Interactive modeling method, device and equipment based on large model and storage medium

Abstract

The invention relates to an interactive modeling method based on a large model, and the method comprises the steps: receiving a natural language instruction inputted by a user, carrying out the analysis of the natural language instruction through a natural language model, and determining the modeling demands of the user; according to the modeling demand, determining a model establishment scheme, and generating a model establishment instruction corresponding to the model establishment scheme; sending a model building instruction corresponding to each modeling step in the model building scheme to a corresponding data processing module in sequence, so that each data processing module carries out model building according to the model building instruction and modeling data; generating a model effect index according to the model establishment result of each data processing module; and feeding back the model establishment result of each data processing module and the model effect to the user.

Description

Technical Field

[0001] The present application relates to the field of interactive modeling technology, and in particular to a large-scale model-based interactive modeling method, apparatus, device, and storage medium. Background Art

[0002] With the advent of the data age, businesses and individuals are faced with increasingly vast amounts of data. Data-driven decision-making has become a crucial means of enhancing competitiveness. However, traditional data modeling processes typically require specialized statistical and programming knowledge, making it difficult for non-professionals to participate. Furthermore, existing data analysis tools often have complex interfaces and high barriers to entry, limiting their widespread adoption.

[0003] In recent years, the rapid development of deep learning and natural language processing technologies has brought new opportunities to the field of data modeling. Large-scale pre-trained models (such as the GPT series and BERT) have demonstrated exceptional performance in understanding and generating natural language, enabling more intuitive and user-friendly interactions. Furthermore, advances in real-time data feedback and intelligent recommendation technologies have opened up new avenues for improving modeling efficiency and accuracy.

[0004] Although some low-code or no-code platforms have emerged, they still lack flexibility and intelligence in interaction.

[0005] Therefore, it is necessary to provide an interactive modeling method, device, equipment and storage medium based on a large model, aiming to lower the modeling threshold and enable more users to conveniently perform data analysis and decision-making. Summary of the Invention

[0006] The present application provides an interactive modeling method, device and storage medium based on a large model to solve the problem that existing data modeling tools often require users to have a deep foundation in statistics and programming, making it difficult for non-professionals to participate in the data modeling process, thereby limiting the popularization of data-driven decision-making.

[0007] In a first aspect, the present application provides an interactive modeling method based on a large model, the method comprising:

[0008] Receiving a natural language instruction input by a user, and parsing the natural language instruction using a natural language model to determine the user's modeling requirements;

[0009] Determine a model building scheme according to the modeling requirements, and generate a model building instruction corresponding to the model building scheme;

[0010] Sending the model building instructions corresponding to each modeling step in the model building scheme to the corresponding data processing modules in sequence, so that each data processing module builds the model according to the model building instructions and modeling data;

[0011] Generate model effect indicators based on the model establishment results of each data processing module;

[0012] Feedback the model building results of each data processing module and the model effect to the user.

[0013] In a second aspect, the present application provides an interactive modeling device based on a large model, the device comprising:

[0014] A receiving module is used to receive a natural language instruction input by a user, and parse the natural language instruction using a natural language model to determine the modeling requirements of the user;

[0015] A generating module, configured to determine a model building scheme according to the modeling requirements, and generate a model building instruction corresponding to the model building scheme;

[0016] A modeling module, configured to sequentially send model building instructions corresponding to each modeling step in the model building scheme to corresponding data processing modules, so that each data processing module builds the model according to the model building instructions and modeling data;

[0017] The analysis module is used to generate model effect indicators based on the model building results of each data processing module;

[0018] Feedback module, used to feed back the model building results of each data processing module and the model effect to the user

[0019] In a third aspect, an electronic device is provided, comprising a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other via the communication bus;

[0020] Memory for storing computer programs;

[0021] The processor is used to implement the steps of the interactive modeling method based on a large model described in any embodiment of the first aspect when executing the program stored in the memory.

[0022] In a fourth aspect, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps of the interactive modeling method based on a large model as described in any embodiment of the first aspect are implemented.

