Software process arrangement method and device based on large model, medium and equipment

Through the software process orchestration method based on large models, nodes are created and adjusted using the process orchestration interface, the poor scalability and development complexity problems caused by hard coding in the existing technology are solved, and flexible orchestration process and more accurate model output are achieved. .

CN119088353APending Publication Date: 2024-12-06SHANDONG LANGCHAO YUNTOU INFORMATION TECH CO LTD
View PDF 0 Cites 2 Cited by

Patent Information

Application Number
CN202411195787.1
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-29
Publication Date
2024-12-06

AI Technical Summary

Technical Problem

The existing software process orchestration method adopts hard-coded methods, resulting in poor scalability and maintainability, high development difficulty and cycle, and a large amount of code is required to be written when dealing with complex business scenarios, resulting in inaccurate results of the model return.

Method used

Provide a software process orchestration method based on large models. By obtaining user needs, it creates input nodes, intermediate nodes and output nodes, including large model nodes, knowledge retrieval nodes and judgment nodes, forming an orchestration process, and drag components and dragging them through the process orchestration interface. Connect the graphics component components for node connection and adjustment until the orchestration process meets the software design requirements.

Benefits of technology

It improves the scalability and maintainability of the software system, reduces development difficulty and cycle, supports customized orchestration processes, optimizes the accuracy of model output results, and has a wide range of applicable scenarios.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119088353A_ABST
    Figure CN119088353A_ABST
Patent Text Reader

Abstract

The invention provides a software process arrangement method and device based on a large model, a medium and equipment. The method comprises the following steps: acquiring a software design requirement of a user; a flow arrangement interface is provided, so that a user creates an input node, at least one intermediate node and an output node on the flow arrangement interface, and the intermediate node comprises at least one of a large model node, a knowledge retrieval node and a judgment node; connecting the created nodes according to software design requirements, wherein an execution path formed from the input node to the output node is an arrangement process; determining functions which can be realized by the arrangement process; judging whether the function meets a software design requirement or not; if yes, outputting the programming process as a target arrangement process; otherwise, reminding the user to adjust the arrangement process, and outputting the adjusted arrangement process as the target arrangement process until the adjusted arrangement process meets the software design requirements. The method has the advantages of flexible arrangement, expandability, and reduction of development difficulty and period.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of software development, and in particular to a software process arrangement method, device, medium and equipment based on a large model. Background Art

[0002] With the rapid development of information technology, the application of big model technology in various fields is becoming more and more extensive. In the development process of big model application software systems, process orchestration is an important link. Process orchestration refers to organizing multiple software modules according to a certain logical order and rules to form a complete software process. Most of the existing process orchestration methods use hard coding to fix the process logic in the code, resulting in poor scalability and maintainability of the software system. In addition, when dealing with complex business scenarios, the existing process orchestration methods often require writing a lot of code, which increases the difficulty and cycle of development. Finally, the scenarios identified by different models are not connected, resulting in inaccurate results returned by the model, requiring personalized logic development, which is time-consuming and labor-intensive and has poor results. Summary of the invention

[0003] In response to at least one of the above technical problems, an embodiment of the present invention provides a software process orchestration method, device, medium, and equipment based on a large model.

[0004] According to the first aspect, the software process arrangement method based on the big model provided by the embodiment of the present invention includes:

[0005] Obtain users' software design requirements;

[0006] A process orchestration interface is provided, so that a user can create an input node, at least one intermediate node and an output node on the process orchestration interface, wherein the intermediate node includes at least one of a large model node, a knowledge retrieval node and a judgment node; and the created nodes are connected according to the software design requirements, and the execution path formed from the input node to the output node is the orchestration process; wherein the input node serves as the starting point of the orchestration process, the output node serves as the end point of the orchestration process, the large model node is used to perform model calculation, the knowledge retrieval node is used to retrieve required knowledge from an external knowledge base, and the judgment node is used to select a subsequent execution path according to input information;

[0007] Determine the functions that the orchestration process can achieve;

[0008] Determine whether the function meets the software design requirements; if so, output the programming process as the target orchestration process; otherwise, remind the user to adjust the orchestration process until the adjusted orchestration process meets the software design requirements and is output as the target orchestration process.

