Network planning construction method, apparatus, device and medium

Abstract

Embodiments of the present application relate to the technical field of artificial intelligence (AI), and provide a network planning construction method, an apparatus, a device and a medium. The method comprises: on the basis of a human-machine interaction mode, collecting a first natural language indicating a network planning construction intent; inputting the first natural language into an AI network large model to obtain a target network construction solution, the target network construction solution comprising network topology information and network configuration information; and when it is detected that a network device corresponding to the network topology information is powered on, configuring the network device according to the network configuration information. The adoption of the technical solution provided by the embodiments of the present application can improve the ability to understand user service-level intents, improve the scalability of intent collection, and improve the adaptability to scenarios and networking.

Description

A method, apparatus, equipment and medium for network planning and construction Technical Field

[0001] This application relates to the field of artificial intelligence technology, and in particular to a method, apparatus, equipment and medium for network planning and construction. Background Technology

[0002] Network planning and construction is one phase in the management services of a Data Center Network (DCN) throughout its entire lifecycle. The process of implementing network planning and construction is essentially the process of configuring network equipment. Currently, the network planning and construction process involves: determining the target network construction plan based on the network language of the user-input diagram and the pre-configured network construction plan, and configuring network equipment according to this target network construction plan.

[0003] This network planning and construction method uses highly technical network language for input intents, which is difficult for users to understand and requires professional network administrators to comprehend, resulting in poor user-friendliness. Furthermore, it can only collect and understand fixed intents. When new or changed intents appear, the device will be unable to understand them and will be unable to configure network devices. In addition, the device can only configure network devices based on fixed intents and pre-configured network construction plans, but the pre-configured network construction plans are limited and cannot adapt to new scenarios or networking requirements on their own.

[0004] Summary of the Invention

[0005] The purpose of this application is to provide a network planning and construction method, apparatus, device, and medium to enhance the understanding of user service-level intentions, improve the scalability of intention collection, and enhance adaptability to different scenarios and network topologies. The specific technical solution is as follows:

[0006] In a first aspect, embodiments of this application provide a network planning and construction method, the method comprising:

[0007] Based on human-computer interaction, the system collects first natural language samples indicating the intentions of network planning and construction.

[0008] The first natural language is input into the Artificial Intelligence (AI) network model to obtain the target network construction scheme, which includes network topology information and network configuration information.

[0009] After the network device corresponding to the network topology information is detected to be powered on, the network device is configured according to the network configuration information.

[0010] In some embodiments, the step of collecting first natural language indicating the intention to plan and construct the network based on human-computer interaction includes:

[0011] Display the human-computer interaction dialog box;

[0012] Receive the user's first natural language input on the human-computer interaction dialog box indicating the network planning and construction intention.

[0013] In some embodiments, the method further includes:

[0014] The first prompt message is displayed on the human-computer interaction dialog box. The first prompt message is used to guide the input of network planning and construction intentions.

[0015] In some embodiments, the step of inputting the first natural language into the AI ​​network large model to obtain the target network construction scheme includes:

[0016] The first natural language is matched with pre-stored knowledge of various network construction schemes to obtain multiple candidate knowledge;

[0017] The first natural language and the various candidate knowledge are input into the AI ​​network big model to obtain the target network construction scheme.

[0018] In some embodiments, the method further includes:

[0019] Based on human-computer interaction, the second natural language of the display method of the network construction plan is collected and indicated;

[0020] The second natural language is input into the AI ​​network model to obtain the target display method;

[0021] An intelligent agent calls the interface corresponding to the target display method to display the target network construction scheme.

[0022] In some embodiments, the step of collecting the second natural language indicating the display method of the network construction scheme based on human-computer interaction includes:

[0023] Display the human-computer interaction dialog box;

[0024] The system receives second natural language input from the user in the human-computer interaction dialog box, indicating how the network construction plan should be displayed.

[0025] In some embodiments, the method further includes:

[0026] A second prompt message is displayed on the human-computer interaction dialog box. The second prompt message is used to guide the input of the display method of the network construction plan.

[0027] In some embodiments, the method further includes:

[0028] The target network construction plan displayed is adjusted based on user actions.

[0029] In some embodiments, the AI ​​network big model outputs multiple target network construction schemes;

[0030] The method further includes:

[0031] From the multiple target network construction schemes, one target network construction scheme is determined;

[0032] The step of configuring the network device according to the network configuration information after detecting that the network device corresponding to the network topology information has been powered on includes:

[0033] After the network device corresponding to the network topology information included in the determined target network construction scheme is powered on, the network device is configured according to the network configuration information included in the determined target network construction scheme.

[0034] In some embodiments, the step of determining a target network construction scheme from the plurality of target network construction schemes includes:

[0035] Based on human-computer interaction, the third natural language of the target network construction plan determined by the instructions is collected;

[0036] The third natural language is input into the AI ​​network model to obtain the determined target network construction scheme.

[0037] In some embodiments, the step of collecting the third natural language of the target network construction scheme determined based on human-computer interaction includes:

[0038] Display the human-computer interaction dialog box;

[0039] The system receives instructions from the user on the human-computer interaction dialog box, specifying the target network construction plan in third natural language.

[0040] In some embodiments, the method further includes:

[0041] A third prompt message is displayed on the human-computer interaction dialog box. The third prompt message is used to guide the input of the determined target network construction plan.

[0042] In some embodiments, the step of configuring the network device according to the network configuration information includes:

[0043] Based on human-computer interaction, collect and instruct network device configuration using the fourth natural language.

[0044] The fourth natural language is input into the AI ​​network model to obtain network device configuration instructions;

[0045] An intelligent agent invokes the interface corresponding to the network device configuration command to configure the network device according to the network configuration information.

[0046] In some embodiments, the step of collecting the fourth natural language indicating the configuration of the network device based on a human-computer interaction method includes:

[0047] Display the human-computer interaction dialog box;

[0048] The system receives instructions from the user in the human-computer interaction dialog box to configure the network device in a fourth natural language.

[0049] In some embodiments, the method further includes:

[0050] A fourth prompt message is displayed on the human-computer interaction dialog box. The fourth prompt message is used to guide the input of execution information for configuring network devices.

[0051] In some embodiments, the method further includes:

[0052] Determine the operation to be performed;

[0053] The prompt information associated with the operation to be performed is displayed on the human-computer interaction dialog box.

[0054] In some embodiments, the method further includes:

[0055] Obtain the sample natural language and the corresponding sample network planning and construction scheme;

[0056] The AI ​​network model is fine-tuned using the sample natural language and the sample network planning and construction scheme.

[0057] Secondly, embodiments of this application provide a network planning and construction apparatus, the apparatus comprising:

[0058] The data acquisition module is used to collect first natural language that indicates the intention of network planning and construction, based on human-computer interaction.

[0059] The first determining module is used to input the first natural language into the AI ​​network big model to obtain a target network construction scheme, wherein the target network construction scheme includes network topology information and network configuration information.

[0060] The configuration module is used to configure the network device according to the network configuration information after the network device corresponding to the network topology information is detected to be powered on.

[0061] In some embodiments, the acquisition module is specifically used for:

[0062] Display the human-computer interaction dialog box;

[0063] Receive the user's first natural language input on the human-computer interaction dialog box indicating the network planning and construction intention.

[0064] In some embodiments, the acquisition module is further configured to:

[0065] The first prompt message is displayed on the human-computer interaction dialog box. The first prompt message is used to guide the input of network planning and construction intentions.

[0066] In some embodiments, the first determining module is specifically used for:

[0067] The first natural language is matched with pre-stored knowledge of various network construction schemes to obtain multiple candidate knowledge;

[0068] The first natural language and the various candidate knowledge are input into the AI ​​network big model to obtain the target network construction scheme.

[0069] In some embodiments, the apparatus further includes: a display module, configured to:

[0070] Based on human-computer interaction, the second natural language of the display method of the network construction plan is collected and indicated;

[0071] The second natural language is input into the AI ​​network model to obtain the target display method;

[0072] An intelligent agent calls the interface corresponding to the target display method to display the target network construction scheme.

[0073] In some embodiments, the display module is specifically used for:

[0074] Display the human-computer interaction dialog box;

[0075] The system receives second natural language input from the user in the human-computer interaction dialog box, indicating how the network construction plan should be displayed.

[0076] In some embodiments, the display module is further configured to:

[0077] A second prompt message is displayed on the human-computer interaction dialog box. The second prompt message is used to guide the input of the display method of the network construction plan.

[0078] In some embodiments, the apparatus further includes: an adjustment module, configured to:

[0079] The target network construction plan displayed is adjusted based on user actions.

[0080] In some embodiments, the AI ​​network big model outputs multiple target network construction schemes;

[0081] The device further includes: a second determining module, used for:

[0082] From the multiple target network construction schemes, one target network construction scheme is determined;

[0083] The configuration module is specifically used for:

[0084] After the network device corresponding to the network topology information included in the determined target network construction scheme is powered on, the network device is configured according to the network configuration information included in the determined target network construction scheme.

[0085] In some embodiments, the second determining module is specifically used for:

[0086] Based on human-computer interaction, the third natural language of the target network construction plan determined by the instructions is collected;

[0087] The third natural language is input into the AI ​​network model to obtain the determined target network construction scheme.

[0088] In some embodiments, the second determining module is specifically used for:

[0089] Display the human-computer interaction dialog box;

[0090] The system receives instructions from the user on the human-computer interaction dialog box, specifying the target network construction plan in third natural language.

[0091] In some embodiments, the second determining module is further configured to:

[0092] A third prompt message is displayed on the human-computer interaction dialog box. The third prompt message is used to guide the input of the determined target network construction plan.

[0093] In some embodiments, the configuration module is specifically used for:

[0094] Based on human-computer interaction, collect and instruct network device configuration using the fourth natural language.

[0095] The fourth natural language is input into the AI ​​network model to obtain network device configuration instructions.

[0096] An intelligent agent invokes the interface corresponding to the network device configuration command to configure the network device according to the network configuration information.

[0097] In some embodiments, the configuration module is specifically used for:

[0098] Display the human-computer interaction dialog box;

[0099] The system receives instructions from the user in the human-computer interaction dialog box to configure the network device in a fourth natural language.

[0100] In some embodiments, the configuration module is further configured to:

[0101] A fourth prompt message is displayed on the human-computer interaction dialog box. The fourth prompt message is used to guide the input of execution information for configuring network devices.

[0102] In some embodiments, the device further includes: a display module, configured to:

[0103] Determine the operation to be performed;

[0104] The prompt information associated with the operation to be performed is displayed on the human-computer interaction dialog box.

[0105] In some embodiments, the apparatus further includes: a fine-tuning module, configured to:

[0106] Obtain the sample natural language and the corresponding sample network planning and construction scheme;

[0107] The AI ​​network model is fine-tuned using the sample natural language and the sample network planning and construction scheme.

[0108] Thirdly, embodiments of this application provide an electronic device, including a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus;

[0109] The memory is used to store computer programs;

[0110] When the processor executes the program stored in the memory, it implements any of the methods described in the first aspect above.

[0111] Fourthly, embodiments of this application provide a computer-readable storage medium storing a computer program, which, when executed by a processor, implements any of the methods described in the first aspect above.

[0112] In another embodiment provided in this application, a computer program product containing instructions is also provided, which, when run on a computer, causes the computer to perform any of the methods described in the first aspect above.

