Network operation and maintenance scheme generation method and device

Abstract

The present application relates to the technical field of communications, and discloses a network operation and maintenance scheme generation method and device. In the method, on the basis of a large language model, a processing flow of a network operation and maintenance task described by means of a natural language is converted into a network operation and maintenance scheme that can be executed by a network operation and maintenance system, and the network operation and maintenance system subsequently executes the network operation and maintenance scheme, so as to complete execution of the network operation and maintenance task, so that a user does not need to write a script of each subtask of the network operation and maintenance task, thereby improving the network operation and maintenance efficiency.

Description

Method and device for generating network operation scheme

[0001] The present application claims priority to the Chinese patent application No. 202411261705.9, filed on September 9, 2024, and entitled "Method and device for generating network operation scheme", the whole content of which is incorporated herein by reference. TECHNICAL FIELD

[0002] The present application relates to the field of communication technology, in particular to a method and device for generating a network operation scheme. BACKGROUND

[0003] For a network composed of multiple network devices, a user (such as an operation and maintenance personnel) ensures the normal operation, efficient performance and security of the network by performing network operation tasks on the network. In related technologies, the execution process of a network operation task can be: for a network operation task to be executed, assuming that the network operation task includes multiple sub-tasks, the user triggers the network operation system to execute the multiple sub-tasks on the network according to the execution order of the multiple sub-tasks in the network operation task by performing certain operations to complete the network operation task, such as, for each sub-task, the user writes the processing flow of the sub-task into a script program, and the script program written is the network operation scheme for the sub-task, and the network operation system executes the network operation scheme to implement the execution of the sub-task on the network.

[0004] However, in the above execution process of the network operation task, the user needs to trigger the network operation system to execute each sub-task of the network operation task by performing certain operations, and the user also needs to write the processing flow of each sub-task into a script program, which increases the execution time of the network operation task and reduces the network operation efficiency. Moreover, if the user's programming expertise is not enough or the user is not familiar with the processing flow of the network operation task / sub-task, the script program will increase the writing time, thereby further increasing the execution time of the network operation task and further reducing the network operation efficiency. SUMMARY

[0005] Embodiments of the present application provide a method and device for generating a network operation scheme, which can improve the network operation efficiency of a network operation system. The technical solution is as follows:

[0006] In a first aspect, a method for generating a network operation scheme is provided, which involves network operation experience including a processing flow obtained by a user performing a network operation task. In the method, network operation experience described by natural language is obtained, and then a network operation scheme for a network operation system to execute is obtained based on a large language model and the network operation experience, and the network operation scheme corresponds to a network operation task.

[0007] The method is based on a large language model, converts a processing flow of a network operation and maintenance task described by natural language into a network operation and maintenance scheme executable by a network operation and maintenance system, and subsequently executes the network operation and maintenance scheme by the network operation and maintenance system to complete the network operation and maintenance task, without the need for the user to write a script program of each subtask of the network operation and maintenance task, thereby improving network operation and maintenance efficiency.

[0008] In a possible implementation, the obtaining, based on the large language model and the network operation and maintenance experience, of the network operation and maintenance scheme for execution by the network operation and maintenance system includes: performing operation and maintenance task analysis on the network operation and maintenance experience based on the large language model to obtain a plurality of subtasks of the network operation and maintenance task and a target dependency relationship; and arranging, based on the large language model, network operation and maintenance sub-schemes corresponding to the plurality of subtasks according to the target dependency relationship to obtain the network operation and maintenance scheme. The target dependency relationship is a dependency relationship between the plurality of subtasks in the processing flow.

[0009] Based on the above possible implementation, the subtasks of the network operation and maintenance task and the dependency relationship between the subtasks are analyzed based on the large language model, and the network operation and maintenance sub-schemes of the subtasks are arranged into the network operation and maintenance scheme according to the target dependency relationship, so that the arranged network operation and maintenance scheme can implement the subtasks in the network operation and maintenance task.

[0010] In a possible implementation, the performing operation and maintenance task analysis on the network operation and maintenance experience based on the large language model to obtain a plurality of subtasks of the network operation and maintenance task and a target dependency relationship includes: decomposing the network operation and maintenance experience based on the large language model to obtain description information of the plurality of subtasks; and performing dependency relationship analysis on the network operation and maintenance experience based on the large language model to obtain the target dependency relationship.

[0011] Based on the above possible implementation, the network operation and maintenance experience is decomposed to accurately obtain each subtask of the network operation and maintenance task, and the dependency relationship analysis is performed on the network operation and maintenance experience to analyze the dependency relationship between each subtask of the network operation and maintenance task, so as to subsequently arrange the network operation and maintenance sub-schemes of the subtasks into an accurate network operation and maintenance scheme according to the dependency relationship.

[0012] In a possible implementation, the performing dependency relationship analysis on the network operation and maintenance experience based on the large language model to obtain a target dependency relationship includes: analyzing, based on the large language model, an execution order of a plurality of subtasks in the network operation and maintenance experience; and determining the target dependency relationship based on the execution order of the plurality of subtasks.

[0013] Based on the above possible implementation manners, the dependency relationship between the sub-tasks of the network operation task can be accurately determined through the execution sequence of the plurality of sub-tasks in the network operation experience, so as to subsequently arrange the network operation sub-schemes of the sub-tasks into an accurate network operation scheme according to the dependency relationship.

[0014] In a possible implementation manner, the network operation experience described by the natural language includes: obtaining the processing flow through a first input box of a task editing interface, and the task editing interface is used for a user to input the network operation experience.

[0015] In a possible implementation manner, the network operation experience further includes task description information of the network operation task, and the network operation experience described by the natural language includes:

[0016] The processing flow is obtained through the first input box of the task editing interface, and the task editing interface is used for the user to input the network operation experience.

[0017] The task description information of the network operation task is obtained through a second input box of the task editing interface.

[0018] Based on the above possible implementation manners, a man-machine interaction interface is provided for the user, so that the user inputs the network operation experience to the background through the man-machine interaction interface, and the background automatically generates the network operation scheme for the network operation task based on the network operation experience input by the user.

[0019] In a possible implementation manner, after the network operation scheme for the network operation system to execute is obtained, the method further includes: outputting the network operation scheme; and in response to a confirmation operation on the network operation scheme, associating the network operation scheme with the task description information of the network operation task, the confirmation operation indicating that the network operation scheme is confirmed to be used to implement the network operation task.

[0020] Based on the above possible implementation manners, the network operation scheme confirmed by the user can be obtained, and the accuracy of the network operation scheme can be ensured, so that the network operation system can run the network operation scheme.

[0021] In a possible implementation manner, after the network operation scheme for the network operation system to execute is obtained, the method further includes: obtaining an input dialogue message, the dialogue message including a network operation intention described by a natural language; in response to the network operation intention matching the task description information of the network operation task, calling the network operation scheme and running the network operation scheme in the network operation system; and outputting an execution result of the network operation task.

[0022] Based on the possible implementation manner, the user only needs to provide the dialogue message in the form of dialogue, without multiple interactions with the network operation and maintenance system, the operation process is simple, and thus the network operation and maintenance efficiency can be improved.

[0023] In a possible implementation manner, after the network operation and maintenance scheme for the network operation and maintenance system to execute is acquired, the method further includes: receiving network alarm information, the network alarm information indicating an abnormal situation occurring in the network; in response to the abnormal situation matching the task description information of the network operation and maintenance task, invoking the network operation and maintenance scheme and running the network operation and maintenance scheme in the network operation and maintenance system; and outputting an execution result of the network operation and maintenance task.

[0024] Based on the possible implementation manner, when the abnormal situation occurs in the network, the network operation and maintenance task related to the abnormal situation can be automatically executed on the network to solve or alleviate the abnormal situation in the network, without waiting for manual processing, and the network operation and maintenance efficiency of the network operation and maintenance system is improved.

[0025] In a possible implementation manner, after the network operation and maintenance scheme for the network operation and maintenance system to execute is acquired, the method further includes: in response to a modification operation on the network operation and maintenance experience, acquiring the modified network operation and maintenance experience; and based on the large language model and the modified network operation and maintenance experience, acquiring the modified network operation and maintenance scheme and updating the network operation and maintenance scheme.

[0026] Based on the possible implementation manner, in the case that the processing flow of the network operation and maintenance task needs to be modified, the user only needs to modify the network operation and maintenance experience, without performing other operations, and thus the network operation and maintenance efficiency can be improved.

[0027] In a possible implementation manner, after the network operation and maintenance scheme for the network operation and maintenance system to execute is acquired, the method further includes: in response to a modification operation on the network operation and maintenance scheme, acquiring the modified network operation and maintenance scheme; and in response to a confirmation operation on the modified network operation and maintenance scheme, associating the modified network operation and maintenance scheme with the task description information of the network operation and maintenance task, the confirmation operation indicating that the network operation and maintenance scheme is confirmed to be used to implement the network operation and maintenance task.

[0028] Based on the possible implementation manner, in the case that the processing flow of the network operation and maintenance task needs to be modified, the user only needs to modify the network operation and maintenance scheme, without performing other operations, and thus the network operation and maintenance efficiency can be improved.

[0029] In a second aspect, a network operation and maintenance scheme generation apparatus is provided, configured to execute the method in the first aspect or any of the optional manners of the first aspect.

[0030] In a third aspect, a computer device is provided, which includes a processor configured to execute program code to cause the computer device to perform the method provided in the first aspect or any of the optional manners of the first aspect.

[0031] In a fourth aspect, a computer readable storage medium is provided, which stores at least one program code, which is read by a processor to cause a computer device to perform the method provided in the first aspect or any of the optional manners of the first aspect.

[0032] In a fifth aspect, a computer program product or computer program is provided, which includes program code stored in a computer readable storage medium, the program code being read by a processor to cause the computer device to perform the method provided in the first aspect or any of the optional manners of the first aspect.

[0033] On the basis of the implementation manners of the aspects provided in the present application, further combinations can be made to provide more implementation manners. BRIEF DESCRIPTION OF DRAWINGS

[0034] FIG. 1 is a schematic diagram of an implementation environment of a network operation system according to an embodiment of the present application;

[0035] FIG. 2 is a flowchart of a network operation scheme generation method according to an embodiment of the present application;

[0036] FIG. 3 is a schematic diagram of a task editing interface according to an embodiment of the present application;

[0037] FIG. 4 is a network operation interaction flowchart based on a large language model according to an embodiment of the present application;

[0038] FIG. 5 is a schematic diagram of a program display interface according to an embodiment of the present application;

[0039] FIG. 6 is a schematic diagram of a result display interface according to an embodiment of the present application;

[0040] FIG. 7 is a flowchart of another network operation scheme generation method according to an embodiment of the present application;

[0041] FIG. 8 is a flowchart of another network operation scheme generation method according to an embodiment of the present application;

[0042] FIG. 9 is a schematic diagram of a conversation interface according to an embodiment of the present application;

[0043] FIG. 10 is a schematic diagram of a task display interface according to an embodiment of the present application;

[0044] Fig. 11 is a structural schematic diagram of a network operation scheme generation device according to an embodiment of the present application;

[0045] Fig. 12 is a structural schematic diagram of a computer device according to an embodiment of the present application. DETAILED DESCRIPTION

[0046] In order to facilitate understanding of the technical solutions proposed in the present application, hereinafter, some of the terms involved in the present application are explained as follows:

[0047] Large language model (LLM): refers to a computer model capable of processing and generating natural language, LLM is an artificial intelligence technology based on deep learning, which can predict the next word or sentence by learning the statistical rules and semantic information of language data, LLM is usually composed of multiple neural networks (such as feedforward neural network, recurrent neural network), wherein, the feedforward neural network is used for feature extraction of input text, and the recurrent neural network is used for modeling of text sequence. Through continuous training on a large amount of text data, LLM can gradually improve the understanding and prediction ability of natural language. With the improvement of LLM ability, LLM can be applied to various application fields, such as natural language processing, machine translation, text generation, robotics, machine learning, speech recognition, image processing, etc., therefore, LLM can also be called multimodal large language model (MLLM).