[0009] In one embodiment, the parameters corresponding to each node include input information, output information, execution status and node type; the input information of each node is the parameters required for the execution of the node, the output information of each node is the result obtained after the node is executed, and the execution status of each node indicates whether the node has been executed. When each node is successfully executed, the execution status of the node is set to have been executed. The node types include five types: input node, output node, large model node, knowledge retrieval node and judgment node.

[0010] In one embodiment, determining the functions that can be achieved by the orchestration process includes:

[0011] Input the problem to be processed into the input node, so that the input node outputs the problem to be processed to the next layer of the post-node of the input node; when executing to any layer between the first layer and the last layer, take this layer as the current layer, and take the node to be executed in this layer as the current node; judge whether all the predecessor nodes in the execution line where the current node is located are successfully executed; if all the predecessor nodes in the execution line where the current node is located are successfully executed, execute the current node; and so on, obtain the result output from the output node;

[0012] According to the result and the problem to be processed, the functions that can be achieved by the orchestration process are determined.

[0013] In one embodiment, the determining of the functions that can be achieved by the orchestration process further includes:

[0014] If there is a predecessor node in the execution line where the current node is located that has not been successfully executed, then return to the previous node of the unsuccessfully executed predecessor node, select a node different from the unsuccessfully executed predecessor node from the next layer of successor nodes of the previous node, and execute the selected node.

[0015] In one embodiment, the determining of the functions that can be achieved by the orchestration process further includes:

[0016] If the previous node of the unsuccessfully executed predecessor node has no other successor nodes except the unsuccessfully executed predecessor node, search upward until a node that can produce a different execution route from the current one is found, and select a node from the next layer of successor nodes of the searched node to execute, so as to enter an execution route different from the current one.

[0017] In one embodiment, the large model node includes multiple sub-nodes, each sub-node represents a different processing step or component in the large model; and / or, the judgment node includes multiple sub-nodes, each sub-node represents a different judgment condition or rule; and / or, the knowledge retrieval node includes multiple sub-nodes, each sub-node represents a different knowledge retrieval condition.

[0018] In one embodiment, the process arrangement interface provides a drag component and a component for connecting graphic parts; the drag component is used by the user to drag the created nodes to change the node positions; the component for connecting graphic parts is used by the user to establish connection relationships between the created nodes.

[0019] According to the second aspect, the software process arrangement device based on the big model provided by the embodiment of the present invention includes:

[0020] Requirements acquisition module, used to obtain users' software design requirements;

[0021] A process orchestration module is used to provide a process orchestration interface, so that the user can create an input node, at least one intermediate node and an output node on the process orchestration interface, wherein the intermediate node includes at least one of a large model node, a knowledge retrieval node and a judgment node; and connect the created nodes according to the software design requirements, and the execution path formed from the input node to the output node is the orchestration process; wherein the input node serves as the starting point of the orchestration process, the output node serves as the end point of the orchestration process, the large model node is used to perform model calculations, the knowledge retrieval node is used to retrieve required knowledge from an external knowledge base, and the judgment node is used to select a subsequent execution path according to input information;

[0022] A function determination module, used to determine the functions that the orchestration process can achieve;

[0023] The first judgment module is used to judge whether the function meets the software design requirements; if so, the programming process is output as a target orchestration process; otherwise, the user is reminded to adjust the orchestration process until the adjusted orchestration process meets the software design requirements and is output as a target orchestration process.

[0024] According to a third aspect, an embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed in a computer, enables the computer to execute the method provided in the first aspect.

[0025] According to a fourth aspect, an embodiment of the present invention provides a computing device, including a memory and a processor, wherein the memory stores executable code, and when the processor executes the executable code, the method provided in the first aspect is implemented.

