A method, apparatus, and related equipment for classifying communication network operation and maintenance intentions.
By combining intent classification agents and intent inspection agents, the problem of low accuracy in manually determining programming languages in network operations and maintenance is solved, and efficient and accurate network operations and maintenance are achieved.
Patent Information
- Application Number
- CN202510631609.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-05-16
- Publication Date
- 2026-11-17
AI Technical Summary
The accuracy of manually determining programming languages by network operations and maintenance personnel is low, resulting in low efficiency in network operations and maintenance work.
By combining an intent classification agent and an intent inspection agent, the system obtains prompts from user input, classifies the intent, and inspects them to ensure the accuracy of the intent classification results.
It improves the accuracy of intent classification, simplifies network operation and maintenance procedures, reduces the difficulty and time of operation and maintenance, and improves operation and maintenance efficiency.
Smart Images

Figure CN123115408A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of communication technology, and in particular to a method, apparatus and related equipment for classifying communication network operation and maintenance intentions. Background Technology
[0002] With the continuous development of communication technology, it has been widely applied in people's lives. To improve communication quality, the number of base stations needs to be increased. This increase in the number of base stations presents significant challenges to network operation and maintenance. Current network operation and maintenance work involves multiple programming languages, requiring network operation and maintenance personnel to manually determine the appropriate programming language and output different languages to handle different problems. However, the accuracy of manually determining the programming language is relatively low. Summary of the Invention
[0003] This application provides a method, apparatus, and related equipment for classifying communication network operation and maintenance intentions, in order to solve the problem of low accuracy in manually determining the type of programming language by network operation and maintenance personnel.
[0004] To solve the above problems, this application is implemented as follows:
[0005] In a first aspect, embodiments of this application provide a method for classifying communication network operation and maintenance intentions, including:
[0006] Get the suggestion words entered by the user;
[0007] The prompt words are input into the intent classification agent for intent classification to obtain target intent information, which is used to represent the network operation and maintenance code generated using the target language.
[0008] At least a portion of the prompt words and the target intent information are input into the intent checking agent to perform intent classification checking, and intent classification result checking information is obtained.
[0009] If the intent classification result check information indicates that the target intent information is correctly classified, the target intent information is determined as the intent classification result of the prompt word.
[0010] Secondly, embodiments of this application provide a communication network operation and maintenance intent classification device, including:
[0011] The acquisition module is used to acquire the prompt words entered by the user;
[0012] The intent classification module is used to input the prompt words into the intent classification agent to classify the intent and obtain the target intent information. The target intent information is used to represent the network operation and maintenance code generated using the target language.
[0013] The intent classification check module is used to input at least part of the prompt words and the target intent information into the intent check agent to perform intent classification check and obtain intent classification result check information.
[0014] The determination module is used to determine the target intent information as the intent classification result of the prompt word when the intent classification result check information indicates that the target intent information is correctly classified.
[0015] Thirdly, embodiments of this application also provide an electronic device, including: a memory, a processor, and a program stored in the memory and executable on the processor; the processor is configured to read the program in the memory to implement the steps in the method described in the first aspect above.
[0016] Fourthly, embodiments of this application also provide a readable storage medium for storing a program, which, when executed by a processor, implements the steps of the method described in the first aspect above.
[0017] Fifthly, embodiments of this application also provide a computer program product, including computer instructions, which, when executed by a processor, implement the steps of the method described in the first aspect above.
[0018] In this embodiment, a prompt word input by the user is obtained; the prompt word is input into an intent classification agent for intent classification to obtain target intent information, which is used to represent network operation and maintenance code generated using a target language; at least a portion of the prompt word and the target intent information are input into an intent checking agent for intent classification checking to obtain intent classification result checking information; if the intent classification result checking information indicates that the target intent information is correctly classified, the target intent information is determined as the intent classification result of the prompt word.