[0023] The above technical solutions provided by the embodiments of the present application have the following advantages over the existing technologies: (1) Lowering the threshold for use: Through conversational modeling interaction, users can describe their needs in natural language without the need for professional programming or statistical knowledge, so that non-professionals can also easily perform data modeling; (2) The system can quickly provide modeling results and model effect indicators during user interaction, helping users to adjust the model in a timely manner and improve modeling efficiency and accuracy; (3) Based on user input and historical behavior analysis, the platform can intelligently recommend suitable models and parameter configurations to meet personalized needs and enhance user experience and modeling effects; (4) The platform integrates a variety of advanced model building algorithms, which are applicable to different types of data (such as numerical values and time series), broadening the application scenarios and making it adaptable to more business needs. This application can significantly improve the user's modeling capabilities and experience, and realize efficient, flexible and intelligent data modeling. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0025] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0026] Figure 1 A flowchart of an interactive modeling method based on a large model provided in an embodiment of the present application;

[0027] Figure 2 is an exemplary flow chart of the modeling method provided in the embodiments of the present application;

[0028] Figure 3 is a flowchart of a method for determining a user's modeling requirements provided by some embodiments of the present application;

[0029] Figure 4 An exemplary structural diagram of the model structure of a large model provided in an embodiment of the present application;

[0030] Figure 5 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0031] To make the purpose, technical solutions, and advantages of the embodiments of this application more clear, the technical solutions in the embodiments of this application will be clearly and completely described below in conjunction with the drawings in the embodiments of this application. Obviously, the described embodiments are part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0032] Figure 1 A flow chart of an interactive modeling method based on a large model provided in an embodiment of the present application. In some embodiments, Figure 1 The process shown can be executed by an electronic device. Figure 1 As shown, the process may include the following operations.

[0033] Step 101: Receive a natural language instruction input by a user, and parse the natural language instruction using a natural language model to determine the modeling requirements of the user.

[0034] User-entered natural language instructions are commands entered in natural language (e.g., Chinese, English, etc.). These instructions reflect the user's needs or expectations for the modeling process or modeling results. For example, a user's natural language instruction might be "Build a classification model to predict product sales trends" or "Build a sales analysis model for the first quarter based on the provided data."

[0035] Natural language models are computational models used to parse and understand natural language text. Examples of natural language models include BERT and GPT. Natural language models can extract key information from natural language instructions, such as the target task and input data type.

[0036] Modeling requirements are specific requirements extracted from natural language instructions input by users. They describe the objectives, data types, algorithms, and other requirements of the model they wish to build. For example, a user may wish to model a set of marketing data to facilitate subsequent prediction and scoring, and to conduct targeted marketing activities.

[0037] In some embodiments, the system may receive natural language instructions input by the user through an interface or other means, and process the natural language instructions using a natural language model (e.g., a model based on BERT or GPT). The model can extract key information from the instructions, such as the type of prediction task, the input data used, the expected output results, etc., and generate specific modeling requirements based on the parsing results. Modeling requirements may include the type of data, the expected task (e.g., regression, classification), the performance indicators of the model (e.g., accuracy, F1 value), etc.

[0038] For example, if a user inputs the instruction "build a classification model to predict product sales trends", the natural language model can determine the modeling requirements after parsing: "Goal: classification task, data type: historical sales data, expected output: sales trend classification label".

[0039] In some embodiments, the receiving of a natural language instruction input by a user, parsing the natural language instruction, and determining the user's modeling requirements may include the following operations.

[0040] S10, determining the instruction type of the natural language instruction; the instruction type includes a requirement type and an instruction type.

[0041] Instruction types refer to the classification of natural language instructions, including requirement types and instruction types. Requirement types refer to the task objectives expressed by users in natural language instructions, which are usually related to the need to build a model. For example, the user wants to build a classification model, regression model, etc. Instruction types refer to the operational instructions given by users in natural language instructions, instructing the system to perform specific steps or methods. For example, "blank instruction, indicating to continue the previous modeling requirement", "start model training", "check data quality", "terminate instruction, indicating to terminate modeling", etc.

[0042] In some embodiments, the input natural language instruction may be parsed using a natural language model to determine whether the instruction is of a requirement type or an instruction type.

[0043] S11 , in response to the instruction type of the natural language instruction being a requirement type, the natural language instruction is parsed using natural language processing technology to determine the modeling requirement of the user.

[0044] Natural language processing (NLP) is a technology used to understand and parse natural language text, including word segmentation, named entity recognition (NER), dependency parsing, and sentiment analysis. Common tools include NLTK, spaCy, and BERT.

