Method, system and equipment for realizing application generation and storage medium

Abstract

The invention discloses a method, a system and equipment for realizing application generation and a storage medium, belongs to the technical field of computer science and information, and solves the technical problem that an existing low-code platform needs a large amount of manual configuration and code writing. The method comprises the steps of receiving a demand of a user and converting the demand into semantics which can be recognized by a large-scale language model through a dialogue between an AI agent and the user, maintaining a dialogue state to manage context information, and obtaining detailed information and configuration data information of a low-code platform API (Application Program Interface); pre-training the large language model, and performing fine tuning according to a business scene; inputting the dialogue content into the fine-tuned large language model to obtain a user intention; decomposing the application creation task into a plurality of sub-tasks according to the user intention; and planning a task execution sequence according to a user intention, calling a low-code platform API and configuration data information to generate a low-code application, and completing service rendering. And the development efficiency and the application intelligence level can be improved.

Description

Technical Field

[0001] The present invention relates to the fields of computer science and information technology, and more specifically, to a method, system, device and storage medium for realizing application generation. Background Art

[0002] As enterprises accelerate their digital transformation, low-code platforms have attracted widespread attention due to their rapid development and deployment capabilities.

[0003] However, the existing low-code platforms still need to be improved in terms of intelligence, especially in terms of automated application generation and intelligent question answering. These platforms usually require a lot of manual configuration and code writing, which limits development efficiency and the level of intelligence of applications. Summary of the invention

[0004] The technical problem to be solved by the present invention is to address the above-mentioned shortcomings of the prior art. One purpose of the present invention is to provide a method for realizing application generation, which can improve the development efficiency of the low-code platform and the intelligence level of the application.

[0005] A second objective of the present invention is to provide a system for realizing application generation.

[0006] A third object of the present invention is to provide a computer device.

[0007] A fourth objective of the present invention is to provide a computer storage medium.

[0008] In order to achieve the above objective 1, the present invention provides a method for implementing application generation, comprising the following steps:

[0009] Step 1. Use AI agents to communicate with users, receive user needs, and convert them into semantics that can be recognized by large language models;

[0010] Step 2. Maintaining a dialogue state by the AI ​​agent to manage context information to store key information and history of the dialogue;

[0011] Step 3. Obtain detailed information of the low-code platform API through the AI ​​agent, and obtain configuration data information for creating an application;

[0012] Step 4. Pre-train the large language model, and fine-tune the pre-trained large language model according to the business scenario to meet the needs of specific tasks;

[0013] Step 5: Input the conversation content between the AI ​​agent and the user into the fine-tuned large language model to obtain the user's intention;

[0014] Step 6. Decompose the application creation task into multiple subtasks according to the user's intention through the AI ​​agent, and the multiple subtasks include: model creation subtask, association model creation subtask, form creation subtask, list creation subtask, process creation subtask, menu creation subtask, permission creation subtask, report and graphic creation subtask;

[0015] Step 7. The AI ​​agent plans the order of task execution according to user intentions, and calls the low-code platform API and configuration data information to complete business model creation, association model creation, form creation, list creation, process creation, menu creation, permission creation, report and graphics creation, generate low-code applications, and complete business rendering.

[0016] As a further improvement, the method further includes step 8, monitoring the operating status of the AI ​​agent in real time and obtaining user feedback information; when the AI ​​agent encounters a situation that it cannot handle, the problem is reported to the developer for decision, or the AI ​​agent is self-repaired according to a preset error handling process.

[0017] Furthermore, step 9 is also included, which performs self-repair and optimization on the AI ​​agent based on the collected feedback information and error logs to improve the response quality and efficiency of the subsequent low-code application generated by the AI ​​agent.

[0018] Furthermore, the AI ​​agent communicates with the user through NLP technology, analyzes the user's semantic needs, and converts the semantic needs into machine-understandable instructions.

[0019] Furthermore, the dialogue state is a structured memory that stores key information and history of the dialogue; enabling the AI ​​agent to maintain contextual consistency in multiple rounds of dialogue and improve the naturalness and accuracy of the interaction; enabling the AI ​​agent to have not only short-term memory and remember current contextual information, but also long-term memory and be able to retain and call up large amounts of information over a long period of time, and accurately understand and remember user needs.