[0019] In this way, the intent classification agent can classify intents based on the prompt words input by the user to obtain target intent information, and the intent checking agent can check the above target intent information. When the intent checking agent checks that the target intent information is classified correctly, the target intent information is determined as the intent classification result of the prompt words, thereby ensuring the accuracy of the intent classification result. Attached Figure Description
[0020] To more clearly illustrate the technical solutions of the embodiments of this application, the drawings used in the description of the embodiments of this application will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0021] Figure 1 This is one of the flowcharts illustrating the communication network operation and maintenance intent classification method provided in the embodiments of this application;
[0022] Figure 2 This is a second flowchart illustrating the communication network operation and maintenance intent classification method provided in the embodiments of this application;
[0023] Figure 3 This is a schematic diagram of the intent classification agent provided in the embodiments of this application;
[0024] Figure 4 This is a schematic diagram of the intent-checking intelligent agent provided in the embodiments of this application;
[0025] Figure 5 This is a schematic diagram of the communication network operation and maintenance intent classification device provided in the embodiments of this application;
[0026] Figure 6 This is a schematic diagram of the structure of the electronic device provided in the embodiments of this application. Detailed Implementation
[0027] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.
[0028] The terms "first," "second," etc., used in the embodiments of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to these processes, methods, products, or devices. Additionally, the use of "and / or" in this application indicates at least one of the connected objects, such as A and / or B and / or C, representing seven possibilities: including A alone, B alone, C alone, and the presence of both A and B, both B and C, both A and C, and the presence of A, B, and C.
[0029] See Figure 1 , Figure 1 This is a flowchart illustrating the communication network operation and maintenance intent classification method provided in the embodiments of this application. Figure 1 The communication network operation and maintenance intent classification method shown can be executed by electronic devices.
[0030] like Figure 1 As shown, the communication network operation and maintenance intent classification method may include the following steps:
[0031] Step 101: Obtain the prompt words entered by the user.
[0032] The specific content of the prompt words is not limited here. Optionally, the prompt words include the questions entered by the user. The intent classification agent can accurately identify the details of the current task and the user's specific needs by recognizing the questions. It should be noted that the questions entered by the user can be intent classification tasks.
[0033] It should be noted that, optionally, the questions input by the users can be problems encountered during the operation and maintenance of the communication network.
[0034] As an optional implementation, the prompt words include at least one of the following: background description information, identity definition information, task description information, example information, user question information, and output requirement information.
[0035] The background description information is used to represent the current intent classification task and the task background. Optionally, when the embodiments of this application are applied to the application scenario of network operation and maintenance code generation, the task background can be the field of wireless communication and the field of network operation and maintenance code generation. That is, the background description information can enable the intent classification agent to accurately understand the task background of the current intent classification task.
[0036] Among them, the identity definition information is used to represent the identity definition information of the intent classification agent, so that the intent classification agent can accurately understand its own role, thereby assisting the execution of the intent classification task and ensuring a high accuracy rate.
[0037] Among them, the task description information is the description information of the intent classification task to be executed by the intent classification agent. The task description information provides a clear, explicit and specific description of the intent classification task to be executed by the intent classification agent, so that the intent classification agent can accurately understand the detailed requirements of the intent classification task.
[0038] The example information can be examples of real intent classification tasks and corresponding classification results. These examples can guide the intent classification agent to learn solutions for such tasks, thereby improving the accuracy of intent classification.
[0039] The user question information can be understood as the questions entered by the user.
[0040] The output requirement information is used to represent the output requirements and output constraints of the target intent information. Optionally, the output requirement information includes the output format and the type of output content, etc.
[0041] In this embodiment of the application, since the prompt words include at least one of the following: background description information, identity definition information, task description information, example information, user question information, and output requirement information, the more content the prompt words include, the more accurate the target intent information can be obtained by the intent classification agent in classifying intent.
[0042] Step 102: Input the prompt word into the intent classification agent to classify the intent and obtain the target intent information. The target intent information is used to represent the network operation and maintenance code generated using the target language.