[0045] Modeling requirements are specific requirements extracted from the user's natural language instructions, indicating the type of model the user expects to build, the data features used, the expected task objectives, etc.

[0046] When the instruction type is a requirement, natural language processing technology is used to analyze the instruction structure and keywords to identify the user's core needs, such as model type and input data requirements. The parsed instructions are then used to further clarify modeling requirements, such as building a regression model, classification model, or using time series data for prediction.

[0047] S12: In response to the instruction type of the natural language instruction being an instruction type, obtaining historical conversation records with the user, and determining the user's modeling requirements based on the historical conversation records.

[0048] Historical conversation records refer to the historical interaction information between users and the system. They include modeling requirements, problem descriptions, and performed operations previously submitted by users. For example, a user may have previously asked how to use a certain algorithm to solve a specific problem.

[0049] Through historical conversation records, the modeling needs of current users can be inferred from past exchanges.

[0050] When the instruction type is command type, the system first accesses the historical conversation records related to the user to extract the user's past needs and questions. By analyzing the content of the historical conversation and combining it with contextual information, the system infers the current user's modeling needs.

[0051] The system can comprehensively determine the final modeling requirements based on the results of natural language instruction parsing and inferences from historical conversation records.

[0052] More information on determining the user's modeling needs can be found in Figure 3 Related description.

[0053] Step 102: Determine a model building scheme according to the modeling requirements, and generate a model building instruction corresponding to the model building scheme.

[0054] A model building plan is a specific plan developed based on modeling requirements, describing the steps for building and training the model, the selected algorithms, the tools and frameworks used, etc. For example, a support vector machine (SVM) is selected as the classification model, and cross-validation is used to optimize hyperparameters.

[0055] Model building instructions refer to the specific operational instructions that guide model building, including data preprocessing step instructions, feature engineering step instructions, and model building step instructions, etc.

[0056] In some embodiments, a solution can be established based on the model to generate corresponding execution instructions.

[0057] In some embodiments, the modeling steps of the model building scheme include: a data preprocessing step for performing data preprocessing on the modeling data provided by the user; a feature engineering step for performing feature analysis and screening on the modeling data after data preprocessing to obtain modeling feature data; and a model building step for calling the corresponding model building algorithm to build the model based on the modeling feature data.

[0058] Data preprocessing refers to the process of cleaning, transforming, and preparing raw data (user-provided modeling data) before modeling. The goal of preprocessing is to ensure data quality and make it suitable for subsequent modeling tasks. Common preprocessing operations include removing missing values, handling outliers, and normalizing data.

[0059] Feature engineering is the process of extracting features from raw data that are useful for modeling, and then filtering and transforming these features. The goal is to improve the effectiveness and performance of the model through feature selection and feature transformation.

[0060] Model building involves using a machine learning algorithm to train a predictive model based on input feature data. The choice of model typically depends on the specific task (e.g., regression, classification, clustering, etc.) and the characteristics of the data. For example, if the goal is to predict housing prices, a linear regression model might be used; if the goal is to distinguish between different types of diseases, a support vector machine (SVM) classification model might be used.

[0061] Step 103 , sending the model building instructions corresponding to each modeling step in the model building solution to the corresponding data processing modules in sequence, so that each data processing module builds the model according to the model building instructions and modeling data.

[0062] The data processing module is the module that processes data and performs model building operations. The data processing module includes data preprocessing modules, feature engineering modules, and model building modules. Each module is responsible for handling specific tasks.

[0063] Model building instructions are specific instructions that describe how to process data, train models, and evaluate them.

[0064] In some embodiments, the various steps required in the model building process (such as data preprocessing, feature engineering, model building, etc.) can be identified based on the model building plan.

[0065] Sending model building instructions means sending the generated model building instructions to each data processing module in the order of steps. Each module interprets the instructions and performs the corresponding tasks.

[0066] For more information about step 103, see Figure 2 Related description.

[0067] Step 104: Generate model effect indicators based on the model building results of each data processing module.

[0068] Model building results refer to the model building results obtained after executing the model building instructions. For example, the trained model parameters, model prediction results, etc.

[0069] Model effectiveness metrics are used to measure model performance, including accuracy, recall, etc. The effectiveness of a model can be evaluated by comparing the difference between the model's output and the actual results.