[0020] Furthermore, the AI ​​agent obtains detailed information of the API by querying the API interface of business model creation, form creation, list creation, process creation, menu creation, permission creation, report and its graphics creation corresponding to the low-code platform design;

[0021] The AI ​​agent obtains the configuration data information for creating the application by calling the database query service, and prepares data for the creation of the low-code application.

[0022] Furthermore, the AI ​​agent establishes a form that conforms to the business through the online form design API provided by the low-code platform. The form is used for data collection and process approval, and to complete business rendering. The form supports online preview. After the low-code application is generated, the user can use the low-code platform model designer to render or fine-tune it.

[0023] In order to achieve the above-mentioned second objective, the present invention provides a system for implementing application generation, comprising:

[0024] The dialogue module is used to communicate with users through AI agents, receive user needs and convert them into semantics that can be recognized by large language models;

[0025] A context management module, for managing context information by maintaining a dialogue state through the AI ​​agent to store key information and history of the dialogue;

[0026] An acquisition module is used to obtain detailed information of the low-code platform API through the AI ​​agent, and to obtain configuration data information for creating an application;

[0027] A training module, used to pre-train the large language model and fine-tune the pre-trained large language model according to business scenarios to meet the needs of specific tasks;

[0028] An intention module, which is used to input the conversation content between the AI ​​agent and the user into a fine-tuned large language model to obtain the user's intention;

[0029] A first creation module is used to decompose the application creation task into multiple subtasks according to the user's intention through the AI ​​agent, and the multiple subtasks include: a model creation subtask, an association model creation subtask, a form creation subtask, a list creation subtask, a process creation subtask, a menu creation subtask, a permission creation subtask, and a report and graphic creation subtask;

[0030] The second creation module is used to plan the order of task execution according to user intentions through the AI ​​agent, and call the low-code platform API and configuration data information to complete business model creation, association model creation, form creation, list creation, process creation, menu creation, permission creation, report and graphics creation, generate low-code applications, and complete business rendering.

[0031] In order to achieve the third objective above, the present invention provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor implements the above-mentioned method for generating an application when executing the computer program.

[0032] In order to achieve the fourth objective above, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program implements the above-mentioned method for generating an application when executed by a processor.

[0033] Beneficial Effects

[0034] Compared with the prior art, the present invention has the following advantages:

[0035] 1. The present invention can improve development efficiency and reduce manual configuration and code writing.

[0036] 2. The present invention can improve the intelligence level of applications and achieve more natural interaction and smarter business process management through LLM technology.

[0037] 3. The present invention can lower the technical threshold, so that non-technical personnel can also quickly generate application programs.

[0038] 4. The present invention can improve the scalability and flexibility of the system and adapt to rapidly changing business needs. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 It is a flow chart of the present invention. DETAILED DESCRIPTION

[0040] The present invention will be further described below with reference to the specific embodiments in the accompanying drawings.

[0041] See also Figure 1 , a method for realizing application generation, comprising the following steps 1 to 7:

[0042] Step 1. Through the AI ​​agent, the user communicates with the user, receives the user's needs and converts them into semantics that can be recognized by the large language model (LLM).

[0043] Specifically, the AI ​​agent communicates with users through NLP technology, analyzes users' semantic needs, and converts semantic needs into machine-understandable instructions.

[0044] Step 2. Manage context information by maintaining a dialogue state through the AI ​​agent to store key information and history of the conversation.

[0045] The dialogue state is a structured memory that stores the key information and history of the dialogue. It enables the AI ​​agent to maintain contextual consistency in multiple rounds of dialogue, improving the naturalness and accuracy of the interaction. It enables the AI ​​agent to have not only short-term memory to remember the current contextual information, but also long-term memory to retain and call large amounts of information over a long period of time, and accurately understand and remember user needs.

[0046] That is, the AI ​​agent receives the dialogue submitted by the user through the memory module, so that it can retain the knowledge learned and the real needs of the user during the interaction. This mechanism allows the agent to reuse the user's previous experience when handling repetitive work, thereby avoiding a large number of repeated interactions by the user. Short-term memory is applicable to learning in all contexts, while long-term memory retains knowledge and interaction memories. For example, the large amount of data and experience accumulated by the agent in a specific industry can provide more professional, in-depth and personalized answers, generate applications more efficiently, and improve user experience.

[0047] Step 3. Obtain detailed information about the low-code platform API through the AI ​​agent, as well as configuration data information for creating an application.

