model stops interacting with the user after extracting all the specific configuration information of the step parameters. The interactive model converts the obtained step parameters and all the specific configuration information into declarative language to obtain the modeling step declarative language command.

[0057] S321. The user answers according to the returned modeling step parameter questions and inputs them into the interactive model.

[0058] S322. The interactive model further extracts the specific configuration information of the required step parameters from the user's answer. According to the obtained specific configuration information, the interactive model generates new modeling step parameter questions again and returns them to the user.

[0059] S323. Repeat steps S321 - S322 until the interactive model stops interacting with the user after extracting all the specific configuration information of the step parameters.

[0060] In a specific implementation, the modeling step parameter questions generated in step S313 are: When the filling method is what, the user answers: Perform mean filling. The interactive model generates again: There are four mean filling methods. Which one do you choose? The user answers: Choose the fourth mean filling as the filling method until the modeling step parameter questions are clearly defined between the interactive model and the user.

[0061] S324. The interactive model converts all the obtained step parameters and all the specific configuration information into declarative language, and summarizes all the converted declarative language to obtain the modeling step declarative language command.

[0062] S33. Perform verification processing on the obtained modeling step declarative language command to obtain the verified modeling step declarative language command. The verification processing includes consistency check of data types, validity check of parameters, and correctness check of logic flow.

[0063] S4. Select the required data model for modeling according to the user's own modeling needs in the computer. Perform modeling processing on the selected data model on the data modeling platform according to the modeling step declarative language command to obtain the modeled data model. Input the industrial data into the modeled data model for training to obtain the trained data model.

[0064] S41. Analyze the category of the required data model for modeling according to the user's own modeling needs using machine learning algorithms, and select the required data model for modeling on the data modeling platform.

[0065] The categories of the required data model for modeling include machine learning models and deep learning models.

[0066] S42. In the computer, perform modeling processing on the selected data model on the data modeling platform according to the modeling step declarative language command after verification processing, and obtain the modeled data model.

[0067] S43. Input industrialized data into the modeled data model, and train the data model according to the modeling step declarative language command after verification processing to obtain the trained data model.

[0068] S5. Deploy the trained data model on the corresponding platform according to the modeling step declarative language command obtained in step S3, obtain new industrialized data in real time, input the new industrialized data into the data model on the platform for processing, and obtain the result of data model processing in real time.

[0069] The deployed platforms include cloud platforms, enterprise internal servers, and edge devices.

[0070] The result of data model processing obtained in real time is provided to the user in the form of a structured report, and the user makes corresponding adjustments and decisions on industrial equipment according to the result output by the data model.

[0071] In specific implementation, after the model is deployed, the user can uniformly call the trained model through declarative language commands for operations such as prediction and reasoning, without paying attention to the specific call syntax of the underlying platform. Through the abstraction layer of the declarative language, this method enables the user to seamlessly migrate and call the model between different platforms, simplifies the complexity of cross-platform operations, and realizes the homogenization and standardization of model operations. The declarative language has the following advantages.

[0072] 1. Platform compatibility and flexibility: The design of the declarative syntax ensures its compatibility on different platforms. Whether it is a streaming modeling platform (such as BML, PAI), operations can be achieved through simple adaptation and parsing of configurations. The platform can flexibly configure various parameters of this declarative language according to its own data processing requirements and modeling processes. In this way, the operation syntax between different platforms remains consistent, greatly improving the collaboration efficiency between platforms.

[0073] 2. Portability and cross-platform operation: Through the unified declarative language, the user does not need to pay attention to the technical details and differences of specific platforms, and only needs to configure according to the standard syntax to smoothly migrate and execute between different platforms. For example, through simple parameter adjustment, the modeling process originally based on the BML platform can be seamlessly migrated to the PAI platform. This cross-platform operation ability greatly improves the portability and platform independence of the modeling process.

[0074] 3. Flexible Expansion and Adaptation: Declarative languages not only support existing mainstream modeling platforms but also have the ability to be flexibly expanded. Through platform customization and extension, they can support the access of emerging modeling platforms. With the evolution and diversification of modeling platforms, the extension mechanism of declarative languages enables this method to always maintain the latest technological adaptability, further enhancing the long-term adaptability of the system.

[0075] Through this design, declarative languages can bridge the technological gaps between different platforms, providing consistent, efficient, and flexible support for the modeling process. This not only meets the current technological requirements but also provides strong guarantees for the future expansion of platforms.