[0070] Step 105: Feedback the model building results of each data processing module and the model effect to the user.

[0071] Feedback refers to displaying or sending the model's results and effects to users so they can make judgments, optimize, or adjust the model. Feedback can include performance indicators of model training and visual displays of results.

[0072] In some embodiments, the model results and effect indicators of the data processing module can be collected, and the model effects can be presented to the user through a suitable interface or method, such as through charts, reports, etc.

[0073] By displaying the accuracy, recall rate and other results of model training to users through charts and providing suggestions for further optimizing the model, users can be provided with a better modeling experience.

[0074] Figure 2 is an exemplary flow chart of the modeling method provided in the embodiments of the present application. In some embodiments, Figure 2 The steps shown can be performed by an electronic device, such as Figure 2 As shown, the process may include the following operations.

[0075] Step 201 : Analyze the model building instructions through a preset model building agent module to identify the model building sub-instructions corresponding to each modeling step in the model building solution.

[0076] The model building agent module is an intermediary module that receives, analyzes, and processes model building instructions. Its purpose is to understand these instructions and break them down into smaller sub-instructions related to specific steps for easier subsequent processing.

[0077] For example, the model building agent module is like a command center. After receiving the complete instructions, it decomposes them into three sub-steps: "data preprocessing", "feature engineering" and "model building", and each sub-step is passed to the corresponding processing module.

[0078] The model proxy module can send each identified sub-instruction to the corresponding data processing module (data preprocessing module, feature engineering module, model building module) for processing.

[0079] Step 202: The model building agent module sends the model building sub-instruction corresponding to the data preprocessing step to the data preprocessing module for processing.

[0080] The data preprocessing module is responsible for processing raw data, cleaning data, and performing preliminary conversions. It performs relevant processing on the data according to the received data preprocessing sub-instructions.

[0081] For example, after receiving the instructions, the data preprocessing module can parse the instructions and perform tasks such as filling missing values, removing duplicate data, and standardizing data.

[0082] Step 203: The execution result of the data preprocessing step and the model building sub-instruction corresponding to the feature engineering step are sent to the feature engineering module for processing through the model building agent module.

[0083] The feature engineering module is responsible for extracting features from data, performing feature selection and transformation. It processes the data according to the received feature engineering sub-instructions and generates appropriate feature data.

[0084] For example, the feature engineering module can perform tasks such as feature selection, feature combination, or principal component analysis (PCA) to select the most useful features for model training from the preprocessed data.

[0085] The feature engineering module receives the processing results (i.e., data after data preprocessing) and the "feature engineering sub-instructions" sent by the model building agent module, and performs feature analysis, selection, and conversion according to the instructions to generate the final modeling feature data.

[0086] After the feature engineering module completes feature processing, it feeds the obtained modeling feature data and its processing results back to the model building agent module, preparing to enter the model building step.

[0087] Step 204: Send the execution result of the feature engineering step and the model building sub-instruction corresponding to the model building step to the model building module for processing through the building agent module.

[0088] The model building module is responsible for building a model using the processed feature data and the model building algorithm. It trains the final prediction model based on the received model building sub-instructions and feature data.

[0089] For example, the model building module selects a suitable algorithm (such as regression analysis, support vector machine, etc.) to build a prediction model based on the feature data and model building sub-instructions provided by the feature engineering module.

[0090] After the model is built, the model building module feeds back the final model and its performance indicators to the model building agent module.

[0091] Step 205: The model building agent module aggregates the results of the data preprocessing step, the feature engineering step, and the model building step to obtain the model building results of each data processing module.

[0092] The model building agent module aggregates the results of the data preprocessing, feature engineering, and model building modules to form the final model building results. The aggregated results include preprocessed data, selected features, trained models, and their evaluation metrics.

[0093] The model building agent module can feed back the complete model building results (including information such as data processing, feature engineering, and training models) to the user or system for further use.

[0094] In this example, a model-building agent coordinates the sequential execution of modeling operations across various modules (data preprocessing, feature engineering, and model building). The agent breaks down model-building instructions and sends them sequentially to the appropriate modules for processing. Ultimately, it aggregates the results of all steps to form a complete model-building process.

[0095] Figure 3 This is a flowchart of a method for determining a user's modeling needs provided by some embodiments of the present application. In some embodiments, Figure 3 The process shown can be executed by an electronic device, such as Figure 3 As shown, the process may include the following operations.