[0026] The software process orchestration method, device, medium, and equipment based on the large model provided by the embodiment of the present invention provide a process orchestration interface, so that the user creates nodes on the process orchestration interface and connects each node to form an orchestration process. Then determine the functions that the orchestration process can achieve. If the functions meet the software design requirements, the programming process is used as the target orchestration process. Otherwise, remind the user to adjust the orchestration process until the adjusted orchestration process meets the software design requirements as the target orchestration process. It can be seen that the user can flexibly create an orchestration process according to different business needs, and there can be multiple types of nodes in the orchestration process. Through the combination of different types of nodes, various business needs can be met. Therefore, the applicable scenarios of the embodiment of the present invention are very wide. In addition, the orchestration process constructed by the embodiment of the present invention is scalable, which reduces the development difficulty and cycle, and can also reduce labor costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 A flowchart of a software process arrangement method based on a large model in one embodiment of the present invention;

[0028] Figure 2 A schematic diagram of an arrangement process in one embodiment of the present invention;

[0029] Figure 3 The present invention is a structural block diagram of a software process orchestration device based on a large model in one embodiment of the present invention. DETAILED DESCRIPTION

[0030] In the first aspect, the embodiment of the present invention provides a software process arrangement method based on a large model, see Figure 1 The method comprises the following steps S110 to S140:

[0031] S110, obtaining the user's software design requirements;

[0032] Among them, software design requirements may include the functions that the software needs to implement, the model architecture adopted by large model nodes, etc.

[0033] S120, providing a process orchestration interface, so that the user can create an input node, at least one intermediate node and an output node on the process orchestration interface, wherein the intermediate node includes at least one of a large model node, a knowledge retrieval node and a judgment node; and connecting the created nodes according to the software design requirements, and the execution path formed from the input node to the output node is the orchestration process; wherein the input node serves as the starting point of the orchestration process, the output node serves as the end point of the orchestration process, the large model node is used to execute model calculation, the knowledge retrieval node is used to retrieve required knowledge from an external knowledge base, and the judgment node is used to select a subsequent execution path according to the input information;

[0034] The large model node may include multiple sub-nodes, and each sub-node represents a different processing step or component in the large model.

[0035] The judgment node may include multiple sub-nodes, and each sub-node represents a different judgment condition or rule.

[0036] The knowledge retrieval node may include multiple sub-nodes, and each sub-node represents a different knowledge retrieval condition.

[0037] In one embodiment, the process arrangement interface provides a drag component and a component for connecting graphic parts; the drag component is used by the user to drag the created nodes to change the node positions; the component for connecting graphic parts is used by the user to establish connection relationships between the created nodes.

[0038] It can be seen that the user can move the created node by dragging the component on the process arrangement node. Connections are established between nodes by connecting the components of the graphic parts.

[0039] S130, determining the functions that can be achieved by the orchestration process;

[0040] S140, determining whether the function meets the software design requirements; if so, outputting the programming process as a target orchestration process; otherwise, reminding the user to adjust the orchestration process until the adjusted orchestration process meets the software design requirements and is output as a target orchestration process.

[0041] That is to say, the user creates an input node and an output node on the process orchestration interface, and creates at least one intermediate node between the input node and the output node, and the intermediate node may include a large model node, a knowledge retrieval node, a judgment node, etc. After the user creates each node as needed, he connects each node on the process orchestration interface. The input node is located in the first layer, the output node is located in the last layer, and the execution path from the input node to the output node is the orchestration process of the software. Then, according to the functions of each node and the connection relationship between each node, determine what functions the orchestration process can achieve, and then find out whether the orchestration process meets the user's software design requirements. If it meets, then the orchestration process can be used as the target orchestration process and applied to the application scenario. If it does not meet, it is necessary to remind the user to adjust the nodes and the connection relationship between the nodes, and after the adjustment, determine whether the adjusted orchestration process meets the user's software design requirements. This is repeated until the adjusted orchestration process meets the user's software design requirements, and then the final orchestration process is used as the target orchestration process and applied in the application scenario.

