Big model-based multi-dimensional semantic blueprint data large screen generation method and system

Abstract

This invention belongs to the field of industrial Internet of Things and artificial intelligence technology, and provides a method and system for generating multidimensional semantic blueprint data dashboards based on a large model. The method includes: acquiring natural language instructions from user data; parsing the acquired natural language instructions based on a large language model to generate a multidimensional semantic blueprint; parsing the generated multidimensional semantic blueprint to obtain a data task queue and a view task queue; executing tasks in the obtained data task queue and view task queue through different expert agents, and updating the task status; performing completion verification and loop closure correction after all tasks are completed to generate the data dashboard; and completing the generation of the multidimensional semantic blueprint data dashboard based on the large model.

Description

Technical Field

[0001] This invention belongs to the field of industrial Internet of Things and artificial intelligence technology, specifically relating to a method and system for generating multidimensional semantic blueprint data dashboards based on large models. Background Technology

[0002] The statements in this section are merely background information related to the present invention and do not necessarily constitute prior art.

[0003] Industrial data visualization dashboards are the core human-machine interface in modern industrial internet and intelligent manufacturing systems. They transform massive and heterogeneous industrial data (such as equipment status, production progress, quality indicators, and energy consumption information) into intuitive graphics, charts, and dashboards, providing managers, engineers, and operators with real-time operational monitoring, anomaly warnings, and decision support.

[0004] In the traditional industrial software development model, building a customized data dashboard is a complex project that relies heavily on human collaboration. The typical process includes: business personnel proposing requirements, product managers writing requirement documents, UI / UX designers designing the interface, front-end engineers implementing the interface and interaction, and back-end / data engineers handling data access, cleaning, modeling, and API development. However, this process has drawbacks such as long cycle time, high communication costs, difficulty in iteration, and high maintenance costs.

[0005] With the explosive growth of Large Language Models (LLM), "one-step" automated content generation tools based on LLM (such as automatically generating PPTs, marketing posters, and simple data reports) have emerged, providing new ideas for lowering the barrier to content creation. Research on using LLM to directly generate data visualization code or interfaces has also arisen, for example, combining Codex or GPT series models to directly output code snippets from visualization libraries such as ECharts and Vega-Lite based on natural language descriptions. However, when facing industrial-grade visualization dashboards with complex data models, rigorous business logic, and highly coupled interactive relationships, fundamental and insurmountable limitations have been exposed: (1) Uncontrollable generation results and the problem of "illusion" Existing "one-step generation" solutions are essentially end-to-end black-box models. The direct output of LLM has high randomness and probability, making it difficult to guarantee that the generated chart types, data binding logic, and calculation methods strictly conform to complex industrial business rules and underlying data constraints. The model may "fabricate" non-existent fields, incorrect aggregation algorithms (e.g., mistaking summation for averaging), or illogical data associations (e.g., associating output with temperature sensors). This "illusion" phenomenon is fatal in industrial monitoring scenarios that require zero errors and high reliability, potentially leading to erroneous decisions and major safety incidents.

[0006] (2) Lack of engineering collaboration capabilities Industrial dashboards are not static content layouts, but rather data-driven software applications with a lifecycle; they involve close collaboration across multiple engineering stages, including data access, cleaning, aggregation, model building, front-end component instantiation, rendering, data binding, and interactive event configuration. Existing generation tools only handle the code synthesis of the final presentation layer, which is a "single-process" and cannot coordinate the complex dependencies between these heterogeneous resources.

[0007] (3) Unable to support incremental demand and closed-loop correction Users find it difficult to effectively intervene and correct during the generation process; once they are dissatisfied with the final generated result, or make fine-grained modification suggestions (such as "change the OEE chart of production line A from a line chart to a dashboard and remove production line B"), there is no response, and all generated results can only be discarded, the input prompts modified, and the generation regenerated; this not only leads to a huge waste of computing resources and time and a very poor user experience, but also fails to meet the reality of continuously evolving needs in industrial production.

