Operation and maintenance question and answer data processing method and system based on large language model

Abstract

The present invention discloses a method and system for processing operation and maintenance question and answer data based on a large language model, the method comprising: obtaining operation and maintenance question data and user parameters of a target operation and maintenance user; inputting the operation and maintenance question into a trained large language model to obtain direct answer data and reference data corresponding to the operation and maintenance question; predicting answer fine-tuning parameters based on the user parameters and the reference data and based on a neural network algorithm; inputting the answer fine-tuning parameters and the direct answer data into the large language model to obtain corrected answer data; the corrected answer data is used to be sent to the terminal of the target operation and maintenance user for display. It can be seen that the present invention can provide operation and maintenance question and answer services to operation and maintenance users more accurately and efficiently, improve the efficiency and convenience of operation and maintenance work, and reduce operation and maintenance costs and operation and maintenance errors.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a method and system for processing operation and maintenance question and answer data based on a large language model. Background Art

[0002] In the field of modern information technology, operations and maintenance play a vital role. It is a key link in ensuring the normal operation of information technology systems, covering aspects such as server management, application deployment, performance monitoring, and troubleshooting. With the continuous advancement of technology and the increasing complexity of business, the importance of operations and maintenance has become increasingly prominent. This has also greatly raised the threshold for operations and maintenance work, requiring operations and maintenance personnel to possess knowledge and skills in systems management, network management, and security management. Existing technologies have proposed standard operations and maintenance information database technologies to meet the data search needs of operations and maintenance personnel, but this technology cannot achieve more intelligent operations and maintenance Q&A, nor can it provide efficient and automated operations and maintenance Q&A to improve the work efficiency of operations and maintenance personnel. It can be seen that existing technologies have shortcomings that need to be addressed urgently. Summary of the Invention

[0003] The technical problem to be solved by the present invention is to provide an operation and maintenance question and answer data processing method and system based on a large language model, which can provide operation and maintenance question and answer services to operation and maintenance users more accurately and efficiently, improve the efficiency and convenience of operation and maintenance work, and reduce operation and maintenance costs and operation and maintenance errors.

[0004] In order to solve the above technical problems, the first aspect of the present invention discloses a method for processing operation and maintenance question and answer data based on a large language model, the method comprising:

[0005] Obtain the operation and maintenance problem data and user parameters of the target operation and maintenance user;

[0006] Input the operation and maintenance question into the trained large language model to obtain direct answer data and reference data corresponding to the operation and maintenance question;

[0007] According to the user parameters and the reference data, based on a neural network algorithm, predicting answer fine-tuning parameters;

[0008] The answer fine-tuning parameters and the direct answer data are input into the large language model to obtain modified answer data; the modified answer data is sent to the terminal of the target operation and maintenance user for display.

[0009] As an optional embodiment, in the first aspect of the present invention, the large language model is trained by the following steps:

[0010] Collecting operation and maintenance standard data and operation and maintenance question and answer template data, cleaning and splitting the operation and maintenance standard data and operation and maintenance question and answer template data, extracting valid question and answer pair data, performing part-of-speech analysis and entity naming task processing on the question and answer pair data, and storing the processed question and answer pair data in a question and answer pair database;

[0011] Noun entities and verb entities are extracted from the question-answer pair database, and related clustering data is analyzed after vectorization conversion of the noun entities and verb entities to obtain multiple sets of mapping pairs. The understanding ability of the PLM pre-trained large language model is evaluated through the multiple sets of mapping pairs, and a knowledge dialogue set is constructed based on the mapping pairs with negative evaluation results; manual answer fine-tuning parameter annotation is performed on the question-answer pair database, and a task dialogue set is constructed based on the question-answer pair data and the manual answer fine-tuning parameter annotation;

[0012] The training corpus and test corpus are constructed through the knowledge dialogue set and the task dialogue set. The training corpus is input into the large language model pre-trained by PLM for instruction fine-tuning, and the test corpus is input into the large language model to verify the accuracy. This process is repeated until the verification result of the large language model meets the preset requirements to obtain a trained large language model.