[0076] Embodiment 2

[0077] In this embodiment, a modeling and development system is designed to execute the method of the present invention.

[0078] D1: Obtain industrialized datasets

[0079] Industrialized data is obtained through the operation logs of industrial machinery and equipment and sensors installed on industrial machinery and equipment.

[0080] D2: Pre-train the interactive model

[0081] An interactive model is used as the medium for interaction between the user and the modeling platform, and the interactive model is pre-trained to transform natural language into a declarative language to obtain the pre-trained interactive model.

[0082] D3: User data upload and requirement confirmation

[0083] At this stage, the user uploads the industrialized data to be analyzed through an interactive web interface. The system supports the upload of multiple data formats, such as CSV, Excel, or direct connection from a database. During the upload process, the system not only checks the file format and compatibility but also pre-scans the data quality, such as checking for common problems like missing data and duplicate records.

[0084] Once the data upload is complete, the user will be required to describe the modeling requirements in detail. The user can express the requirements by filling out a form or directly having a natural language conversation with the system. The system uses natural language processing technology to understand the specific requirements of the user. The system will parse these inputs, determine the user's analysis goals, and automatically identify the data features related to these goals.

[0085] After the requirements are confirmed, the system further interacts with the user to ensure that all necessary information is collected completely. This may include asking the user if there are specific data processing rules or preferences, such as data normalization methods or specific data exclusion criteria. This process ensures that the system fully understands the user's analysis goals and methodological requirements, so that subsequent data processing and modeling steps can be properly configured.

[0086] D4: Requirement Analysis and Data Processing Pipeline Configuration

[0087] Based on the requirements and data characteristics collected in the previous step, the system automatically recommends a preliminary data processing pipeline. This pipeline is based on the results of data science best practices and previous data exploration, including data cleaning, variable transformation, and preliminary data exploration and analysis steps. The system will display a summary of these steps and explain the necessity and expected effects of each step to the user through a visual interface.

[0088] Next, the system will ask the user to review and approve the recommended data processing process. At this stage, the user can modify or extend the pipeline through a graphical process editor. For example, the system may default to recommending filling missing values with the mean, and the user can choose to change it to filling with the median or adopting a more complex interpolation method. The interactive interface provided by the system allows the user to drag and drop different processing modules, adjust their order, or add new modules to meet specific business needs.

[0089] In addition, the system will provide detailed parameter configuration options for each data processing step. The user can finely adjust the parameters of each step, such as selecting different normalization techniques or adjusting the threshold for outlier detection. The system provides real-time feedback when the user makes these adjustments, predicting the possible impact of the adjustments on the model performance. This includes evaluating the changes in data quality, feature distribution, and the possible impact on the final modeling results.

[0090] D5: Declarative Language Generation and Modeling Process Construction

[0091] After confirming the data processing pipeline, the system will switch to the stage of generating declarative language commands. The core task of this stage is to convert the data processing steps and their configurations agreed by the user into a declarative language, which will be used to guide the subsequent automated modeling and execution process. The system first converts the user's configuration input into an internal data processing script, and then these scripts are compiled into executable declarative language commands, which precisely describe the operation details of each data processing step, including the operation type, required parameters, and the order of operations.

[0092] To ensure the accurate execution of these operations, the system will strictly verify the generated declarative language script. The verification process includes checking the consistency of data types, the validity of parameters, and the correctness of the logical flow. Once the verification is completed, the system will display the final modeling flowchart to the user, which intuitively represents the entire process from data input to model training. The user can also make final modifications at this stage, such as reordering operations or adding additional data processing steps.

[0093] Furthermore, the system of this embodiment also provides a simulation running function, allowing the user to preview the execution results of the declarative language command script to ensure that each step works as expected before actual operation. This not only improves the transparency of the process but also increases the user's confidence in the entire modeling process. The results of the simulation running help the user understand the specific impact of each operation on the data, so as to make more precise adjustments when needed.

[0094] D6: Data Modeling and Training

[0095] After the preparations in the previous steps, the system now begins the actual data modeling and training phase. The file of declarative language commands already contains all the necessary data processing and modeling instructions. The system automatically loads the industrial data according to these instructions, applies all the preprocessing steps, and starts the model training process. During this process, the system uses advanced machine learning algorithms to select the most suitable model according to the user's business needs, such as regression models, classification models, or more complex deep learning models.