[0008] Therefore, the fundamental flaw of existing technologies lies in their adoption of a "content generation paradigm," which focuses on the rapid, one-way synthesis of the final product while neglecting the determinism, controllability, collaboration, and maintainability required for industrial software development. They lack an intermediate structured representation that combines business readability and machine executableness as a control center to break down the "black box" between undeterministic requirements and deterministic software engineering products. Therefore, the industry urgently needs a new technological solution that can transform unstructured natural language requirements into structured, executable software engineering tasks, and support closed-loop correction and collaborative construction. Summary of the Invention

[0009] To address the aforementioned issues, this invention proposes a method and system for generating large-scale data dashboards based on a multidimensional semantic blueprint model. It employs a multidimensional semantic blueprint to transform natural language requirements into a structured, executable intermediate representation. Through an engineering-based decomposition mechanism of "blueprint → dual task queues," the blueprint is automatically decomposed into logically separate data task queues and view task queues with dependencies. Based on state synchronization and asynchronous collaboration among multiple agents, data expert agents and dashboard production expert agents collaborate asynchronously through listening to shared states. A bidirectional tracking mapping from blueprint nodes to engineering products is established, achieving a closed-loop maintenance system for lossless blueprint-level correction. This further solves the problems of uncontrollable generation results, lack of engineering collaboration, and inability to achieve closed-loop correction in existing technologies, realizing efficient, controllable, and intelligent generation and maintenance of industrial-grade data dashboards.

[0010] According to some embodiments, the first solution of the present invention provides a method for generating a large-scale multidimensional semantic blueprint data dashboard based on a large model, which adopts the following technical solution: A method for generating multidimensional semantic blueprint data dashboards based on large models includes: Natural language commands to retrieve user data; Based on the natural language instructions obtained from parsing the large language model, a multidimensional semantic blueprint is generated. The generated multidimensional semantic blueprint is parsed to obtain the data task queue and the view task queue; Different expert agents execute tasks in the obtained data task queue and view task queue respectively, and update the task status; After all tasks are completed, a completion verification and closed-loop correction are performed to generate a data dashboard; the generation of a multi-dimensional semantic blueprint data dashboard based on a large model is completed.

[0011] As a further technical limitation, when a user needs to modify the generated data dashboard, an incremental change set is calculated based on the differences in the multidimensional semantic blueprint, and incremental task instructions are issued to the relevant agents to drive the agents to perform local updates and achieve closed-loop correction.

[0012] As a further technical limitation, the specific process of the closed-loop correction is as follows: When a user edits a locked multidimensional semantic blueprint, a new version of the blueprint is generated; By comparing the differences between the old and new versions of the blueprint, and combining the two-way mapping relationship between blueprint nodes and downstream engineering products, the incremental change set is automatically calculated, which includes at least the task items that need to be added, deleted, and updated. Based on the incremental change set, incremental, reduction, or update task instructions are issued only to the relevant data expert agents and / or large screen production expert agents. The relevant intelligent agents execute precise local updates or resource reclamation based on the received instructions, realize automatic reconstruction in response to changes in requirements, and complete automatic correction.

[0013] As a further technical limitation, at least one data expert agent executes tasks in the data task queue, interacts with the data platform to complete data queries or processing, and updates the task status; at least one large screen creation expert agent listens to the view task queue, executes view tasks after confirming that the data dependencies of the tasks are met, instantiates UI components and completes data binding and style configuration, and updates the task status.

[0014] As a further technical limitation, after acquiring natural language instructions, the acquired natural language instructions are parsed based on a large language model to generate or update a multidimensional semantic blueprint. The multidimensional semantic blueprint is then rendered in real time as a low-fidelity visual sketch for user preview and interaction. Once the user confirms, the multidimensional semantic blueprint is locked.

[0015] Furthermore, the multidimensional semantic blueprint and the low-fidelity visualization sketch are updated in tandem through a two-way data binding mechanism.