[0048] The AI ​​agent obtains detailed information of the API by querying the API interfaces for business model creation, form creation, list creation, process creation, menu creation, permission creation, report creation and its graphics creation corresponding to the low-code platform design.

[0049] The AI ​​agent obtains the configuration data information for creating the application by calling the database query service, and prepares data for the creation of low-code applications.

[0050] Step 4. Pre-train the large language model and fine-tune the pre-trained large language model according to the business scenario to meet the needs of specific tasks.

[0051] During the pre-training phase, large language models learn the statistical laws, semantic relationships, and grammatical structures of a language on a large amount of unlabeled text data.

[0052] In this embodiment, Baidu's mature LLM large model can be directly used, and the fine-tuning stage is trained on a specific data set according to the specific business scenario.

[0053] Step 5. Input the conversation content between the AI ​​agent and the user into the fine-tuned large language model to obtain the user's intention.

[0054] Step 6. Use the AI ​​agent to decompose the application creation task into multiple subtasks based on user intent. The multiple subtasks include: model creation subtask, associated model creation subtask, form creation subtask, list creation subtask, process creation subtask, menu creation subtask, permission creation subtask, report and graphics creation subtask.

[0055] Step 7. Use the AI ​​agent to plan the order of task execution according to user intent, and call the low-code platform API and configuration data information to complete business model creation, association model creation, form creation, list creation, process creation, menu creation, permission creation, report and graphics creation, generate low-code applications, and complete business rendering.

[0056] The AI ​​agent uses the online form design API provided by the low-code platform to create a form that meets the business requirements. The form is used for data collection and process approval, and to complete business rendering. The form supports online preview. After the low-code application is generated, users can use the low-code platform model designer to render or fine-tune it.

[0057] Furthermore, the method further includes step 8, monitoring the running status of the AI ​​agent in real time and obtaining user feedback information. When the AI ​​agent encounters a situation that it cannot handle, it reports the problem to the developer for decision-making, or the AI ​​agent is self-repaired according to a preset error handling process.

[0058] By designing prompts to guide large model disassembly tasks, accessing private data center APIs and work report application APIs, and extracting historical report features of developers to form a long memory library, the intelligent agent can automatically generate and submit work reports, achieving full automation.

[0059] Furthermore, step 9 is also included, in which the AI ​​agent is self-repaired and optimized according to the collected feedback information and error logs to improve the response quality and efficiency of the low-code application generated by the AI ​​agent in the subsequent sequence. That is, the AI ​​agent is continuously self-repaired and optimized according to the collected feedback information and error logs, so that the description information of the AI ​​agent is as accurate and clear as possible, so as to improve the accuracy of large model recognition and realize online debugging and optimization.

[0060] A system for implementing application generation, comprising:

[0061] The dialogue module is used to communicate with users through AI agents, receive user needs and convert them into semantics that can be recognized by large language models;

[0062] The context management module is used to manage context information by maintaining a dialogue state through the AI ​​agent to store key information and history of the dialogue;

[0063] The acquisition module is used to obtain detailed information about the low-code platform API through the AI ​​agent, as well as configuration data information for creating applications;

[0064] The training module is used to pre-train large language models and fine-tune the pre-trained large language models according to business scenarios to meet the needs of specific tasks;

[0065] The intent module is used to input the conversation content between the AI ​​agent and the user into the fine-tuned large language model to obtain the user's intent;

[0066] The first creation module is used to decompose the application creation task into multiple subtasks according to the user's intention through the AI ​​agent. The multiple subtasks include: model creation subtask, association model creation subtask, form creation subtask, list creation subtask, process creation subtask, menu creation subtask, permission creation subtask, report and graphic creation subtask;

[0067] The second creation module is used to plan the order of task execution according to user intentions through AI agents, and call the low-code platform API and configuration data information to complete business model creation, association model creation, form creation, list creation, process creation, menu creation, permission creation, report and graphics creation, generate low-code applications, and complete business rendering.

[0068] A computer device includes a memory and a processor. The memory stores a computer program. When the processor executes the computer program, the above-mentioned method for realizing application generation is implemented.

[0069] A computer-readable storage medium stores a computer program, which implements the above-mentioned method for generating an application when executed by a processor.

[0070] The above are only preferred embodiments of the present invention. It should be pointed out that for those skilled in the art, several modifications and improvements can be made without departing from the structure of the present invention, which will not affect the effect of the implementation of the present invention and the practicality of the patent.