[0113] Beneficial effects of the embodiments in this application:

[0114] In the technical solution provided in this application embodiment, natural language is used to express user intent, namely, network planning and construction intent. Natural language is a language that users can easily understand and does not require a professional knowledge background. Therefore, using natural language to express user intent and realize network planning and construction increases the user's ability to understand business-level intents. Furthermore, in the technical solution provided in this application embodiment, an AI network big data model is used to analyze and process the natural language expressing user intent (i.e., the first natural language) to obtain a target network construction scheme that meets the network planning and construction intent. Because the AI ​​network big data model has strong natural language processing capabilities, even if new or changed intents appear, the AI ​​network big data model can analyze and obtain the required target network construction scheme, improving the scalability of intent collection and the adaptability to scenarios and network topologies. Attached Figure Description

[0115] The accompanying drawings, which are provided to further illustrate this application and form part of this application, illustrate exemplary embodiments of this application and are used to explain this application, but do not constitute an undue limitation of this application.

[0116] Figure 1 is a schematic diagram of a hierarchical generational definition of a data center self-intelligent network;

[0117] Figure 2 is a schematic diagram of the five general task phases of information and communication network operation and management activities;

[0118] Figure 3 is a schematic diagram of the interface during the intent acquisition stage;

[0119] Figure 4 is a schematic diagram of the interface during the scheme generation stage;

[0120] Figure 5 is a schematic diagram of the interface during the simulation phase of the scheme.

[0121] Figure 6 is a flowchart illustrating a network planning and construction method provided in an embodiment of this application.

[0122] Figure 7 is a detailed schematic diagram of step S62 provided in an embodiment of this application;

[0123] Figure 8 is a schematic diagram of a knowledge base establishment process provided in an embodiment of this application;

[0124] Figure 9 is a flowchart illustrating a target network construction scheme based on an AI network large model provided in an embodiment of this application.

[0125] Figure 10 is a schematic diagram of the intent analysis stage processing logic provided in an embodiment of this application;

[0126] Figure 11 is a flowchart illustrating a network construction scheme provided in an embodiment of this application;

[0127] Figure 12 is a schematic diagram of a graphical representation using an intelligent agent framework provided in an embodiment of this application;

[0128] Figure 13 is a schematic diagram of a first process for configuring a network device according to an embodiment of this application;

[0129] Figure 14 is a detailed schematic diagram of step S131 provided in an embodiment of this application;

[0130] Figure 15 is a schematic diagram of a second process for configuring a network device according to an embodiment of this application;

[0131] Figure 16 is a schematic diagram of a network planning and construction method framework provided in an embodiment of this application;

[0132] Figures 17a to 17c are schematic diagrams of a first type of SDN controller interface provided in the embodiments of this application;

[0133] Figures 18a and 18b are second schematic diagrams of the SDN controller interface provided in the embodiments of this application;

[0134] Figures 19a to 19e are third schematic diagrams of the SDN controller interface provided in the embodiments of this application;

[0135] Figures 20a and 20b are the fourth schematic diagrams of the SDN controller interface provided in the embodiments of this application;

[0136] Figure 21 is a schematic diagram of a network planning and construction device provided in an embodiment of this application;

[0137] Figure 22 is a schematic diagram of an electronic device provided in an embodiment of this application. Detailed Implementation

[0138] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided with reference to the accompanying drawings and embodiments. Obviously, the described embodiments are merely some embodiments of this application, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments in this application are within the scope of protection of this application.

[0139] For ease of understanding, the terms appearing in the embodiments of this application are explained below.

[0140] Self-Intelligent Network: It aims to build automated and intelligent operation and maintenance capabilities for the entire network lifecycle, providing consumers and vertical industry customers with new network and information communication technology (ICT) services with "zero waiting time, zero failure, and zero contact", and creating digital and intelligent operation and maintenance capabilities of "self-configuration, self-repair, and self-optimization" for intelligent network operation and maintenance.

[0141] Large-scale AI network models are machine learning models with extremely large parameters (typically over a billion) and massive computing resources. They are capable of processing massive amounts of data and completing various complex tasks, such as natural language processing and image recognition. Large-scale AI network models can also be called large-scale generative artificial intelligence (AIGC) network models.

[0142] Currently, DCNs are gradually evolving into self-intelligent networks, forming data center self-intelligent networks, which will be referred to as such from now on. The goal of data center self-intelligent networks is to gradually reduce and eventually eliminate manual operation of the network, guiding customers through the definition of hierarchical capabilities, and gradually evolving towards the vision of unattended DCN self-intelligent networks. Figure 1 shows the hierarchical generational definition of data center self-intelligent networks.

[0143] Level 1 (L1) Manual Processing: The data center's intelligent network relies primarily on experience and is entirely dependent on manual processing.

[0144] Level 2 (L2) tool collaboration: Data center self-intelligent networks use domain tools to help people improve efficiency within their domain.

[0145] Level 3 (L3) network automation: The data center's self-intelligent network enables automated processing of network intent, with some systems providing auxiliary analysis and human decision-making.

[0146] Level 4 (L4) network intelligence: The data center's intelligent network is integrated with applications, and technologies such as AI and machine learning (ML) are used to achieve intelligent processing of application intent, system-assisted analysis, and human decision-making.

[0147] Level 5 (L5) Fully Autonomous Intelligent Network: The data center autonomous intelligent network realizes a fully lights-out data center, where the system performs analysis and decision-making, achieving automated end-to-end autonomy.

[0148] The achievement of the goal of data center self-intelligent networks is inseparable from the development of network automation and intelligence technologies. Data center self-intelligent network classification can be divided into two dimensions.

[0149] Dimension 1: DCN full lifecycle management services, including planning and construction, network operation, monitoring and maintenance, and operation optimization.

[0150] Dimension Two: Information and Communication Network Operation and Management Activities, including intent management, perception, analysis, decision-making, and execution.

[0151] Data center self-intelligent networks require the execution of information and communication network operation and management activities at every management service stage throughout their entire lifecycle. As shown in Figure 2, the five general task stages of information and communication network operation and management activities are intent management, perception, analysis, decision-making, and execution, as mentioned in Dimension Two above.

[0152] Intent Management Phase: Understanding the customer's business and management / operations intent and translating it into specific network configurations and policies. Intent management can support both low-level traditional functions, such as orchestrating any necessary configuration operations on the network, and higher-level abstract information understanding capabilities, such as supporting open, modifiable capabilities, allowing customers to adjust the solution according to their actual network topology. Intent management also includes a complete record of the business intent operation process, ensuring traceability and queryability.

[0153] Perception Phase: Real-time monitoring and observation of the DCN to detect anomalies in network services or multi-dimensional issues such as Service-Level Agreement (SLA) problems, triggering network analysis and localization. This includes collecting raw network data and performing necessary preprocessing (e.g., data cleaning, enhancement, statistics) to monitor and perceive network information (including network performance, network anomalies, network events, etc.) and achieve visualized presentation.

[0154] Analysis phase: Analyze the current state of DCN, and based on historical data, combine it with the customer's intent to perform network analysis, and generate options and suggestions for operational actions and execution strategies that can meet the customer's intent.

[0155] Decision-making phase: Based on the options and suggestions provided in the analysis phase, determine the most suitable, feasible network operations and strategies that meet the client's intentions and needs.

[0156] Execution phase: Generate executable network operations and policies for customers who have made decisions, automatically implement and deploy them to the production network (DCN) in the data center, and also include business verification of the intent after network implementation.

[0157] In Figure 2, the enterprise system / customer inputs intents. The intelligent network manages these intents, understands them, and performs perception, analysis, decision-making, and execution. It operates on the managed objects and provides feedback to the enterprise system / customer. The managed objects of the DCN are its various applications, devices, and network endpoint devices.

[0158] Currently, the planning and construction phase of the DCN full lifecycle management service (i.e., Dimension 1) is as follows: Through the interface entry point fixed by the Software Defined Network (SDN) controller, based on the network language of the user-input diagram and the pre-configured network construction plan, the target network construction plan is determined, and network devices are configured according to this plan. The network planning and construction process consists of four stages: intent acquisition, plan generation, plan simulation, and plan implementation, as detailed below.

[0159] 1) Intent Acquisition Phase

[0160] Electronic devices (i.e., devices used for network planning and construction) collect user input in network language (i.e., user requirements for network construction). This network language is a language expressing the network architecture, arranged according to business resources, such as the number of spine nodes, leaf nodes, and boundary leaf nodes, as well as the network topology. During this stage, users can input their intended content according to the prompts displayed on the electronic device and the options provided within the display window.

[0161] As shown in Figure 3, the electronic device displays the number of spine nodes, leaf nodes, and border leaf nodes, as well as prompts indicating the network topology. Based on the prompts, the user enters the corresponding quantities in the boxes below the number of spine nodes, leaf nodes, and border leaf nodes, and clicks the right arrow button in the box below the network topology to select the network topology, i.e., whether the network topology is a single machine or a multichassis link aggregation group (M-LAG).

[0162] After completing the network architecture construction and inputting the intent content (i.e., network language) in the above manner, the next step of scheme generation can be carried out.

[0163] 2) Solution Generation Stage

[0164] The electronic device automatically converts the user's input network construction intent (i.e., intent content) into a pre-configured network construction scheme and presents it to the user. That is, it automatically generates network topology (such as interconnection devices, link information, etc.) and network device roles; automatically allocates device management IP, Virtual Extensible Local Area Network Tunnel Endpoints (VTEP) IP, and other information from the Internet Protocol (IP) address pool; and automatically generates network device interconnection routing configurations, etc., and finally obtains the network configuration of the network devices (i.e., network construction scheme). As shown in the scheme generation stage diagram in Figure 4, the topology of the network construction scheme recommended by the electronic device to the user includes leaf nodes 1 to 10 and spine nodes 1 to 2.

[0165] 3) Simulation Phase

[0166] Before the network configuration is sent to the production network for network device configuration, electronic devices can simulate network connectivity and configuration consistency in advance based on the simulated network, and simulate whether there are routing loops, routing black holes and other problems.

[0167] Figure 5 shows the simulation phase interface. The simulation phase interface displays the simulation evaluation results in the lower right corner, including configuration generation, simulation environment construction, and simulation network verification. Configuration generation includes basic configuration generation and Border Gateway Protocol (BGP) configuration generation, while simulation network verification includes connectivity and configuration verification. Users can click the "View" button to see the detailed simulation evaluation results for each aspect.

[0168] In Figure 5, the current network construction plan does not meet the connectivity requirement, i.e. the simulation evaluation fails. The electronic equipment can correct the network construction plan until the simulation evaluation is successful.

[0169] After the simulation evaluation is successful, the user can click the "Continue" button in the upper right corner, and the electronic device will respond to the user's click of the "Continue" button to enter the implementation phase of the solution.

[0170] 4) Implementation phase

[0171] Network cabling is performed based on the device and link information output from the network construction plan. After cabling is completed, users can click the "Start Implementation" button to deploy and implement the plan, thereby automating the process of bringing devices online (i.e., configuring network devices) and managing network devices.

[0172] This network planning and construction method uses highly technical network language as input, such as the number of spine nodes, leaf nodes, and boundary leaf nodes, and whether the network topology is standalone or M-LAG. This network language is difficult for users to understand and requires professional network administrators. Furthermore, electronic devices lack the ability to understand the user's truly relevant business-level intentions, such as the number of computing resources needing to be connected to the network, network reliability requirements, and specific bandwidth requirements. In other words, the existing network planning and construction method is poorly user-friendly.

[0173] Furthermore, existing network planning and construction methods can only collect and understand fixed intents. When new or changed intents emerge, electronic devices will be unable to understand them and will be unable to configure network devices. For example, as the network scale increases, it is necessary to support an additional layer of access layer network device roles (i.e., access nodes). However, the current network collection and construction only supports spine nodes, leaf nodes, and boundary leaf nodes, and lacks the ability to automatically extend network intents.