[0016] Furthermore, the locked multidimensional semantic blueprint is parsed and automatically decomposed to obtain a logically separated data task queue and a view task queue with dependencies. Each view task item in the view task queue explicitly records the identifier of its dependent data task. A view task is only allowed to execute when all its dependent data tasks are marked as completed in the global task state.

[0017] According to some embodiments, the second aspect of the present invention provides a multi-dimensional semantic blueprint data large-screen generation system based on a large model, which adopts the following technical solution: A system for generating multidimensional semantic blueprint data dashboards based on large models includes: The blueprint generation and interaction module is used to receive natural language commands from users, call a large language model to parse the natural language commands, generate or update multidimensional semantic blueprints, and provide a graphical user interface for displaying and editing the low-fidelity visual sketches corresponding to the multidimensional semantic blueprints, so as to realize the two-way linkage between multidimensional semantic blueprints and low-fidelity visual sketches. The task decomposition and scheduling engine is used to parse locked multidimensional semantic blueprints to obtain data task queues and view task queues, and calculate incremental change sets based on blueprint changes. At least one data expert agent is used to listen to and pull tasks from the data task queue, perform data queries or processing, and update task status. At least one large screen production expert intelligent agent is used to listen to the tasks in the view task queue, pull the task after confirming that the data dependency of the task is satisfied, instantiate the UI component, complete the data binding and style configuration, and update the task status. The version management library is used to store different versions of the multidimensional semantic blueprint and maintain the bidirectional mapping relationship between each node in the blueprint and the downstream generated products.

[0018] As a further technical limitation, a multi-dimensional semantic blueprint data dashboard generation system based on a large model also includes a component generation planning and best practice knowledge base, which is used to store the optimal configuration rules, style templates and interaction modes of different visualization component types in different business scenarios; the dashboard production expert intelligent agent queries the knowledge base to obtain recommended configurations when instantiating UI components.

[0019] According to some embodiments, the third solution of the present invention provides a multi-dimensional semantic blueprint data large screen generation system based on a large model, which adopts the following technical solution: A system for generating multidimensional semantic blueprint data dashboards based on large models includes: The acquisition module is configured to retrieve natural language instructions for user data; The generation module is configured to generate a multidimensional semantic blueprint based on the natural language instructions obtained from parsing the large language model. The parsing module is configured to parse the generated multidimensional semantic blueprint, resulting in a data task queue and a view task queue. The execution module is configured to execute tasks in the obtained data task queue and view task queue through different expert agents, and update the task status. The generation module is configured to perform completion verification and closed-loop correction after all tasks are completed, and generate a data dashboard; it completes the generation of a multi-dimensional semantic blueprint data dashboard based on a large model.

[0020] Compared with the prior art, the beneficial effects of the present invention are as follows: This invention employs a multi-dimensional semantic blueprint to transform natural language requirements into a structured, executable intermediate representation. Through an engineering-based decomposition mechanism of "blueprint → dual task queues," the blueprint is automatically decomposed into logically separate data task queues and view task queues with dependencies. Based on state synchronization and asynchronous multi-agent collaboration, data expert agents and dashboard production expert agents engage in dependency-driven asynchronous collaboration by monitoring shared states. A bidirectional tracking mapping from blueprint nodes to engineering products is established, enabling a closed-loop maintenance system for lossless blueprint-level correction. This further solves the problems of uncontrollable generation results, lack of engineering collaboration, and inability to achieve closed-loop correction in existing technologies, realizing efficient, controllable, and intelligent generation and maintenance of industrial-grade data dashboards. Attached Figure Description

[0021] The accompanying drawings, which form part of this embodiment, are used to provide a further understanding of this embodiment. The illustrative embodiments and their descriptions are used to explain this embodiment and do not constitute an improper limitation of this embodiment.