Claims

1. A method for realizing application generation, characterized in that: The following steps are involved: Step 1. Use AI agents to communicate with users, receive user needs, and convert them into semantics that can be recognized by large language models; Step 2. Maintaining a dialogue state by the AI ​​agent to manage context information to store key information and history of the dialogue; Step 3. Obtain detailed information of the low-code platform API through the AI ​​agent, and obtain configuration data information for creating an application; Step 4. Pre-train the large language model, and fine-tune the pre-trained large language model according to the business scenario to meet the needs of specific tasks; Step 5: Input the conversation content between the AI ​​agent and the user into the fine-tuned large language model to obtain the user's intention; Step 6. Decompose the application creation task into multiple subtasks according to the user's intention through the AI ​​agent, and the multiple subtasks include: model creation subtask, association model creation subtask, form creation subtask, list creation subtask, process creation subtask, menu creation subtask, permission creation subtask, report and graphic creation subtask; Step 7. The AI ​​agent plans the order of task execution according to user intentions, and calls the low-code platform API and configuration data information to complete business model creation, association model creation, form creation, list creation, process creation, menu creation, permission creation, report and graphics creation, generate low-code applications, and complete business rendering.

2. A method for realizing application generation according to claim 1, characterized in that: The method also includes step 8, monitoring the operating status of the AI ​​agent in real time and obtaining user feedback information; when the AI ​​agent encounters a situation that it cannot handle, the problem is reported to the developer for decision-making, or the AI ​​agent is self-repaired according to a preset error handling process.

3. A method for realizing application generation according to claim 1, characterized in that: It also includes step 9, which performs self-repair and optimization on the AI ​​agent based on the collected feedback information and error logs to improve the response quality and efficiency of the subsequent low-code application generated by the AI ​​agent.

4. A method for realizing application generation according to claim 1, characterized in that: The AI ​​agent communicates with the user through NLP technology, analyzes the user's semantic needs, and converts the semantic needs into machine-understandable instructions.

5. A method for realizing application generation according to claim 1, characterized in that: The dialogue state is a structured memory that stores key information and history of the dialogue; it enables the AI ​​agent to maintain contextual consistency in multiple rounds of dialogue and improve the naturalness and accuracy of the interaction; it enables the AI ​​agent to have not only short-term memory and remember current contextual information, but also long-term memory and can retain and call large amounts of information over a long period of time, and accurately understand and remember user needs.

6. A method for realizing application generation according to claim 1, characterized in that: The AI ​​agent obtains detailed information of the API by querying the API interface for business model creation, form creation, list creation, process creation, menu creation, permission creation, report and its graphics creation corresponding to the low-code platform design; The AI ​​agent obtains the configuration data information for creating the application by calling the database query service, and prepares data for the creation of the low-code application.

7. A method for realizing application generation according to claim 1, characterized in that: The AI ​​agent establishes a form that conforms to the business through the online form design API provided by the low-code platform. The form is used for data collection and process approval, and to complete business rendering. The form supports online preview. After the low-code application is generated, users can use the low-code platform model designer to render or fine-tune it.

8. A system for implementing application generation, characterized in that: include: The dialogue module is used to communicate with users through AI agents, receive user needs and convert them into semantics that can be recognized by large language models; A context management module, for managing context information by maintaining a dialogue state through the AI ​​agent to store key information and history of the dialogue; An acquisition module is used to obtain detailed information of the low-code platform API through the AI ​​agent, and to obtain configuration data information for creating an application; A training module, used to pre-train the large language model and fine-tune the pre-trained large language model according to business scenarios to meet the needs of specific tasks; An intention module, which is used to input the conversation content between the AI ​​agent and the user into a fine-tuned large language model to obtain the user's intention; A first creation module is used to decompose the application creation task into multiple subtasks according to the user's intention through the AI ​​agent, and the multiple subtasks include: a model creation subtask, an association model creation subtask, a form creation subtask, a list creation subtask, a process creation subtask, a menu creation subtask, a permission creation subtask, and a report and graphic creation subtask; The second creation module is used to plan the order of task execution according to user intentions through the AI ​​agent, and call the low-code platform API and configuration data information to complete business model creation, association model creation, form creation, list creation, process creation, menu creation, permission creation, report and graphics creation, generate low-code applications, and complete business rendering.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the method for realizing application generation described in any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, a method for realizing application generation as described in any one of claims 1 to 7 is implemented.

Images