[0174] Furthermore, electronic devices can only be configured with network equipment based on fixed intentions and pre-configured network construction schemes. However, these pre-configured network construction schemes are limited and cannot adapt to new scenarios or networking requirements. In other words, the ability to automatically generate network construction schemes is limited to automatically converting fixed network intentions (such as the number of network devices and the network configuration provided to the user) into network construction schemes. For new network configurations and requirements, new code needs to be developed to implement the adaptation and conversion logic.

[0175] To enhance the understanding of user business-level intents, improve the scalability of intent collection, and increase adaptability to different scenarios and network topologies, this application provides a network planning and construction method applicable to electronic devices such as computers, servers, and SDN controllers. For ease of description, the following description will use electronic devices as the execution subject and will not be considered limiting.

[0176] Referring to Figure 6, which is a schematic flowchart of a network planning and construction method provided in an embodiment of this application, the method includes the following steps.

[0177] Step S61: Based on human-computer interaction, collect the first natural language indicating the intention of network planning and construction.

[0178] Step S62: Input the first natural language into the AI ​​network big model to obtain the target network construction plan, which includes network topology information and network configuration information.

[0179] Step S63: After the network device corresponding to the network topology information is detected to be powered on, configure the network device according to the network configuration information.

[0180] In the technical solution provided in this application embodiment, natural language is used to express user intent, namely, network planning and construction intent. Natural language is a language that users can easily understand and does not require a professional knowledge background. Therefore, using natural language to express user intent and realize network planning and construction increases the user's ability to understand business-level intents. Furthermore, in the technical solution provided in this application embodiment, an AI network big data model is used to analyze and process the natural language expressing user intent (i.e., the first natural language) to obtain a target network construction scheme that meets the network planning and construction intent. Because the AI ​​network big data model has strong natural language processing capabilities, even if new or changed intents appear, the AI ​​network big data model can analyze and obtain the required target network construction scheme, improving the scalability of intent collection and the adaptability to scenarios and network topologies.

[0181] The network planning and construction method provided in this application embodiment may include four stages: the intent perception stage corresponding to step S61, the intent analysis stage and the intent decision stage corresponding to step S62, and the intent execution stage corresponding to step S63.

[0182] In step S61 above, the first natural language is any natural language used to indicate the intention of network planning and construction. The first natural language can be text or speech, etc., and there is no limitation on the form of the first natural language.

[0183] Electronic devices can provide users with a variety of human-computer interaction methods. Users can input their first natural language into the electronic device through various human-computer interaction methods, and then the electronic device can collect the first natural language input by the user based on these various human-computer interaction methods.

[0184] In one example, if the electronic device provides a human-computer interaction dialog box, then step S61 above can be: displaying the human-computer interaction dialog box; receiving the first natural language input by the user on the human-computer interaction dialog box indicating the intention of network planning and construction.

[0185] For example, an electronic device displays a human-computer interaction dialog box. The user enters text in the dialog box, and the electronic device detects the text and treats it as the first natural language.

[0186] For example, an electronic device displays a human-computer interaction dialog box. The user inputs voice on the dialog box, and the electronic device can receive the user's voice input through a voice receiver (such as a microphone) and use the voice directly as the first natural language; or, the voice is recognized, converted into text, and then used as the first natural language.

[0187] Human-computer interaction dialog boxes can intuitively display the interaction between users and electronic devices, improving the user-friendliness of network planning and construction.

[0188] When using a human-computer interaction dialog box for human-computer interaction, the electronic device can also display prompts (i.e., first prompts) on the dialog box to guide the input of network planning and construction intentions. The first prompts may include the specific content that the user needs to input into the dialog box, such as "the number of physical servers to be connected" or "uplink / downlink bandwidth convergence ratio." The first prompts may also include requirements for the user's input, such as "please use natural language to input your network planning and construction needs." The content of the first prompts is not limited here. By displaying the first prompts on the human-computer interaction dialog box, the electronic device can guide the user to input their network planning and construction intentions, more accurately meeting user needs and facilitating user operation.

[0189] In another example, the electronic device can also receive natural language input by the user on other devices and use that natural language as the first natural language. For instance, if the electronic device is a personal computer (PC) and the other device is a mobile phone, the user can input natural language on the mobile phone, which then sends the natural language to the PC. The PC then uses the received natural language as the first natural language. No restrictions are placed on the human-computer interaction method between the electronic device and the user.

[0190] In step S62 above, the AI ​​network big model can be a general AI network big model, or it can be an AI network big model pre-trained based on a general AI network big model and combined with a large amount of DCN corpus; there is no limitation on this. For the AI ​​network big model, the training process is as follows: input the DCN corpus into the AI ​​network big model; after the AI ​​network big model obtains the DCN corpus, it will learn the contextual knowledge in the DCN corpus.

[0191] In this embodiment of the application, the DCN corpus may include: DCN networking information, topology information, device information, link information, port information, security information, routing information, network policy information, reliability information, consistency information, stability information, performance load information, capacity information, system information, and resource information, etc. The DCN corpus is not limited here.

[0192] Network topology information refers to the topological structure between network devices in the planned network construction. Network topology information may include network architecture information, interconnection device information, device role information, link information, and port information, etc. Network configuration information is the information required to configure network devices. Network configuration information may include network device information, IP and VTEP IP resource pools, IP addresses of each network device and VTEP IP addresses, and routing information, etc. Network device information may include the type, model, and quantity of network devices, etc. The content included in network topology information and network configuration information is not limited here.

[0193] In this embodiment of the application, the network device can be a switch or a router, etc., and the network device is not limited here.

[0194] The electronic device inputs the collected first natural language into the AI ​​network big model, so that the AI ​​network big model outputs a network construction plan that meets the user's network planning and construction intentions based on the first natural language. The electronic device uses the network construction plan output by the AI ​​network big model as the target network construction plan.

[0195] In this embodiment, the electronic device can call the interface (i.e., the first interface) corresponding to the AI ​​network big model to input the first natural language into the AI ​​network big model. The interface can be an application programming interface (API), so the first interface is the API corresponding to the AI ​​network big model. Here, the method by which the electronic device inputs the first natural language into the AI ​​network big model is not limited.

[0196] In step S63 above, the electronic device can detect in real time whether the network devices in the planned network (i.e., the network devices corresponding to the network topology information) are powered on. If a network device is detected to be powered on, the electronic device can configure an IP address from the IP resource pool, a VTEP IP address from the VTEP IP resource pool, and configure routing information for the powered-on network device, etc. The configuration content is not limited here. If no network device is detected to be powered on, the electronic device continues to detect until a network device is detected to be powered on.

[0197] In this embodiment, the electronic device can configure all network devices together after detecting that all network devices are powered on. Alternatively, the electronic device can configure any network device directly after detecting that any network device is powered on, without waiting for all network devices to be powered on; this is not limited.

[0198] In this embodiment, after the electronic device detects that the network device is powered on, it can use an AI network big data model to check whether the network device information, network architecture information, interconnected device information, device role information, link information, and port information are correct, based on the network topology and network configuration information in the target network construction plan. For example, the electronic device can use the AI ​​network big data model to check whether the type, model, location in the network architecture, and other network devices connected to the powered-on network device are correct. If any of the above information is found to be incorrect, it indicates that the network configuration personnel have made an error in the process of network cabling based on the network device information, network architecture information, interconnected device information, device role information, link information, and port information (i.e., the process of powering on the network device). The electronic device can then provide feedback to the user, allowing the user to adjust the cabled network until all network devices are detected to be correctly powered on.

[0199] In some embodiments, referring to FIG7, FIG7 is a detailed schematic diagram of step S62 provided in the embodiments of this application. Step S62 may include the following steps.

[0200] Step S71: Match the first natural language with the knowledge of multiple pre-stored network construction schemes to obtain multiple candidate knowledge.

[0201] Step S72: Input the first natural language and multiple candidate knowledge into the AI ​​network large model to obtain the target network construction scheme.

[0202] In the technical solution provided in this application embodiment, the electronic device pre-establishes a solution knowledge base including knowledge of various network construction schemes. The solution knowledge base with rich network experience serves as an effective supplement to the AI ​​network big model in recommending network construction schemes in planning and construction scenarios. This allows the AI ​​network big model to recommend target network construction schemes based on candidate knowledge selected from the solution knowledge base and network planning and construction intentions, thereby improving the accuracy of the recommended network construction schemes.

[0203] In step S71 above, the knowledge of the network construction scheme is related to the network planning and construction scenario. It can be the text information of the actual network construction scheme or the text information on which the network construction scheme is based, and there is no limitation on this.

[0204] In this embodiment, the electronic device can establish a solution knowledge base in the following way: It acquires experience and guidance documents related to network planning and construction scenarios as raw knowledge; it divides the raw knowledge into multiple text fragments; it vectorizes each text fragment; and it stores each vectorized text fragment as knowledge of a network commissioning solution in a database, thus obtaining knowledge of various network construction solutions and forming a solution knowledge base, as illustrated in Figure 8, which is a schematic diagram of the solution knowledge base establishment process. The raw knowledge may include solution scenarios, business requirements, design principles, design experience, implementation cases, solution value, etc. The content of the raw knowledge and the method by which the electronic device establishes the solution knowledge base are not limited here.

[0205] In this embodiment of the application, the vectorization process can be as follows: based on the pre-stored correspondence between characters and feature values, determine the feature value corresponding to each character in the text segment, and concatenate the feature values ​​corresponding to the characters in character order to obtain the vector corresponding to the text segment, thus realizing the vectorization process of the text segment.

[0206] Another method of vectorization is to preprocess the text segments, removing invalid characters, conjunctions, and other useless strings, such as removing commas, spaces, and words like "and" and "but". Based on the pre-stored correspondence between characters and feature values, determine the feature value corresponding to each character in the preprocessed text segment, and concatenate the feature values ​​corresponding to the characters in character order to obtain the vector corresponding to the text segment, thus realizing the vectorization of the text segment.

[0207] In this embodiment of the application, vectorization processing can also be implemented in other ways, such as converting text segments into vectors represented by 0 and 1.

[0208] After acquiring the first natural language, the electronic device performs vectorization processing on it. Here, the electronic device can treat the first natural language as a text segment and use the aforementioned text segmentation vectorization processing method to convert the first natural language into a vector represented by 0s and 1s, thus obtaining the vectorized first natural language.

[0209] Electronic devices can use similarity algorithms such as cosine similarity and Euclidean distance to perform similarity matching between the vectorized first natural language and the knowledge of various network construction schemes in the scheme knowledge base. Based on the matching results, the first preset number of vectors that are most similar to the first natural language are obtained. In other words, the knowledge of various network construction schemes that are most similar to the network planning and construction intention is obtained. This knowledge of various network construction schemes is the multiple candidate knowledge. Here, the value of the preset number is not limited.

[0210] In this embodiment of the application, corresponding to the network construction scheme knowledge in the scheme knowledge base, the various candidate knowledge obtained by the electronic device can be the text information of the actual network construction scheme, or the text information on which the network construction scheme is based, and there is no limitation on this.

[0211] In this embodiment, the electronic device can continuously update the solution knowledge base. Specifically, the electronic device can acquire new experience and guidance documents (i.e., original knowledge) related to new network planning and construction scenarios, segment and vectorize the new original knowledge, and store it in the database, thus updating the solution knowledge base. Furthermore, the electronic device can use the updated solution knowledge base for matching to obtain richer and more accurate candidate knowledge, without limitation.