[0022] Figure 1 This is a flowchart of the method for generating a large-scale multidimensional semantic blueprint data screen based on a large model in Embodiment 1 of the present invention; Figure 2 This is a structural block diagram of the multidimensional semantic blueprint data large screen generation system based on a large model in Embodiment 2 of the present invention; Figure 3 This is a structural block diagram of the multidimensional semantic blueprint data large screen generation system based on a large model in Embodiment 3 of the present invention. Detailed Implementation

[0023] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0024] It should be noted that the following detailed descriptions are exemplary and intended to provide further illustration of the invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0025] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of exemplary embodiments according to the invention. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0026] In this invention, terms such as "upper," "lower," "left," "right," "front," "back," "vertical," "horizontal," "side," and "bottom" indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. These terms are used only to facilitate the description of the structural relationships of the various components or elements of this invention and do not specifically refer to any component or element in this invention. They should not be construed as limiting the invention.

[0027] In this invention, terms such as "fixed connection," "connected," and "linked" should be interpreted broadly, indicating a fixed connection, an integral connection, or a detachable connection; a direct connection or an indirect connection through an intermediate medium. Those skilled in the art can determine the specific meaning of these terms in this invention based on the specific circumstances, and they should not be construed as limitations on the invention.

[0028] Where there is no conflict, the embodiments and features in the embodiments of the present invention can be combined with each other.

[0029] Example 1 Embodiment 1 of this invention introduces a method for generating a large-scale multidimensional semantic blueprint data dashboard based on a large model.

[0030] like Figure 1 The method for generating a multidimensional semantic blueprint data dashboard based on a large model, as shown, includes: Natural language commands to retrieve user data; Based on the natural language instructions obtained from parsing the large language model, a multidimensional semantic blueprint is generated. The generated multidimensional semantic blueprint is parsed to obtain the data task queue and the view task queue; Different expert agents execute tasks in the obtained data task queue and view task queue respectively, and update the task status; After all tasks are completed, a completion verification and closed-loop correction are performed to generate a data dashboard; the generation of a multi-dimensional semantic blueprint data dashboard based on a large model is completed.

[0031] This embodiment uses a "multidimensional semantic blueprint" as the control center, a structured intermediate representation layer that lies between natural language and the final code / interface, combining business semantics and engineering constraints. The multidimensional semantic blueprint serves as the sole source of facts and task contract driving the entire automated software engineering process, achieving auditable and deterministic transformation from natural language instructions to the final application. This embodiment employs an engineering decomposition mechanism of "blueprint -> dual task queue," automatically decomposing a blueprint into two logically separate but dependent task lists: data construction and view rendering, and explicitly managing the dependencies between them. This embodiment uses a state-synchronized multi-agent asynchronous collaboration method, where data expert agents and large-screen production expert agents collaborate asynchronously and dependently by monitoring the shared task queue state, reducing system coupling and improving concurrency efficiency and robustness. It supports a closed-loop maintenance system for lossless blueprint-level correction, establishing a bidirectional tracking mapping from blueprint nodes to generated artifacts (data queries, UI components), enabling blueprint-level modifications to automatically and accurately trigger incremental, decremental, or update operations in downstream engineering artifacts.

[0032] As one or more implementation methods, this embodiment first constructs a progressive blueprint, specifically: Receive natural language commands from users; The Large Language Model (LLM) is used to parse instructions and generate or update the initial "multidimensional semantic blueprint". The generated multidimensional semantic blueprint is a structured data containing multidimensional information such as business nodes, data constraints, layout attributes, and interaction logic. The sketch renderer immediately renders the blueprint as a low-fidelity visual sketch for the user to preview; Users can directly manipulate the sketch (such as dragging and adjusting the layout, modifying component types) or edit the blueprint text at this stage, and update the other side synchronously to achieve "blueprint-sketch" linkage; After the user confirms that everything is correct, the blueprint status will be set to "locked".

[0033] It should be noted that this embodiment solidifies vague requirements into an executable "technology contract" by constructing an incremental blueprint.