[0042] In one embodiment, the parameters corresponding to each node include input information, output information, execution status and node type; the input information of each node is the parameters required for the execution of the node, the output information of each node is the result obtained after the node is executed, and the execution status of each node indicates whether the node has been executed. When each node is successfully executed, the execution status of the node is set to have been executed. The node types include five types: input node, output node, large model node, knowledge retrieval node and judgment node.

[0043] It can be seen that each node performs operations according to the input information, outputs the results as output information, and sets the execution status to successfully executed. The node type is used to mark the type of the node.

[0044] In one embodiment, the determining of the functions that can be implemented by the orchestration process in S130 may include the following S131 and S132:

[0045] S131, input the problem to be processed to the input node, so that the input node outputs the problem to be processed to the next layer of the post-node of the input node; when executing to any layer between the first layer and the last layer, take this layer as the current layer, and take the node to be executed in this layer as the current node; judge whether all the predecessor nodes in the execution line where the current node is located are successfully executed; if all the predecessor nodes in the execution line where the current node is located are successfully executed, execute the current node; and so on, obtain the result output from the output node;

[0046] During the process execution, we first start from the input node, and traverse its post-nodes according to the connection relationship of the nodes. Specifically, the problem to be processed is input to the input node, and the input node outputs the problem to be processed to the post-node of the second layer, which can be a judgment node, a knowledge retrieval node or a large model node. If it is a judgment node, the judgment node will choose the subsequent execution path according to the input information. If it is a knowledge retrieval node, the knowledge retrieval node will perform information retrieval based on the input information, obtain the retrieval results, and input the retrieval results to the post-node of the next layer. If it is a large model node, the large model node will input the input information to the preset large model, and the preset large model will perform model calculation to obtain the calculation results, and input the calculation results to the post-node of the next layer. It can be seen that it will be executed from the first layer to the bottom according to the execution path, and finally reach the output node.

[0047] Among them, when executing to the i-th layer between the first layer and the last layer, the i-th layer is taken as the current layer, and the node to be executed in the i-th layer is taken as the current node; it is judged whether all the predecessor nodes in the execution line where the current node is located are successfully executed. Only when all the predecessor nodes in the execution line where the current node is located are successfully executed, the current node will be executed.

[0048] S132: Determine the functions that can be achieved by the orchestration process according to the result and the problem to be processed.

[0049] Among them, the pending issues are issues used to determine the functions of the orchestration process, rather than issues that need to be handled by the entire orchestration process in the application scenario.

[0050] It can be seen that based on the results output by the output node and the pending problems input to the input node, we can know what functions the entire orchestration process can achieve, and then determine whether the function meets the user's software design requirements.

[0051] In one embodiment, the determining of the functions that can be implemented by the orchestration process in S130 further includes the following S133:

[0052] S133. If there is a predecessor node in the execution line where the current node is located that has not been successfully executed, return to the previous node of the predecessor node that has not been successfully executed, select a node different from the predecessor node that has not been successfully executed from the next layer of successor nodes of the previous node, and execute the selected node.

[0053] For example, if there is a predecessor node a in the execution line where the current node is located that is not successfully executed, it will return to the previous layer predecessor node b of this predecessor node a. Under the premise that the previous layer predecessor node b has multiple next layer post nodes, a node c different from the predecessor node a is selected from the multiple next layer post nodes, so as to execute node c and enter a different path branch. In this way, another path can be selected when there is a problem with the execution of the predecessor node a.

[0054] In one embodiment, the determining of the functions that can be implemented by the orchestration process in S130 may further include S134:

[0055] S134. If the previous node of the unsuccessfully executed predecessor node has no other successor node except the unsuccessfully executed predecessor node, search upward until a node that can generate a different execution line from the current execution line is found, and select a node from the next layer of successor nodes of the searched node to execute, so as to enter a line branch different from the current execution line.