[0048] Network operation: is the abbreviation of network operation and maintenance, which refers to a series of work such as planning, deployment, monitoring, maintenance, optimization and fault elimination of network infrastructure, network equipment, network system and network application.

[0049] Network operation task: is a task performed for network operation and maintenance, and the operation and maintenance personnel performs network operation task on the network to ensure the stability, security, availability and efficiency of the network, so as to meet the needs of network daily operation and business development. There are various types of network operation tasks in the network, such as planning, deployment, monitoring, maintenance, optimization and fault elimination, which are respectively used for network planning, business deployment, performance monitoring, performance maintenance, performance optimization and fault elimination of at least one of network infrastructure, network equipment, network system and network application. The task types of different network operation tasks may be different or the same, and the present application does not limit the task type of the network operation task.

[0050] In order to make the purpose, technical solutions and advantages of the present application more clear, the embodiments of the present application will be further described in detail below with reference to the drawings.

[0051] FIG. 1 is a schematic diagram of an implementation environment of a network operation system according to an embodiment of the present application. Referring to FIG. 1, the implementation environment includes a network 11, a network operation system 12, and a terminal 13. The network 11 includes at least one network device 111. FIG. 1 illustrates an example in which the network 11 includes two network devices 111. However, the number of network devices 111 in the network 11 is not limited to two, and can be one or more than two. In this embodiment of the present application, the number of network devices 111 in the network 11 is not limited.

[0052] The network device 111 can be a network element device, a computing device, or a storage device. In this embodiment of the present application, the type of the network device 111 is not limited. In the case where the network 11 includes a plurality of network devices 111, if the network device 111 is a network element device, the network 11 is a communication network for providing a communication function. If the network device 111 is a computing device, the network 11 is a computing device cluster for providing a computing function. If the network device 111 is a storage device, the network 11 is a storage device cluster for providing a data storage function. The network 11 can be deployed in a cloud (private cloud or public cloud), or can not be deployed in a cloud. In this embodiment of the present application, the deployment environment of the network 11 is not limited.

[0053] The network operation system 12 is configured to provide network operation services for the network devices 111 in the network 11. For example, the network operation system 12 performs network operation tasks on the network 11 to ensure normal operation, efficient performance, and security of the network 11. The network operation system 12 can be a network cloud engine (NCE). In other embodiments, the network operation system 12 can be another system for managing and maintaining the network 11, such as an operation support system (OSS), an element management system (EMS), or a network management system (NMS). In this embodiment of the present application, the type of the network operation system 12 is not limited.

[0054] As shown in FIG. 1, the network operation system 12 includes a network management module 121, a control module 122, and a large language model 123. The network management module 121 has the ability to perform network operation on the network 11, and is configured to manage and / or maintain resources of the network 11, such as hardware resources, software resources, service resources, and the like. The hardware resources refer to hardware resources of the network 11, the software resources refer to software resources of the network 11, and the service resources refer to service resources provided by the network 11, such as computing resources, storage resources, communication resources, and the like. The communication resources refer to resources related to communication of the network 11, such as network bandwidth, network latency, network packet loss, network infrastructure, and the like. The network infrastructure refers to all hardware and software used to support network communication.

[0055] The terminal 13 is an electronic device used by a user (e.g., an operator of the network 11), such as at least one of a smartphone, a tablet computer, a personal computer, a laptop computer, and a desktop computer. In this embodiment, the type and number of the terminal 13 are not limited.

[0056] The terminal 13 is installed with an application (referred to as a network operation application) supporting network operation. The control module 122 is configured to provide background computing services for the network operation application. The user can trigger the control module 122 to perform a network operation task on the network 11 through the network operation application in the terminal 122. The control module 122 can call the network management module 121 to perform the network operation task on the network 11.

[0057] Exemplarily, a processing flow of a user performing a network operation and maintenance task through a natural language description is referred to as a network operation and maintenance experience, the user can input the network operation and maintenance experience in a network operation and maintenance application of the terminal 13, the terminal 13 imports the input network operation and maintenance experience into the control module 122 through the network operation and maintenance application, the control module 122 compiles the imported network operation and maintenance experience into a network operation and maintenance scheme executable by the network operation and maintenance system 12 through the large language model 123, and then the control module 122 invokes the network management module 121 to run the network operation and maintenance scheme, so that the network management module 121 performs the network operation and maintenance task on the network 11, thereby the user provides the network operation and maintenance experience to the network operation and maintenance system 12, and the network operation and maintenance system 12 automatically completes the network operation and maintenance task. Alternatively, after the network operation and maintenance experience is compiled into the network operation and maintenance scheme, the control module 122 can also issue the network operation and maintenance scheme to the network management module 121, and establish a corresponding relationship between the network operation and maintenance task and the network operation and maintenance scheme in the network management module 121, in the case that the network equipment 111 reports a network alarm information to the network management module 121, if the network alarm information indicates that an abnormal situation of the network 11 is related to the network operation and maintenance task, the network management module 121 runs the network operation and maintenance scheme corresponding to the network operation and maintenance task to perform the network operation and maintenance task on the network 11 to solve the abnormal situation.

[0058] The large language model 123 is a language model applied in the field of network operation and maintenance, and is trained based on the network operation and maintenance experience and a basic large language model in the communication field, so that the large language model 123 has the ability to analyze subtasks of a network operation and maintenance task and a dependency relationship between the subtasks based on the network operation and maintenance experience, to compile a processing flow of the network operation and maintenance task into a network operation and maintenance scheme executable by the network management module 121 based on a function library supported by the network management module 121 and the dependency relationship between the subtasks, and the like.

[0059] Next, based on the implementation environment introduced above, the generation process of the network operation and maintenance scheme is introduced.

[0060] FIG. 2 is a flowchart of a network operation and maintenance scheme generation method provided by an embodiment of the present application, and the method includes the following steps.

[0061] 201. The terminal obtains a network operation and maintenance experience described by a natural language, and the network operation and maintenance experience includes a processing flow obtained by a user performing a network operation and maintenance task.

[0062] The network operation task is any network operation task for a network (such as the network 11), which can or can not include subtasks, the subtasks are network operation tasks of a smaller granularity than the network operation task, the task types of different subtasks in the plurality of subtasks can be the same or different, and the number of subtasks and the task types in the network operation task are not limited in the embodiments of the present application.

[0063] The processing flow obtained by the user performing the network operation task refers to the processing flow of the user performing the network operation task in the network operation system, which is the experience of the user performing the network operation task on the network. Therefore, the network operation experience is the experience of the user for the network operation task, and the network operation experience is described by natural language. For example, the network operation experience includes flow description information of the network operation task, which is the processing flow of the user performing the network operation task described by natural language, and the flow description information can be text information. In the network operation experience, each subtask of the network operation task and the execution order of each subtask in the processing flow of the network operation task are described by natural language. For example, the network operation experience includes description information of each subtask and the execution order of each subtask, and the description information of the subtask is used to describe the subtask by natural language. For example, it is assumed that the network device in the network includes a base station, and the network operation task is "single base station service withdrawal", the purpose of which is to diagnose the root cause of the single base station fault, and the network operation task "single base station service withdrawal" includes subtasks 1-5. The description information of subtask 1 is "query the base station access network element according to the base station IP", the description information of subtask 2 is "query the base station access network element state", the description information of subtask 3 is "perform network element offline diagnosis to determine the network element offline reason", the description information of subtask 4 is "query all connected link ETH_LOS alarms of the offline network element, and compare the occurrence time of the ETH_LOS alarms", and the description information of subtask 5 is "query yyyyy". The network operation experience of the network operation task "single base station service withdrawal" includes:

[0064] 1. Query the base station access network element according to the base station IP;

[0065] 2. Query the base station access network element state:

[0066] The network element state is offline, perform network element offline diagnosis, and execute 3;

[0067] The network element state is online, and execute 5;

[0068] 3. Perform network element offline diagnosis to determine the network element offline reason:

[0069] The network element DCN is offline, execute 5;

[0070] The network element is offline due to network failure, and execute 4;

[0071] 4. Query the ETH_LOS alarms of all connected links of the offline network element, and compare the occurrence time of the ETH_LOS alarms:

[0072] Xxxxxx;

[0073] 5. Query yyyy.

[0074] For a network operation task, the user can write the processing flow of the network operation task into the network operation experience according to his own operation experience through natural language. Alternatively, the user can also query the network operation experience corresponding to the network operation task from the runbook of the network. The runbook records the network operation experience corresponding to a plurality of network operation tasks related to the network.

[0075] The user can input the network operation experience into the terminal, and the terminal acquires the network operation experience input by the user. In some embodiments, the terminal acquires the processing flow obtained by the user performing the network operation task through a first input box of a task editing interface, wherein the task editing interface is used for the user to input the network operation experience. For example, the following steps A1 and A2.

[0076] Step A1, the terminal displays a task editing interface, and the task editing interface includes a first input box, which is used to input the flow description information of the network operation task.

[0077] The task editing interface is a kind of human-computer interaction interface of the network operation application, which is used to input the network operation experience. The network operation experience includes the flow description information of the network operation task and the task information, wherein the task information is used to indicate the network operation task, and the task information includes at least one of the task identifier of the network operation task, the task type of the network operation task and the task description information. The task identifier is used to represent the network operation task, which can be the name of the network operation or the task number of the network operation task, etc. The task description information is a brief introduction of the network operation task, which is used to simply describe the network operation task, such as describing the purpose of the network operation task or the problem that can be solved by the network operation task. In other embodiments, the network operation experience does not include the task information, and the network operation experience is the flow description information of the network operation task. The task information of the network operation experience and the network operation experience together constitute the information of the network operation task.

[0078] For example, FIG. 3 is a schematic diagram of a task editing interface provided by an embodiment of the present application. As shown in FIG. 3, the task editing interface 300 includes a first input box 31, a second input box 32, a third input box 33, a fourth input box 34, and a compiling control 35. The second input box 32 is used to input task description information of a network operation and maintenance task. The third input box 33 is used to input a task identifier of the network operation and maintenance task. The fourth input box 34 is used to input a task type of the network operation and maintenance task.

[0079] A user can trigger the terminal to display the task editing interface 300 through a network operation and maintenance application in the terminal, so that the user inputs network operation and maintenance experience in the task editing interface 300. For example, the user inputs flow description information 311 of a network operation and maintenance task “single base station withdrawal” in the first input box 31. For example, the user edits the flow description information 311 in the first input box 31, or the user copies flow description information 311 of the network operation and maintenance task “single base station withdrawal” in a running manual to the first input box 31. The user inputs task description information 321 of the network operation and maintenance task “single base station withdrawal” in the second input box 32, so as to simply describe the network operation and maintenance task “single base station withdrawal”. The user also inputs a name of the network operation and maintenance task in the third input box 33 as a task identifier. The user also selects a task type “fault diagnosis” from a plurality of task types associated with the fourth input box 34 as a task type of the network operation and maintenance task “single base station withdrawal”, and inputs the task type to the fourth input box 34.

[0080] In step A2, the terminal acquires the information input in the first input box as network operation and maintenance experience corresponding to the network operation and maintenance task.

