Intelligent fund management method and system based on large model

Abstract

The invention provides an intelligent fund management method and system based on a large model, and relates to the technical field of computers, and the method comprises the steps: sorting basic corpus documents of large model learning; selecting a suitable internet large model to carry out corpus training; business system functions are arranged for function training of a large model; receiving a demand described by a natural language of a user, converting the demand into a function corresponding to a service system by the large model, and sending an operation instruction; after receiving the operation instruction of the large model, the service system identifies whether an operation process conforming to the operation instruction exists or not, and if yes, the corresponding operation process is executed, and the service system is driven to operate; if the corresponding operation process is not recognized, the service system forms a prompt process for a user to judge whether to execute the operation according to the prompt process or not; if the user confirms the prompt process, calling a browser bottom layer API interface to obtain page content and elements, and driving a browser robot to execute operation according to the prompt process until the operation instruction of the large model is input completely.

Description

Technical Field

[0001] The present invention relates to the field of computer technology, and in particular to an intelligent fund management method and system based on a large model. Background Art

[0002] In the process of using the treasury cloud system, human errors are prone to occur due to the complexity and repetitiveness of the financial system, resulting in inefficiency. Due to its professional nature, treasury software requires users who are familiar with financial and capital processes to operate it, which increases the difficulty for new users to get started.

[0003] In order to solve the above problems, the current treasury system converts manual operations into control codes of RPA robots through programming and records them. Then, the RPA robot program controls the computer's mouse, keyboard and other input devices to replace manual operations. However, RPA robots are difficult to use. Their operating procedures require professional personnel to develop programs and handle operations such as simulated keyboard input, verification code recognition, button and information clicks, and file parsing. The development and debugging process is long, and due to simulated input, the probability of error is high. Summary of the invention

[0004] The purpose of the present invention is to provide an intelligent fund management method and system based on a large model to solve the problems of low intelligence level, repetitive and cumbersome operations and high development cost of the current fund management system.

[0005] To solve the above problems, the present invention first provides an intelligent fund management method based on a big model, comprising the following steps: arranging the corresponding contents of user needs and system operation modes as the basic corpus documents for big model learning; selecting an applicable Internet big model or an independently deployed big model according to customer needs for corpus training; arranging the functions in the business system that can be driven by the big model and forming training data for the big model to perform function training; receiving the needs described in natural language by the user, the big model converts the natural language description into the corresponding functions of the business system and issues operation instructions; after the business system receives the operation instruction of the big model, it identifies whether there is an operation process that matches it, and if so, executes the corresponding operation process to drive the business system to operate; if no matching operation process is identified, the business system forms a prompt process for the user to judge whether to perform the operation according to the prompt process; if the user confirms the prompt process, the driver is called to obtain the page content and elements, and the matching elements are found according to the path indicated by the big model, and the browser robot is driven to perform the operation according to the prompt process until the operation instruction input of the big model is completed; if the user does not agree with the prompt process, or cannot continue to execute the operation instruction, and the operation instruction is ambiguous, the user is prompted to manually intervene in the operation.

[0006] Furthermore, the content of the basic corpus document includes: business concepts, proper nouns, common parameters, and requirement descriptions.

[0007] Furthermore, after receiving the user's needs described in natural language, the big model extracts keywords from the natural language description, matches the keywords with the functions they can drive, identifies the operation type, operation object and operation parameters, and calls the corresponding business code to issue operation instructions.

[0008] The present invention also provides an intelligent fund management system based on a big model, including: a corpus training module, which selects an applicable Internet big model or an independently deployed big model according to customer needs, organizes the corresponding contents of user needs and system operation methods to form a basic corpus document, and the big model performs corpus training by learning the basic corpus document; a function training module, which organizes and forms training data according to the functions that can be driven by the big model in the business system, and the big model performs function training through the training data; a conversion module, which receives the user's natural language description of the needs, and the big model converts the natural language description into the corresponding function of the business system and issues an operation instruction; an execution module, the business system receives the big model and sends the corresponding function to the big model. After receiving the operation instruction of the large model, it identifies whether there is an operation process that matches it. If so, it executes the corresponding operation process and drives the business system to perform the operation. If no matching operation process is identified, the business system forms a prompt process for the user to judge whether to perform the operation according to the prompt process. If the user confirms the prompt process, the browser's underlying API interface is called to obtain the page content and elements, and the matching elements are found according to the path indicated by the large model. The browser robot is driven to perform the operation according to the prompt process until the operation instruction input of the large model is completed. If the user does not agree with the prompt process, or cannot continue to execute the operation instruction, or the operation instruction is ambiguous, the user is prompted to manually intervene in the operation.

[0009] Furthermore, the content of the basic corpus document includes: business concepts, proper nouns, common parameters, and requirement descriptions.

[0010] Furthermore, after receiving the user's needs described in natural language, the big model extracts keywords from the natural language description, matches the keywords with the functions they can drive, identifies the operation type, operation object and operation parameters, and calls the corresponding business code to issue operation instructions.