[0056] For example, node a is located at the fourth layer, and node b is located at the fourth layer. If the previous layer of the previous node b of the previous node a has only one node a, you need to continue searching upwards. The previous layer of the previous node of node b is node d, which is located at the third layer, and node d also has only one subsequent node, which is node b. It can be seen that you need to continue searching upwards. The previous layer of the previous node of node d is node e, and node e has subsequent nodes d and node f, so go back to node e and execute the subsequent node f, thereby entering different line branches.

[0057] It can be seen that when the current node is executed, the output information of the previous node of the node is used as the input information of the current node, and the corresponding operation is performed according to the node type. If it is a knowledge retrieval node, the required information is retrieved from the external knowledge base; if it is a large model node, large-scale calculation is called; if it is a judgment node, the subsequent execution path is selected according to the input information. After a node is successfully executed, the status of the node is set to successfully executed. If the execution fails, it is necessary to backtrack upward to find a node that can enter the next line branch.

[0058] It can be seen that the embodiment of the present invention provides a flexible and scalable process orchestration method, which implements complex process orchestration by defining different types of nodes and allowing arbitrary connections between these nodes.

[0059] In actual scenarios, users can perform the following operations on the process orchestration interface:

[0060] First, according to business needs, create at least one large model node, each of which represents a software module with complex business logic. The large model node can be a deep learning model, a machine learning model, or other types of artificial intelligence models, which are used to process input data and output results. The large model node can include multiple sub-nodes, each of which represents a processing step or component in the model. Here, the large model node can be used to process the question "Which province does Jinan belong to?", but because the answer of the large model may be biased, other pre-judgment nodes, knowledge retrieval nodes, etc. are required for preparatory processing.

[0061] Next, create at least one judgment node, each of which is used to judge the input data, and select different subsequent nodes to execute according to the judgment result. The judgment node can make judgments based on specific conditions or rules of the input data, such as data range, data type, data value, etc., and select corresponding subsequent nodes for execution. The judgment node may include multiple child nodes, each of which represents a judgment condition or rule. Here, it is judged whether the input question contains Jinan City. If it contains Jinan City, it can be passed to the Jinan City Knowledge Base Retrieval Node through configuration means, so as to retrieve a more accurate answer.

[0062] Then, at least one knowledge retrieval node is created, each knowledge retrieval node is used to retrieve relevant information from the knowledge base, and select different subsequent nodes to execute according to the retrieval results. The knowledge retrieval node can retrieve knowledge, rules or information related to the input data from the knowledge base through a query language or API interface, and select the corresponding subsequent node for execution. The knowledge retrieval node may include multiple child nodes, each of which represents a knowledge retrieval condition or query. Multiple knowledge retrieval nodes may be created, for example, a knowledge retrieval node corresponding to the Jinan knowledge base and a knowledge retrieval node corresponding to the Qingdao knowledge base. By configuring different scenario databases, different answer retrieval returns are performed, thereby increasing the accuracy of the answer.

[0063] Next, according to business needs, the created input nodes, large model nodes, judgment nodes, knowledge retrieval nodes, and output nodes are organized in a certain logical order and rules to form a self-orchestrated process. That is, according to business processes, data processing logic or specific rules, each node can be connected in a certain order and relationship to form a complete process. Specifically, the connection relationship and parameter configuration between nodes can be defined and edited through a graphical interface or an orchestration language. For example, see Figure 2 , a process orchestration built by the user on the process orchestration interface. The second-level judgment node is to judge whether the question contains the word Jinan. The third-level judgment node is to judge whether the question contains the word Qingdao. The fifth-level judgment node is to judge whether the result of the previous level is empty.