[0096] Step 301: Acquire modeling data provided by the user.

[0097] Modeling data is a collection of raw data provided by users for model training and validation. The data can include features, labels, or target variables.

[0098] The uploaded modeling data can be saved to the database or memory for subsequent processing.

[0099] Step 302: extract the data type, data format and key features of the modeling data.

[0100] Data type refers to the classification of data, such as numeric data, text data, date data, etc. For example, the data type of the customer age field is numeric, while the data type of the customer name field is text.

[0101] Data format refers to the specific representation of data, such as CSV, JSON, SQL database table, etc.

[0102] Data key features are the most representative variables or information in the data, which usually have a greater impact on model training and prediction results.

[0103] For example, in a model to predict housing prices, number of rooms, area, location, etc. may be key features.

[0104] Step 303: Based on the data type, data format and key features of the data, a prediction modeling requirement is obtained using a modeling command prediction model.

[0105] Modeling commands refer to instructions used to guide the model building process, including data preprocessing, feature selection, model selection, etc.

[0106] Predictive modeling needs refer to the system predicting the next modeling steps based on existing data and model requirements, such as which model to choose and which features to use.

[0107] In some embodiments, the system can predict modeling requirements based on data type and features, inferring the required modeling algorithm or method based on the data type, format, and key features. Based on the prediction results, the system generates corresponding modeling commands, including data processing steps, feature selection methods, and model types.

[0108] Step 304: parse the natural language instruction using the natural language model to obtain initial modeling requirements.

[0109] The initial modeling requirements are determined by the natural language model parsing the instructions input by the user.

[0110] Step 305: The predicted modeling requirement and the initial modeling requirement are integrated to determine the user's modeling requirement.

[0111] Fusion involves comparing and integrating the predictive modeling requirements with the initial modeling requirements to ensure that the final modeling requirements meet the user's needs. For example, if the predictive modeling requirement is a "regression model" and the user's initial requirement is a "classification prediction," the system will determine the final classification model based on the fused requirements and make appropriate adjustments.

[0112] In some embodiments, the system may determine the user's modeling requirements by fusing the predicted modeling requirements and the initial modeling requirements through collaborative filtering.

[0113] Collaborative filtering fusion refers to the fusion of predictive modeling requirements and initial modeling requirements through collaborative filtering methods to obtain the final user modeling requirements.

[0114] In some embodiments, collaborative filtering can be implemented using the following formula (1).

[0115]

[0116] Where R represents the user's modeling requirement obtained by fusion, D is the predicted modeling requirement, L is the initial modeling requirement, α is the weight coefficient corresponding to the predicted modeling requirement, and β is the weight coefficient corresponding to the initial modeling requirement. is the synergistic effect term, γ is the weight coefficient corresponding to the collaborative filtering term, Represents element-wise multiplication.

[0117] By integrating predictive modeling requirements with initial modeling requirements, the modeling solution can be automatically optimized based on user needs and data characteristics. This approach improves compatibility between different modeling requirements, enabling the system to adjust the modeling solution based on actual data characteristics, thereby achieving modeling results that better meet user expectations.

[0118] Figure 4 The exemplary structural diagram of the model structure of the large model provided in the embodiment of the present application. Figure 4 As shown, the model structure of the large model can include a data preprocessing module, a feature engineering module, a model building module and a model building agent module.

[0119] The data preprocessing module includes a data preprocessing user agent, a data preprocessing assistant, a data preprocessing manager and a data preprocessing agent.

[0120] Preprocessing User Agent: This agent is responsible for receiving user input instructions and interacting with the user. It passes the user's requirements to the data preprocessing manager.

[0121] Data Preprocessing Assistant: This agent is responsible for analyzing and planning the data preprocessing workflow. It comprehensively evaluates the input from the user agent and calls specific Python code to guide the subsequent steps.

[0122] Data Preprocessing Manager: As the manager of the data preprocessing module, this component collects and summarizes the information within the module, including user input and execution status, and passes the sorted data to the data preprocessing agent.

[0123] Data preprocessing agent: This agent is responsible for further analyzing and summarizing the data received from the data preprocessing module manager, and interacting it with the model building agent module (agents), so that the model building agent module (agents) have more information to call the appropriate module agent to handle these tasks.