[0013] As an optional embodiment, in the first aspect of the present invention, the answer fine-tuning parameters or the manual answer fine-tuning parameter annotations include at least one of an answer tone parameter, an answer information amount parameter, an answer citation material amount parameter and an answer text style parameter.

[0014] As an optional implementation manner, in the first aspect of the present invention, the user parameters include user position, user department, user working experience, user gender and user educational background.

[0015] As an optional embodiment, in the first aspect of the present invention, predicting the answer fine-tuning parameters based on the user parameters and the reference data and a neural network algorithm includes:

[0016] Based on the user parameters, multiple similar user historical operation and maintenance question and answer records are screened from multiple candidate user historical operation and maintenance question and answer records;

[0017] All the similar user historical operation and maintenance question and answer records and the reference data are input into the trained fine-tuning parameter prediction neural network to obtain the output answer fine-tuning parameters.

[0018] As an optional embodiment, in the first aspect of the present invention, the filtering out of a plurality of similar user historical operation and maintenance question and answer records from a plurality of candidate user historical operation and maintenance question and answer records based on the user parameters includes:

[0019] For each candidate user's historical operation and maintenance question and answer record, obtain the candidate user parameters and multiple question and answer record texts corresponding to the candidate user's historical operation and maintenance question and answer record;

[0020] Calculating a first similarity between the candidate user parameter and the user parameter;

[0021] Calculating an average of similarities between all the question and answer record texts and the current question and answer data to obtain a second similarity; the current answer data includes the operation and maintenance question data and the direct answer data;

[0022] Calculate the product of the first similarity and the second similarity to obtain the record parameters corresponding to the candidate user's historical operation and maintenance question and answer record;

[0023] Sort all the candidate user's historical operation and maintenance question and answer records from largest to smallest according to the record parameters to obtain a record sequence;

[0024] All the candidate user historical operation and maintenance question and answer records whose first preset number in the record sequence and whose record parameters are greater than a parameter threshold are screened out to obtain a plurality of similar user historical operation and maintenance question and answer records.

[0025] As an optional embodiment, in the first aspect of the present invention, the fine-tuning parameter prediction neural network is a CNN neural network, which is trained by a training data set including a plurality of training user operation and maintenance question and answer record data and corresponding reference material annotations and answer fine-tuning parameter annotations, and is trained based on the cross entropy loss function and gradient descent algorithm until convergence.

[0026] As an optional embodiment, in the first aspect of the present invention, inputting the answer fine-tuning parameters and the direct answer data into the large language model to obtain the modified answer data includes:

[0027] Based on the preset language organization rules, determining the fine-tuning command sentence according to the answer fine-tuning parameters;

[0028] Combining the direct answer data and the fine-tuning command sentence into an input command sentence;

[0029] The input command sentence is input to the large language model to obtain corrected answer data output by the large language model.

[0030] A second aspect of an embodiment of the present invention discloses an operation and maintenance question and answer data processing system based on a large language model, the system comprising:

[0031] The acquisition module is used to obtain the operation and maintenance problem data and user parameters of the target operation and maintenance user;

[0032] An input module, configured to input the operation and maintenance question into a trained large language model to obtain direct answer data and reference data corresponding to the operation and maintenance question;

[0033] A prediction module, configured to predict answer fine-tuning parameters based on the user parameters and the reference data and a neural network algorithm;

[0034] A correction module is used to input the answer fine-tuning parameters and the direct answer data into the large language model to obtain corrected answer data; the corrected answer data is used to be sent to the terminal of the target operation and maintenance user for display.