[0064] When the above orchestration process is used to solve a problem in an application scenario, for example, the question is "Which province does Jinan belong to?", each node will be executed in sequence according to the execution path in the orchestration process, and the final result will be output. In addition, the input data can be a data set, data stream, or other type of data source. The input data will be passed to the input node, and then each node will be executed in sequence according to the order and rules of the orchestration scenario. Each node processes according to the input data and internal logic, and the output result is used as the input of the subsequent node. Finally, the result of the output node is the final output result of the entire process.

[0065] The embodiments of the present invention have the following beneficial effects:

[0066] (1) Improve the scalability and maintainability of the software system and reduce the development difficulty and cycle;

[0067] (2) Supports custom orchestration processes, which can be flexibly adjusted according to different business needs;

[0068] (3) The accuracy of the model’s output results is optimized through the combination of large model nodes, judgment nodes, and knowledge retrieval nodes.

[0069] In a second aspect, an embodiment of the present invention provides a software process arrangement device based on a large model, see Figure 3 , the device 100 comprises:

[0070] The requirement acquisition module 110 is used to acquire the software design requirements of the user;

[0071] The process orchestration module 120 is used to provide a process orchestration interface, so that the user can create an input node, at least one intermediate node and an output node on the process orchestration interface, wherein the intermediate node includes at least one of a large model node, a knowledge retrieval node and a judgment node; and connect the created nodes according to the software design requirements, and the execution path formed from the input node to the output node is the orchestration process; wherein the input node serves as the starting point of the orchestration process, the output node serves as the end point of the orchestration process, the large model node is used to perform model calculation, the knowledge retrieval node is used to retrieve required knowledge from an external knowledge base, and the judgment node is used to select a subsequent execution path according to the input information;

[0072] Function determination module 130, used to determine the functions that the orchestration process can achieve;

[0073] The first judgment module 140 is used to judge whether the function meets the software design requirements; if so, the programming process is output as the target orchestration process; otherwise, the user is reminded to adjust the orchestration process until the adjusted orchestration process meets the software design requirements and is output as the target orchestration process.

[0074] In one embodiment, the parameters corresponding to each node include input information, output information, execution status and node type; the input information of each node is the parameters required for the execution of the node, the output information of each node is the result obtained after the node is executed, and the execution status of each node indicates whether the node has been executed. When each node is successfully executed, the execution status of the node is set to have been executed. The node types include five types: input node, output node, large model node, knowledge retrieval node and judgment node.

[0075] In one embodiment, the function determination module is specifically used to: input the problem to be processed into the input node, so that the input node outputs the problem to be processed to the next layer of post-nodes of the input node; when executing to any layer between the first layer and the last layer, use this layer as the current layer and the node to be executed in this layer as the current node; determine whether all the predecessor nodes in the execution line where the current node is located have been successfully executed; if all the predecessor nodes in the execution line where the current node is located have been successfully executed, execute the current node; and so on, to obtain the result output from the output node; based on the result and the problem to be processed, determine the function that the orchestration process can achieve.

[0076] In one embodiment, the function determination module is specifically used for: if there is a predecessor node in the execution line where the current node is located that is not successfully executed, then return to the previous node of the unsuccessfully executed predecessor node, select a node different from the unsuccessfully executed predecessor node from the next layer of successor nodes of the previous node, and execute the selected node.

[0077] In one embodiment, the function determination module is specifically used for: if the previous node of the unsuccessfully executed predecessor node has no other successor node except the unsuccessfully executed predecessor node, then searching upward until a node that can generate a different execution line from the current execution line is found, and selecting a node from the next layer of successor nodes of the searched node to execute, so as to enter an execution line different from the current execution line.

[0078] In one embodiment, the large model node includes multiple sub-nodes, each sub-node represents a different processing step or component in the large model; and / or, the judgment node includes multiple sub-nodes, each sub-node represents a different judgment condition or rule; and / or, the knowledge retrieval node includes multiple sub-nodes, each sub-node represents a different knowledge retrieval condition.