[0124] The feature engineering module includes the feature engineering user agent, feature engineering assistant, feature engineering manager, and feature engineering agent.

[0125] Feature Engineering User Agent: Similar to the Data Preprocessing User Agent, this agent receives user input related to feature engineering and passes instructions to the Feature Engineering Manager.

[0126] Feature Engineering Assistant: This agent is similar to the Data Preprocessing Assistant and is responsible for analyzing and planning the operational process of feature engineering. It also feeds back input from the Feature Engineering User Agent and calls the corresponding Python code.

[0127] Feature Engineering Manager: The manager of the feature engineering module, responsible for collecting and summarizing information within the feature engineering, including user requirements and module status, and then passing this information to the feature engineering agent.

[0128] Feature Engineering Agent: This agent is responsible for further analyzing and summarizing the data received from the Feature Engineering Module Manager, interacting with the model building agent module (agents), so that the model building agent module (agents) have more information to call the appropriate module agent to handle these tasks.

[0129] The model building module includes the model building user agent, the model building assistant, the model building manager and the model building agent.

[0130] Model building user agent: Similar to the data preprocessing user agent, this agent receives user input related to model building and passes instructions to the model building manager.

[0131] Model Building Assistant: This agent is similar to the Data Preprocessing Assistant and is responsible for analyzing and planning the operational process of model building, such as which algorithm to use to build the model, determining whether optimization is needed based on model indicators, etc. It also feeds back input from the model building user agent and calls the corresponding Python code.

[0132] Model Building Manager: The manager of the feature engineering module, responsible for collecting and summarizing information within the model building, including user requirements and module status, and then passing this information to the model building agent.

[0133] Model building agent: This agent is responsible for further analyzing and summarizing the data received from the model building module manager, and interacting with the model building agent modules (agents) so that the model building agent modules (agents) have more information to call the appropriate module agents to handle these tasks.

[0134] Throughout the system, each agent and the model-building agent module interact through a clear messaging and coordination mechanism. Users submit instructions through their respective user agents, which are aggregated by the model-building agent module and ultimately broken down into small, executable tasks, which are then assigned to the appropriate agents. This structure ensures the system's efficiency and flexibility, enabling rapid response to user needs and automated execution of complex data processing and feature engineering tasks.

[0135] The embodiment of the present application further provides an interactive modeling device based on a large model, the device comprising:

[0136] A receiving module is used to receive a natural language instruction input by a user, and parse the natural language instruction using a natural language model to determine the modeling requirements of the user;

[0137] A generating module, configured to determine a model building scheme according to the modeling requirements, and generate a model building instruction corresponding to the model building scheme;

[0138] A modeling module, configured to sequentially send model building instructions corresponding to each modeling step in the model building scheme to corresponding data processing modules, so that each data processing module builds the model according to the model building instructions and modeling data;

[0139] The analysis module is used to generate model effect indicators based on the model building results of each data processing module;

[0140] The feedback module is used to feed back the model building results of each data processing module and the model effect to the user.

[0141] like Figure 5 As shown, an embodiment of the present application provides an electronic device, including a processor 111, a communication interface 112, a memory 113 and a communication bus 114, wherein the processor 111, the communication interface 112, and the memory 113 communicate with each other through the communication bus 114.

[0142] Memory 113, for storing computer programs;

[0143] In one embodiment of the present application, the processor 111 is configured to execute a program stored in the memory 113 to implement the large model-based interactive modeling method provided by any of the aforementioned method embodiments, including:

[0144] An embodiment of the present application also provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the steps of the interactive modeling method based on a large model as provided in any of the aforementioned method embodiments are implemented.

[0145] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article, or device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or elements inherent to such process, method, article, or device. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of other identical elements in the process, method, article, or device comprising the element.

[0146] The foregoing is merely a list of specific embodiments of the present application, intended to enable those skilled in the art to understand or implement the present application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application is not limited to the embodiments shown herein, but is intended to conform to the broadest scope consistent with the principles and novel features of the present application.

Claims

1. An interactive modeling method based on a large model, characterized in that: The method comprises: Receiving a natural language instruction input by a user, and parsing the natural language instruction using a natural language model to determine the user's modeling requirements; Determine a model building scheme according to the modeling requirements, and generate a model building instruction corresponding to the model building scheme; Sending the model building instructions corresponding to each modeling step in the model building scheme to the corresponding data processing modules in sequence, so that each data processing module builds the model according to the model building instructions and modeling data; Generate model effect indicators based on the model establishment results of each data processing module; Feedback the model building results of each data processing module and the model effect to the user.