[0043] The specific form of the target intent information is not limited here. Optionally, the target intent information can be text information; alternatively, the target intent information can be image information.
[0044] As an optional implementation, the target intent information is an intent classification index. This simplifies the output content and word count of the target intent information, thereby reducing the illusion of a large language model in the intent classification agent and improving the accuracy of the target intent information.
[0045] It should be noted that the intent classification agent can include a large language model, which is used to achieve intent classification. The large language model can be a pre-trained model.
[0046] Optionally, the intent category index may include at least one. For example, the intent category index may include 1, 2, and 3. When the intent category index includes 1, it can be used to indicate that the target language can be SQL, that is, the network operation and maintenance code is generated using the Structured Query Language (SQL). When the intent category index includes 2, it can be used to indicate that the target language can be Python, that is, the network operation and maintenance code is generated using Python. When the intent category index includes 3, it can be used to indicate that the target language can be shell, that is, the network operation and maintenance code is generated using shell.
[0047] It should be noted that SQL, Python, and shell cover most of the code-related tasks in network operations and maintenance. Thus, by selecting output intent category index as 1, 2, or 3, the target language for generating network operations and maintenance code can be accurately identified.
[0048] It should be noted that the implementation method of this application can be more conveniently switched between generating network operation and maintenance code using SQL, generating network operation and maintenance code using Python, and generating network operation and maintenance code using shell through natural language interaction. This simplifies the operation and maintenance process, lowers the threshold for operation and maintenance and usage, and makes network operation and maintenance work more convenient and efficient.
[0049] In addition, if a new language for generating network operation and maintenance code emerges in network operation and maintenance work, the types of intent classification indexes can be increased. For example, the types of intent classification indexes can be increased to 4. When the intent classification index is 4, it can be indicated that the target language can be the aforementioned new language, that is, the new language is used to generate network operation and maintenance code.
[0050] Optionally, see Figure 3 , Figure 3 This diagram illustrates an intent classification agent. After inputting prompt words into the intent classification agent, the large language model within the agent can perform planning and action processing. The planning process can be understood as the thinking and decision-making process undertaken by the large language model in the intent classification agent to achieve the purpose of intent classification. The action processing can be understood as the large language model in the intent classification agent, based on the prompt words, utilizing the powerful semantic understanding, knowledge implication, and reasoning capabilities of the large language model to output the category to which the user's question belongs.
[0051] Step 103: Input at least part of the prompt words and the target intent information into the intent checking agent to perform intent classification checking and obtain intent classification result checking information.
[0052] Among them, the intent checking agent can perform intent classification checks on the target intent information based on at least part of the content in the prompt words. By performing intent classification checks on the target intent information through the intent checking agent, the accuracy of the target intent information can be further guaranteed.
[0053] Step 104: If the intent classification result check information indicates that the target intent information is correctly classified, the target intent information is determined as the intent classification result of the prompt word.
[0054] In this embodiment of the application, through steps 101 to 104, the intent classification agent can classify intents based on the prompt words input by the user to obtain target intent information, and the intent checking agent can check the target intent information. When the intent checking agent checks that the target intent information is classified correctly, the target intent information is determined as the intent classification result of the prompt words, thereby ensuring the accuracy of the intent classification result.
[0055] It should be noted that, optionally, after the target intent information is determined to be the intent classification result of the prompt word, the corresponding operation can also be executed according to the target intent information. For example, if the user question included in the prompt word can be the average value of the target file, then the target intent information can be 1, and SQL can be used to generate network operation and maintenance code, which is used to analyze the average value of the target file.
[0056] It should be noted that the operation corresponding to the prompt word is executed according to the target intent information, and the operation can be generated into network operation and maintenance code using the target language. This allows the embodiments of this application to effectively reduce the difficulty of writing network operation and maintenance code in the field of wireless communication, reduce the steps and time of code writing, simplify the operation and maintenance process and operation, and improve operation and maintenance efficiency.