[0212] In step S72 above, the electronic device uses the obtained candidate knowledge as context and adds it to the prompt template along with the first natural language. The added prompt template is then input into the AI ​​network big model, so that the AI ​​network big model performs contextual reasoning such as information combination and elimination based on the network planning and construction intention, obtains and outputs a recommended network construction scheme, and the electronic device uses the network construction scheme output by the AI ​​network big model as the target network construction scheme.

[0213] In this embodiment of the application, the electronic device can call the request interface to input the first natural language and multiple candidate knowledge (i.e., the added prompt template) into the AI ​​network big model, without limitation.

[0214] Figure 9 shows the complete process by which an electronic device obtains a target network construction scheme based on an AI network big data model. The electronic device collects the user's input network planning and construction intent (i.e., the first natural language), vectorizes this intent, and then retrieves information from a scheme knowledge base. Specifically, it performs similarity matching between the vectorized network planning and construction intent and vectorized text segments in the scheme knowledge base, obtaining the top N (a preset number) network construction schemes with the highest similarity to the intent (i.e., candidate knowledge). The electronic device then calls a request API (i.e., a request interface), inputting the obtained top N scheme knowledge and the network planning and construction intent into the AI ​​network big data model. The AI ​​network big data model then performs contextual reasoning based on the top N most similar scheme knowledge and the network planning and construction intent, ultimately outputting a recommended scheme, i.e., the target network construction scheme.

[0215] In this embodiment, if the first natural language fails to match the knowledge of multiple network construction schemes (i.e. no candidate knowledge is obtained), it indicates that the similarity between the knowledge of multiple network construction schemes and the first natural language is too low. The electronic device can directly input the first natural language (i.e., the network planning and construction intention input by the user) into the AI ​​network big model. The AI ​​network big model only performs contextual reasoning based on the first natural language and pre-learned knowledge to obtain the target network construction scheme. This further improves the ability of the network planning and construction method provided in this embodiment to automatically generate network construction schemes in different scenarios and improves the adaptability to scenarios and network topologies.

[0216] Here, the knowledge learned in advance can be used to input the AI ​​large language model into electronic devices during the training and pre-training phases.

[0217] In some embodiments, the electronic device can fine-tune a large AI network model. The electronic device acquires sample natural language and corresponding sample network planning and construction schemes; it then uses these to fine-tune the large AI network model. The sample natural language and corresponding sample network planning and construction schemes are samples of a pre-constructed network planning and construction scenario. Each sample is a labeled data point; the content of the sample is sample natural language, and the label of the sample is the actual network construction scheme corresponding to that sample natural language, i.e., the sample network planning and construction scheme.

[0218] Electronic devices can pre-construct a sample set (i.e., a labeled dataset) of network planning and construction scenarios by using samples from multiple network planning and construction scenarios (i.e., multiple samples of natural language and corresponding sample network planning and construction schemes). This sample set actually consists of multiple question-answer pairs, where the questions are sample natural language and the answers are sample network planning and construction schemes. The electronic device feeds the sample set into the AI ​​network model, enabling the AI ​​network model to learn from these question-answer pairs and fine-tune the AI ​​network model.

[0219] In this embodiment of the application, the electronic device can continuously update the sample set of the network planning and construction scenario. That is, the electronic device can continuously acquire new sample natural language and corresponding sample network planning and construction schemes to fine-tune the AI ​​network model.

[0220] Electronic devices can also continuously update DCN corpora. By utilizing the updated DCN corpora, rich network planning and construction scenarios, and continuously updated solution knowledge base, the AI ​​network big model can be trained and updated. Here, the method of training and updating the AI ​​network big model is not limited.

[0221] By applying the technical solution provided in this application, the AI ​​network model is fine-tuned using a labeled dataset from a network planning and construction scenario. This makes the AI ​​network model more adaptable to the network planning and construction scenario, enabling the electronic device to self-evolve and evolve when configuring network devices in such a scenario. Furthermore, the electronic device can also use a continuously updated solution knowledge base, new DCN corpus, and rich network planning and construction scenarios to train and update the AI ​​network model, achieving self-evolution and evolution of the network construction solution recommendation and generation capabilities under the network planning and construction scenario, further enhancing the electronic device's ability to self-evolve and evolve.

[0222] Figure 10 illustrates a schematic diagram of the intent analysis phase processing logic, taking network planning and construction scenario intent management using electronic devices as an example. In a data center intelligent network, the network planning and construction scenario intent management system receives network planning and construction intent (i.e., first natural language) input by the client (i.e., user) and performs intent perception, intent analysis, intent decision-making, and intent execution. In the intent analysis phase, the network planning and construction scenario intent management system analyzes and extracts features (i.e., vectorizes) the network planning and construction intent, obtaining network planning and construction intent features (i.e., vectorized first natural language), and inputs these features into a large network model (such as an AI network model).

[0223] In this embodiment, when deploying the AI ​​network big model, a solution knowledge base and an AI agent framework can also be deployed to assist the AI ​​network big model in its operation. The network planning and construction scenario intent management system outputs a network construction plan (i.e., the target network construction plan) based on the solution knowledge base, the AI ​​agent, and the AI ​​network big model.

[0224] The solution knowledge base assists the AI ​​network model in performing contextual reasoning with limited but highly relevant knowledge to arrive at the target network construction solution. The intelligent agent framework can leverage the toolset to call the rich APIs of the SDN controller, enabling graphical representation of network planning and construction scenarios, as well as other capabilities. The intelligent agent framework can continuously enrich the toolset for interaction between network planning and construction scenarios and the SDN controller, enhancing the intent execution capabilities of network planning and construction scenarios.

[0225] As shown in Figure 10, the training method for the large model can be as follows: the general large model is trained using the DCN corpus, and then the trained large model is fine-tuned by combining it with the labeled dataset of the network planning and construction scenario, so as to finally obtain a reliable AI network large model and release it.

[0226] In some embodiments, electronic devices can demonstrate network construction schemes through human-computer interaction. Referring to Figure 11, which is a flowchart illustrating a network construction scheme according to an embodiment of this application, the process may include the following steps.

[0227] Step S111: Based on the human-computer interaction method, collect the second natural language of the display method of the network construction plan.

[0228] Step S112: Input the second natural language into the AI ​​network model to obtain the target display method.

[0229] Step S113: The intelligent agent calls the interface corresponding to the target display method to display the target network construction scheme.

[0230] In the technical solution provided in this application embodiment, based on the strong natural language processing capability of the AI ​​network big model, the AI ​​network big model is used to analyze and process the natural language (i.e., the second natural language) expressing the display method of user needs to obtain the target display method, which improves interactivity. Furthermore, an intelligent agent is used to call the corresponding API to realize the display of the target network construction scheme in the target display method.

[0231] In step S111 above, the second natural language is any natural language used to indicate the display method of the network construction plan. The second natural language can be text or voice, etc. The display method of the network construction plan can include displaying it in the form of text and displaying it in the form of graphics, etc. There are no limitations on the form of the second natural language and the display method of the network construction plan.

[0232] Electronic devices can provide users with a variety of human-computer interaction methods. Users can input second natural language into the electronic device through various human-computer interaction methods, and then the electronic device can collect the second natural language input by the user based on the various human-computer interaction methods.

[0233] When the electronic device provides a human-computer interaction dialog box, step S111 above can be: displaying the human-computer interaction dialog box; receiving the second natural language input by the user on the human-computer interaction dialog box indicating the display method of the network construction scheme. In this case, the electronic device can also display prompt information (i.e., second prompt information) on the human-computer interaction dialog box to guide the input of the display method of the network construction scheme, such as "Please enter the display method," etc. The content of the second prompt information is not limited here. Through the human-computer interaction dialog box, the interaction between the user and the electronic device can be displayed intuitively. By displaying prompt information on the human-computer interaction dialog box, the user can be guided to input, thus meeting the user's needs and facilitating user operation.

[0234] In this embodiment of the application, the method by which the electronic device collects the second natural language based on human-computer interaction is similar to the method of collecting the first natural language described above. For details, please refer to the relevant description of step S61 above.

[0235] In step S112 above, the electronic device inputs the collected second natural language into the AI ​​network big model, so that the AI ​​network big model outputs a display method that meets the user's needs based on the second natural language, and the electronic device uses this display method as the target display method.

[0236] In this embodiment of the application, the electronic device can call the interface (i.e., the first interface) corresponding to the AI ​​network big model to input the second natural language into the AI ​​network big model, and there are no limitations on this.

[0237] In step S113 above, the electronic device uses an intelligent agent to call the interface corresponding to the target display method (i.e., the second interface) to display the target network construction scheme based on the target display method.

[0238] For example, if the target is displayed graphically (i.e., graphical display), the electronic device uses an intelligent agent to call the corresponding graphical display interface (i.e., API) to display the target network construction plan graphically, that is, to display the network topology information (such as network architecture information, link information, and network device port information) included in the target network construction plan graphically. If the target is displayed textually, the electronic device uses an intelligent agent to call the corresponding textual display interface to display the network topology information and network configuration information included in the target network construction plan in textual form.

[0239] Figure 12 shows a schematic diagram of a graphical display using an intelligent agent framework. The electronic device obtains a user's request for a graphical display of the network construction plan (i.e., the second natural language), inputs the request into the intelligent agent framework, and the AI ​​network big model performs reasoning and decision-making on the request, outputting a graphical display method. The intelligent agent selects a pre-set tool to call the visualization solution API (i.e., the second interface) and outputs a graphically displayed network construction plan.

[0240] In this embodiment, the electronic device can execute steps S111 to S113 after obtaining the target network construction scheme to display the obtained target network construction scheme. Alternatively, the electronic device can execute steps S111 to S112 before obtaining the target network construction scheme to obtain the target display method, and then execute step S113 after obtaining the target network construction scheme to display the obtained target network construction scheme. For example, when collecting the first natural language, the electronic device can simultaneously collect the second natural language, inputting both the first and second natural languages ​​into the AI ​​network large model to obtain the target network construction scheme and the target display method; the electronic device can also collect the second natural language after collecting the first natural language. The timing of the execution of steps S111 to S113 is not limited here.

[0241] In some embodiments, after performing step S113 to display the target network construction scheme, the electronic device can further adjust the displayed target network construction scheme based on user operations. User operations can include drag-and-drop and editing, and are not limited to any particular operation. The electronic device can provide users with multiple human-computer interaction methods. Users can operate the displayed target network construction scheme through these methods, and the electronic device can then collect user operations based on these methods and adjust the displayed target network construction scheme accordingly.

[0242] For example, when the target network construction scheme is displayed graphically, the electronic device can perform human-computer interaction based on the graphical interface, collecting drag-and-drop and editing operations performed by the user on the displayed target network construction scheme in the graphical interface, and adjusting the displayed target network construction scheme accordingly based on the collected operations. In this embodiment, the electronic device can also adjust the displayed target network construction scheme based on a human-computer interaction dialog box, and this is not limited.

[0243] The electronic device adjusts the displayed target network construction scheme based on user operations, making the adjusted target network construction scheme more in line with the user's intent and meet the user's expectations. Then, the electronic device can execute step S63 above, and after detecting that the network device corresponding to the network topology information included in the displayed target network construction scheme has been powered on, configure the network device according to the network configuration information included in the displayed target network construction scheme.

[0244] In some embodiments, in step S62 above, the AI ​​network big model can output multiple target network construction schemes. Referring to Figure 13, Figure 13 is a schematic diagram of the first process of configuring network devices provided in the embodiment of this application, which may include the following steps.

[0245] Step S131: Select one target network construction scheme from multiple target network construction schemes.

[0246] Step S132: After detecting that the network device corresponding to the network topology information included in the determined target network construction scheme is powered on, configure the network device according to the network configuration information included in the determined target network construction scheme.