[0081] The terminal acquires the flow description information 311 input by the user in the first input box 31 as network operation and maintenance experience. Alternatively, the terminal also acquires task description information of the network operation and maintenance task through the second input box 32 of the task editing interface, and acquires the task description information and the flow description information 311 input in the first input box 31 as the task description information of the network operation and maintenance task.

[0082] For example, as shown in FIG. 3, after the user inputs information in each input box of the task editing interface 300, the user triggers the compiling control 35 in the task editing interface 300, for example, by clicking, sliding, or other operations. In this embodiment of the present application, the triggering operation is not limited. The terminal acquires the information input in each input box of the task editing interface 300 in response to the triggering operation, takes the information input in the third input box 33 as the task identifier, takes the information input in the first input box 31 as the flow description information, takes the information input in the fourth input box 34 as the task type, takes the information input in the second input box 32 as the task description information, takes the acquired task identifier, task type, and task description information as the task information of the network operation and maintenance task, or takes the acquired flow description information as the network operation and maintenance experience input by the user, or takes the task information and the acquired flow description information as the network operation and maintenance experience input by the user.

[0083] The above steps A1 and A2 provide a man-machine interactive interface for the user to input the network operation and maintenance experience to the background, so that the background automatically generates the network operation and maintenance scheme for the network operation and maintenance task based on the network operation and maintenance experience input by the user.

[0084] 202. The terminal sends the network operation and maintenance experience described by natural language to the control module in the network operation and maintenance system.

[0085] For example, as shown in FIG. 3, the user triggers the compiling control 35 in the task editing interface 300, and the terminal acquires the flow description information as the network operation and maintenance experience, generates a compiling request based on the acquired task information and the network operation and maintenance experience, and sends the compiling request to the server, where the compiling request includes the task information and the network operation and maintenance experience, to indicate that the network operation and maintenance experience is compiled.

[0086] At least one of the second input box 32, the third input box 33, and the fourth input box 34 in the task editing interface 300 is optional. In other embodiments, the task editing interface 300 does not include the at least one, and accordingly, the user does not input information in the at least one after the terminal displays the task editing interface 300, and the terminal acquires the task information from the task editing interface 300 does not include the information input in the at least one.

[0087] 203. The control module acquires the network operation and maintenance experience described by natural language.

[0088] For example, the control module receives the compiling request sent by the terminal, and acquires the network operation and maintenance experience and the task information from the compiling request.

[0089] The steps 201 to 203 are described by taking the terminal providing the network operation experience to the control module as an example. In some other embodiments, the control module is preconfigured with network operation experiences corresponding to at least one network operation task, and the control module can obtain the network operation experience from the preconfigured network operation experience without obtaining the network operation experience from the terminal. In this case, the terminal does not need to perform the steps 201 and 202.

[0090] 204. The control module obtains, based on the large language model and the network operation experience, a network operation scheme for execution by the network operation system, the network operation scheme corresponding to the network operation task.

[0091] The network operation scheme is a compilation result of the network operation experience, and the network operation scheme is used to describe the processing flow of the network operation task in a language (such as a computer language) recognizable by the network operation system, so that the network operation scheme can be executed by the network operation system. In the case where the network operation task includes a plurality of subtasks, the network operation scheme includes a plurality of network operation sub-schemes corresponding to the plurality of subtasks, each network operation sub-scheme corresponding to a subtask, and each network operation sub-scheme corresponding to the subtask is used to implement the corresponding subtask.

[0092] In a possible implementation, the control module compiles the network operation experience into the network operation scheme based on the language model, and the compilation process is implemented by, for example, the following steps B1 to B3.

[0093] B1. The control module performs operation task analysis on the network operation experience based on the large language model to obtain a plurality of subtasks of the network operation task and a target dependency relationship, the target dependency relationship being a dependency relationship between the plurality of subtasks in the processing flow described in the network operation experience.

[0094] The target dependency relationship includes a branch execution order corresponding to each first subtask in the plurality of subtasks, the first subtask being a first executed subtask in the execution flow of the network operation task, and there being at least one first subtask. The branch execution order corresponding to any first subtask is used to represent a dependency relationship between the first subtask and other subtasks in the plurality of subtasks except the first subtask, and the first subtask and the other subtasks form a branch task in the network operation task. Taking the above network operation task “single base station withdrawal” as an example, it can be known from the network operation experience of the “single base station withdrawal” that the subtask 1 is the first subtask, the subtask 1 corresponds to the branch execution orders 1 and 2, the branch execution order 1 is subtask 1→subtask 2→subtask 3→subtask 4, and the subtask 1, the subtask 2, the subtask 3, and the subtask 4 form a branch task 1; the branch execution order 2 is subtask 1→subtask 2→subtask 5, and the subtask 1, the subtask 2, and the subtask 5 form a branch task 2.

[0095] The operation and maintenance task analysis of the network operation and maintenance experience includes decomposing the network operation and maintenance experience to obtain description information of multiple sub-tasks. For example, the control module decomposes the network operation and maintenance experience based on a large language model to obtain description information of multiple sub-tasks.

[0096] The large language model can split the input network operation and maintenance experience into description information of multiple sub-tasks according to a task decomposition rule, and each sub-task description information is used to describe the corresponding sub-task. The task decomposition rule refers to the rule of splitting the network operation and maintenance experience into multiple description information, such as the task analysis rule of circular decomposition, that is, each sentence, information or step in the network operation and maintenance experience is split into a description information of a sub-task. For example, the description information of each step is split into a description information of a sub-task, and the description information of “querying the base station access network element” in the network operation and maintenance experience of “single base station withdrawal” is taken as a description information of a sub-task, and the description information of “querying the base station access network element state” in the network operation and maintenance experience is taken as another description information of a sub-task. In actual application process, different application scenarios can be preset with different task decomposition rules, and in this application embodiment, the task decomposition rule is not limited.

[0097] For example, the control module is preset with a task decomposition rule, and the control module concatenates the network operation and maintenance experience and the task decomposition rule into first prompt information according to the prompt template that can be recognized by the large language model, the first prompt information is used to prompt the large language model to decompose the network operation and maintenance experience into description information of multiple sub-tasks according to the task decomposition rule, the control module inputs the first prompt information into the large language model, the large language model decomposes the input network operation and maintenance experience into description information of multiple sub-tasks according to the prompt of the first prompt information based on the task decomposition rule, and outputs the description information of multiple sub-tasks, and the control module obtains the description information of multiple sub-tasks output by the large language model.

[0098] The operation and maintenance task analysis of the network operation and maintenance experience also includes dependency analysis of the network operation and maintenance experience to obtain a target dependency relationship. For example, the control module analyzes the execution order of multiple sub-tasks in the network operation and maintenance experience based on a large language model, such as identifying the step number corresponding to the description information of multiple sub-tasks through the large language model, and taking the identified step number as the execution order of the corresponding sub-task. For example, the description information “querying the base station access network element according to the base station IP” in the network operation and maintenance experience of the network operation and maintenance task “single base station withdrawal” corresponds to the step number 1 as the execution order of the corresponding sub-task.

[0099] After analyzing the execution sequence of the plurality of sub-tasks, the control module determines the target dependency relationship based on the execution sequence of the plurality of sub-tasks. For example, the control module analyzes the description information of the plurality of sub-tasks based on the execution sequence of the plurality of sub-tasks by using the large language model to obtain the target dependency relationship. Taking the network operation task "single base station off-service" as an example, by analyzing the description information "query base station access network element state" of sub-task 2, it can be known that sub-task 2 is executed in step 2, and if the execution result of sub-task 2 is "network element state is offline", step 3 is executed, and sub-task 3 is executed in step 3. Therefore, there is a dependency relationship between sub-task 2 and sub-task 3.

[0100] The control module can input the network operation experience into the large language model to analyze the dependency relationship. For example, the first prompt information is also used to prompt the large language model to analyze the dependency relationship of the network operation experience. After the network operation experience is decomposed into a plurality of sub-task description information, according to the prompt of the first prompt information, the large language model analyzes the context of the description information of the plurality of sub-tasks based on the execution sequence of the plurality of sub-tasks to obtain the target dependency relationship, and outputs the description information of the plurality of sub-tasks and the target dependency relationship.

[0101] The above scheme decomposes the network operation experience to accurately obtain each sub-task of the network operation task, analyzes the dependency relationship between each sub-task of the network operation task through dependency relationship analysis, and then arranges the network operation sub-scheme of the sub-task into an accurate network operation scheme according to the dependency relationship. The above scheme can accurately determine the dependency relationship between the sub-tasks of the network operation task through the execution sequence of the plurality of sub-tasks in the network operation experience, so as to arrange the network operation sub-scheme of the sub-task into an accurate network operation scheme according to the dependency relationship.

[0102] Step B2, the control module obtains a plurality of network operation sub-schemes corresponding to the plurality of sub-tasks based on the description information of the plurality of sub-tasks.

[0103] Any network operation sub-solution corresponds to a sub-task, and is a network operation solution for implementing the corresponding sub-task. Any network operation sub-solution can be implemented by calling at least one function supported by the network management module in the network operation system. The network management module has a function library, the function library includes a plurality of functions, the network management module supports running the plurality of functions, each function corresponds to a network operation task, different functions correspond to different network operation tasks, and each function is used to implement the corresponding network operation task. The granularity of the network operation tasks corresponding to different functions can be different or the same. For example, some functions correspond to network operation tasks that do not include sub-tasks, and some functions correspond to network operation tasks that include a plurality of sub-tasks. Any function can be an application programming interface (API) function, or a tool, which is an agent or a chain interface used to interact with the world. In this embodiment of the present application, the type of the function is not limited.

[0104] The functions in the function library can be recorded by a function list, the function list includes the function identifier of each function in the function library and the description information of the network operation task corresponding to each function, and the function identifier of any function is used to represent the function. The function identifier can be the name of the function.

[0105] The control module can query the functions supported by the network management module, and the network management module sends the function list to the network management module based on the query of the control module, so that the control module obtains the network operation solution based on the function list. In some embodiments, after the control module obtains the function list from the network management module, it can query the function list from the network management module again. For example, the control module periodically queries the function list from the network management module, and uses the latest queried function list when generating a network operation solution for any network operation task, so that the control module can generate a network operation solution based on the functions in the updated function library, the generated network operation solution can be successfully run by the network management module, the network operation system is improved in network operation efficiency, and the network operation system is improved in network operation efficiency. In some embodiments, the control module does not need to obtain the function list from the network management module, but configures the function list for the control module by the user, and the control module obtains the function list configured by the user.

[0106] For the network operation task corresponding to the network operation experience, the functions required to implement the network operation task are referred to as first functions, and the first functions have at least one. For example, in the case where the network operation task does not include subtasks, the first function is the function corresponding to the network operation task in the function library, and in the case where the network operation task includes multiple subtasks, the first function has multiple, and each first function is the function corresponding to a subtask of the network operation task in the function library. The calling process of each first function is regarded as a network operation sub-scheme corresponding to a subtask, so that any network operation scheme corresponding to a subtask can implement the corresponding subtask.

[0107] Each subtask may or may not include sub-subtasks, and sub-subtasks are network operation tasks of smaller granularity than subtasks. In the case where a subtask includes sub-subtasks, the subtask can be completed by multiple first functions, and each first function is used to complete a sub-subtask. In the case where a subtask does not include sub-subtasks, the subtask is completed by one first function. Therefore, each network operation sub-scheme describes the calling process of one or more first functions through a language recognizable by the network operation system, and the calling process of one or more first functions is used to reflect the execution process of the corresponding subtask. In the case where the network operation scheme is a script program, the network operation sub-scheme is a program segment (i.e., a script program segment) in the script program.