[0035] As an optional implementation, in the second aspect of the present invention, the large language model is trained by the following steps:

[0036] Collecting operation and maintenance standard data and operation and maintenance question and answer template data, cleaning and splitting the operation and maintenance standard data and operation and maintenance question and answer template data, extracting valid question and answer pair data, performing part-of-speech analysis and entity naming task processing on the question and answer pair data, and storing the processed question and answer pair data in a question and answer pair database;

[0037] Noun entities and verb entities are extracted from the question-answer pair database, and related clustering data is analyzed after vectorization conversion of the noun entities and verb entities to obtain multiple sets of mapping pairs. The understanding ability of the PLM pre-trained large language model is evaluated through the multiple sets of mapping pairs, and a knowledge dialogue set is constructed based on the mapping pairs with negative evaluation results; manual answer fine-tuning parameter annotation is performed on the question-answer pair database, and a task dialogue set is constructed based on the question-answer pair data and the manual answer fine-tuning parameter annotation;

[0038] The training corpus and test corpus are constructed through the knowledge dialogue set and the task dialogue set. The training corpus is input into the large language model pre-trained by PLM for instruction fine-tuning, and the test corpus is input into the large language model to verify the accuracy. This process is repeated until the verification result of the large language model meets the preset requirements to obtain a trained large language model.

[0039] As an optional embodiment, in the second aspect of the present invention, the answer fine-tuning parameters or the manual answer fine-tuning parameter annotations include at least one of an answer tone parameter, an answer information amount parameter, an answer citation material amount parameter and an answer text style parameter.

[0040] As an optional implementation, in the second aspect of the present invention, the user parameters include user position, user department, user working experience, user gender and user educational background.

[0041] As an optional embodiment, in the second aspect of the present invention, the prediction module predicts a specific method of fine-tuning the answer parameters based on the user parameters and the reference data using a neural network algorithm, including:

[0042] Based on the user parameters, multiple similar user historical operation and maintenance question and answer records are screened from multiple candidate user historical operation and maintenance question and answer records;

[0043] All the similar user historical operation and maintenance question and answer records and the reference data are input into the trained fine-tuning parameter prediction neural network to obtain the output answer fine-tuning parameters.

[0044] As an optional embodiment, in the second aspect of the present invention, the specific manner in which the prediction module selects multiple similar user historical operation and maintenance question and answer records from multiple candidate user historical operation and maintenance question and answer records based on the user parameters includes:

[0045] For each candidate user's historical operation and maintenance question and answer record, obtain the candidate user parameters and multiple question and answer record texts corresponding to the candidate user's historical operation and maintenance question and answer record;

[0046] Calculating a first similarity between the candidate user parameter and the user parameter;

[0047] Calculating an average of similarities between all the question and answer record texts and the current question and answer data to obtain a second similarity; the current answer data includes the operation and maintenance question data and the direct answer data;

[0048] Calculate the product of the first similarity and the second similarity to obtain the record parameters corresponding to the candidate user's historical operation and maintenance question and answer record;

[0049] Sort all the candidate user's historical operation and maintenance question and answer records from largest to smallest according to the record parameters to obtain a record sequence;

[0050] All the candidate user historical operation and maintenance question and answer records whose first preset number in the record sequence and whose record parameters are greater than a parameter threshold are screened out to obtain a plurality of similar user historical operation and maintenance question and answer records.

[0051] As an optional embodiment, in the second aspect of the present invention, the fine-tuning parameter prediction neural network is a CNN neural network, which is trained by a training data set including multiple training user operation and maintenance question and answer record data and corresponding reference material annotations and answer fine-tuning parameter annotations, and is trained based on the cross entropy loss function and gradient descent algorithm until convergence.

[0052] As an optional embodiment, in the second aspect of the present invention, the specific manner in which the correction module inputs the answer fine-tuning parameter and the direct answer data into the large language model to obtain the corrected answer data includes:

[0053] Based on the preset language organization rules, determining the fine-tuning command sentence according to the answer fine-tuning parameters;

[0054] Combining the direct answer data and the fine-tuning command sentence into an input command sentence;

[0055] The input command sentence is input to the large language model to obtain corrected answer data output by the large language model.

[0056] A third aspect of the present invention discloses another operation and maintenance question and answer data processing system based on a large language model, the system comprising:

[0057] a memory storing executable program code;

[0058] a processor coupled to the memory;

[0059] The processor calls the executable program code stored in the memory to execute some or all of the steps in the operation and maintenance question and answer data processing method based on a large language model disclosed in the first aspect of the present invention.