[0057] As an optional implementation, after inputting at least a portion of the prompt words and the target intent information into the intent checking agent for intent classification checking, and obtaining intent classification result checking information, the method further includes:
[0058] If the intent classification result check information indicates that the target intent information is incorrectly classified, the prompt word is corrected;
[0059] The corrected prompt is determined as the next input to the intent classification agent, and the process returns to inputting the prompt into the intent classification agent for intent classification to obtain the target intent information.
[0060] The specific method for correcting the prompt words is not limited here. Optionally, users can correct the prompt words through at least one of the following methods: voice input, touch input, or press input.
[0061] In addition, the specific content of the correction prompt is not limited here. Optionally, any one of the following can be corrected: background description information, identity definition information, task description information, example information, user question information, and output requirement information.
[0062] In this embodiment, if the target intent information is misclassified, the corrected prompt word can be input as a new prompt word into the intent classification agent for another intent classification. In this way, by performing intent classification again, the accuracy of the target intent information output by the intent classification agent can be further improved.
[0063] As an optional implementation, determining the modified prompt word as the next input to the intent classification agent, and returning to the step of inputting the prompt word into the intent classification agent for intent classification to obtain target intent information, includes:
[0064] Determine the number of corrections required for the prompt word;
[0065] If the number of corrections is less than a preset value, the corrected prompt word is determined as the next input of the intent classification agent, and the process returns to inputting the prompt word into the intent classification agent for intent classification to obtain the target intent information.
[0066] In this embodiment, when the number of corrections to the prompt word is less than a preset value, it indicates that the number of times the intent classification agent re-classifies the intent is also less than the preset value. Thus, the intent classification agent only re-classifies the intent again when the number of times it re-classifies the intent is also less than the preset value. This avoids the intent classification agent from entering an infinite loop of intent classification, thereby strictly limiting the number of times the intent classification agent outputs, reducing the illusion of a large language model in the intent classification agent, improving the accuracy of the target intent information output by the intent classification agent, and realizing accurate understanding and classification of user intent in natural language interaction mode.
[0067] It should be noted that when the number of corrections is greater than or equal to the preset value, the prompt word corresponding to the number of corrections = preset value - 1 can be directly input into the intent classification agent for intent classification. The obtained target intent information is then used as the final intent classification result.
[0068] As an optional implementation, the step of inputting at least a portion of the prompt word and the target intent information into the intent inspection agent for intent classification inspection to obtain intent classification result inspection information includes:
[0069] Acquire memory information, which includes the historical prompt words input by the user and the historical intent output by the intent classification agent in response to the historical prompt words;
[0070] At least a portion of the prompt words, the memory information, and the target intent information are input into the intent checking agent to perform intent classification checking, thereby obtaining intent classification result checking information.
[0071] Among these, memory information can be stored as contextual features in the intent-checking agent, and such as Figure 4 As shown, at least some of the prompts input into the intent checking agent may include background description information, identity definition information, task description information, output requirement information, and details of the preceding question.
[0072] The details of the preceding questions can be obtained from the user's questions. Optionally, the user's questions can be linked and connected with the target intent information, so that the large language model in the intent checking agent can understand the user's specific needs and the decisions made by the intent classification agent. This allows the intent checking agent to make accurate evaluations and judgments on the decisions made by the intent classification agent.
[0073] It should be noted that the large language model in the intent checking agent and the large language model in the intent classification agent can be different models. The large language model in the intent checking agent is mainly used for intent classification checking, while the large language model in the intent classification agent is mainly used for intent classification.
[0074] Optionally, see Figure 4 , Figure 4 This is a schematic diagram of an intent checking agent. After at least some of the memory information, target intent information, and prompt words are input into the intent checking agent, the intent checking agent can perform planning processing and action processing respectively. The planning processing can be understood as the thinking and decision-making process made by the large language model in the intent checking agent in order to achieve the purpose of intent classification and checking, while the action processing can be understood as the process of intent classification and checking of the target intent information by the large language model in the intent checking agent.