[0108] For ease of description, the description information of the subtasks extracted from the network operation experience is referred to as first description information, and the description information of the network operation tasks in the function list is referred to as second description information.

[0109] The control module obtains the network operation sub-schemes corresponding to the multiple subtasks based on the large language model, the first description information of the multiple subtasks, and the function list. For example, for the first description information of any subtask in the multiple subtasks, the control module inputs the first description information of the subtask and the function list into the large language model, and generates the network operation sub-scheme corresponding to the subtask based on the input first description information and the function list.

[0110] The large language model generates a network operation and maintenance sub-solution. For example, the large language model queries the function list for second description information that matches the first description information. If any second description information matches the first description information, the large language model generates a network operation and maintenance sub-solution corresponding to the sub-task based on a function identifier corresponding to the second description information in the function list. The network operation and maintenance sub-solution includes a calling flow of a function represented by the function identifier, and the function represented by the function identifier is a first function for implementing the sub-task. The function list is queried for second description information that matches the first description information. This includes calculating the similarity between the first description information and each second description information in the function list, and determining the second description information with the highest similarity to the first description information as the second description information that matches the first description information. The similarity between the first description information and any second description information represents the degree of similarity between the first description information and any second description information.

[0111] For example, the control module concatenates the first description information of any sub-task and the function list into second prompt information according to a prompt template that can be recognized by the large language model. The second prompt information is used to prompt the large language model to generate a network operation and maintenance sub-solution for the sub-task described by the first description information based on the function list. The control module inputs the second prompt information into the large language model. The large language model generates a network operation and maintenance sub-solution corresponding to the sub-task based on the input first description information and function list according to the prompt of the second prompt information. The control module obtains the network operation and maintenance sub-solution output by the large language model.

[0112] The above describes the process of obtaining a network operation and maintenance sub-solution corresponding to a sub-task. In a similar manner, the control module can obtain network operation and maintenance sub-solutions corresponding to multiple sub-tasks of a network operation and maintenance task.

[0113] Taking triggering a large language model generation network operation scheme by an artificial intelligence (AI) assistant (copilot) in a control module as an example, the AI assistant in the control module is shown in FIG. 1. The above steps are introduced through the network operation process shown in FIG. 4. The AI assistant runs in the control module. It is assumed that the network operation task is a live network fault diagnosis task, and the user inputs the live network fault diagnosis experience (i.e., network operation experience) into the control module through the terminal. After the AI assistant obtains the live network fault diagnosis experience, it uses the task decomposition rule of piece-by-piece circular decomposition to splice the task decomposition rule and the live network fault diagnosis experience into a prompt template to obtain prompt information. The prompt information is input into the large language model, and the large language model processes the live network fault diagnosis experience in the prompt information and returns description information of multiple sub-processes (i.e., sub-tasks) and the dependency relationship between the multiple sub-processes to the AI assistant. The AI assistant also queries the network management module for available functions (API functions and / or tools) in the network management module to obtain a function list for subsequent splicing of prompt information when selecting functions for sub-processes. For the description information of each sub-process generated by piece-by-piece circular decomposition, the AI assistant splices the description information of the sub-process and the function list into prompt information, and inputs the prompt information into the large language model. The large language model generates a network operation sub-scheme corresponding to the sub-process based on the prompt information. The network operation sub-scheme includes the calling process of the API function and / or tool, and the network operation sub-scheme satisfies the runbook syntax description of the sub-process. The large language model responds to the AI assistant: the network operation sub-scheme that satisfies the runbook syntax description. In this way, the AI assistant can obtain network operation sub-schemes corresponding to each of the multiple sub-processes. The AI assistant splices the multiple network operation sub-schemes corresponding to the multiple sub-processes into a complete runbook process scheme (i.e., a network operation scheme).

[0114] The above control module obtains one network operation sub-scheme at a time. In other embodiments, the control module obtains multiple network operation sub-schemes at a time through the large language model. For example, the control module splices the first description information of multiple sub-tasks and the function list into third prompt information according to the prompt template that can be recognized by the large language model. The third prompt information is used to prompt the large language model to generate a network operation scheme for each description information indicated by the network operation task based on the functions recorded in the function list. The control module inputs the third prompt information into the large language model. The large language model generates a corresponding network operation sub-scheme for each first sub-description information based on the functions recorded in the function list according to the prompt of the third prompt information, outputs multiple network operation sub-schemes corresponding to the multiple sub-tasks, and the control module obtains the multiple network operation sub-schemes corresponding to the multiple sub-tasks output by the large language model.

[0115] Step B3, the control module arranges the network operation sub-solutions corresponding to the plurality of sub-tasks according to the target dependency relationship based on the large language model to obtain the network operation solution.

[0116] The arrangement of the network operation sub-solutions corresponding to the plurality of sub-tasks is to arrange the processing flow of the network operation task.

[0117] The control module concatenates the target dependency relationship and the network operation sub-solutions corresponding to the plurality of sub-tasks into fourth prompt information according to the prompt template recognizable by the large language model, and the fourth prompt information is used to prompt the large language model to concatenate the network operation sub-solutions according to the target dependency relationship. The control module inputs the fourth prompt information into the large language model, and the large language model concatenates the input network operation sub-solutions into one network operation solution according to the prompt of the fourth prompt information and the target dependency relationship, so that the execution order of the plurality of network operation sub-solutions meets the target dependency relationship, and outputs the network operation solution. The control module acquires the network operation solution output by the large language model.

[0118] The above steps B1 to B3 analyze the sub-tasks of the network operation task and the dependency relationship between the sub-tasks based on the large language model, arrange the network operation sub-solutions of the sub-tasks into the network operation solution according to the target dependency relationship, so that the arranged network operation solution can implement the sub-tasks in the network operation task.

[0119] In some other embodiments, the control module can also input the network operation experience into the large language model, and output the target dependency relationship and the network operation sub-solution corresponding to each sub-task of the network operation task by the large language model. For example, the control module concatenates the network operation experience, the task decomposition rule, and the function list into fifth prompt information according to the prompt template recognizable by the large language model, and the fifth prompt information is used to prompt the large language model to decompose the network operation experience into the description information of the sub-tasks according to the task decomposition rule; analyze the target dependency relationship between each sub-task of the corresponding network operation task based on the network operation experience analysis; generate the network operation sub-solution corresponding to each sub-task based on the functions recorded in the function list, and the control module inputs the fifth prompt information into the large language model. The large language model processes the network operation experience based on the prompt of the fifth prompt information, and outputs the target dependency relationship and the network operation sub-solution corresponding to the plurality of sub-tasks.

[0120] In some embodiments, the control module can also input the description information of the plurality of sub-tasks and the target dependency relationship into the large language model together, and output the network operation and maintenance scheme by the large language model. For example, the control module splices the target dependency relationship, the description information of the plurality of sub-tasks, and the function list into the sixth prompt information according to the prompt template recognizable by the large language model, and the sixth prompt information is used to prompt the large language model to generate a network operation and maintenance sub-scheme for each sub-task described by the description information based on the function list; and splice the generated network operation and maintenance sub-schemes into the network operation and maintenance scheme based on the target dependency relationship. The control module inputs the sixth prompt information into the large language model, and the large language model processes the description information of the plurality of sub-tasks based on the prompt of the sixth prompt information, and outputs the network operation and maintenance scheme.

[0121] In some embodiments, the control module can also output the network operation and maintenance experience to the large language model, and output the network operation and maintenance scheme by the large language model. For example, the control module splices the network operation and maintenance experience, the task decomposition rule, and the function list into the seventh prompt information according to the prompt template recognizable by the large language model, and the seventh prompt information is used to prompt the large language model to decompose the network operation and maintenance experience into the description information of the sub-tasks according to the task decomposition rule; analyze the target dependency relationship between the plurality of sub-tasks corresponding to the network operation and maintenance task based on the network operation and maintenance experience; generate a network operation and maintenance sub-scheme for each sub-task based on the function list, and splice the generated network operation and maintenance sub-schemes into the network operation and maintenance scheme based on the target dependency relationship. The control module inputs the seventh prompt information into the large language model, and the large language model processes the network operation and maintenance experience based on the prompt of the seventh prompt, and outputs the network operation and maintenance scheme.

[0122] After obtaining the network operation and maintenance scheme, the control module also associates the network operation and maintenance scheme with the task description information of the network operation and maintenance task, to indicate that the network operation and maintenance scheme corresponds to the network operation and maintenance task. For example, the control module maintains a scheme list used to record the network operation and maintenance scheme, and the control module can establish a corresponding relationship between the scheme identifier of the network operation and maintenance scheme and the task information of the network operation and maintenance scheme in the scheme list, to associate the network operation and maintenance scheme with the task description information in the task information. For example, the scheme identifier of the network operation and maintenance scheme and the task information of the network operation and maintenance scheme are added in the scheme list, and the scheme identifier corresponds to the task information. The scheme identifier of the network operation and maintenance scheme is used to represent the network operation and maintenance scheme.

[0123] In some embodiments, after obtaining the network operation and maintenance scheme, the control module outputs the network operation and maintenance scheme to the terminal, and the control module executes the step of associating the network operation and maintenance scheme with the task description information of the network operation and maintenance task in response to a confirmation operation on the network operation and maintenance scheme. The confirmation operation indicates that the network operation and maintenance scheme is confirmed to be used to implement the network operation and maintenance task.

[0124] For example, the control module sends a confirmation message to the terminal, the confirmation message indicating whether to use the network operation scheme for the network operation task, the confirmation message including the network operation scheme. After receiving the confirmation message, the terminal displays the network operation scheme in the confirmation message.

[0125] For example, the terminal displays the program display interface 500 after receiving the confirmation message, and displays the network operation scheme in the program display interface 500. For example, the terminal displays the network operation scheme in the program display interface 500 in a graphical manner. For example, taking the network operation task “nceip_naie_test_0318” as an example, the terminal displays the processing flowchart 51 of the network operation task “nceip_naie_test_0318” in the program display interface 500, and the processing flowchart 51 is an image-based network operation scheme. The processing flowchart 51 includes a start node 511, a plurality of task nodes 512, and an end node 513, wherein the start node 511 represents the starting point of the network operation scheme, and the end node 513 represents the end point of the network operation scheme. Each task node 512 represents a subtask in the network operation task “nceip_naie_test_0318”, for example, the task node “llm_alarm_id” represents the subtask “llm_alarm_id” in the network operation task “nceip_naie_test_0318”, and the task node “prompt_summary” represents the subtask “prompt_summary” in the network operation task “nceip_naie_test_0318”. The plurality of task nodes 512 are arranged according to the target dependency relationship in the processing flowchart 51, so that the user can intuitively understand the subtasks in the network operation task and the dependency relationship between the subtasks, thereby intuitively understanding the entire processing flow of the network operation task and increasing the explainability of the processing flow of the network operation task.

[0126] Each task node 512 is associated with a network operation sub-scheme corresponding to a sub-task in the network operation scheme, and the program display interface 500 further includes a program display window 52, which is configured to display the network operation sub-scheme associated with the task node 512. The user can trigger an operation on any task node 512, and the terminal displays the network operation sub-scheme associated with the task node 512 in the program display window 52 in response to the triggering operation, so that the user can view or modify the network operation sub-scheme associated with the task node 512 in the program display window 52. Taking the task node "prompt_summary" as an example, it is assumed that the user triggers an operation on the task node "prompt_summary", and the terminal displays the network operation sub-scheme 521 corresponding to the sub-task "prompt_summary" in the program display window 52 in response to the triggering operation.