[0034] As one or more implementation methods, this embodiment decomposes the obtained multidimensional semantic blueprint into a task queue; specifically: The task scheduling engine parses the locked blueprint; The engine automatically breaks down the blueprint into two logically separate but dependent task queues: the data task queue and the view task queue. The data task queue contains task items such as the dataset identifier to be created, required fields, aggregation calculation logic, and data source mapping relationships. The view task queue contains task items such as the type of visualization component to be instantiated, layout coordinates, and style configuration. Each view task will explicitly reference the data task ID it depends on. Dependencies between tasks are explicitly managed, and a view task is only allowed to execute when all data tasks it depends on are marked as "complete".

[0035] As one or more implementation methods, this embodiment performs asynchronous cooperation of multiple agents based on state synchronization, specifically: (1) Data task processing branch The data expert agent proactively pulls tasks to be processed from the data task queue; Based on the business semantic description in the blueprint, map it with the backend data platform and execute or generate corresponding data query and processing scripts; Once the task is completed, update the status of the data task in the queue to "Done" and output the standardized data access interface.

[0036] (2) View Task Processing Branch Large screen production expert intelligent agent listening view task queue; The task scheduling engine will only assign a view task to the large-screen intelligent agent when all the dependent data tasks of a view task are in the "Done" state. When the large-screen intelligent agent performs a task, it will query the component knowledge base to obtain best practices, then create the corresponding component on the canvas, obtain the interface from the data intelligent agent and complete the data binding, apply styles, and configure interactive events; Once a view task is completed, its status is marked as "Done".

[0037] As one or more implementation methods, after all tasks in this embodiment are completed, the coverage of business nodes in the blueprint with rendered components and bound data sources is compared to generate an integrity report.

[0038] It should be noted that when a user needs to modify the generated large screen, a closed-loop correction process is initiated. Specifically: Users can directly edit locked blueprints (such as deleting or modifying nodes); Based on the mapping relationship between blueprint nodes and engineering artifacts maintained in the version control repository, the differences are automatically calculated; Based on the differences, the task scheduling engine only issues incremental, decremental, or update tasks to the relevant agents (such as only updating the affected data queries and corresponding components), driving them to perform partial updates or resource reclamation, thus avoiding global reconstruction.

[0039] This embodiment takes the generation of a "real-time production line monitoring dashboard" as an example to provide a detailed introduction to the multi-dimensional semantic blueprint data dashboard generation method based on a large model proposed in this embodiment: (1) Blueprint construction and anchoring User input: "Display today's real-time output, equipment status (running / shutdown), and OEE for production lines A and B, with anomaly alerts"; Generate a multi-dimensional semantic blueprint, which includes: a "KPI Overview" group (embedded with output and OEE cards), an "Equipment Status Panel" (matrix diagram), and an "Anomaly Scrolling List".

[0040] Combining the generated multidimensional semantic blueprint and the sketch renderer, a wireframe sketch containing these three regions is generated. The user finds the "Device Status" not intuitive enough and drags it directly onto the sketch to change it to "Topology Map"; simultaneously, the "Component Type: Matrix Diagram" in the blueprint is updated to "Component Type: Topology Map"; after the user confirms everything is correct, the blueprint is locked.

[0041] (2) Task decomposition and collaboration The task engine parses the blueprint and creates tasks. For example: Data Task 1: Obtain the real-time production flow (ID: D001) of production lines A and B. Data Task 2: Calculate the real-time OEE (ID: D002) for production lines A and B. Data Task 3: Acquire device heartbeat and status event streams. (ID: D003) View Task 1: Create a KPI card group, depending on [D001, D002] (ID: V001); View Task 2: Create a production line equipment topology diagram, dependent on [D003] (ID: V002); View Task 3: Create an exception list, dependent on [D003] (ID: V003); The data agent processes D001, D002, and D003 in parallel; when D003 is completed, it notifies the task queue.