[0011] According to the big model-based intelligent fund management method and system provided by the present invention, by training the big model so that the big model can sequence the system functions and operation procedures, the big model can be used to convert the user's natural language description into a call to the business system function, so that the user can control the system through natural language, which reduces the difficulty of system operation and improves the convenience of system operation. For users who are not familiar with the system functions, natural language can also be used to guide and operate the functions, which reduces the threshold for using the business system. It also changes the current process that any function development requires targeted customized development, fully utilizes the learning and adaptability of the big model, and converts the user's changing business needs into existing functions of the business system, saving development costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.

[0013] Figure 1 A flow chart of an intelligent fund management method based on a large model provided by an embodiment of the present invention;

[0014] Figure 2 Operation diagram of the intelligent funds management system provided by an embodiment of the present invention (I);

[0015] Figure 3 Operation diagram of the intelligent fund management system provided by an embodiment of the present invention (II);

[0016] Figure 4 Operation diagram of the intelligent funds management system provided by the embodiment of the present invention (III);

[0017] Figure 5 Operation diagram of the intelligent funds management system provided by an embodiment of the present invention (IV). DETAILED DESCRIPTION

[0018] In order to make the above-mentioned objects, features and advantages of the present invention more clearly understood, the specific embodiments of the present invention are described in detail below in conjunction with the accompanying drawings. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0019] With the rapid development of artificial intelligence models, they can not only understand natural language descriptions, but also have the ability to identify and understand user behavior. Based on this technology, combined with RPA operation robots, it is applied to treasury system data acquisition and daily operations, and the user's simple and repetitive daily operation processes are identified and sorted out, and some system operations are automatically executed, or handed over to browser robots or RPA robots for operation, improving work efficiency and accuracy.

[0020] In the past, treasury systems used programming to convert manual operations into control codes for RPA robots and then recorded them. The RPA robot program then controlled the computer's mouse, keyboard, and other input devices to replace manual operations. The operation procedures of RPA robots require professional personnel to develop the program, and they also need to handle operations such as simulated keyboard input, verification code recognition, button and information clicks, and file parsing. The development and debugging process is long, and due to simulated input, the probability of error is high.

[0021] The intelligent treasury system based on the big model of the present invention, on the basis of training the business system functions of the big model, allows users to describe their business demands through natural language, and uses the big model to convert the user's natural language into specific operation steps, and call specific functions for operation. For actions that cannot be directly operated, combined with the browser interface or RPA programming interface, the operation steps are learned and solidified, and the treasury system page is automatically operated to realize the user's business demands.

[0022] Example 1

[0023] This embodiment provides an intelligent fund management method based on a large model, such as Figure 1 As shown, the following steps are included:

[0024] Organize the correspondence between user requirements and system operation methods, including business concepts, proper nouns, common parameters, and requirement descriptions, which serve as the basic corpus document for large model learning;

[0025] According to customer needs, select the applicable Internet big model or independently deployed big model for corpus training;

[0026] Organize the functions in the business system that can be driven by the big model and generate training data for the big model to perform functional training;

[0027] The big model receives the user's natural language description of the needs, and converts the natural language description into the corresponding functions of the business system. Specifically, the big model extracts keywords from the natural language description, matches the keywords with the functions they can drive, identifies the operation type, operation object and operation parameters, and calls the corresponding business code to issue operation instructions;

[0028] After receiving the operation instruction from the big model, the business system identifies whether there is an operation process that matches it. If so, it executes the corresponding operation process to drive the business system to operate;

[0029] If no matching operation process is identified, the business system will generate a prompt process for the user to decide whether to perform the operation according to the prompt process;

[0030] If the user confirms the prompt process, the driver is called to obtain the page content and elements. According to the path indicated by the large model, the matching elements are found, and the browser robot is driven to perform operations according to the prompt process until the operation instructions of the large model are input;

[0031] If the user does not agree with the prompt process, or is unable to continue to execute the operation instruction, or the operation instruction is ambiguous, the user will be prompted to manually intervene.

[0032] Example 2

[0033] The present invention also provides an intelligent fund management system based on a large model, comprising:

[0034] The corpus training module organizes the corresponding content of user needs and system operation methods into a basic corpus document, which includes business concepts, proper nouns, common parameters, and demand descriptions. The large model performs corpus training by learning the basic corpus document. The training process is as follows: Figure 2 As shown;

[0035] Specifically, according to customer needs, choose the applicable Internet big model or the independently deployed big model. The Internet big model can be connected to YiQianwen (Ali), WenxinYiYan (Baidu), and Xinghuo (iFLYTEK). If it is deployed independently, the customer needs to provide a server with high GPU configuration to provide the computing power of the AI ​​big model, such as NVIDIA A100, Huawei Ascend and other GPUs.