[0127] FIG. 5 is a processing flow chart 51 including a plurality of task nodes 512 as an example. In some other embodiments, if the network operation task does not include sub-tasks, the processing flow chart 51 includes one task node 512, which represents the network operation task. The network operation sub-scheme associated with the task node 512 is the network operation scheme.

[0128] The confirmation operation on the network operation scheme displayed by the terminal can be a click operation on a confirmation control (not shown in the figure) in the program display interface 500, or a language instruction, or an edited confirmation instruction, and the present embodiment does not limit the use of the confirmation operation. If the user performs a confirmation operation on the network operation scheme, the terminal sends a confirmation response to the control module in response to the confirmation operation, and the control module executes the step of associating the network operation scheme with the task description information of the network operation task based on the confirmation response.

[0129] By confirming the network operation scheme with the user, the control module can obtain the network operation scheme confirmed by the user, and can ensure the accuracy of the network operation scheme, so that the network operation system can run the network operation scheme.

[0130] The method provided by the present embodiment converts the processing flow of the network operation task described by natural language into a network operation scheme executable by the network operation system based on the large language model, and subsequently executes the network operation scheme by the network operation system to complete the network operation task, without the need for the user to write a script program for each sub-task of the network operation task, thereby improving the network operation efficiency.

[0131] In some embodiments, after obtaining the network operation scheme corresponding to the network operation task, or after the user confirms the network operation scheme, the control module can use the network operation scheme in the network operation system, such as the following steps C1 to C3.

[0132] Step C1, the control module invokes the network operation scheme and runs the network operation scheme in the network operation system to perform the network operation task on the network.

[0133] Wherein, running the network operation scheme in the network operation system includes running the network operation scheme in the network management module. The control module can send the network operation scheme to the network management module, and the network management module runs the network operation scheme in the network operation system.

[0134] For example, the control module sends the network operation scheme corresponding to the network operation task and the task information to the network management module, the network management module receives the network operation scheme and the task information, and the network management module associates the network operation scheme with the task description information of the network operation task in the task information, such as establishing the corresponding relationship between the scheme identifier of the network operation scheme and the task information of the network operation scheme in the scheme list.

[0135] The network management module invokes the network operation scheme corresponding to the network operation task, and runs the network operation scheme in the network operation system to perform the network operation task on the network. In the process of running the network operation scheme, the network management module runs each network operation sub-scheme in the network operation scheme according to the target dependency relationship to perform the corresponding sub-task. For example, according to the execution order of any first sub-task in the target dependency relationship, the network management module first runs the network operation sub-scheme corresponding to the first sub-task to perform the first sub-task, and obtains the execution result of the first sub-task, then takes the execution result of the first sub-task as the input parameter of the next sub-task after the first sub-task in the execution order of the branch, runs the network operation sub-scheme corresponding to the next sub-task to perform the next sub-task, and obtains the execution result of the next sub-task, and so on, until the last sub-task in the execution order of the branch is executed, and the execution result of the last sub-task is obtained. The execution result of the last sub-task is an execution result of the network operation task. Wherein, the process of running the network operation sub-scheme corresponding to any sub-task can be: calling the first function in the network operation sub-scheme according to the function calling process defined by the network operation sub-scheme.

[0136] Step C2, the control module outputs the execution result of the network operation task to the terminal.

[0137] Wherein, the execution result includes the execution result of each sub-task of the network operation task.

[0138] The control module obtains the execution result of each subtask after each subtask is executed, and sends the execution result of each subtask to the terminal. The terminal displays the execution result of each subtask, so that the user can view the execution result of each subtask in the terminal in a timely manner and learn the execution progress of the network operation task. Alternatively, the control module can also send the execution result of each subtask of the network operation task to the terminal after obtaining the execution result of each subtask of the network operation task.

[0139] Step C3: The terminal receives and displays the execution result of the network operation task.

[0140] The terminal can display the execution result of each subtask according to the execution order of each subtask in the network operation experience, so as to graphically reflect the processing flow of the network operation task, so that the user can understand the subtasks in the network operation task and the execution result of each subtask in detail, and the explainability of the processing flow of the network operation task can be improved.

[0141] Taking FIG. 6 as an example, the terminal displays a result display interface 600. In the result display interface 600, the relevant information of each subtask is displayed, including the description information 61 of each subtask and the execution result 62 of each subtask. The description information 61 of each subtask is the description information extracted from the network operation experience by the large language model. The relevant information of each subtask is arranged in the result display interface 600 according to the execution order of each subtask, so that the user can understand the processing flow of the network operation task in detail, and the explainability of the processing flow can be improved. In some embodiments, the terminal also displays the reason 63 for executing each subtask in the result display interface 600, further improving the explainability of the processing flow of the network operation task.

[0142] The method provided by the embodiments of the present application can execute the network operation task on the network by executing the network operation scheme in the network operation system. In this process, the network operation task is automatically completed by the network operation system without human intervention, so that the execution efficiency of the network operation task can be improved, and the network operation efficiency of the network operation system can be improved.

[0143] In other embodiments, the control module does not execute the above steps of calling the network operation scheme and executing the network operation scheme in the network operation system immediately after obtaining the network operation scheme corresponding to the network operation task, but executes the steps when an abnormal situation related to the network operation task occurs in the network. Next, taking FIG. 7 as an example, this situation is introduced.

[0144] FIG. 7 is a flowchart of a method for generating another network operation scheme according to an embodiment of the present application. The method comprises the following steps.

[0145] 701. The terminal acquires the network operation experience described in natural language, wherein the network operation experience comprises a processing flow acquired by a user performing a network operation task.

[0146] The step 701 is the same as the step 201, and thus the step 701 will not be described herein.

[0147] 702. The terminal sends the network operation experience described in natural language to a control module in a network operation system.

[0148] The step 702 is the same as the step 202, and thus the step 702 will not be described herein.

[0149] 703. The control module acquires the network operation experience described in natural language.

[0150] The step 703 is the same as the step 203, and thus the step 703 will not be described herein.

[0151] 704. The control module acquires, based on a large language model and the network operation experience, a network operation scheme for execution by the network operation system, wherein the network operation scheme corresponds to the network operation task.

[0152] The step 704 is the same as the step 204, and thus the step 704 will not be described herein.

[0153] 705. The control module sends the network operation scheme to a network management module in the network operation system.

[0154] In some embodiments, the control module executes the step 705 after the user confirms the network operation scheme. The process of confirming the network operation scheme by the user is as described above, and thus the process will not be described herein. For example, as shown in FIG. 4, the AI assistant displays the runbook flow scheme (i.e., the network operation scheme) to the terminal used by the user after the AI assistant splices the network operation sub-schemes corresponding to the plurality of sub-flows into the runbook flow scheme. After the user confirms to use the runbook flow scheme, the AI assistant sends the runbook flow scheme to the network management module, so as to save the runbook flow scheme confirmed by the user in the network management module.

[0155] 706. The network management module associates the network operation scheme with the task description information of the network operation task.

[0156] The control module can send the task information of the network operation task and the network operation scheme to the network management module together, the network management module receives the task information and the network operation scheme, stores the network operation scheme, such as persistently storing the network operation scheme, so as to avoid loss of the network operation scheme.

[0157] The network management module can also maintain a scheme list of the network, after storing the network operation scheme, the network management module can also establish a corresponding relationship between the scheme identifier of the network operation scheme and the task information of the network operation scheme in the scheme list, so as to associate the network operation scheme with the task description information in the task information, so as to query the network operation scheme based on the scheme list subsequently.

[0158] In some embodiments, the control module sends the network operation scheme corresponding to each network operation task to the network management module, the network management module stores the network operation scheme, and records the network operation scheme in the scheme list, as the number of network operation schemes stored by the network management module increases, the number of network operation schemes recorded in the scheme list also increases.

[0159] 707、The network management module receives network alarm information, the network alarm information indicating an abnormal situation occurring in the network.

[0160] The abnormal situation can be any situation that has a negative impact on the stability, security, availability and efficiency of the network in terms of network operation, such as any abnormal situation in the planning, deployment, monitoring, maintenance, optimization and troubleshooting of the network infrastructure, network devices, network systems or network applications, such as a failure (such as a single base station withdrawal) occurring in the network.

[0161] The network alarm information can be reported by a network device in the network, for example, after determining an abnormal situation occurring in the network, the network device sends network alarm information to the control module based on the abnormal situation, and the control module receives the network alarm information sent by the network device.

[0162] 708、The network management module responds to the matching of the abnormal situation and the task description information of the network operation task, invokes the network operation scheme and runs the network operation scheme in the network operation system, to perform the network operation task on the network.

[0163] The matching of the abnormal situation and the task description information of the network operation task includes that the correlation between the abnormal situation and the task description information is greater than or equal to a first threshold, the correlation represents the degree of correlation between the abnormal situation and the network operation task described by the task description information, the higher the degree of correlation, the better the network operation task described by the task description information can relieve or solve the abnormal situation.

[0164] After receiving the network alarm information, the network management module determines a target network operation and maintenance scheme based on the large language model, the network alarm information, and the task list. The target network operation and maintenance scheme is a network operation and maintenance scheme corresponding to the task description information matching the abnormal situation.

[0165] For example, the network management module concatenates the network alarm information and the scheme list into eighth prompt information according to the prompt template recognizable by the large language model. The eighth prompt information is used to prompt the large language model to select the target network operation and maintenance scheme from the network operation and maintenance schemes recorded in the scheme list according to the preset condition. The network management module inputs the eighth prompt information into the large language model. The large language model identifies the abnormal situation indicated by the network alarm information based on the prompt of the eighth prompt information, determines the relevance between the identified abnormal situation and each task description information in the task list, and outputs the relevance between the task description information and the abnormal situation and the scheme identifier corresponding to the task description information in the scheme list in response to the relevance between any task description information and the abnormal situation being greater than or equal to a first threshold (i.e., the task description information matches the abnormal situation). The relevance corresponds to the scheme identifier. The large language model can output at least one relevance and the scheme identifier corresponding to each relevance in the at least one relevance. If the large language model outputs one relevance and one corresponding scheme identifier, the network management module determines the network operation and maintenance scheme indicated by the scheme identifier as the target network operation and maintenance scheme. If the large language model outputs multiple relevances, the network management module determines the network operation and maintenance scheme indicated by the scheme identifier corresponding to the highest relevance in the multiple relevances as the target network operation and maintenance scheme.

[0166] After determining the target network operation and maintenance scheme, the network management module invokes the target network operation and maintenance scheme and runs the target network operation and maintenance scheme in the network operation and maintenance system. For example, the network management module invokes the network operation and maintenance scheme corresponding to the network operation and maintenance task and runs the network operation and maintenance scheme in the network operation and maintenance system. The specific process can be referred to the above step C1.

[0167] Taking FIG. 4 as an example, the network device reports the network state to the network management module, and the network state represents an abnormal situation in the network. The network management module splices the network state and the scheme list to the prompt template to obtain the prompt information, and the prompt information prompts the large language model to help select the network operation and maintenance scheme according to the network state and the scheme list. The large language model selects the network operation and maintenance scheme for the network state from the network operation and maintenance schemes recorded in the scheme list based on the input prompt information, and returns the scheme identifier of the selected network operation and maintenance scheme to the network management module. The network management module obtains the scheme identifier and executes the network operation and maintenance scheme indicated by the scheme identifier. For example, in the process of executing the network operation and maintenance scheme, the network management module sends an instruction to the network device, the instruction is involved when the network management module calls the first function in the network operation and maintenance scheme, the network device receives and executes the instruction, and returns the execution result of the instruction to the network management module. The network management module returns the execution result of the instruction to the user for viewing, for example, the network management module sends the execution result of the instruction to the AI assistant, the AI assistant returns the execution result of the instruction to the terminal, and the user views the execution result of the instruction on the terminal. The execution result includes the current information or state of the network and / or the network device after the execution of the instruction. The execution result of the instruction is also the execution result of the network operation and maintenance task.