[0042] The large-screen intelligent agent sees that the dependencies of V002 and V003 on D003 have been satisfied, and then begins to execute, creating a topology graph and list, and binding data from D003.

[0043] (3) Closed-loop correction One day later, the user wanted to remove the monitoring for production line B, specifically: Users can directly edit the blueprint and delete all nodes related to "Production Line B".

[0044] When a blueprint change is detected, the differences are automatically calculated: the query conditions for data tasks D001 and D002 need to be changed (remove B), and the configurations for view tasks V001 and V002 need to be updated. Only incremental tasks updating D001 and D002 are issued to the data agent, instead of rebuilding the entire process. After the data update, the rebinding of V001 and V002 is automatically triggered. View task V003 is unaffected and continues to run. The entire correction process is fast and accurate, and users don't need to worry about how to query backend data or modify frontend code.

[0045] This embodiment breaks through the black-box paradigm of existing "one-step generation" AI tools, taking the "multi-dimensional semantic blueprint" of the structured intermediate representation layer, which combines business readability and machine executableness, as the core control center and the sole source of facts. The user's natural language requirements are first "frozen" into this deterministic, auditable, and editable technical contract, and all subsequent data acquisition, component rendering, and layout adjustments strictly follow the blueprint definition. This fundamentally eliminates the randomness and uncontrollability issues (i.e., the "illusion" phenomenon) brought about by directly generating the final code from a large language model, ensuring that the generated large-screen application strictly meets user expectations in terms of business logic, data constraints, and interaction rules, and satisfies the stringent requirements of zero errors in industrial scenarios.

[0046] This embodiment employs an engineering-based decomposition mechanism of "blueprint → dual task queues" and "asynchronous collaboration of multiple agents based on state synchronization." The task scheduling engine automatically decomposes a complex blueprint into logically separated, clearly defined data task queues and view task queues, managed through explicit dependencies. Data expert agents and large-screen production expert agents do not need to communicate directly; instead, they collaborate asynchronously and through dependencies by listening to and updating a shared global state table. This achieves decoupling and engineering-based collaboration between data logic and view logic, significantly improving system concurrency efficiency and reliability.

[0047] This embodiment achieves closed-loop maintenance that supports lossless blueprint-level correction by establishing a version management library and maintaining a two-way fine-grained mapping between "blueprint nodes and downstream products." When user requirements change, there is no need to discard all results and regenerate; the differences between the old and new blueprints are automatically calculated and transformed into incremental, decremental, or update instructions for downstream engineering products (data queries, UI components). After receiving the instructions, each intelligent agent executes precise local operations, completely changing the predicament of "a change in one part affects the whole" in the traditional development model. This enables industrial large screens to be continuously iterated and maintained like mature software engineering with extremely low cost and extremely high response speed, greatly extending the life cycle of large screen applications and reducing the total cost of ownership for enterprises.

[0048] The bidirectional linkage mechanism between the "low-fidelity visual sketch" and the "multi-dimensional semantic blueprint" in this embodiment provides users with timely visual feedback. Users do not need to understand the complex blueprint data structure; they can directly modify the sketch through intuitive operations such as dragging and clicking, and the underlying blueprint will be automatically updated synchronously. This greatly enhances the efficiency of human-computer collaboration, enabling business experts who do not understand programming or databases to independently and efficiently complete the construction and iteration of industrial-grade data dashboards, truly realizing the democratization and democratization of data visualization capabilities.

[0049] Example 2 Embodiment 2 of the present invention introduces a multi-dimensional semantic blueprint data large screen generation system based on a large model, which is used to execute the multi-dimensional semantic blueprint data large screen generation method based on a large model introduced in Embodiment 1.