[0247] In the technical solution provided in this application embodiment, when the AI ​​network big model outputs multiple target network construction schemes, the electronic device can determine the target network construction scheme that best matches the user's network planning and construction intention from the multiple target network construction schemes, and configure the network device according to the network configuration information included in the determined target network construction scheme, thereby further improving the accuracy of network construction scheme generation and recommendation.

[0248] In step S131 above, the electronic device determines a target network construction scheme from multiple target network construction schemes output by the AI ​​network large model. For example, the electronic device can receive information about the target network construction scheme indicated by the user input, and determine a target network construction scheme based on the user input information. Alternatively, the electronic device can randomly determine a target network construction scheme from multiple target network construction schemes; or, the electronic device can determine a target network construction scheme from multiple target network construction schemes according to preset rules. The preset rules can be set according to actual conditions, and the preset rules could be: selecting the network construction scheme with the best scalability. The method by which the electronic device determines a target network construction scheme is not limited here.

[0249] In step S132 above, the determined target network construction scheme is the target network construction scheme used to configure the network devices. The electronic device continuously monitors whether the network devices corresponding to the network topology information included in the determined target network construction scheme are powered on. After detecting that the network devices are powered on, it configures the powered-on network devices according to the network configuration information included in the determined target network construction scheme. For specific configuration methods, please refer to the relevant description in step S63 above.

[0250] In this embodiment, before executing step S131, the electronic device can display multiple target network construction schemes output by the AI ​​network large model. The specific display method can be found in the relevant description in Figure 11 above. After displaying the multiple target network construction schemes, the electronic device can also fine-tune the displayed schemes, determine one fine-tuned target network construction scheme from among the fine-tuned schemes, and configure the network device according to the determined fine-tuned target network construction scheme.

[0251] In some embodiments, the electronic device can determine a target network construction scheme based on a human-computer interaction method. Referring to Figure 14, which is a detailed schematic diagram of step S131 provided in an embodiment of this application, it may include the following steps.

[0252] Step S141: Based on human-computer interaction, collect the third natural language of the target network construction scheme determined by the instruction.

[0253] Step S142: Input the third natural language into the AI ​​network large model to obtain the determined target network construction scheme.

[0254] In the technical solution provided in this application embodiment, based on the strong natural language processing capability of the AI ​​network big model, the AI ​​network big model is used to analyze and process the natural language (i.e., third natural language) that expresses user needs and determines the target network construction scheme, thereby obtaining the determined target network construction scheme, improving interactivity, making the determined target network construction scheme more in line with user expectations, and improving user experience.

[0255] In step S141 above, the third natural language is any natural language used to indicate the determined target network construction scheme. The third natural language can be text or speech, etc., and there is no limitation on the form of the third natural language.

[0256] Electronic devices can provide users with a variety of human-computer interaction methods. Users can input third natural language into the electronic device through various human-computer interaction methods, and then the electronic device can collect the third natural language input by the user based on the various human-computer interaction methods.

[0257] When the electronic device provides a human-computer interaction dialog box, step S141 above can be: displaying the human-computer interaction dialog box; receiving the user's input of a third natural language indicating the determined target network construction scheme on the human-computer interaction dialog box. In this case, the electronic device can also display prompts (i.e., third prompts) on the human-computer interaction dialog box to guide the input of the determined target network construction scheme. For example, the third prompts could be "Please enter the determined network construction scheme" or "Please enter your preference for the overall requirements of the network construction scheme," etc. The content of the third prompts is not limited here. The human-computer interaction dialog box allows for a direct demonstration of the interaction between the user and the electronic device. By displaying prompts on the dialog box, the user can be guided to input relevant information about the determined target network construction scheme, thus meeting the user's needs and facilitating user operation.

[0258] In this embodiment of the application, the method by which the electronic device collects the third natural language based on human-computer interaction is similar to the method of collecting the first natural language described above. For details, please refer to the relevant description of step S61 above.

[0259] In step S142 above, the electronic device inputs the collected third natural language into the AI ​​network big model, so that the AI ​​network big model outputs a target network construction scheme based on the third natural language, and the electronic device uses the target network construction scheme as the determined target network construction scheme.

[0260] In this embodiment of the application, the electronic device can call the interface (i.e., the first interface) corresponding to the AI ​​network big model to input the third natural language into the AI ​​network big model, and there are no limitations on this.

[0261] In this embodiment of the application, depending on the timing of the execution of steps S141 and S142, there are two specific cases.

[0262] Scenario 1: After obtaining multiple target network construction schemes, the electronic device executes steps S141 and S142 to collect third natural language and obtain the determined target network construction scheme.

[0263] Scenario 2: Before obtaining multiple target network construction schemes, the electronic device executes step S141 to collect a third natural language. For example, the electronic device can collect the third natural language while collecting the first natural language. In this case, the electronic device can simultaneously execute steps S62 and S142, inputting both the first and third natural language into the AI ​​network model to obtain a determined target network construction scheme. That is, when the user's input network planning and construction intention has a tendency towards the overall requirements of the network construction scheme, the electronic device can input all the user's input, including both the first and third natural language, into the AI ​​network model, causing the AI ​​network model to output a target network construction scheme, thus obtaining a determined target network construction scheme. The timing of the execution of steps S141 and S142 is not limited here.

[0264] The electronic device can also receive user actions such as clicking and switching to select a target network construction scheme, and respond to the user's actions by determining one target network construction scheme from multiple options. The method by which the electronic device determines a target network construction scheme is not limited here.

[0265] In some embodiments, electronic devices can configure network devices based on human-computer interaction. Referring to Figure 15, which is a schematic diagram of a second process for configuring network devices according to an embodiment of this application, step S63 above may include the following steps.

[0266] Step S151: Based on human-computer interaction, collect the fourth natural language of the instruction configuration network device.

[0267] Step S152: Input the fourth natural language into the AI ​​network model to obtain network device configuration instructions.

[0268] Step S153: The intelligent agent calls the interface corresponding to the network device configuration command to configure the network device according to the network configuration information.

[0269] In the technical solution provided in this application embodiment, based on the strong natural language processing capability of the AI ​​network big model, the AI ​​network big model is used to analyze and process the natural language (i.e., the fourth natural language) expressing network device configuration to obtain network device configuration instructions, and an intelligent agent is used to call the corresponding API to realize the configuration of network devices according to the target network construction plan.

[0270] In step S151 above, the fourth natural language is any natural language used to instruct on configuring the network device. The fourth natural language can be text or voice, etc., and there is no limitation on the form of the fourth natural language.

[0271] Electronic devices can provide users with a variety of human-computer interaction methods. Users can input a fourth natural language into the electronic device through various human-computer interaction methods, and then the electronic device can collect the fourth natural language input by the user based on the various human-computer interaction methods.

[0272] When the electronic device provides a human-computer interaction dialog box, step S151 above can be: displaying the human-computer interaction dialog box; receiving instructions from the user in a fourth natural language input on the human-computer interaction dialog box to configure the network device. In this case, the electronic device can also display prompts (i.e., fourth prompts) on the human-computer interaction dialog box to guide the input of execution information for configuring the network device. For example, the fourth prompt could be "Please enter whether to configure the network device," etc. The content of the fourth prompt is not limited here. Through the human-computer interaction dialog box, the interaction between the user and the electronic device can be displayed intuitively. By displaying prompts on the human-computer interaction dialog box, the user can be guided to input relevant information on whether to configure the network device, thus meeting the user's needs and facilitating user operation.

[0273] In some embodiments, the electronic device can also determine the operation to be performed and display related prompts on the human-computer interaction dialog box. The electronic device can detect in real time the interface displayed to the user (or the stage of the scheme execution) and the content already entered by the user (or the operation already performed), determine the operation to be performed by the electronic device, and display related prompts on the human-computer interaction dialog box based on the operation to be performed. For example, when the electronic device detects that the interface displayed to the user is an intent decision interface (i.e., in the intent decision stage), and detects that the content the user last entered on the human-computer interaction dialog box was natural language indicating the display method of the network construction scheme (i.e., the second natural language), the electronic device can determine that the operation to be performed is: to determine a target network construction scheme from the target network construction schemes (in the case of multiple target network construction schemes output by the AI ​​network large model), and display a third prompt on the human-computer interaction dialog box to guide the input of the determined target network construction scheme.

[0274] In this embodiment of the application, the method by which the electronic device collects the fourth natural language based on human-computer interaction is similar to the method of collecting the first natural language described above. For details, please refer to the relevant description of step S61 above.

[0275] In step S152 above, the electronic device inputs the collected fourth natural language into the AI ​​network big model, so that the AI ​​network big model outputs network device configuration instructions based on the fourth natural language.

[0276] In this embodiment of the application, the electronic device can call the interface (i.e., the first interface) corresponding to the AI ​​network big model to input the fourth natural language into the AI ​​network big model, and there are no limitations on this.

[0277] In step S153 above, the electronic device uses an intelligent agent to call the interface (i.e., the third interface) corresponding to the network device configuration command and configure the network device using the target network construction scheme.

[0278] In this embodiment of the application, after determining the target network construction scheme for configuring the network device, the electronic device can execute steps S151 to S153 after detecting that the network device is powered on, collect the fourth natural language, obtain the network device configuration instruction, and configure the network device; the electronic device can also execute steps S151 to S152 before detecting that the network device is powered on, collect the fourth natural language, obtain the network device configuration instruction, and execute step S153 after detecting that the network device is powered on to configure the network device, and there is no limitation on this.

[0279] In this embodiment of the application, the electronic device can display a dialog box icon on the interface. After the user clicks the dialog box icon, the electronic device responds to the user's click operation by calling the front-end general API (i.e., the fourth interface) corresponding to the click operation to display the human-computer interaction dialog box.

[0280] In this embodiment of the application, the electronic device can also continuously update the toolset for human-computer interaction, that is, continuously update the interface, so that the electronic device can adopt intelligent agents, call richer interfaces, and improve the ability of intelligent agents to configure network devices in network planning and construction scenarios.

[0281] The network planning and construction method provided in this application will be described in detail below based on the network planning and construction framework shown in Figure 16 and the SDN controller interfaces shown in Figures 17a-17c, 18a-18e, 19a-19e, and 20a-20b. The example of an electronic device acting as an SDN controller and the SDN controller interacting with the user via a human-computer interaction dialog box is not intended to limit the scope of the application.

[0282] Figure 16 is a schematic diagram of a network planning and construction method framework provided in an embodiment of this application. As shown in Figure 16, in the data center intelligent network, the network planning and construction scenario intent management system (i.e., the SDN controller) on the control plane of the intelligent network receives the network planning and construction intent (i.e., the first natural language) input by the enterprise system / customer (i.e., the user), and performs intent perception, intent analysis, intent decision-making, and intent execution based on the AI ​​network big model, solution knowledge base, and intelligent agent. After determining the network construction scheme (i.e., the target network construction scheme), the SDN controller distributes the configuration to the forwarding plane of the intelligent network and collects information from the forwarding plane of the intelligent network. The forwarding plane of the intelligent network is a leaf-spine architecture, including multiple spine nodes, leaf nodes, and servers. Network devices are configured according to the target network construction scheme, forwarding paths between communicating servers are opened, and servers interact through spine nodes and leaf nodes.

[0283] The following describes in detail the four stages of network planning and construction provided in this application, with reference to the SDN controller interfaces shown in Figures 17a-17c, 18a-18e, 19a-19e, and 20a-20b.