[0168] In some other embodiments, the network device can directly send the network alarm information to the control module, or the network device sends the network alarm information to the network management module, and the network management module forwards the network alarm information to the control module. After the control module receives the network alarm information, the control module responds to the matching between the abnormal situation indicated by the network alarm information and the task description information of any network operation and maintenance task, calls the network operation and maintenance scheme corresponding to the network operation and maintenance task, and runs the network operation and maintenance scheme in the network operation and maintenance system to obtain the execution result of the network operation and maintenance task. The control module outputs the execution result of the network operation and maintenance task to the terminal.

[0169] In the network alarm scenario, for any module in the network management module or the control module, after the module obtains the execution result of the network operation and maintenance task, the module can record the association relationship between the execution result and the network alarm information without actively outputting the execution result of the network operation and maintenance task to the terminal. When the user queries the network alarm information, the execution result of the network operation and maintenance task is output to the terminal.

[0170] The method provided by the embodiments of the present application can not only achieve the beneficial effects of the method embodiment of FIG. 2, but also trigger the network operation and maintenance system to automatically execute the network operation and maintenance task related to the abnormal situation in the network when the abnormal situation occurs in the network, so as to solve or alleviate the abnormal situation in the network without waiting for manual processing, thereby improving the network operation and maintenance efficiency of the network operation and maintenance system.

[0171] In some embodiments, an abnormal situation that occurs in the network more than or equal to a frequency threshold is referred to as a high-frequency abnormal situation, and for a network operation task for solving the high-frequency abnormal situation, the user triggers the terminal to perform the above step 701 to trigger the execution of the above steps 702 to 705 to store the network operation scheme corresponding to the network operation task in the network management module. Alternatively, after the control module performs the above step 704 and obtains the network operation scheme corresponding to a certain network operation task, the control module determines whether the abnormal situation solved by the network operation task is a high-frequency abnormal situation according to the frequency threshold. If it is a high-frequency abnormal situation, the control module stores the network operation scheme corresponding to the network operation task in the network management module. If it is not a high-frequency abnormal situation (the frequency of occurrence of the abnormal situation in the network is less than the frequency threshold), the control module does not store the network operation scheme corresponding to the network operation task in the network management module, so that the network management module can have more storage space to store network operation schemes for solving high-frequency abnormal situations. In this way, when a high-frequency abnormal situation occurs in the network, the network management module can quickly find a network operation scheme for solving the currently occurring high-frequency abnormal situation from the stored network operation schemes, and solve the currently occurring high-frequency abnormal situation by executing the queried network operation scheme in the network operation system, thereby further improving the network operation efficiency.

[0172] In other embodiments, after the control module obtains the first script program of the network operation task, the user can also trigger the network management module to run the first script program through a dialogue message. Next, taking Figure 8 as an example, this case is introduced.

[0173] Figure 8 is a flowchart of another network operation method based on a large language model provided by an embodiment of the present application, which includes the following steps.

[0174] 801. The terminal obtains network operation experience described in natural language, and the network operation experience includes a processing flow obtained by the user performing a network operation task.

[0175] Wherein, the step 801 is the same as the step 201, and here, the embodiment of the present application will not repeat the step 801.

[0176] 802. The terminal sends the network operation experience described in natural language to the control module in the network operation system.

[0177] Wherein, the step 802 is the same as the step 202, and here, the embodiment of the present application will not repeat the step 802.

[0178] 803. The control module obtains the network operation experience described in natural language.

[0179] In this step 803, the network operation task is obtained based on the network operation experience, and the network operation task corresponds to the network operation scheme. The step 803 is the same as the step 203, and details are not repeated here.

[0180] 804、The control module obtains a network operation scheme for the network operation system to execute based on the large language model and the network operation experience, and the network operation scheme corresponds to the network operation task.

[0181] The network operation scheme corresponds to the network operation task, which means that the network operation scheme is associated with the task description information of the network operation task. The step 804 is the same as the step 204, and details are not repeated here.

[0182] 805、The terminal sends a conversation message to the control module, and the conversation message includes a network operation intent described by natural language.

[0183] The network operation intent is the user's intention in network operation, i.e., the network operation task the user expects to perform.

[0184] The user interacts with the AI assistant in the network operation system through the terminal, and in the conversation process, the user interacts with the AI assistant through the conversation message to trigger the control module where the AI assistant is located to execute the network operation scheme in the network operation system. For example, the network operation application running in the terminal provides a conversation interface, which is used to provide network operation services for the user in a conversation manner. For example, the user sends a conversation message to the AI assistant of the control module through the conversation interface, and the AI assistant analyzes the conversation message to trigger the network management module to execute the network operation scheme related to the conversation message by using the resources of the control module.

[0185] For example, the terminal displays the conversation interface 900, the user inputs the conversation message 92 in the message input box 91 in the conversation interface 900, the conversation message 92 includes a network operation intent 921 described by natural language, the user performs a sending operation on the conversation message 92, the terminal acquires the conversation message 92 input by the user in response to the sending operation, sends the conversation message 92 to the second account 94 in the conversation interface 900 through the first account 93, the first account 93 is an account used by the user to log in to the network operation application, and the second account 94 is an account used by the AI assistant to log in to the network operation application. Sending the conversation message 92 to the second account 94 means that the terminal sends the conversation message 92 to the control module running the AI assistant.

[0186] 806、The control module responds to the matching of the network operation intent in the conversation message and the task description information of the network operation task, calls the network operation scheme, and runs the network operation scheme in the network operation system to perform the network operation task on the network.

[0187] The matching of the network operation intention and the task description information includes that a correlation between the network operation intention and the task description information is greater than or equal to a second threshold value, and the correlation indicates a degree of association (or a degree of similarity) between the network operation intention and the task description information.

[0188] After receiving the dialogue message, the control module determines, based on the large language model and the task description information of the plurality of network operation tasks, a target network operation scheme from a plurality of network operation schemes, the plurality of network operation schemes being each network operation scheme recorded in the scheme list, the target network operation scheme being a network operation scheme in the plurality of network operation schemes that matches the dialogue message, and the description information corresponding to the target network operation scheme matching the network operation intention.

[0189] For example, the control module concatenates the dialogue message and the scheme list into ninth prompt information according to a prompt template recognizable by the large language model, the scheme list including the task description information of the plurality of network operation tasks. The ninth prompt information is used to prompt the large language model to select, from the task list, the task description information corresponding to the network operation intention in the dialogue message. The control module inputs the ninth prompt information into the large language model. The large language model performs operation intention analysis on the dialogue message based on the prompt of the ninth prompt, obtains the network operation intention of the dialogue message, determines the correlation between the network operation intention and each task description information in the scheme list, and in response to the correlation between any task description information and the network operation intention being greater than or equal to the second threshold value (i.e., the task description information matches the network operation intention), outputs the correlation between the task description information and the network operation intention and the scheme identifier corresponding to the task description information in the scheme list. The large language model can output at least one correlation and the scheme identifier corresponding to each correlation in the at least one correlation. If the large language model outputs one correlation and one corresponding scheme identifier, the network management module determines the network operation scheme indicated by the scheme identifier as the target network operation scheme. If the large language model outputs a plurality of correlations, the network management module determines the network operation scheme indicated by the scheme identifier corresponding to the highest correlation in the plurality of correlations as the target network operation scheme.

[0190] After determining the target network operation scheme, the control module invokes the target network operation scheme and runs the target network operation scheme in the network operation system. For example, the control module invokes the network operation scheme corresponding to the network operation task and runs the network operation scheme in the network operation system as the network operation scheme involved in step 804 above. The specific process can be referred to step C1 above.

[0191] In some embodiments, before the control module invokes the target network operation and maintenance scheme and runs the target network operation and maintenance scheme in the network operation and maintenance system, the control module sends the target network operation and maintenance scheme to the terminal in the form of a session message, and the terminal displays the target network operation and maintenance scheme in the form of a session message in a session interface. If the user performs a confirmation operation on the target network operation and maintenance scheme in the session interface, the control module performs the steps of invoking the target network operation and maintenance scheme and running the target network operation and maintenance scheme in the network operation and maintenance system in response to the confirmation operation.

[0192] 807. The control module outputs the execution result of the network operation and maintenance task to the terminal.

[0193] In this step 808, the control module sends the execution result of the network operation and maintenance task to the terminal in the form of a session message. That is, the session message sent by the control module includes the execution result of the network operation and maintenance task.

[0194] 808. The terminal receives and displays the execution result of the network operation and maintenance task.

[0195] In this step 808, the control module sends the execution result of the network operation and maintenance task to the terminal in the form of a session message. That is, the session message sent by the control module includes the execution result of the network operation and maintenance task.

[0196] For example, in FIG. 9, after the terminal sends the session message 92 to the second account 94 in the session interface 900, the terminal receives the session message 95 returned by the control module. The session message 95 includes the execution result of the network operation and maintenance task 951. The terminal sends the session message 95 to the first account 93 in the form of a session message sent by the second account 94. In the session interface 900, the session message 95 sent by the second account 94 to the first account 93 is displayed. In the session message 95, the execution result 953 of each subtask is displayed in the order of execution of each subtask 952 in the network operation and maintenance task 951. In some embodiments, the reason 954 for executing each subtask 952 is also displayed in the session message 95.

[0197] In some embodiments, if there is no task description information in the network operation system that matches the network operation intent in the dialogue message, and there is no network operation solution in the network operation system that matches the dialogue message, the control module can prompt the user to input network operation experience for realizing the network operation intent through the terminal, trigger the control module to generate a network operation solution for the network operation intent according to the flow shown in FIG. 2, call the generated network operation solution and run the network operation solution in the network operation system to complete the network operation intent. Or, trigger the control module to execute the step of determining the target network operation solution for the dialogue message again, call the corresponding determined network operation solution and run the network operation solution in the network operation system. Or, the control module designs a processing flow for the network operation intent based on the large language model, arranges the designed processing flow into a network operation solution, calls the arranged network operation solution and runs the new network operation solution in the network operation system to obtain the execution result of the network operation task, and provides the execution result to the user as a reference result.

[0198] The method provided by the embodiments of the present application determines the task description information matching the network operation intent in the dialogue message through the large language model, calls the network operation solution corresponding to the network operation task described by the task description information to execute the network operation task, and provides the execution result of the network operation task to the terminal after the execution of the network operation task is completed. In this process, the user only needs to provide the dialogue message in the form of dialogue, without the need for multiple interactions with the network operation system, the operation process is simple, and the network operation efficiency can be improved.

[0199] The user can also modify the generated network operation solution to meet the user's modification requirements for the processing flow of the network operation task. For example, the control module acquires the modified network operation solution in response to the modification operation on the network operation solution, and associates the modified network operation solution with the task description information of the network operation task in response to the confirmation operation on the modified network operation solution. The network operation solution is any network operation solution involved in the above embodiments.