[0050] like Figure 2 The system shown is a multidimensional semantic blueprint data dashboard generation system based on a large model, comprising: The blueprint generation and interaction module is used to receive natural language commands from users, call a large language model to parse the natural language commands, generate or update multidimensional semantic blueprints, and provide a graphical user interface for displaying and editing the low-fidelity visual sketches corresponding to the multidimensional semantic blueprints, so as to realize the two-way linkage between multidimensional semantic blueprints and low-fidelity visual sketches. The task decomposition and scheduling engine is used to parse locked multidimensional semantic blueprints to obtain data task queues and view task queues, and calculate incremental change sets based on blueprint changes. At least one data expert agent is used to listen to and pull tasks from the data task queue, perform data queries or processing, and update task status. At least one large screen production expert intelligent agent is used to listen to the tasks in the view task queue, pull the task after confirming that the data dependency of the task is satisfied, instantiate the UI component, complete the data binding and style configuration, and update the task status. The version management library is used to store different versions of the multidimensional semantic blueprint and maintain the bidirectional mapping relationship between each node in the blueprint and the downstream generated products.

[0051] As a further technical limitation, the multi-dimensional semantic blueprint data large screen generation system based on a large model in this embodiment includes a component generation planning and best practice knowledge base, which is used to store the optimal configuration rules, style templates and interaction modes of different visualization component types in different business scenarios; the large screen production expert intelligent agent queries the knowledge base to obtain recommended configurations when instantiating UI components.

[0052] Example 3 Embodiment 3 of this invention introduces a multidimensional semantic blueprint data large screen generation system based on a large model.

[0053] like Figure 3 The system shown is a multidimensional semantic blueprint data dashboard generation system based on a large model, comprising: The acquisition module is configured to retrieve natural language instructions for user data; The generation module is configured to generate a multidimensional semantic blueprint based on the natural language instructions obtained from parsing the large language model. The parsing module is configured to parse the generated multidimensional semantic blueprint, resulting in a data task queue and a view task queue. The execution module is configured to execute tasks in the obtained data task queue and view task queue through different expert agents, and update the task status. The generation module is configured to perform completion verification and closed-loop correction after all tasks are completed, and generate a data dashboard; it completes the generation of a multi-dimensional semantic blueprint data dashboard based on a large model.

[0054] The detailed steps are the same as those of the method for generating a large-scale multidimensional semantic blueprint data dashboard based on a large model provided in Example 1, and will not be repeated here.

[0055] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0056] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

[0057] The above description is merely a preferred embodiment of this practice and is not intended to limit the scope of this practice. Various modifications and variations can be made to this practice by those skilled in the art. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of this practice should be included within the protection scope of this practice.

Claims

1. A method for generating a large-scale multidimensional semantic blueprint data dashboard based on a large model, characterized in that, include: Natural language commands to retrieve user data; Based on the natural language instructions obtained from parsing the large language model, a multidimensional semantic blueprint is generated. The generated multidimensional semantic blueprint is parsed to obtain the data task queue and the view task queue; Different expert agents execute tasks in the obtained data task queue and view task queue respectively, and update the task status; After all tasks are completed, a completion check and closed-loop correction will be performed to generate a data dashboard. Complete the generation of a multidimensional semantic blueprint data dashboard based on a large model.

2. The method for generating a large-scale multidimensional semantic blueprint data dashboard based on a large model as described in claim 1, characterized in that, When a user needs to modify the generated data dashboard, the incremental change set is calculated based on the differences in the multidimensional semantic blueprint, and incremental task instructions are issued to the relevant agents to drive the agents to perform local updates and achieve closed-loop correction.

3. The method for generating a large-scale multidimensional semantic blueprint data dashboard based on a large model as described in claim 1, characterized in that, The specific process of the closed-loop correction is as follows: When a user edits a locked multidimensional semantic blueprint, a new version of the blueprint is generated; By comparing the differences between the old and new versions of the blueprint, and combining the two-way mapping relationship between blueprint nodes and downstream engineering products, the incremental change set is automatically calculated, which includes at least the task items that need to be added, deleted, and updated. Based on the incremental change set, incremental, reduction, or update task instructions are issued only to the relevant data expert agents and / or large screen production expert agents. The relevant intelligent agents execute precise local updates or resource reclamation based on the received instructions, realize automatic reconstruction in response to changes in requirements, and complete automatic correction.