[0096] The model training session includes multiple sub-steps, such as feature selection, model parameter optimization, and cross-validation, etc. Each step is precisely controlled by the declarative language. The system monitors the model training process according to the preset performance metrics, such as accuracy, recall, and loss function values, etc., and adjusts the training strategy in real-time to optimize the model performance. For example, if the phenomenon of overfitting is found during the training process, the system may automatically adjust the regularization parameter or suggest modifying the complexity of the model.

[0097] After the model training is completed, the system will automatically perform model evaluation, using the preset test data set to verify the performance of the model. The evaluation results will detail the performance of the model on various metrics and be presented to the user in a visual way, such as confusion matrices, ROC curves, etc. According to the evaluation results, the user can decide whether to accept the current model or continue to optimize. The system supports iterative training of the model. The user can update the model parameters or adjust the data processing flow based on the previous training results and train again to obtain a better model.

[0098] D7: Model Deployment and Cross-Platform Invocation

[0099] After completing model training and validation, the system enters the model deployment phase. The key task in this phase is to deploy the trained model into the production environment for use in actual business. The system configures the automated deployment process using a declarative language to ensure that the model can run seamlessly on any specified platform, including cloud platforms, on-premises enterprise servers, or edge devices. The system adjusts the model configuration according to the characteristics of the target platform, such as memory usage, processing speed, and data access methods.

[0100] Once the model is deployed, users can call the model through a unified API interface for real-time prediction or batch processing tasks. The API interface provided by the system follows the declarative language specification, enabling users to perform the model on different platforms using the same calling method without concerning themselves with the specific implementation details of the underlying platform. In addition, the system also supports model monitoring and maintenance, collecting real-time operation data and performance metrics to help users monitor the model status and automatically trigger the tuning or update process when needed.

[0101] To enhance the availability and stability of the model, the system also implements a fault recovery mechanism and a load balancing function. These functions ensure that the model can still run stably and provide continuous services under high load or limited computing resources. Users can configure the running parameters of the model, such as the processing priority and resource allocation of requests, through the control panel to adapt to different business requirements and operating environments.

[0102] D8: Model Evaluation and Iterative Update

[0103] After the model is deployed, the system will enter the stage of continuous performance evaluation and iterative update. The system will regularly evaluate the model using real-time business data to ensure that its prediction accuracy is consistent with the actual business needs. This evaluation process includes statistical analysis of the model output, comparison with historical data, and concept drift detection, etc.

[0104] If the system detects a decline in model performance or a change in data concepts, such as changes in market dynamics or user behavior patterns, the system will automatically propose suggestions for model updates. The update process can be parameter fine-tuning, model retraining, or adopting a new modeling method. The declarative language tools provided by the system allow users to quickly modify the model configuration and redeploy the updated model without in-depth understanding of the internal structure or training details of the model.

[0105] Furthermore, the system of this embodiment also supports evaluating the effects of different model versions through A / B testing to help users select the best model configuration. Through continuous monitoring and iteration, the system ensures that the model always adapts to the latest business needs and data environment, maximizing its business value and application effect. Users can track the update and performance status of the model through a visual dashboard to obtain key insights in a timely manner and make business decisions.

[0106] The present invention realizes automated model optimization and iteration. By automatically evaluating the model performance by the system and providing optimization suggestions, users can quickly adjust the model parameters or datasets without manual intervention, thereby realizing the continuous optimization of the model. This process has a high degree of automation, can effectively improve the accuracy and stability of the model, and reduces the workload of manual debugging.

[0107] The present invention improves the efficiency of model deployment and invocation. The provided unified invocation mechanism enables the trained model to be seamlessly migrated and invoked between different platforms. Users only need to use the unified declarative language interface, and the system can automatically complete the interface adaptation with different platforms, greatly improving the efficiency of model deployment and cross-platform invocation, and reducing the technical complexity of platform switching.

[0108] In summary, through the innovative application of natural language processing, declarative language, and seamless support for cross-platform operations, the present invention significantly improves the degree of automation of data modeling, model optimization, and deployment, reduces the technical threshold for users, and enhances the flexibility of multi-platform operations.

[0109] The method of the present invention provides an efficient, convenient, and intelligent data modeling and application solution for users in the industrial field.

[0110] The above is only a specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention can easily think of changes or substitutions, which should all be covered within the protection scope of the present invention.