[0036] Function training module, according to the business system can be driven by the big model to form training data, the big model through the training data for functional training, the training process is as follows Figure 3 As shown;

[0037] The conversion module receives the user's natural language description of the needs. The big model converts the natural language description into the corresponding functions of the business system and issues operation instructions. Specifically, it can extract keywords from the natural language description, match the keywords with the functions they can drive, identify the operation type, operation object and operation parameters, and call the corresponding business code to issue the operation instruction;

[0038] Execution module: After the business system receives the operation instruction of the large model, it identifies whether there is an operation process that matches it. If so, it executes the corresponding operation process to drive the business system to operate, such as Figure 4 As shown;

[0039] If no matching operation process is identified, the business system generates a prompt process for the user to decide whether to perform the operation according to the prompt process. Figure 4 As shown;

[0040] If the user confirms the prompt process, the browser's underlying API interface is called to obtain the page content and elements. According to the path indicated by the large model, the matching elements are found, and the browser robot is driven to perform operations according to the prompt process until the operation instructions of the large model are input. Figure 5 As shown;

[0041] If the user does not agree with the prompt process, or is unable to continue to execute the operation instruction, or the operation instruction is ambiguous, the user will be prompted to manually intervene.

[0042] According to the big model-based intelligent fund management method and system provided by the present invention, by training the big model so that the big model can sequence the system functions and operation procedures, the big model can be used to convert the user's natural language description into a call to the business system function, so that the user can control the system through natural language, which reduces the difficulty of system operation and improves the convenience of system operation. For users who are not familiar with the system functions, natural language can also be used to guide and operate the functions, which reduces the threshold for using the business system. It also changes the current process that any function development requires targeted customized development, fully utilizes the learning and adaptability of the big model, and converts the user's changing business needs into existing functions of the business system, saving development costs.

[0043] In the description of this embodiment, it should be noted that those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment method can be implemented by instructing the control device through a computer program, and the program can be stored in a computer-readable storage medium. When the program is executed, it may include the processes of the above-mentioned method embodiments, wherein the storage medium may be a memory, a disk, an optical disk, etc.

[0044] Although the present invention is disclosed as above, the present invention is not limited thereto. Any person skilled in the art can make various changes and modifications without departing from the spirit and scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the scope defined by the claims.

[0045] Finally, it should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the presence of other identical elements in the process, method, article or device including the elements.

[0046] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0047] The above description of the disclosed embodiments enables one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to one 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 invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. An intelligent fund management method based on a large model, characterized in that: The following steps are involved: Organize the correspondence between user requirements and system operation methods as the basic corpus document for large model learning; According to customer needs, select the applicable Internet big model or independently deployed big model for corpus training; Organize the functions in the business system that can be driven by the big model and generate training data for the big model to perform functional training; The big model receives the user's natural language description of the needs, converts the natural language description into the corresponding functions of the business system, and issues operation instructions; After receiving the operation instruction from the big model, the business system identifies whether there is an operation process that matches it. If so, it executes the corresponding operation process to drive the business system to operate; If no matching operation process is identified, the business system will generate a prompt process for the user to decide whether to perform the operation according to the prompt process; If the user confirms the prompt process, the driver is called to obtain the page content and elements. According to the path indicated by the large model, the matching elements are found, and the browser robot is driven to perform operations according to the prompt process until the operation instructions of the large model are input; If the user does not agree with the prompt process, or is unable to continue to execute the operation instruction, or the operation instruction is ambiguous, the user will be prompted to manually intervene.

2. The intelligent fund management method based on a large model according to claim 1 is characterized in that: The content of the basic corpus document includes: business concepts, proper nouns, common parameters, and requirement descriptions.

3. The intelligent fund management method based on a large model according to claim 1 is characterized in that: After receiving the user's natural language description of the needs, the big model extracts keywords from the natural language description, matches the keywords with the functions they can drive, identifies the operation type, operation object and operation parameters, and calls the corresponding business code to issue operation instructions.

4. An intelligent fund management system based on a large model, characterized in that: include: The corpus training module selects the applicable Internet big model or independently deployed big model according to customer needs, and organizes the corresponding content of user needs and system operation methods into basic corpus documents. The big model performs corpus training by learning the basic corpus documents; Function training module: It organizes training data based on the functions in the business system that can be driven by the big model. The big model performs function training based on the training data. The conversion module receives the user's natural language description of the needs. The big model converts the natural language description into the corresponding functions of the business system and issues operation instructions. Execution module: After the business system receives the operation instruction of the large model, it identifies whether there is an operation process that matches it. If so, it executes the corresponding operation process to drive the business system to operate; If no matching operation process is identified, the business system will generate a prompt process for the user to decide whether to perform the operation according to the prompt process; If the user confirms the prompt process, the browser's underlying API interface is called to obtain the page content and elements. According to the path indicated by the large model, the matching elements are found, and the browser robot is driven to perform operations according to the prompt process until the operation instructions of the large model are input; if the user does not agree with the prompt process, or cannot continue to execute the operation instructions, or the operation instructions are ambiguous, the user is prompted to manually intervene.

5. The intelligent fund management system based on a large model according to claim 4 is characterized in that: The content of the basic corpus document includes: business concepts, proper nouns, common parameters, and requirement descriptions.

6. The intelligent fund management system based on a large model according to claim 4 is characterized in that: After receiving the user's natural language description of the needs, the big model extracts keywords from the natural language description, matches the keywords with the functions they can drive, identifies the operation type, operation object and operation parameters, and calls the corresponding business code to issue operation instructions.

Images