4. The method for generating a large-scale multidimensional semantic blueprint data dashboard based on a large model as described in claim 1, characterized in that, At least one data expert agent executes tasks in the data task queue, interacts with the data platform to complete data queries or processing, and updates task status. The view task queue is monitored by at least one large-screen production expert intelligent agent. After confirming that the data dependencies of the task are met, the view task is executed, UI components are instantiated and data binding and style configuration are completed, and the task status is updated.

5. The method for generating a large-scale multidimensional semantic blueprint data dashboard based on a large model as described in claim 1, characterized in that, After acquiring natural language instructions, the acquired natural language instructions are parsed based on a large language model to generate or update a multidimensional semantic blueprint. The multidimensional semantic blueprint is then rendered in real time as a low-fidelity visual sketch for user preview and interaction. Once the user confirms, the multidimensional semantic blueprint is locked.

6. The method for generating a large-scale multidimensional semantic blueprint data dashboard based on a large model as described in claim 5, characterized in that, The multidimensional semantic blueprint and the low-fidelity visualization sketch are updated and linked through a two-way data binding mechanism.

7. The method for generating a large-scale multidimensional semantic blueprint data dashboard based on a large model as described in claim 5, characterized in that, After parsing and locking the multidimensional semantic blueprint, it is automatically decomposed to obtain a logically separated data task queue and a view task queue with dependencies; each view task item in the view task queue explicitly records the identifier of the data task it depends on. A view task is only allowed to execute if all its dependent data tasks are marked as completed in the global task state.

8. A system for generating multidimensional semantic blueprint data dashboards based on large models, used to execute the method for generating multidimensional semantic blueprint data dashboards based on large models as described in any one of claims 1-7, characterized in that, include: The blueprint generation and interaction module is used to receive natural language commands from users, call a large language model to parse the natural language commands, generate or update multidimensional semantic blueprints, and provide a graphical user interface for displaying and editing the low-fidelity visual sketches corresponding to the multidimensional semantic blueprints, so as to realize the two-way linkage between multidimensional semantic blueprints and low-fidelity visual sketches. The task decomposition and scheduling engine is used to parse locked multidimensional semantic blueprints to obtain data task queues and view task queues, and calculate incremental change sets based on blueprint changes. At least one data expert agent is used to listen to and pull tasks from the data task queue, perform data queries or processing, and update task status. At least one large screen production expert intelligent agent is used to listen to the tasks in the view task queue, pull the task after confirming that the data dependency of the task is satisfied, instantiate the UI component, complete the data binding and style configuration, and update the task status. The version management library is used to store different versions of the multidimensional semantic blueprint and maintain the bidirectional mapping relationship between each node in the blueprint and the downstream generated products.

9. The system for generating a large-scale multidimensional semantic blueprint data dashboard based on a large model as described in claim 8, characterized in that, It also includes a component generation plan and best practice knowledge base, which stores the optimal configuration rules, style templates and interaction modes of different visualization component types in different business scenarios; the large screen production expert intelligent agent queries the knowledge base to obtain recommended configurations when instantiating UI components.

10. A system for generating large-scale multidimensional semantic blueprint data dashboards based on large models, characterized in that: include: The acquisition module is configured to retrieve natural language instructions for user data; The generation module is configured to generate a multidimensional semantic blueprint based on the natural language instructions obtained from parsing the large language model. The parsing module is configured to parse the generated multidimensional semantic blueprint, resulting in a data task queue and a view task queue. The execution module is configured to execute tasks in the obtained data task queue and view task queue through different expert agents, and update the task status. The generation module is configured to perform completion verification and closed-loop correction after all tasks are completed, and then generate a data dashboard. Complete the generation of a multidimensional semantic blueprint data dashboard based on a large model.

Images