[0079] In one embodiment, the process arrangement interface provides a drag component and a component for connecting graphic parts; the drag component is used by the user to drag the created nodes to change the node positions; the component for connecting graphic parts is used by the user to establish connection relationships between the created nodes.

[0080] It is understandable that the explanation, specific implementation, beneficial effects, examples, etc. of the relevant contents in the device provided in the embodiment of the present invention can be found in the corresponding parts of the method provided in the first aspect, and will not be repeated here.

[0081] In a third aspect, an embodiment of the present invention provides a computer-readable medium, on which computer instructions are stored. When the computer instructions are executed by a processor, the processor executes the method provided in the first aspect.

[0082] Specifically, a system or device equipped with a storage medium can be provided, on which software program code that implements the functions of any of the above-mentioned embodiments is stored, and a computer (or CPU or MPU) of the system or device can be enabled to read and execute the program code stored in the storage medium.

[0083] In this case, the program code itself read from the storage medium can realize the function of any one of the above-mentioned embodiments, and thus the program code and the storage medium storing the program code constitute a part of the present invention.

[0084] The storage medium embodiments for providing the program code include a floppy disk, a hard disk, a magneto-optical disk, an optical disk (such as CD-ROM, CD-R, CD-RW, DVD-ROM, DVD-RAM, DVD-RW, DVD+RW), a magnetic tape, a non-volatile memory card, and a ROM. Alternatively, the program code can be downloaded from a server computer by a communication network.

[0085] In addition, it should be clear that the functions of any of the above embodiments can be implemented not only by executing the program code read by the computer, but also by enabling an operating system operating on the computer to complete part or all of the actual operations based on instructions from the program code.

[0086] In addition, it can be understood that the program code read from the storage medium is written to a memory provided in an expansion board inserted into the computer or to a memory provided in an expansion module connected to the computer, and then based on the instructions of the program code, a CPU installed on the expansion board or expansion module is enabled to perform part or all of the actual operations, thereby realizing the functions of any of the above-mentioned embodiments.

[0087] It is understandable that the explanation, specific implementation methods, beneficial effects, examples, etc. of the relevant contents in the computer-readable medium provided in the embodiment of the present invention can be found in the corresponding parts of the method provided in the first aspect, and will not be repeated here.

[0088] In a fourth aspect, an embodiment of the present specification provides a computing device, including a memory and a processor, wherein the memory stores executable code, and when the processor executes the executable code, it implements the method in any embodiment of the specification.

[0089] It is understandable that the explanation, specific implementation, beneficial effects, examples, etc. of the relevant contents in the computing device provided in the embodiment of the present invention can be found in the corresponding parts of the method provided in the first aspect, and will not be repeated here.

[0090] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0091] Those skilled in the art should be aware that in one or more of the above examples, the functions described in the present invention can be implemented by hardware, software, widgets, or any combination thereof. When implemented by software, these functions can be stored in a computer-readable medium or transmitted as one or more instructions or codes on a computer-readable medium.

[0092] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made on the basis of the technical solution of the present invention should be included in the scope of protection of the present invention.

Claims

1. A software process arrangement method based on a large model, characterized in that: include: Obtain users' software design requirements; Providing a process arrangement interface, so that a user can create an input node, at least one intermediate node and an output node on the process arrangement interface, wherein the intermediate node includes at least one of a large model node, a knowledge retrieval node and a judgment node; And according to the software design requirements, the created nodes are connected, and the execution path formed from the input node to the output node is the orchestration process; wherein the input node is used as the starting point of the orchestration process, the output node is used as the end point of the orchestration process, the large model node is used to perform model calculation, the knowledge retrieval node is used to retrieve required knowledge from an external knowledge base, and the judgment node is used to select a subsequent execution path according to the input information; Determine the functions that the orchestration process can achieve; Determine whether the function meets the software design requirements; if so, output the programming process as the target orchestration process; otherwise, remind the user to adjust the orchestration process until the adjusted orchestration process meets the software design requirements and is output as the target orchestration process.