[0284] 1) Intention perception stage

[0285] Figures 17a and 17c are schematic diagrams of a first type of SDN controller interface provided in the embodiments of this application. Figures 17a and 17c are all SDN controller interfaces in the intent-aware stage. As shown in Figures 17a and 17c, the SDN controller interface mainly includes a stage bar 171, a model selection icon 176, a display bar 177, and an interaction icon 178. Among them, the stage bar 171 includes an intent-aware stage tab 172, an intent analysis stage tab 173, an intent decision stage tab 174, and an intent execution stage tab 175, which correspond to the four stages of network planning and construction, respectively. The intent-aware stage tab 172 is in bold, indicating that the current stage is intent-aware.

[0286] When displaying the SDN controller interface shown in Figure 17a, the user can click the model selection icon 176. In response to this click, the SDN controller displays several selectable models for intent analysis (i.e., recommending network construction schemes). After the user selects a model, the SDN controller can switch between models used for intent analysis. In Figure 17a, the currently used model is the AI ​​network large model.

[0287] Users can click the interactive icon 178 (i.e., the dialog box icon) on the SDN controller interface. In response to the user's click on the interactive icon 178, the SDN controller calls the front-end general API (i.e., the fourth interface) corresponding to the user's click on the interactive icon 178, and displays the human-computer natural language interaction interface (i.e., the human-computer interaction dialog box) in the display bar 177, that is, the human-computer natural language interaction interface 179 shown in Figure 17b. The human-computer natural language interaction interface 179 may include a confirmation button 1710.

[0288] The SDN controller can detect operations to be performed in real time or triggered by events. Events that trigger the detection of operations to be performed can include: user clicks on the interface, completion of operations by the SDN controller or the AI ​​network's large model (such as intent analysis events), etc.

[0289] When the SDN controller detects that the current stage is the intent-aware stage and determines that the user has not yet performed any operation, the SDN controller can determine that the operation to be performed is the business intent of collecting network planning and construction scenarios (i.e., network planning and construction intent). Then, the SDN controller displays the prompt information associated with the operation to be performed (i.e., the first prompt information) on the human-machine natural language interaction interface 179, as shown in the prompt information displayed in the human-machine natural language interaction interface 179 as shown in Figure 17b.

[0290] Users can input natural language (i.e., the first natural language) expressing the business intent of the network planning and construction scenario in the human-computer natural language interaction interface 179 according to the prompts in the interface, as shown in Figure 17c. The user clicks the confirmation button 1710; the SDN controller responds to this by using the natural language from the human-computer natural language interaction interface 179 as the collected business intent of the network planning and construction scenario. The prompts and business intents shown in Figures 17b and 17c are merely examples and are not intended to be limiting.

[0291] In this embodiment, the interactive icon 178 and the human-computer natural language interaction interface 179 can be floating on the display bar 177. That is, the user can drag the interactive icon 178 and the human-computer natural language interaction interface 179 on the SDN controller interface. The SDN controller responds to the user's drag operation by calling the front-end general API corresponding to the drag operation to move the interactive icon 178 and the human-computer natural language interaction interface 179.

[0292] The user can click the interactive icon 178 again or click the blank space in the display bar 177. In response to the user's click on the interactive icon 178 again or the user's click on the blank space in the display bar 177, the SDN controller calls the front-end general API corresponding to the click operation and closes the human-computer natural language interaction interface 179.

[0293] The human-machine natural language interaction interface 179 on the SDN controller is not limited to the SDN controller interface in the intent perception stage. That is, the human-machine natural language interaction interface 179 can also be displayed on the SDN controller interface in the intent analysis stage, intent decision stage, and intent execution stage.

[0294] The SDN controller provides a human-machine natural language interaction interface on the SDN controller interface of the data center's self-intelligent network. Through API calls, it integrates the ability of AI network large models and realizes the collection of user input of business intent in network planning and construction scenarios in a natural language manner. The human-machine natural language interaction interface is pre-set with prompts for inputting business intent in network planning and construction scenarios, guiding users to make efficient input of business intent.

[0295] 2) Intent Analysis Phase

[0296] In this embodiment of the application, the intent analysis stage can be entered in the following ways.

[0297] In Method 1, after the user clicks the confirmation button 1710 during the intent perception phase, the network planning and construction scheme jumps to the intent analysis phase.

[0298] Specifically, after the user inputs the business intent of the network planning and construction scenario in the human-computer natural language interaction interface 179 and clicks the confirmation button 1710, the SDN controller responds to the user's click of the confirmation button 1710 by calling the front-end general API that redirects to the corresponding interface, and then redirects to the SDN controller interface in the intent analysis phase, thus entering the intent analysis phase. At this time, the business intent of the network planning and construction scenario is: the natural language displayed in the human-computer natural language interaction interface 179 when the user clicks the confirmation button 1710.

[0299] Alternatively, users can click the Intent Analysis Phase tab (173) to jump to the Intent Analysis Phase of the network planning and construction scheme.

[0300] Specifically, after the user inputs the business intent of the network planning and construction scenario in the human-computer natural language interaction interface 179 and clicks the confirmation button 1710, the user can click the intent analysis phase tab 173. In response to the user's click, the SDN controller calls the front-end general API that redirects to the corresponding interface, navigating to the SDN controller interface for the intent analysis phase, thus entering the intent analysis phase. At this point, the business intent of the network planning and construction scenario is: the natural language displayed in the human-computer natural language interaction interface 179 when the user clicks the confirmation button 1710.

[0301] After entering the intent analysis phase, the SDN controller calls the API (i.e., the first interface) corresponding to the AI ​​network big model, inputs the natural language displayed in the human-computer natural language interaction interface 179 into the AI ​​network big model, and performs intent analysis.

[0302] Figures 18a and 18b are a second schematic diagram of the SDN controller interface provided in the embodiments of this application. Figures 18a and 18b are all SDN controller interfaces in the intent analysis phase, and the intent analysis phase tab 173 is in bold.

[0303] The human-computer natural language interaction interface 179 can also be displayed on the SDN controller interface during the intent analysis phase. For details on how to open and close the human-computer natural language interaction interface 179, please refer to the relevant description in the intent perception phase.

[0304] During the intent analysis phase, the SDN controller can display the analysis progress 181 on the display bar 177, as shown in Figure 18a. The current analysis progress is 80%. In Figure 18a, the SDN controller has closed the human-machine natural language interaction interface 179.

[0305] When the human-computer natural language interaction interface 179 is open, the human-computer natural language interaction interface 179 can display historical question and answer information, such as the historical question information 182 (i.e. business intent) of the user input AI network big model as shown in Figure 18b.

[0306] In this embodiment, the SDN controller can detect that the current stage is the intent analysis stage and determine the business intent collected in the network planning and construction scenario. If the intent analysis is not yet complete (i.e., the analysis progress has not reached 100%), the SDN controller can display a "Please wait" prompt on the human-machine natural language interaction interface 179. If the intent analysis is complete (i.e., the analysis progress has reached 100%), meaning the AI ​​network big model has output a recommended network construction plan, the SDN controller can determine that the operation to be performed is to enter the intent decision stage. Correspondingly, the SDN controller displays prompts on the human-machine natural language interaction interface 179 guiding the user to enter the intent decision stage, as shown in the prompts displayed on the human-machine natural language interaction interface 179 in Figure 18b.

[0307] 3) Intent Decision-Making Stage

[0308] In this embodiment of the application, after the intent analysis is completed, the intent decision-making stage can be entered in the following ways.

[0309] In Method 1, after the user clicks the confirmation button 1710 during the intent analysis phase, the network planning and construction scheme jumps to the intent decision phase.

[0310] Specifically, the user can input the natural language to enter the intent decision stage in the human-computer natural language interaction interface 179 and click the confirmation button 1710. In response to the user's click of the confirmation button 1710, the SDN controller calls the front-end general API to jump to the corresponding interface and jumps to the SDN controller interface of the intent decision stage, thus entering the intent decision stage.

[0311] Method 2: The user clicks the Intent Decision Stage tab 174, and the network planning and construction scheme jumps to the Intent Decision Stage.

[0312] Users can click the Intent Decision Stage tab 174. In response to the user's click, the SDN controller calls the front-end general API that redirects to the corresponding interface, and then redirects to the SDN controller interface for the Intent Decision Stage, thus entering the Intent Decision Stage.

[0313] Figures 19a to 19e are a third schematic diagram of the SDN controller interface provided in the embodiments of this application. Figures 19a to 19e are all SDN controller interfaces in the intent decision stage, and the intent decision stage tab 174 is in bold.

[0314] The human-computer natural language interaction interface 179 can also be displayed on the SDN controller interface during the intent decision-making phase. For details on how to open and close the human-computer natural language interaction interface 179, please refer to the relevant description of the intent perception phase.

[0315] When the human-computer natural language interaction interface 179 is opened, the historical question-and-answer information displayed on the human-computer natural language interaction interface 179 may include: the historical answer information 191 output by the AI ​​network big model as shown in Figure 19a (i.e., the scheme description of the target network construction scheme recommended by the AI ​​network big model).

[0316] In this embodiment of the application, the business intent of the user inputting the network planning and construction scenario into the AI ​​network big model may or may not include a preference for the overall requirements of the network construction scheme; the content of the historical answer information 191 is determined based on whether the historical question information 182 has a preference requirement.

[0317] For example, if the business intent in a network planning and construction scenario includes a preference for the overall requirements of the network construction solution, taking the preference for high reliability and good scalability at the device and link levels as an example, the AI ​​network big data model can generate a network construction solution based on the business intent of the network planning and construction scenario that requires high reliability and good scalability at the device and link levels. This network construction solution is a network construction solution with high reliability and good scalability at the device and link levels. The historical response information output by the AI ​​network big data model shows the solution description, as shown in the historical responses in Figure 19a. The solution described in Information 191 is as follows: The network architecture (i.e., network architecture information) adopts a three-layer network architecture consisting of boundary leaf nodes, spine nodes, and leaf nodes; the number of devices (i.e., network device information) is: boundary leaf nodes – 2 box switches, spine nodes – 2 chassis switchers, leaf nodes – 10 box switches; the network configuration (i.e., link information) is: 2 leaf nodes form an M-LAG group, and the server is dual-homed to the M-LAG; the port information is: ports 1 and 2 of the switches are interconnected with other switches, and ports 3 and 4 serve as the internal control link (Intra-Portal Link, IPL) and peer-link of the M-LAG; the resource pools are: device management IP resource pool: 172.168.1.0 / 24; VTEP IP resource pool: 172.168.2.0 / 24.

[0318] For example, if the business intent in a network planning and construction scenario does not include a preference for the overall requirements of the network construction plan, the AI ​​network big data model can generate multiple network construction plans based on the business intent of the network planning and construction scenario. In this case, the historical response information output by the AI ​​network big data model can include descriptions of these multiple network construction plans.

[0319] In this embodiment, the SDN controller can detect that the current stage is the intent decision stage and determine that the AI ​​network big model has output a recommended network construction scheme. Therefore, it can determine that the operation to be performed is to display the network construction scheme. Accordingly, the SDN controller displays the prompt information associated with the operation to be performed (i.e., the second prompt information) on the human-computer natural language interaction interface 179, as shown in Figure 19a.

[0320] Users can input their preferred expression in natural language (i.e., second natural language) on the human-computer natural language interaction interface 179 based on the prompts, and then click the confirmation button 1710. The following example illustrates this: the user inputs natural language for graphical display on the human-computer natural language interaction interface 179. After the user inputs the natural language for graphical display, corresponding historical information can be added to the historical question and answer information of the human-computer natural language interaction interface 179, as shown in the historical question information 192 in Figure 19b.