[0060] The fourth aspect of the present invention discloses a computer storage medium, which stores computer instructions. When the computer instructions are called, they are used to execute some or all of the steps in the operation and maintenance question and answer data processing method based on a large language model disclosed in the first aspect of the present invention.

[0061] Compared with the prior art, the embodiments of the present invention have the following beneficial effects:

[0062] The present invention can obtain direct answer data and reference data corresponding to operation and maintenance questions based on the trained large language model, and then predict answer fine-tuning parameters based on user parameters and reference data, so as to correct the direct answer data based on the large language model to obtain more accurate corrected answer data, thereby being able to provide operation and maintenance question and answer services to operation and maintenance users more accurately and efficiently, improve the efficiency and convenience of operation and maintenance work, and reduce operation and maintenance costs and operation and maintenance errors. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0064] Figure 1 This is a flow chart of a method for processing operation and maintenance question and answer data based on a large language model disclosed in an embodiment of the present invention.

[0065] Figure 2 This is a structural diagram of an operation and maintenance question and answer data processing system based on a large language model disclosed in an embodiment of the present invention.

[0066] Figure 3 This is a structural diagram of another operation and maintenance question and answer data processing system based on a large language model disclosed in an embodiment of the present invention. DETAILED DESCRIPTION

[0067] In order to enable those skilled in the art to better understand the solutions of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0068] The terms "first," "second," and so on, in the description and claims of the present invention and the accompanying drawings are used to distinguish between different objects, not to describe a specific order. Furthermore, the terms "including," "having," and any variations thereof, are intended to cover non-exclusive inclusions. For example, a process, method, apparatus, product, or device comprising a series of steps or elements is not limited to the listed steps or elements but may optionally include steps or elements not listed, or may optionally include other steps or elements inherent to the process, method, product, or device.

[0069] References herein to "embodiments" mean that a particular feature, structure, or characteristic described in connection with the embodiments may be included in at least one embodiment of the present invention. The appearance of this phrase in various places in the specification does not necessarily refer to the same embodiment, nor does it constitute a separate or alternative embodiment that is mutually exclusive of other embodiments. It is understood, both explicitly and implicitly, by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0070] The present invention discloses a large language model-based operation and maintenance question and answer data processing method and system. This method can obtain direct answer data and reference data corresponding to operation and maintenance questions based on a trained large language model. It then predicts answer fine-tuning parameters based on user parameters and reference data, and uses the large language model to correct the direct answer data to obtain more accurate corrected answer data. This method can provide operation and maintenance question and answer services to operation and maintenance users more accurately and efficiently, improving the efficiency and convenience of operation and maintenance work, and reducing operation and maintenance costs and errors. These are described in detail below.

[0071] Example 1

[0072] See also Figure 1 , Figure 1 This is a flow chart of a method for processing operation and maintenance question and answer data based on a large language model disclosed in an embodiment of the present invention. Figure 1 The described operation and maintenance question and answer data processing method based on a large language model can be applied to a data processing system / data processing device / data processing server (wherein the server includes a local processing server or a cloud processing server). Figure 1 As shown, the operation and maintenance question and answer data processing method based on the large language model may include the following operations:

[0073] 101. Obtain the operation and maintenance problem data and user parameters of the target operation and maintenance user.

[0074] 102. Input the operation and maintenance questions into the trained large language model to obtain direct answer data and reference data corresponding to the operation and maintenance questions.

[0075] 103. Based on user parameters and reference data, the answer fine-tuning parameters are predicted based on the neural network algorithm.

[0076] 104. Input the answer fine-tuning parameters and the direct answer data into the large language model to obtain the corrected answer data.

[0077] Optionally, the corrected answer data is sent to a terminal of a target operation and maintenance user for display.

[0078] It can be seen that the above-mentioned embodiment of the invention can obtain direct answer data and reference data corresponding to operation and maintenance questions based on the trained large language model, and then predict the answer fine-tuning parameters based on user parameters and reference data, so as to correct the direct answer data based on the large language model to obtain more accurate corrected answer data, thereby being able to provide operation and maintenance question and answer services to operation and maintenance users more accurately and efficiently, improve the efficiency and convenience of operation and maintenance work, and reduce operation and maintenance costs and operation and maintenance errors.