[0200] For example, the control module sends a confirmation message of a network operation and maintenance solution to the terminal, the terminal displays the network operation and maintenance solution in the confirmation message, such as the program display interface 500 shown in FIG. 5 displays the process flow diagram 51, the user triggers an operation on any task node 512 in the process flow diagram 51, triggering the terminal to display the network operation and maintenance sub-solution associated with the task node 512 in the program display window 52, the user can modify the network operation and maintenance sub-solution in the program display window 52, if the user has modification requirements for the network operation and maintenance sub-solution associated with other task nodes, the user can modify the network operation and maintenance sub-solution associated with other task nodes in a similar manner. After the modification is completed, the user confirms the modified process flow diagram 51, in response to the confirmation operation, for any task node 512, if the user modifies the network operation and maintenance sub-solution associated with the task node 512, the network operation and maintenance sub-solution associated with the task node 512 in the network operation and maintenance solution is replaced by the network operation and maintenance sub-solution modified by the user, to obtain a modified network operation and maintenance solution. The terminal sends a confirmation response including the modified network operation and maintenance solution to the control module, the control module compares the network operation and maintenance solution in the confirmation response with the network operation and maintenance solution sent to the terminal after receiving the confirmation response, if the two network operation and maintenance solutions are the same, it means that the user does not modify the network operation and maintenance solution, then the control module associates the network operation and maintenance solution with the task description information of the network operation and maintenance task, if the two script programs are not the same, it means that the user modifies the network operation and maintenance solution, the network operation and maintenance solution in the confirmation response is obtained as the modified network operation and maintenance solution, and the modified network operation and maintenance solution is associated with the task description information of the network operation and maintenance task.

[0201] In some other embodiments, the terminal sends a request to the control module to modify a certain network operation and maintenance solution, the control module returns a network operation and maintenance solution to the terminal based on the request, the user modifies the network operation and maintenance solution on the terminal, the terminal sends the modified network operation and maintenance solution to the control module, and the control module associates the modified network operation and maintenance solution with the task description information of the network operation and maintenance task in response to the confirmation operation of the modified network operation and maintenance solution. The process is the same as the process in which the control module associates the network operation and maintenance solution with the task description information of the network operation and maintenance task in response to the confirmation operation of the network operation and maintenance solution, which will not be described here.

[0202] Subsequently, in the case that the task description information matches the network operation and maintenance intention in a certain dialogue message of the terminal, or in the case that the task description information matches the abnormal situation indicated by the network alarm information, the control module calls the modified network operation and maintenance solution and runs the modified network operation and maintenance solution in the network operation and maintenance system.

[0203] In other embodiments, the user can also trigger the control module to regenerate the network operation scheme by modifying the network operation experience corresponding to the network operation task. Next, steps D1 to D2 are introduced in combination to describe this case.

[0204] Step D1, the control module obtains the modified network operation experience in response to the modification operation on the network operation experience corresponding to the network operation task.

[0205] The network operation experience is the network operation experience corresponding to the existing network operation scheme in the network operation system.

[0206] The user can perform a modification operation on the network operation experience on the terminal, and the control module obtains the modified network operation experience from the terminal. Taking FIG. 10 as an example, the network operation application also provides a task display interface 1000, which is used to display network operation tasks with network operation schemes in the network operation system. The display method is, for example, the task display interface 1000 includes a task display table 1001, and the task display table 1001 includes task identifiers 1002 of a plurality of network operation tasks with network operation schemes. The task display table 1001 also includes an editing control 1003 corresponding to the task identifier 1002, and the editing control 1003 is used to trigger a task editing interface of the corresponding network operation task. Taking the network operation task “single base station withdrawal” as an example, the user triggers the editing control 1003 corresponding to the network operation task “single base station withdrawal” in the task display table 1001, and the terminal displays the task editing interface 300 of the network operation task “single base station withdrawal” in response to the triggering operation. The user can modify the network operation experience 311 of the network operation task “single base station withdrawal” in the first input box 31 in the task editing interface 300. After the modification is completed, the user triggers the compilation control 35 in the task editing interface 300, and the terminal sends a compilation request to the control module in response to the triggering operation. The compilation request includes the modified network operation experience in the first input box 31. The control module receives the compilation request and obtains the modified network operation experience of “single base station withdrawal” from the network operation experience in the compilation request.

[0207] Step D2, the control module obtains the modified network operation scheme based on the large language model and the modified network operation experience, and updates the network operation scheme corresponding to the network operation task.

[0208] For the modified network operation and maintenance scheme, in a similar manner to step 204, the control module obtains a network operation and maintenance scheme for the network operation and maintenance system to execute based on the large language model and the modified network operation and maintenance experience. The obtained network operation and maintenance scheme is the modified network operation and maintenance experience of the network operation and maintenance task. The control module replaces the stored network operation and maintenance scheme corresponding to the network operation and maintenance task with the modified network operation and maintenance scheme, to update the network operation and maintenance scheme corresponding to the network operation and maintenance task. In some other embodiments, the control module replaces the stored network operation and maintenance scheme corresponding to the network operation and maintenance task with the modified network operation and maintenance scheme in response to a confirmation operation on the modified network operation and maintenance scheme.

[0209] Subsequently, in a case where the task description information of the network operation and maintenance task matches a network operation and maintenance intention in a certain dialogue message of the terminal, or in a case where the task description information matches an abnormal situation indicated by the network alarm information, the control module invokes the modified network operation and maintenance scheme and runs the modified network operation and maintenance scheme in the network operation and maintenance system.

[0210] For the network operation and maintenance system involved in each of the above embodiments, the control module in the network operation and maintenance system can be deployed on at least one first computer device, and the method steps performed by the control module in the above method embodiments can be executed by the at least one first computer device. The first computer device can be a personal computer, a laptop computer, a desktop computer, a server, etc. In this embodiment of the present application, the device type and the number of the first computer device are not limited.

[0211] The network management module in the network operation and maintenance system can be deployed on at least one second computer device, and the method steps performed by the network management module in the above method embodiments can be executed by the at least one second computer device. The second computer device can be a personal computer, a laptop computer, a desktop computer, a server, a network element device, or a network element management device, etc. In this embodiment of the present application, the device type and the number of the second computer device are not limited. The first computer device and the second computer device can be different computer devices, or can be the same computer device.

[0212] The large language model in the network operation and maintenance system can be deployed on at least one third computer device, and the at least one third computer device is used to run the large language model. The third computer device can be a desktop computer or a server, etc. In this embodiment of the present application, the device type and the number of the third computer device are not limited.

[0213] ​The third computer device is different from the first computer device and the second computer device. In the process of implementing the method for generating the network operation scheme, the control module in the first computer device and the network management module in the second computer device can call the large language model in the third computer device in the case of having a use demand for the large language model. For example, in the generation of the network operation scheme, the control module sends the network operation experience to the third computer device, and the third computer device compiles the network operation experience into the network operation scheme through the large language model, and returns the network operation scheme to the control module.

[0214] In other embodiments, the large language model can also be deployed in the first computer device or the second computer device. In this embodiment of the application, the deployment position of the large language model is not limited. The large language model can be inferred and trained by the computer device where the large language model is located, or can be inferred and trained by other computer devices other than the computer device where the large language model is located. In this embodiment of the application, the device used for inferring and training the large language model is not limited.

[0215] The above embodiments are described by taking the control module and the network management module as different modules as an example. In other embodiments, the control module and the network management module can also be integrated into one module, and the module can implement the method steps performed by the control module and the network management module in the above method embodiments.

[0216] The above describes the method of the embodiment of the application, and the device of the embodiment of the application is described below. It should be understood that the device described below has the same inventive concept of the above network operation method based on the large language model, and can achieve the beneficial effects achieved by the above method embodiments. The device and the equipment related to the application will be described below with reference to FIGS. 11 to 12. It should be understood that the technical features described in the above method embodiments are also applicable to the following device embodiments.

[0217] FIG. 11 is a structural schematic diagram of a network operation scheme generation device. As shown in FIG. 11, the device 1100 includes:

[0218] A first acquisition unit 1101 is configured to acquire network operation experience described by natural language. The network operation experience includes a processing flow obtained by a user performing a network operation task.

[0219] A second acquisition unit 1102 is configured to acquire, based on a large language model and the network operation experience, a network operation scheme for a network operation system to execute. The network operation scheme corresponds to the network operation task.

[0220] Optionally, the second acquisition unit 1002 includes:

[0221] The analysis unit is configured to analyze the network operation experience based on the large language model to obtain a plurality of subtasks and a target dependency relationship of the network operation task, the target dependency relationship being a dependency relationship between the plurality of subtasks in a processing flow.

[0222] The arrangement unit is configured to arrange a plurality of network operation sub-solutions corresponding to the plurality of subtasks based on the large language model and the target dependency relationship to obtain the network operation solution.

[0223] Optionally, the analysis unit includes:

[0224] The disassembling subunit is configured to disassemble the network operation experience based on the large language model to obtain description information of the plurality of subtasks.

[0225] The analysis subunit is configured to analyze the dependency relationship of the network operation experience based on the large language model to obtain the target dependency relationship.

[0226] Optionally, the analysis subunit is configured to:

[0227] analyze an execution order of the plurality of subtasks in the network operation experience based on the large language model; and

[0228] determine the target dependency relationship based on the execution order of the plurality of subtasks.

[0229] Optionally, the first obtaining unit is configured to obtain the processing flow through a first input box of a task editing interface, the task editing interface being used for a user to input the network operation experience.

[0230] Optionally, the first obtaining unit is configured to obtain the processing flow through a first input box of a task editing interface, the task editing interface being used for a user to input the network operation experience; and obtain task description information of the network operation task through a second input box of the task editing interface.

[0231] Optionally, the apparatus 1100 further includes:

[0232] The output unit is configured to output the network operation solution.

[0233] The association unit is configured to associate the network operation solution with the task description information of the network operation task in response to a confirmation operation on the network operation solution, the confirmation operation indicating that the network operation solution is used to implement the network operation task.

[0234] Optionally, the apparatus 1100 further includes:

[0235] The third obtaining unit is configured to obtain an input dialogue message, the dialogue message including a network operation intention described through natural language.

[0236] The calling unit is configured to, in response to the network operation intention matching the task description information of the network operation task, call the network operation scheme and run the network operation scheme in the network operation system.

[0237] The output unit is configured to output an execution result of the network operation task.

[0238] Optionally, the apparatus 1100 further includes:

[0239] The receiving unit is configured to receive network alarm information, the network alarm information indicating an abnormal situation occurring in the network.

[0240] The calling unit is configured to, in response to the abnormal situation matching the task description information of the network operation task, call the network operation scheme and run the network operation scheme in the network operation system.

[0241] The output unit is configured to output an execution result of the network operation task.

[0242] Optionally, the apparatus 1100 further includes:

[0243] The fourth obtaining unit is configured to, in response to a modification operation on the network operation experience, obtain the modified network operation experience.

[0244] The second obtaining unit 1102 is further configured to, based on the large language model and the modified network operation experience, obtain the modified network operation scheme and update the network operation scheme.

[0245] Optionally, the apparatus 1100 further includes:

[0246] The fifth obtaining unit is configured to, in response to a modification operation on the network operation scheme, obtain the modified network operation scheme.

[0247] The associating unit is configured to, in response to a confirmation operation on the modified network operation scheme, associate the modified network operation scheme with the task description information of the network operation task, the confirmation operation indicating that the network operation scheme is confirmed to be used to implement the network operation task.

[0248] It should be understood that the apparatus 1100 corresponds to the control module in the network operation system described above, and each module in the apparatus 1100 and the other operations and / or functions described above are respectively used to implement various steps and methods performed by the control module in the method embodiments, and specific details can be referred to the method embodiments described above. For the sake of brevity, they will not be described here. In some other embodiments, the apparatus 1100 also corresponds to the network management module in the network operation system, and the apparatus 1100 is also used to perform the method steps performed by the network management module in the generation method of the network operation scheme described above. For the sake of brevity, they will not be described here. Specific details can be referred to the method embodiments described above.