[0075] It should be noted that the large language model in the intent inspection agent can utilize powerful semantic understanding, knowledge implication, and reasoning capabilities to accurately determine whether the target intent information is correctly classified. This enables accurate understanding and correction of user intent in natural language interaction mode, further improving the accuracy of intent classification results in the intent classification agent.
[0076] In this embodiment, at least a portion of the memory information, target intent information, and prompt words are input into the intent checking agent to perform intent classification checking, thereby obtaining intent classification result checking information. This makes the intent checking agent more accurate in performing intent classification checking, and further improves the accuracy of the determined intent classification result checking information.
[0077] It should be noted that, in order to more fully illustrate the above embodiments, specific embodiments are described below, which can be found in [reference needed]. Figure 2 The flowchart shown is as follows. Figure 2 This is a flowchart illustrating the communication network operation and maintenance intent classification method provided in the embodiments of this application, as shown below. Figure 2 As shown, Figure 2The user question in the prompt can be understood as the user question information in the prompt, while the operation type, including generating SQL, generating Python, and generating shell, can refer to the target intent information in the above embodiment. The limit on the number of checks can refer to whether the number of times the prompt is corrected exceeds a preset threshold, and the classification result can be understood as the intent classification result in the above embodiment.
[0078] Example 1
[0079] 1. The user's question is: The upgrade time is December 12, 2023, at 15:00. Please help me analyze the average value of the QoS flow establishment success rate after the upgrade. According to... Figure 2 The process shown in the diagram involves the user question first being categorized by an intent classification agent. The intent classification agent categorizes the user's intent as "1" based on the user question, other content in the prompts, and strictly follows the output format.
[0080] 2. Category 1 corresponds to "Generate SQL" in the operation categories. The intent checking agent based on the large language model and intelligent agent technology will take at least part of the prompt words, such as the user question and the output of the intent classification agent, as input, and combine it with the background information in the prompt words to determine that the current classification result "satisfies" the user intent. That is, the intent classification result checking information indicates that the target intent information is classified correctly.
[0081] 3. Because it satisfies the user's intent, the operation corresponding to the intent category is "generating SQL" as the final output of the intent classification agent. This final output can also be understood as the final output corresponding to the user's question in the prompt.
[0082] Example 2
[0083] 1. The user's input question is: For the KPI file "Indicator Table.xlsx", analyze the average "QoS Flow Establishment Success Rate" across different sites and present it in a bar chart format, according to... Figure 2 The process shown in the diagram involves the user question first being categorized by an intent classification agent. The intent classification agent categorizes the intent based on the user question and, according to the output format in the prompt, classifies the user's intent as "1".
[0084] 2. Category 1 corresponds to "Generate SQL" in the operation categories. The intent checking agent based on the large language model and intelligent agent technology will take at least part of the prompt words, such as the user question and the output of the intent classification agent, as input, and combine it with the background information in the prompt words to judge the current intent classification result. Since the SQL language can calculate the average value but cannot perform graphing operations, the intent checking agent cannot judge the user question through tool calls. Therefore, the intent checking agent judges the current classification result as "not satisfying" the user intent, that is, the intent classification result check information indicates that the target intent information is incorrectly classified.
[0085] 3. When the intent checking agent finds that the classification result of the intent classification agent does not meet the user's intent, it will check the number of checks. This number of checks can also be understood as the number of times the prompt words are corrected. When the number of checks is 0, which is less than the limit of the number of checks, i.e. less than the preset value, the user's question and the check result of the intent checking agent (i.e., the intent classification result check information) can be returned to the intent classification agent.
[0086] 4. The intent classification agent re-determines the category based on the user's question and the intent checking agent's results. This time, the user's intent is classified as "2".
[0087] 5. Category 2 corresponds to "Generate Python" in the operation category. The intent checking agent based on the large language model and agent technology will take at least part of the prompt words, such as the user question and the output of the intent classification agent, as input, and combine it with the background information in the prompt words and the previous judgment results in the memory information to judge the current intent classification result. Since the user question can be satisfied by writing simple Python code, the judgment result of the intent checking agent is "satisfy" the user intent.