[0079] As an optional embodiment, in the above steps, the large language model is trained by the following steps:

[0080] Collect standard operation and maintenance data and operation and maintenance question and answer template data, clean and split the data, extract valid question and answer pair data, perform part-of-speech analysis and entity naming task processing on the question and answer pair data, and store the processed question and answer pair data in the question and answer pair database;

[0081] Extract noun and verb entities from the question-answer pair database, vectorize them, and analyze related clustering data to obtain multiple mapping pairs. Use these mapping pairs to evaluate the understanding ability of the PLM pre-trained large language model. Build a knowledge dialogue set based on mapping pairs with negative evaluation results. Optionally, manually fine-tune parameters for the question-answer pair database, and construct a task dialogue set based on the question-answer pair data and the manually fine-tuned parameter annotations.

[0082] The training corpus and test corpus are constructed through the knowledge dialogue set and the task dialogue set. The training corpus is input into the large language model pre-trained by PLM for instruction fine-tuning, and the test corpus is input into the large language model to verify the accuracy. This process is repeated until the verification result of the large language model meets the preset requirements, and a trained large language model is obtained.

[0083] It can be seen that through the above optional embodiments, the training details of the large language model are limited, and the large language model can be fully trained based on the collected operation and maintenance data and manual answer fine-tuning parameter annotations to obtain a model that can accurately predict answer data and correct answer data, thereby helping to provide operation and maintenance question and answer services to operation and maintenance users more accurately and efficiently, improve the efficiency and convenience of operation and maintenance work, and reduce operation and maintenance costs and errors.

[0084] As an optional embodiment, in the above steps, the answer fine-tuning parameters or manual answer fine-tuning parameter annotations include at least one of the answer tone parameters, answer information volume parameters, answer citation material volume parameters and answer text style parameters.

[0085] It can be seen that through the above optional embodiments, the answer fine-tuning parameters or the content of the answer fine-tuning parameter annotations are limited, which can more accurately characterize the characteristics of the answers that users may be more acceptable to, so as to obtain more accurate corrected answer data in the future, and assist in providing operation and maintenance question and answer services to operation and maintenance users more accurately and efficiently, improve the efficiency and convenience of operation and maintenance work, and reduce operation and maintenance costs and operation and maintenance errors.

[0086] As an optional embodiment, in the above steps, the user parameters include user position, user department, user working experience, user gender and user education background.

[0087] It can be seen that through the above optional embodiments, the content of user parameters is limited, which can fully characterize the user's characteristics, so as to facilitate the subsequent accurate prediction of fine-tuning parameters, and assist in providing operation and maintenance question and answer services to operation and maintenance users more accurately and efficiently, thereby improving the efficiency and convenience of operation and maintenance work, and reducing operation and maintenance costs and operation and maintenance errors.

[0088] As an optional embodiment, in the above steps, based on the user parameters and reference data, and based on the neural network algorithm, predicting the answer fine-tuning parameters includes:

[0089] Based on user parameters, multiple similar user historical operation and maintenance question and answer records are screened from multiple candidate user historical operation and maintenance question and answer records;

[0090] All similar user historical operation and maintenance question and answer records and reference data are input into the trained fine-tuning parameter prediction neural network to obtain the output answer fine-tuning parameters.

[0091] It can be seen that through the above optional embodiments, multiple similar user historical operation and maintenance question and answer records can be screened out from multiple candidate user historical operation and maintenance question and answer records based on user parameters, so as to accurately predict fine-tuning parameters based on the trained fine-tuning parameter prediction neural network combined with similar user question and answer records and reference data, thereby assisting in providing operation and maintenance question and answer services to operation and maintenance users more accurately and efficiently, improving the efficiency and convenience of operation and maintenance work, and reducing operation and maintenance costs and operation and maintenance errors.