[0321] Furthermore, the user inputs natural language to express the graphical display method in the human-computer natural language interaction interface 179 and clicks the confirmation button 1710. In response to the user's click of the confirmation button 1710, the SDN controller calls the API (i.e., the first interface) corresponding to the AI ​​network big model, inputs the natural language input by the user to the AI ​​network big model, and causes the AI ​​network big model to output the corresponding display method. Then, using an intelligent agent, the SDN controller calls the SDN controller API (i.e., the second interface / visualization scheme API) corresponding to the display method to graphically display the network construction scheme, as shown in Figures 19c and 19d.

[0322] Figure 19c shows the interface diagram of the intent analysis phase when the business intent in a network planning and construction scenario includes a preference for the overall requirements of the network construction plan. Figure 19d shows the interface diagram of the intent analysis phase when the business intent in a network planning and construction scenario does not include a preference for the overall requirements of the network construction plan. In Figures 19c and 19d, the SDN controller disables the human-machine natural language interaction interface 179. Additionally, the display bar 177 may include a scheme description 193 presenting the network construction plan in text form.

[0323] The difference between Figures 19c and 19d is that when the business intent of the network planning and construction scenario does not include a preference for the overall requirements of the network construction scheme, display bar 177 also includes tabs 194 corresponding to various network construction schemes. Different tabs correspond to different network construction schemes. As shown in Figure 19d, display bar 177 includes tabs 194 corresponding to four network construction schemes: optimal scalability, optimal reliability, optimal security, and optimal cost. Under different tabs, display bar 177 graphically displays different network construction schemes. Currently, display bar 177 is located under tab 194 corresponding to the optimal scalability network construction scheme, and the SDN controller graphically displays the optimal scalability network construction scheme. Figures 19c and 19d include two schematic diagrams of graphically displayed network construction schemes. In Figure 19d, the graphically displayed network construction scheme also includes a firewall (FW) module and its corresponding ports.

[0324] In Figure 19d, users can click on tabs 194 corresponding to different network construction schemes. In response to the user's click on the corresponding tab 194, the SDN controller calls the corresponding front-end general API and displays the corresponding network construction scheme to the user in the display bar 177, so that the user can view it by himself.

[0325] In this embodiment of the application, if the user does not input the display method, the SDN controller can display the network construction plan in a preset display method. For example, the network construction plan can be displayed in text form, as shown in Figure 19e, where the network construction plan description 193 is displayed in text form in display bar 177.

[0326] In addition, users can drag and edit the graphically displayed network construction plan in the display bar 177. The SDN controller responds to the user's drag and edit operations and fine-tunes the network construction plan.

[0327] After demonstrating the network construction plan, the SDN controller can determine that the operation to be performed is to determine the network construction plan. Then, prompts (i.e., third prompts) guiding the input for determining the network construction plan can be displayed on the human-machine natural language interaction interface 179.

[0328] Users can input the network construction plan based on the prompts (i.e., the third prompt) displayed on the human-computer natural language interaction interface 179, and then click the confirmation button 1710. In response to the user clicking the confirmation button 1710, the SDN controller calls the API (i.e., the first interface) corresponding to the AI ​​network model, inputting the user's natural language input into the AI ​​network model, causing the AI ​​network model to output the determined network construction plan. Using an intelligent agent, the SDN controller can call the corresponding API to display the determined network construction plan in the display bar 177.

[0329] If the displayed network construction plan does not meet expectations, users can continue to input in a third language until the displayed network construction plan meets their expectations. This operation is the same as the effect described above, where the user clicks on tab 194 corresponding to different network construction plans to switch between them.

[0330] After determining the network construction plan, the SDN controller can determine that the operation to be performed is to configure the network device. Then, prompts (i.e., the fourth prompts) guiding the execution of network device configuration can be displayed on the human-machine natural language interaction interface 179, as shown in the prompts in the human-machine natural language interaction interface 179 in Figure 19b.

[0331] In this embodiment, when the business intent in a network planning and construction scenario includes a preference for the overall requirements of the network construction plan, the AI ​​network big data model outputs a network construction plan. In this case, the SDN controller can omit the display process of the third prompt information and directly display the fourth prompt information.

[0332] In this embodiment, the default processing flow of the SDN controller in the intent decision-making phase is as follows: first, the network construction scheme output by the AI ​​network large model is displayed; then, the network construction scheme for configuring network devices is determined; and finally, the network devices are configured. The SDN controller can also change the processing flow based on user input. For example, when the human-computer natural language interaction interface 179 displays prompts guiding the display method (i.e., the second prompt), the user can input natural language to determine the network construction scheme (i.e., the third natural language) to confirm the network construction scheme; subsequently, when the human-computer natural language interaction interface 179 displays prompts guiding the input of execution information for configuring network devices (i.e., the fourth prompt), the user can input the display method (i.e., the second natural language) to display the network construction scheme. This is not limited.

[0333] In actual execution, the default processing flow of the SDN controller in the intent decision phase can be set according to actual needs.

[0334] 4) Intent Execution Phase

[0335] In this embodiment of the application, the intent execution phase can be entered in the following ways.

[0336] Method 1: When the user clicks the "Start Implementation" button (195), the network planning and construction scheme jumps to the "Intent Execution" stage.

[0337] For example, in Figure 19d, the SDN controller interface may include a "Start Implementation" button 195. When a user clicks the "Start Implementation" button 195, the SDN controller responds by calling the front-end general API that redirects to the corresponding interface, navigating to the SDN controller interface in the intent execution phase, and configuring network devices according to the determined network construction plan. Alternatively, in Figures 19c and 19e, the SDN controller interface may also include a "Start Implementation" button 195 (not shown in the figures).

[0338] In this embodiment, the SDN controller can use the user's click of the Start Implementation button 195 as an instruction to execute network device configuration. After jumping to the SDN controller interface in the intent decision stage, the SDN controller can directly configure the network device according to the determined network construction plan; alternatively, it can input an instruction to execute network device configuration. After receiving the instruction to execute network device configuration, the SDN controller then configures the network device according to the determined network construction plan.

[0339] Method 2: The user enters the fourth natural language in the human-computer natural language interaction interface 179 and clicks the confirmation button 1710. The network planning and construction scheme then jumps to the intent execution stage.

[0340] Specifically, the user can input natural language instructions (i.e., the fourth natural language) to configure network devices in the human-machine natural language interaction interface 179 and click the confirmation button 1710. In response to the user clicking the confirmation button 1710, the SDN controller calls the API (i.e., the first interface) corresponding to the AI ​​network big model, inputting the user's natural language into the AI ​​network big model. This causes the AI ​​network big model to output the corresponding network device configuration instructions. Then, using an intelligent agent, the SDN controller calls the front-end general API that redirects to the corresponding interface, jumping to the SDN controller interface in the intent execution phase. Entering the intent execution phase, it calls the API (i.e., the third interface) corresponding to the network device configuration instructions and configures the network devices according to the determined network construction plan.

[0341] Method 3: When the user clicks the Intent Execution Phase tab 175, the network planning and construction scheme will jump to the Intent Execution Phase.

[0342] Specifically, during the intent decision phase, if the user does not input a fourth natural language, the user can click the intent execution phase tab 175. In response to the user's click on the intent execution phase tab 175, the SDN controller calls the front-end general API that jumps to the corresponding interface, jumps to the SDN controller interface of the intent execution phase, enters the intent execution phase, and configures the network devices according to the determined network construction plan.

[0343] In this embodiment, the SDN controller can use the user's click on the intent execution phase tab 175 as an instruction to configure the network device. After jumping to the SDN controller interface in the intent execution phase, the SDN controller can directly configure the network device according to the determined network construction plan; alternatively, it can input an instruction to configure the network device. After receiving the instruction to configure the network device, the SDN controller will then configure the network device according to the determined network construction plan.

[0344] For example, after navigating to the SDN controller interface in the intent execution phase, method two is used: inputting a fourth natural language in the human-machine natural language interaction interface 179 to obtain instructions for configuring network devices.

[0345] For example, a "Start Implementation" button can be set on the SDN controller interface during the intent execution phase, as shown by the "Start Implementation" button 201 in display bar 177 of Figure 20a. After navigating to the SDN controller interface during the intent execution phase, the user clicks the "Start Implementation" button 201. In response to the user's click of the "Start Implementation" button 201, the SDN controller generates network device configuration instructions, calls the API (i.e., the third interface) corresponding to the network device configuration instructions, and configures the network devices according to the determined network construction plan.

[0346] In this embodiment of the application, the determined network construction scheme is the network construction scheme displayed on the display bar 177 when the user clicks the start implementation button 201, the confirmation button 1710, or the intention execution stage tab 175. It can be determined by the user by clicking the tab 194 corresponding to different network construction schemes in Figure 19d, or by the user by inputting natural language (i.e., third natural language) to determine the network construction scheme in the human-computer natural language interaction interface 179.

[0347] Figures 20a and 20b are the fourth schematic diagrams of the SDN controller interface provided in the embodiments of this application. Figures 20a and 20b are all SDN controller interfaces in the intent execution phase, and the intent execution phase tab 175 is in bold.

[0348] During the execution of the intent (i.e., configuring the network device), the SDN controller can display the execution progress 202 on display bar 177, as shown in Figure 20b. The current execution progress is 70%, not yet 100%, indicating that the intent execution is not yet complete. At this time, the SDN controller can display a "Please wait" message on the human-machine interface 179. Once the intent execution is complete (i.e., the execution progress reaches 100%), the SDN controller confirms that the network device configuration is complete.

[0349] The technical solution provided in this application introduces an AI network big model. Based on the human-computer natural language interaction capability of the AI ​​network big model, it solves the problem that the data center's intelligent network lacks the ability to understand user business-level intentions when collecting network planning and construction intentions.

[0350] The AI ​​network model is trained by a large DCN corpus, and fine-tuned by combining labeled datasets of network planning and construction scenarios (i.e., sample natural language and corresponding sample network planning and construction schemes). The AI ​​network model is then deployed in the production environment to improve the ability to understand network planning and construction intentions, and to solve the problem that the data center intelligent network cannot understand new and changed network intentions in the network planning and construction scenario.

[0351] By combining an AI network big model with a solution knowledge base, the private domain knowledge of configuring network devices in the network planning and construction scenarios of the AI ​​network big model is expanded, thereby improving the accuracy of the AI ​​network big model in recommending and generating solutions for intelligent network planning and construction scenarios.

[0352] By leveraging the intelligent agent framework and calling the rich APIs of the SDN controller based on the toolset, a graphical representation of network construction solutions can be achieved.

[0353] By continuously enriching the network corpus and labeled datasets for network planning and construction scenarios of the data center's self-intelligent network, the training and fine-tuning of the AI ​​network large model are achieved. By continuously updating the deployment method of the AI ​​network large model in the background, the self-intelligent network of the data center can achieve self-evolution and evolution of its network construction scheme recommendation and generation capabilities in network planning and construction scenarios. By segmenting and vectorizing new knowledge and storing it in a vector database, the scheme knowledge base is continuously supplemented and improved. By continuously enriching the toolset for interaction between the network planning and construction scenario and the SDN controller through the intelligent agent framework, the intent execution capability of the network planning and construction scenario is improved, ultimately achieving self-evolution and evolution of the network scheme recommendation and generation capabilities in the network planning and construction scenario.

[0354] Corresponding to the above-described network planning and construction method, this application provides a network planning and construction apparatus. Referring to Figure 21, which is a schematic diagram of a network planning and construction apparatus provided in this application, the apparatus includes:

[0355] The data acquisition module 211 is used to acquire first natural language that indicates the intention of network planning and construction based on human-computer interaction.