2. The method according to claim 1, characterized in that The parameters corresponding to each node include input information, output information, execution status and node type; the input information of each node is the parameters required for the execution of the node, the output information of each node is the result obtained after the node is executed, and the execution status of each node indicates whether the node has been executed. When each node is successfully executed, the execution status of the node is set to executed. The node types include five types: input node, output node, large model node, knowledge retrieval node and judgment node.

3. The method according to claim 1, characterized in that The determining of the functions that can be achieved by the orchestration process includes: Input the problem to be processed into the input node, so that the input node outputs the problem to be processed to the next layer of the post-node of the input node; when executing to any layer between the first layer and the last layer, take this layer as the current layer, and take the node to be executed in this layer as the current node; judge whether all the predecessor nodes in the execution line where the current node is located are successfully executed; if all the predecessor nodes in the execution line where the current node is located are successfully executed, execute the current node; and so on, obtain the result output from the output node; According to the result and the problem to be processed, the functions that can be achieved by the orchestration process are determined.

4. The method according to claim 3, characterized in that The determining of the functions that the orchestration process can achieve also includes: If there is a predecessor node in the execution line where the current node is located that has not been successfully executed, then return to the previous node of the unsuccessfully executed predecessor node, select a node different from the unsuccessfully executed predecessor node from the next layer of successor nodes of the previous node, and execute the selected node.

5. The method according to claim 4, characterized in that The determining of the functions that the orchestration process can achieve also includes: If the previous node of the unsuccessfully executed predecessor node has no other successor nodes except the unsuccessfully executed predecessor node, search upward until a node that can produce a different execution route from the current one is found, and select a node from the next layer of successor nodes of the searched node to execute, so as to enter an execution route different from the current one.

6. The method according to claim 1, characterized in that The large model node includes multiple sub-nodes, each sub-node represents a different processing step or component in the large model; and / or, the judgment node includes multiple sub-nodes, each sub-node represents a different judgment condition or rule; and / or, the knowledge retrieval node includes multiple sub-nodes, each sub-node represents a different knowledge retrieval condition.

7. The method according to claim 1, characterized in that The process arrangement interface provides a drag component and a component for connecting graphic parts; the drag component is used by the user to drag the created nodes to change the node positions; the component for connecting graphic parts is used by the user to establish connection relationships between the created nodes.

8. A software process arrangement device based on a large model, characterized in that: include: Requirements acquisition module, used to obtain users' software design requirements; A process orchestration module, used to provide a process orchestration interface, so that a user can create an input node, at least one intermediate node and an output node on the process orchestration interface, wherein the intermediate node includes at least one of a large model node, a knowledge retrieval node and a judgment node; And according to the software design requirements, the created nodes are connected, and the execution path formed from the input node to the output node is the orchestration process; wherein the input node is used as the starting point of the orchestration process, the output node is used as the end point of the orchestration process, the large model node is used to perform model calculation, the knowledge retrieval node is used to retrieve required knowledge from an external knowledge base, and the judgment node is used to select a subsequent execution path according to the input information; A function determination module, used to determine the functions that the orchestration process can achieve; The first judgment module is used to judge whether the function meets the software design requirements; if so, the programming process is output as a target orchestration process; otherwise, the user is reminded to adjust the orchestration process until the adjusted orchestration process meets the software design requirements and is output as a target orchestration process.

9. A computer-readable storage medium, characterized in that: A computer program is stored thereon, and when the computer program is executed in a computer, the computer is caused to execute the method according to any one of claims 1 to 7.

10. A computing device, characterized in that The method comprises a memory and a processor, wherein the memory stores executable codes, and when the processor executes the executable codes, the method according to any one of claims 1 to 7 is implemented.

Citation Information

Cited By

  • Intelligent agent flow arrangement system, method and device

    CN120410437A

  • Process management method, system and equipment for large model service and medium

    CN120930774A