Claims

1. A modeling method for an industrial data processing model based on the binding of natural language interaction and declarative language, characterized in that, It includes the following steps: S1. Obtain industrialization data through the operation logs of industrial machinery and equipment and sensors installed on the industrial machinery and equipment; S2. Use an interactive model as the medium for interaction between the user and the data modeling platform, and perform pre-training on the interactive model to transform natural language into declarative language, obtaining a pre-trained interactive model; S3. The user inputs the industrialization data and their own modeling requirements into the pre-trained interactive model for analysis and processing. The interactive model conducts interactive processing with the user based on the analysis and processing results to obtain a modeling step declarative language command; S4. Select the required data model for modeling according to the user's own modeling requirements in the computer. Perform modeling processing on the selected data model on the data modeling platform according to the modeling step declarative language command to obtain a completed modeling data model. Input the industrialization data into the completed modeling data model for training to obtain a trained data model; S5. Deploy the trained data model on the corresponding platform according to the modeling step declarative language command obtained in step S3, obtain new industrialization data in real time, input the new industrialization data into the data model on the platform for processing, and obtain the results of data model processing in real time.

2. The modeling method of an industrialization data processing model based on natural language interaction and declarative language binding according to claim 1, characterized in that: The interactive model in step S2 is a GPT model or a DeepSeek model; The data modeling platform in step S2 includes a BML platform and a PAI platform.

3. A modeling method for an industrial data processing model based on natural language interaction and declarative language binding according to claim 1, characterized in that, The specific content of step S3 is as follows: S31. The user inputs the industrialization data and their own modeling requirements into the pre-trained interactive model for analysis and processing, obtains step parameters, generates a modeling step parameter problem, and returns it to the user; S32. The user answers according to the returned modeling step parameter problem and inputs it into the interactive model. The user and the interactive model interact continuously until the interactive model stops interacting with the user after extracting all the specific configuration information of the step parameters. The interactive model transforms the obtained step parameters and all the specific configuration information into declarative language to obtain a modeling step declarative language command; S33. Perform verification processing on the obtained modeling step declarative language command to obtain a verified modeling step declarative language command.

4. A modeling method for an industrial data processing model based on natural language interaction and declarative language binding according to claim 3, characterized in that The specific content of step S31 is as follows: S311. The user inputs the industrialization data and their own modeling requirements into the pre-trained interactive model; S312. The interactive model extracts the step parameters required for modeling from the input data; The step parameters include the category of the data model for modeling, filling method, normalization, and target variable selection; S313. According to the extracted step parameters required for modeling, the interactive model generates a modeling step parameter problem and returns the modeling step parameter problem to the user.

5. A modeling method for an industrial data processing model based on natural language interaction and declarative language binding according to claim 3, characterized in that The specific content of step S32 is as follows: S321. The user answers according to the returned modeling step parameter problem and inputs it into the interactive model; S322. The interactive model further extracts the specific configuration information of the required step parameters from the user's answer. Based on the obtained specific configuration information, the interactive model generates a new modeling step parameter question again and returns it to the user; S323. Repeat steps S321 - S322 until the interactive model stops interacting with the user after extracting all the specific configuration information of the step parameters; S324. The interactive model converts all the obtained step parameters and all the specific configuration information into declarative language, and summarizes all the converted declarative language to obtain the modeling step declarative language command.

6. A modeling method of an industrial data processing model based on natural language interaction and declarative language binding according to claim 3, characterized in that: The verification process in step S33 includes consistency check of data types, validity check of parameters, and correctness check of logic flow.

7. A modeling method for an industrial data processing model based on natural language interaction and declarative language binding according to claim 1, characterized in that, Step S4 is specifically as follows: S41. Analyze the category of the data model required for modeling by using a machine learning algorithm according to the user's own modeling needs, and select the data model required for modeling on the data modeling platform; S42. Perform modeling processing on the selected data model on the data modeling platform according to the modeling step declarative language command after verification processing to obtain a modeled data model; S43. Input the industrial data into the modeled data model, and train the data model according to the modeling step declarative language command after verification processing to obtain a trained data model.

8. A modeling method of an industrial data processing model based on natural language interaction and declarative language binding according to claim 5, characterized in that: The categories of the data models required for modeling in step 41 include machine learning models and deep learning models.

9. A modeling method of an industrial data processing model based on natural language interaction and declarative language binding according to claim 1, characterized in that: The platforms deployed in step S5 include cloud platforms, enterprise internal servers, and edge devices.

Images