[0356] The first determining module 212 is used to input the first natural language into the AI ​​network big model to obtain the target network construction scheme, which includes network topology information and network configuration information.

[0357] Configuration module 213 is used to configure the network device according to the network configuration information after the network device corresponding to the network topology information is detected to be powered on.

[0358] In the technical solution provided in this application embodiment, natural language is used to express user intent, namely, network planning and construction intent. Natural language is a language that users can easily understand and does not require a professional knowledge background. Therefore, using natural language to express user intent and realize network planning and construction increases the user's ability to understand business-level intents. Furthermore, in the technical solution provided in this application embodiment, an AI network big data model is used to analyze and process the natural language expressing user intent (i.e., the first natural language) to obtain a target network construction scheme that meets the network planning and construction intent. Because the AI ​​network big data model has strong natural language processing capabilities, even if new or changed intents appear, the AI ​​network big data model can analyze and obtain the required target network construction scheme, improving the scalability of intent collection and the adaptability to scenarios and network topologies.

[0359] In some embodiments, the acquisition module 211 is specifically used for:

[0360] Display the human-computer interaction dialog box;

[0361] It receives the user's intention to plan and build the network in the first natural language input on the human-computer interaction dialog box.

[0362] In some embodiments, the acquisition module 211 is further configured to:

[0363] The first prompt message is displayed on the human-computer interaction dialog box. The first prompt message is used to guide the input of network planning and construction intentions.

[0364] In some embodiments, the first determining module 212 described above is specifically used for:

[0365] By matching the first natural language with pre-stored knowledge of various network construction schemes, multiple candidate knowledge can be obtained;

[0366] By inputting the first natural language and multiple candidate knowledge into the AI ​​network model, the target network construction scheme is obtained.

[0367] In some embodiments, the above-described apparatus further includes: a display module, configured to:

[0368] Based on human-computer interaction, the second natural language of the display method of the network construction plan is collected and indicated;

[0369] Inputting the second natural language into the AI ​​network model yields the target display method;

[0370] The intelligent agent calls the interface corresponding to the target display method to display the target network construction plan.

[0371] In some embodiments, the above-described display module is specifically used for:

[0372] Display the human-computer interaction dialog box;

[0373] The second natural language receives instructions from users on how to display the network construction plan through a human-computer interaction dialog box.

[0374] In some embodiments, the above-described display module is further configured to:

[0375] A second prompt message is displayed on the human-computer interaction dialog box. This second prompt message is used to guide the input of the network construction plan display method.

[0376] In some embodiments, the above-described apparatus further includes: an adjustment module, configured to:

[0377] The target network construction plan displayed is adjusted based on user actions.

[0378] In some embodiments, the AI ​​network big model outputs multiple target network construction schemes;

[0379] The above-mentioned device further includes: a second determining module, used for:

[0380] From multiple target network construction schemes, determine one target network construction scheme;

[0381] The aforementioned configuration module 213 is specifically used for:

[0382] After the network devices corresponding to the network topology information included in the determined target network construction scheme are powered on, the network devices are configured according to the network configuration information included in the determined target network construction scheme.

[0383] In some embodiments, the second determining module is specifically used for:

[0384] Based on human-computer interaction, the third natural language of the target network construction plan determined by the instructions is collected;

[0385] By inputting the third natural language into the AI ​​network model, the determined target network construction scheme is obtained.

[0386] In some embodiments, the second determining module is specifically used for:

[0387] Display the human-computer interaction dialog box;

[0388] The third natural language receives instructions from the user on the human-computer interaction dialog box to determine the target network construction plan.

[0389] In some embodiments, the second determining module described above is further configured to:

[0390] A third prompt message is displayed on the human-computer interaction dialog box. This third prompt message is used to guide the input of the determined target network construction plan.

[0391] In some embodiments, the configuration module 213 is specifically used for:

[0392] Based on human-computer interaction, collect and instruct network device configuration using the fourth natural language.

[0393] Inputting the fourth natural language into the AI ​​network model yields network device configuration instructions;

[0394] An intelligent agent is used to call the interface corresponding to the network device configuration command in order to configure the network device according to the network configuration information.

[0395] In some embodiments, the configuration module 213 is specifically used for:

[0396] Display the human-computer interaction dialog box;

[0397] The fourth natural language receives instructions from the user in the human-computer interaction dialog box to configure network devices.

[0398] In some embodiments, the configuration module 213 is further configured to:

[0399] The fourth prompt message is displayed on the human-computer interaction dialog box. The fourth prompt message is used to guide the input of execution information for configuring network devices.

[0400] In some embodiments, the above-described apparatus further includes: a display module, configured to:

[0401] Determine the operation to be performed;

[0402] Display the prompt information associated with the operation to be performed on the human-computer interaction dialog box.

[0403] In some embodiments, the above-described apparatus further includes: a fine-tuning module, configured to:

[0404] Obtain the sample natural language and the corresponding sample network planning and construction scheme;

[0405] By utilizing sample natural language and sample network planning and construction schemes, the large-scale AI network model is fine-tuned.

[0406] This application embodiment also provides an electronic device, as shown in FIG22, including a processor 221, a communication interface 222, a memory 223 and a communication bus 224, wherein the processor 221, the communication interface 222 and the memory 223 communicate with each other through the communication bus 224;

[0407] The aforementioned memory 223 is used to store computer programs;

[0408] When the processor 221 executes the program stored in the memory 223, it implements any of the above-mentioned network planning and construction methods.

[0409] The communication bus mentioned in the above electronic devices can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. This communication bus can be divided into address bus, data bus, control bus, etc. For ease of illustration, only one thick line is used to represent it in the diagram, but this does not mean that there is only one bus or one type of bus.

[0410] Communication interface 222 is used for communication between the above-mentioned electronic device and other devices.

[0411] The memory 223 may include random access memory (RAM) or non-volatile memory, such as at least one disk storage device. Optionally, the memory 223 may also be at least one storage device located remotely from the aforementioned processor 221.

[0412] The processor 221 mentioned above can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0413] In another embodiment provided in this application, a computer-readable storage medium is also provided, wherein a computer program is stored in the computer-readable storage medium, and the computer program, when executed by a processor, implements any of the above-described network planning and construction methods.

[0414] In another embodiment provided in this application, a computer program product containing instructions is also provided, which, when run on a computer, causes the computer to execute any of the network planning and construction methods described in the above embodiments.

[0415] In the above embodiments, implementation can be achieved entirely or partially through software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented entirely or partially in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., solid state disk (SSD)).

[0416] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0417] The various embodiments in this specification are described in a related manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the embodiments for devices, electronic devices, storage media, and program products are basically similar to the method embodiments, so the descriptions are relatively simple; relevant parts can be referred to the descriptions of the method embodiments.

[0418] The above description is only a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A network planning and construction method, characterized in that, The method includes: Based on human-computer interaction, the system collects first natural language samples indicating the intentions of network planning and construction. The first natural language is input into the AI ​​network model to obtain the target network construction scheme, which includes network topology information and network configuration information. After the network device corresponding to the network topology information is detected to be powered on, the network device is configured according to the network configuration information.

2. The method according to claim 1, characterized in that, The step of collecting first natural language indicating the intention of network planning and construction based on human-computer interaction includes: Display the human-computer interaction dialog box; Receive the user's first natural language input on the human-computer interaction dialog box indicating the network planning and construction intention.

3. The method according to claim 2, characterized in that, The method further includes: The first prompt message is displayed on the human-computer interaction dialog box. The first prompt message is used to guide the input of network planning and construction intentions.

4. The method according to claim 1, characterized in that, The step of inputting the first natural language into the AI ​​network model to obtain the target network construction scheme includes: The first natural language is matched with pre-stored knowledge of various network construction schemes to obtain multiple candidate knowledge; The first natural language and the various candidate knowledge are input into the AI ​​network big model to obtain the target network construction scheme.

5. The method according to claim 1, characterized in that, The method further includes: Based on human-computer interaction, the second natural language of the display method of the network construction plan is collected and indicated; The second natural language is input into the AI ​​network model to obtain the target display method; An intelligent agent calls the interface corresponding to the target display method to display the target network construction scheme.

6. The method according to claim 5, characterized in that, The step of collecting the second natural language of the display method of the network construction scheme based on human-computer interaction includes: Display the human-computer interaction dialog box; The system receives a second natural language input from the user on the human-computer interaction dialog box, indicating how the network construction plan should be displayed.

7. The method according to claim 6, characterized in that, The method further includes: A second prompt message is displayed on the human-computer interaction dialog box. The second prompt message is used to guide the input of the display method of the network construction plan.

8. The method according to any one of claims 5-7, characterized in that, The method further includes: The target network construction plan displayed is adjusted based on user actions.

9. The method according to claim 1, characterized in that, The AI ​​network model outputs multiple target network construction schemes; The method further includes: From the multiple target network construction schemes, one target network construction scheme is determined; The step of configuring the network device according to the network configuration information after detecting that the network device corresponding to the network topology information has been powered on includes: After the network device corresponding to the network topology information included in the determined target network construction scheme is powered on, the network device is configured according to the network configuration information included in the determined target network construction scheme.

10. The method according to claim 9, characterized in that, The step of determining a target network construction scheme from the plurality of target network construction schemes includes: Based on human-computer interaction, the third natural language of the target network construction plan determined by the instructions is collected; The third natural language is input into the AI ​​network model to obtain the determined target network construction scheme.

11. The method according to claim 10, characterized in that, The step of collecting the third natural language of the target network construction scheme determined by the instruction based on human-computer interaction includes: Display the human-computer interaction dialog box; The system receives instructions from the user on the human-computer interaction dialog box, specifying the target network construction plan in third natural language.

12. The method according to claim 11, characterized in that, The method further includes: A third prompt message is displayed on the human-computer interaction dialog box. The third prompt message is used to guide the input of the determined target network construction plan.

13. The method according to claim 1, characterized in that, The step of configuring the network device according to the network configuration information includes: Based on human-computer interaction, collect and instruct network device configuration using the fourth natural language. The fourth natural language is input into the AI ​​network model to obtain network device configuration instructions; An intelligent agent invokes the interface corresponding to the network device configuration command to configure the network device according to the network configuration information.

14. The method according to claim 13, characterized in that, The step of collecting instructions in the fourth natural language for configuring network devices based on human-computer interaction includes: Display the human-computer interaction dialog box; The system receives instructions from the user in the human-computer interaction dialog box to configure the network device in a fourth natural language.

15. The method according to claim 14, characterized in that, The method further includes: A fourth prompt message is displayed on the human-computer interaction dialog box. The fourth prompt message is used to guide the input of execution information for configuring network devices.

16. The method according to any one of claims 3, 7, 12, and 15, characterized in that, The method further includes: Determine the operation to be performed; The prompt information associated with the operation to be performed is displayed on the human-computer interaction dialog box.

17. The method according to claim 1, characterized in that, The method further includes: Obtain the sample natural language and the corresponding sample network planning and construction scheme; The AI ​​network model is fine-tuned using the sample natural language and the sample network planning and construction scheme.

18. A network planning and construction device, characterized in that, The device includes: The data acquisition module is used to collect first natural language that indicates the intention of network planning and construction, based on human-computer interaction. The first determining module is used to input the first natural language into the artificial intelligence (AI) network model to obtain a target network construction scheme, wherein the target network construction scheme includes network topology information and network configuration information. The configuration module is used to configure the network device according to the network configuration information after the network device corresponding to the network topology information is detected to be powered on.

19. An electronic device, characterized in that, It includes a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus; The memory is used to store computer programs; When the processor executes the program stored in the memory, it implements the method described in any one of claims 1-17.

20. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the method described in any one of claims 1-17.

Images