[0249] It should be understood that the apparatus 1100 is only exemplified by the above division of the functional modules when generating the network operation and maintenance scheme, and in actual application, the above functions can be completed by different functional modules according to needs, that is, the internal structure of the apparatus 1100 is divided into different functional modules to complete all or part of the functions described above. In addition, the apparatus 1100 provided in the above embodiments and the above method embodiments belong to the same concept, and the specific implementation process is detailed in the above method embodiments, which will not be described here.

[0250] Fig. 12 shows a structural schematic diagram of a computer device provided by an embodiment of the present application. The computer device 1200 shown in Fig. 12 is configured as a terminal or a server in the foregoing application environment. As shown in Fig. 12, the computer device 1200 includes a bus 1202, a processor 1204, a memory 1206 and a communication interface 1208. The processor 1204, the memory 1206 and the communication interface 1208 communicate through the bus 1202. The computer device 1200 can be a server or a terminal device. It should be understood that the number of processors and memories in the computer device 1200 is not limited by the present application. The bus 1202 can be a peripheral component interconnect (PCI) bus or an extended industry standard architecture (EISA) bus, etc. The bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, only one line is shown in Fig. 12, but it does not mean that there is only one bus or only one type of bus. The bus 1204 can include a path for transmitting information between various components (for example, the memory 1206, the processor 1204, the communication interface 1208) of the computer device 1200. The processor 1204 can include any one or more of a central processing unit (CPU), a graphics processing unit (GPU), a microprocessor (MP), or a digital signal processor (DSP), etc. The memory 1206 can include a volatile memory, such as a random access memory (RAM). The processor 1204 can also include a non-volatile memory, such as a read-only memory (ROM), a flash memory, a mechanical hard disk drive (HDD) or a solid state drive (SSD). The memory 1206 stores executable program code. The processor 1204 executes the executable program code to implement the method steps performed by the control module and / or the network management module in the method for generating a network operation scheme in the foregoing method embodiments. Taking the computer device 1200 configured as a first computer device as an example, the processor 1204 executes the executable program code to implement the method steps performed by the control module in the method for generating a network operation scheme.For example, the computer device 1200 is configured as the second computer device, the processor 1204 executes the executable program code to implement the method steps performed by the network management module in the generation method of the network operation and maintenance scheme. The communication interface 1208 uses a transceiver module such as but not limited to a network interface card and a transceiver to implement the communication between the computer device 1200 and other devices or communication networks.

[0251] In an exemplary embodiment, a computer readable storage medium is also provided, for example, a memory including program code, the program code being executable by a processor of a computer device to complete the generation method of the network operation and maintenance scheme. For example, in the case that the computer device can be configured as the first computer device, the program code is executable by the processor of the computer device to complete the method steps performed by the control module in the generation method of the network operation and maintenance scheme. For another example, in the case that the computer device can be configured as the second computer device, the program code is executable by the processor of the computer device to complete the method steps performed by the network management module in the generation method of the network operation and maintenance scheme. The computer readable storage medium is a non-transitory computer readable storage medium, such as a read-only memory (ROM), a random access memory (RAM), a compact disc read-only memory (CD-ROM), a magnetic tape, a floppy disk, and an optical data storage device, etc.

[0252] The embodiments of the present application also provide a computer program product or a computer program, the computer program product or the computer program including program code stored in a computer readable storage medium, the processor of a computer device reading the program code from the computer readable storage medium, and the processor executing the program code to make the computer device execute the generation method of the network operation and maintenance scheme. For example, in the case that the computer device can be configured as the first computer device, the program code is executable by the processor of the computer device to complete the method steps performed by the control module in the generation method of the network operation and maintenance scheme. For another example, in the case that the computer device can be configured as the second computer device, the program code is executable by the processor of the computer device to complete the method steps performed by the network management module in the generation method of the network operation and maintenance scheme.

[0253] In addition, the embodiment of the present application further provides a device, which can be a chip, a component or a module. The device can include a processor and a memory connected to each other. The memory is used to store computer-executable instructions. When the device is running, the processor can execute the computer-executable instructions stored in the memory, so that the chip executes the network operation scheme generation method provided in the above method embodiments.

[0254] The device, the equipment, the computer readable storage medium, the computer program product or the chip provided by the embodiment can achieve the beneficial effects of the corresponding method provided above, and the beneficial effects of the corresponding method provided above are not repeated here.

[0255] In the description of the present application, unless otherwise specified, " / " means "or", for example, A / B can mean A or B. "And / or" in this paper is only a description of the association between the associated objects, which means that there can be three relationships, for example, A and / or B, which can mean that A exists alone, A and B exist together, and B exists alone. In addition, "at least one" means one or more, and "multiple" means two or more. "First", "second", etc. do not limit the quantity and execution order, and "first", "second", etc. do not necessarily mean different.

[0256] In the present application, the word "exemplary" or "for example" is used to mean serving as an example, instance, or illustration. Any embodiment or design scheme described as "exemplary" or "for example" in the present application should not be interpreted as more preferred or more advantageous than other embodiments or design schemes. Rather, the use of "exemplary" or "for example" is intended to present relevant concepts in a concrete manner.

[0257] It should be noted that the information (including but not limited to user equipment information, user personal information, etc.), data (including but not limited to data for analysis, stored data, displayed data, etc.) and signals involved in the present application are authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data need to comply with relevant laws, regulations and standards of relevant countries and regions. For example, the network operation experience and dialogue messages involved in the present application are obtained under full authorization.

[0258] All the optional technical solutions described above can be combined to form optional embodiments of the present disclosure, which will not be repeated here.

[0259] The above is only optional embodiments of the present application, and does not limit the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principles of the present application shall be included in the protection scope of the present application.

Claims

1. A method for generating a network operation scheme, characterized in that, The method comprises: acquiring network operation and maintenance experience described in natural language, the network operation and maintenance experience comprising a processing flow obtained by a user performing a network operation and maintenance task; based on a large language model and the network operation and maintenance experience, acquiring a network operation and maintenance scheme for a network operation and maintenance system to execute, the network operation and maintenance scheme corresponding to the network operation and maintenance task.

2. The method of claim 1, wherein, The acquisition of the network operation and maintenance scheme for the network operation and maintenance system to execute based on the large language model and the network operation and maintenance experience comprises: based on the large language model, performing operation and maintenance task analysis on the network operation and maintenance experience to obtain a plurality of subtasks of the network operation and maintenance task and a target dependency relationship, the target dependency relationship being a dependency relationship between the plurality of subtasks in the processing flow; based on the large language model, arranging network operation and maintenance sub-schemes corresponding to the plurality of subtasks according to the target dependency relationship to obtain the network operation and maintenance scheme.

3. The method of claim 2, wherein, The operation and maintenance task analysis on the network operation and maintenance experience based on the large language model to obtain the plurality of subtasks of the network operation and maintenance task and the target dependency relationship comprises: based on the large language model, disassembling the network operation and maintenance experience to obtain description information of the plurality of subtasks; based on the large language model, performing dependency relationship analysis on the network operation and maintenance experience to obtain the target dependency relationship.

4. The method of claim 3, wherein, The dependency relationship analysis on the network operation and maintenance experience based on the large language model to obtain the target dependency relationship comprises: based on the large language model, analyzing the execution order of the plurality of subtasks in the network operation and maintenance experience; based on the execution order of the plurality of subtasks, determining the target dependency relationship.

5. The method according to any one of claims 1-4, characterized in that, The acquisition of the network operation and maintenance experience described in natural language comprises: acquiring the processing flow through a first input box of a task editing interface, the task editing interface being used by the user to input the network operation and maintenance experience.

6. The method according to any one of claims 1-4, characterized in that, The network operation and maintenance experience further comprises task description information of the network operation and maintenance task, and the acquisition of the network operation and maintenance experience described in natural language comprises: acquiring the processing flow through a first input box of a task editing interface, the task editing interface being used by the user to input the network operation and maintenance experience; acquiring the task description information of the network operation and maintenance task through a second input box of the task editing interface.

7. The method according to any one of claims 1 to 6, characterized in that, After the acquisition of the network operation and maintenance scheme for the network operation and maintenance system to execute, the method further comprises: outputting the network operation and maintenance scheme; in response to a confirmation operation of the network operation and maintenance scheme, associating the network operation and maintenance scheme with task description information of the network operation and maintenance task, the confirmation operation indicating that the network operation and maintenance scheme is confirmed to be used to implement the network operation and maintenance task.

8. The method according to any one of claims 1 to 7, characterized in that, After the acquisition of the network operation and maintenance scheme for the network operation and maintenance system to execute, the method further comprises: acquiring an input dialogue message, the dialogue message comprising a network operation and maintenance intent described in natural language; in response to the network operation and maintenance intent matching the task description information of the network operation and maintenance task, invoking the network operation and maintenance scheme and running the network operation and maintenance scheme in the network operation and maintenance system; output an execution result of the network operation task.

9. The method according to any one of claims 1-8, characterized in that, After the network operation scheme for execution by the network operation system is acquired, the method further includes: receiving network alarm information, the network alarm information indicating an abnormal situation occurring in the network; in response to the abnormal situation matching task description information of the network operation task, invoking the network operation scheme and running the network operation scheme in the network operation system; outputting an execution result of the network operation task.

10. A network operation scheme generation apparatus characterized by comprising: The device includes: a first acquisition unit configured to acquire network operation experience described in natural language, the network operation experience including a processing flow obtained by a user performing a network operation task; a second acquisition unit configured to acquire, based on a large language model and the network operation experience, a network operation scheme for execution by a network operation system, the network operation scheme corresponding to the network operation task.

11. The apparatus of claim 10, wherein, The second acquisition unit includes: an analysis unit configured to analyze, based on the large language model, an execution order of the multiple subtasks in the network operation experience; and an arrangement unit configured to arrange, based on the large language model and the target dependency relationship, network operation sub-schemes corresponding to the multiple subtasks to obtain the network operation scheme.

12. The apparatus of claim 11, wherein, The analysis unit is configured to: analyze, based on the large language model, an execution order of the multiple subtasks in the network operation experience; and determine the target dependency relationship based on the execution order of the multiple subtasks.

13. The apparatus of claim 12, wherein, The first acquisition unit is configured to: acquire the processing flow through a first input box of a task editing interface, the task editing interface being used by the user to input the network operation experience. The first acquisition unit is configured to:

14. The apparatus of any one of claims 10-13, wherein, acquire the processing flow through a first input box of a task editing interface, the task editing interface being used by the user to input the network operation experience; and acquire task description information of the network operation task through a second input box of the task editing interface.

15. The apparatus of any one of claims 10-13, wherein, The device further includes: an output unit configured to output the network operation scheme; and an association unit configured to associate, in response to a confirmation operation on the network operation scheme, the network operation scheme with task description information of the network operation task, the confirmation operation indicating confirmation that the network operation scheme is used to implement the network operation task.

16. The apparatus of any one of claims 10-15, wherein, The device further includes: a third acquisition unit configured to acquire an input dialogue message, the dialogue message including a network operation intent described in natural language. ​ 17. The apparatus of any one of claims 10-16, wherein, ​ ​ The calling unit is configured to, in response to the network operation intention matching the task description information of the network operation task, call the network operation scheme and run the network operation scheme in the network operation system. The output unit is configured to output an execution result of the network operation task.

18. The apparatus of any one of claims 10-17, wherein, The apparatus further includes: The receiving unit is configured to receive network alarm information, the network alarm information indicating an abnormal situation occurring in a network. The calling unit is configured to, in response to the abnormal situation matching the task description information of the network operation task, call the network operation scheme and run the network operation scheme in the network operation system. The output unit is configured to output an execution result of the network operation task.

Images