[0088] 6. Because it satisfies the user's intent, the operation corresponding to the intent category, "Generate Python", is taken as the final output of the intent classification agent. This final output can also be understood as the final output corresponding to the user's question in the prompt.
[0089] Example 3
[0090] 1. The user's input question was: Run the script program with default configuration parameters. (Following...) Figure 2 The process shown in the diagram involves the user's question first being categorized by an intent classification agent. Based on the user's question and other content included in the prompt, the intent classification agent categorizes the user's intent as "3".
[0091] Category 2, Category 3 corresponds to "Generate Shell" in the operation categories. The intent checking agent based on the large language model and agent technology will take at least part of the prompt words, such as the user question and the output of the intent classification agent, as input, and combine it with the background information in the prompt words to determine whether the current classification result "satisfies" the user's intent.
[0092] 3. Because it satisfies the user's intent, the operation corresponding to the intent category, "generate shell", is taken as the final output of the intent classification agent. This final output can also be understood as the final output corresponding to the user's question in the prompt.
[0093] See Figure 5 , Figure 5 This is a structural diagram of the communication network operation and maintenance intent classification device 500 provided in the embodiments of this application. The communication network operation and maintenance intent classification device 500 includes:
[0094] The acquisition module 501 is used to acquire the prompt words input by the user;
[0095] The intent classification module 502 is used to input the prompt words into the intent classification agent to classify the intent and obtain target intent information. The target intent information is used to represent the network operation and maintenance code generated using the target language.
[0096] The intent classification check module 503 is used to input at least part of the content of the prompt word and the target intent information into the intent check agent to perform intent classification check and obtain intent classification result check information.
[0097] The determination module 504 is used to determine the target intent information as the intent classification result of the prompt word when the intent classification result check information indicates that the target intent information is correctly classified.
[0098] As an optional implementation, the communication network operation and maintenance intention classification device 500 further includes:
[0099] The correction module is used to correct the prompt word when the intent classification result check information indicates that the target intent information is classified incorrectly.
[0100] The return module is used to determine the corrected prompt word as the next input of the intent classification agent, and return to execute the step of inputting the prompt word into the intent classification agent for intent classification to obtain target intent information.
[0101] As an optional implementation, the return module includes:
[0102] A determination submodule is used to determine the number of corrections for the prompt word;
[0103] The return submodule is used to determine the corrected prompt word as the next input of the intent classification agent when the number of corrections is less than a preset value, and return to execute the step of inputting the prompt word into the intent classification agent for intent classification to obtain target intent information.
[0104] As an optional implementation, the intent classification check module includes:
[0105] The acquisition submodule is used to acquire memory information, which includes the historical prompt words input by the user and the historical intent output by the intent classification agent in response to the historical prompt words;
[0106] The classification check submodule is used to input at least part of the content of the prompt words, the memory information and the target intent information into the intent check agent to perform intent classification check and obtain intent classification result check information.
[0107] As an optional implementation, the target intent information is an intent classification index.
[0108] As an optional implementation, the prompt words include at least one of the following: background description information, identity definition information, task description information, example information, user question information, and output requirement information.
[0109] The communication network operation and maintenance intention classification device 500 can achieve the following in the embodiments of this application: Figure 1 The various processes in the method embodiments, and the ways to achieve the same beneficial effects, will not be repeated here to avoid repetition.
[0110] This application also provides an electronic device. Please refer to [link to relevant documentation]. Figure 6 The electronic device may include a processor 601, a memory 602, and a program 6021 stored in the memory 602 and executable on the processor 601. When the program 6021 is executed by the processor 601, it can achieve... Figure 1 Any steps in the corresponding method embodiments and the achievement of the same beneficial effects will not be repeated here.