2. The method according to claim 1, characterized in that The receiving of a natural language instruction input by a user, parsing the natural language instruction, and determining the user's modeling requirements includes: Determining the instruction type of the natural language instruction; the instruction type includes a requirement type and an instruction type; In response to the instruction type of the natural language instruction being a requirement type, parsing the natural language instruction using natural language processing technology to determine the modeling requirement of the user; In response to the instruction type of the natural language instruction being an instruction type, a historical conversation record with the user is obtained, and a modeling requirement of the user is determined based on the historical conversation record.

3. The method according to claim 1, characterized in that The modeling steps of the model building scheme include: A data preprocessing step, for performing data preprocessing on the modeling data provided by the user; Feature engineering step, used to perform feature analysis and screening on the modeling data after data preprocessing to obtain modeling feature data; The model building step is used to call the corresponding model building algorithm to build the model according to the modeling feature data.

4. The method according to claim 3, characterized in that The model building instructions corresponding to each modeling step in the model building scheme are sent to the corresponding data processing modules in sequence, so that each data processing module builds the model according to the model building instructions and the modeling data, including: Analyzing the model building instructions through a preset model building agent module to identify model building sub-instructions corresponding to each modeling step in the model building solution; The model building agent module sends the model building sub-instructions corresponding to the data preprocessing step to the data preprocessing module for processing; The execution result of the data preprocessing step and the model building sub-instruction corresponding to the feature engineering step are sent to the feature engineering module for processing through the model building agent module; Sending the execution results of the feature engineering step and the model building sub-instructions corresponding to the model building step to the model building module for processing through the building agent module; The model building agent module aggregates the results of the data preprocessing step, the feature engineering step, and the model building step to obtain the model building results of each data processing module.

5. The method according to claim 1, wherein The receiving of a natural language instruction input by a user and parsing the natural language instruction using a natural language model to determine the modeling requirements of the user includes: Acquiring modeling data provided by the user; Extracting the data type, data format and key data features of the modeling data; Based on the data type, data format and key features of the data, using a modeling command prediction model to obtain a prediction modeling requirement; Parsing the natural language instructions using the natural language model to obtain initial modeling requirements; The predictive modeling requirement and the initial modeling requirement are integrated to determine the user's modeling requirement.

6. The method according to claim 5, characterized in that The fusing of the prediction modeling requirement and the initial modeling requirement to determine the user's modeling requirement includes: Based on the predicted modeling requirement and the initial modeling requirement, they are fused by collaborative filtering to determine the user's modeling requirement.

7. The method according to claim 6, characterized in that The step of fusing the predicted modeling requirement and the initial modeling requirement through collaborative filtering to determine the user's modeling requirement includes: The collaborative filtering is implemented by the following formula: Where R represents the user's modeling requirement obtained by fusion, D is the predicted modeling requirement, L is the initial modeling requirement, α is the weight coefficient corresponding to the predicted modeling requirement, and β is the weight coefficient corresponding to the initial modeling requirement. is the synergistic effect term, γ is the weight coefficient corresponding to the collaborative filtering term, Represents element-wise multiplication.

8. An interactive modeling device based on a large model, characterized in that: The device comprises: A receiving module is used to receive a natural language instruction input by a user, and parse the natural language instruction using a natural language model to determine the modeling requirements of the user; A generating module, configured to determine a model building scheme according to the modeling requirements, and generate a model building instruction corresponding to the model building scheme; A modeling module, configured to sequentially send model building instructions corresponding to each modeling step in the model building scheme to corresponding data processing modules, so that each data processing module builds the model according to the model building instructions and modeling data; The analysis module is used to generate model effect indicators based on the model building results of each data processing module; The feedback module is used to feed back the model building results of each data processing module and the model effect to the user.

9. An electronic device, characterized in that: It includes a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other via the communication bus; Memory for storing computer programs; The processor is configured to implement the steps of the interactive modeling method based on a large model as described in any one of claims 1 to 7 when executing the program stored in the memory.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the interactive modeling method based on a large model are implemented.

Images