[0111] Those skilled in the art will understand that all or part of the steps of the methods described in the above embodiments can be implemented by hardware related to program instructions, and the program can be stored in a readable medium. This application also provides a readable storage medium storing a computer program, which, when executed by a processor, can implement the above-described methods. Figure 1 Any step in the corresponding method embodiment can achieve the same technical effect, and will not be repeated here to avoid repetition.
[0112] The storage medium may be a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.
[0113] This application also provides a computer program product, including computer instructions, which, when executed by a processor, can perform the above-described functions. Figure 1 Any step in the corresponding method embodiment can achieve the same technical effect, and will not be repeated here to avoid repetition.
[0114] The above description represents the preferred embodiments of this application. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles described in this application, and these improvements and modifications should also be considered within the scope of protection of this application.
Claims
1. A method for classifying operation and maintenance intents of a communication network, characterized in that, include: Get the suggestion words entered by the user; The prompt words are input into the intent classification agent for intent classification to obtain target intent information, which is used to represent the network operation and maintenance code generated using the target language. At least a portion of the prompt words and the target intent information are input into the intent checking agent to perform intent classification checking, and intent classification result checking information is obtained. If the intent classification result check information indicates that the target intent information is correctly classified, the target intent information is determined as the intent classification result of the prompt word.
2. The method of claim 1, wherein, After inputting at least a portion of the prompt words and the target intent information into the intent inspection agent for intent classification inspection, and obtaining intent classification result inspection information, the method further includes: If the intent classification result check information indicates that the target intent information is incorrectly classified, the prompt word is corrected; The corrected prompt is determined as the next input to the intent classification agent, and the process returns to inputting the prompt into the intent classification agent for intent classification to obtain the target intent information.
3. The method of claim 2, wherein, The step of determining the corrected prompt as the next input to the intent classification agent and returning to the step of inputting the prompt into the intent classification agent for intent classification to obtain target intent information includes: Determine the number of corrections required for the prompt word; If the number of corrections is less than a preset value, the corrected prompt word is determined as the next input of the intent classification agent, and the process returns to inputting the prompt word into the intent classification agent for intent classification to obtain the target intent information.
4. The method of claim 1, wherein, The step of inputting at least a portion of the prompt words and the target intent information into the intent inspection agent for intent classification inspection, and obtaining intent classification result inspection information, includes: Acquire memory information, which includes the historical prompt words input by the user and the historical intent output by the intent classification agent in response to the historical prompt words; At least a portion of the prompt words, the memory information, and the target intent information are input into the intent checking agent to perform intent classification checking, thereby obtaining intent classification result checking information.
5. The method according to any one of claims 1 to 4, characterized in that, The target intent information is an intent classification index.
6. The method according to any one of claims 1 to 4, characterized in that, The prompts include at least one of the following: background description information, identity definition information, task description information, example information, user question information, and output requirement information. 7.A device for classifying operation and maintenance intents of a communication network, characterized in that, include: The acquisition module is used to acquire the prompt words entered by the user; The intent classification module is used to input the prompt words into the intent classification agent to classify the intent and obtain the target intent information. The target intent information is used to represent the network operation and maintenance code generated using the target language. The intent classification check module is used to input at least part of the prompt words and the target intent information into the intent check agent to perform intent classification check and obtain intent classification result check information. The determination module is used to determine the target intent information as the intent classification result of the prompt word when the intent classification result check information indicates that the target intent information is correctly classified.
8. An electronic device comprising: A memory, a processor, and a program stored in the memory and executable on the processor; characterized in that the processor is configured to read the program in the memory to implement the steps in the communication network operation and maintenance intent classification method as described in any one of claims 1 to 6.
9. A readable storage medium for storing a program, characterized in that, When the program is executed by the processor, it implements the steps in the communication network operation and maintenance intent classification method as described in any one of claims 1 to 6.
10. A computer program product, characterised in that, It includes computer instructions, which, when executed by a processor, implement the steps in the communication network operation and maintenance intent classification method as described in any one of claims 1 to 6.