[0092] As an optional embodiment, in the above step, based on user parameters, screening out multiple similar user historical operation and maintenance question and answer records from multiple candidate user historical operation and maintenance question and answer records includes:

[0093] For each candidate user's historical operation and maintenance question and answer record, obtain the candidate user parameters and multiple question and answer record texts corresponding to the candidate user's historical operation and maintenance question and answer record;

[0094] Calculating a first similarity between the candidate user parameters and the user parameters;

[0095] Calculate the average of the similarities between all question and answer record texts and the current question and answer data to obtain a second similarity; optionally, the current answer data includes operation and maintenance question data and direct answer data;

[0096] Calculate the product of the first similarity and the second similarity to obtain the record parameters corresponding to the candidate user's historical operation and maintenance question and answer record;

[0097] Sort all candidate user historical operation and maintenance question and answer records from largest to smallest according to the record parameters to obtain a record sequence;

[0098] All candidate user historical operation and maintenance question and answer records whose first preset number in the record sequence and whose record parameters are greater than the parameter threshold are screened out to obtain multiple similar user historical operation and maintenance question and answer records.

[0099] It can be seen that through the above optional embodiments, user question and answer records can be screened by comparing the similarity between the user parameters and question and answer record text in the candidate user's historical operation and maintenance question and answer records and the current user parameters and question and answer data, so as to facilitate the subsequent accurate prediction and fine-tuning of parameters, and assist in providing operation and maintenance question and answer services to operation and maintenance users more accurately and efficiently, improve the efficiency and convenience of operation and maintenance work, and reduce operation and maintenance costs and operation and maintenance errors.

[0100] As an optional embodiment, in the above steps, the fine-tuning parameter prediction neural network is a CNN neural network, which is trained by a training data set including a plurality of training user operation and maintenance question and answer record data and corresponding reference material annotations and answer fine-tuning parameter annotations, and is trained based on the cross entropy loss function and gradient descent algorithm until convergence.

[0101] It can be seen that through the above optional embodiments, the network details and training details of the fine-tuning parameter prediction neural network are defined to accurately predict the answer fine-tuning parameters, assisting in providing operation and maintenance question and answer services to operation and maintenance users more accurately and efficiently, improving the efficiency and convenience of operation and maintenance work, and reducing operation and maintenance costs and operation and maintenance errors.

[0102] As an optional embodiment, in the above step, inputting the answer fine-tuning parameters and the direct answer data into the large language model to obtain the modified answer data includes:

[0103] Based on the preset language organization rules, a fine-tuning command statement is determined according to the answer fine-tuning parameters;

[0104] The direct answer data and the fine-tuning command statement are combined into an input command statement;

[0105] An input command sentence is input to the large language model to obtain corrected answer data output by the large language model.

[0106] It can be seen that through the above-mentioned optional embodiments, the fine-tuning command statement can be determined according to the answer fine-tuning parameters through the preset language organization rules, so as to combine the direct answer data to form the input command statement, so that the large language model can directly understand and execute the fine-tuning command, obtain more accurate corrected answer data, and provide operation and maintenance question and answer services to operation and maintenance users more accurately and efficiently, improve the efficiency and convenience of operation and maintenance work, and reduce operation and maintenance costs and operation and maintenance errors.

[0107] Example 2

[0108] See also Figure 2 , Figure 2This is a schematic diagram of the structure of an operation and maintenance question and answer data processing system based on a large language model disclosed in an embodiment of the present invention. Figure 2 The described operation and maintenance question-answering data processing system based on a large language model can be applied to a data processing system / data processing device / data processing server (wherein the server includes a local processing server or a cloud processing server). Figure 2 As shown, the operation and maintenance question and answer data processing system based on the large language model may include:

[0109] The acquisition module 201 is used to acquire the operation and maintenance problem data and user parameters of the target operation and maintenance user.

[0110] The input module 202 is used to input the operation and maintenance questions into the trained large language model to obtain direct answer data and reference data corresponding to the operation and maintenance questions.

[0111] The prediction module 203 is used to predict the answer fine-tuning parameters based on the user parameters and reference data and the neural network algorithm.

[0112] The correction module 204 is used to input the answer fine-tuning parameters and the direct answer data into the large language model to obtain the corrected answer data.

[0113] Optionally, the corrected answer data is sent to a terminal of a target operation and maintenance user for display.

[0114] It can be seen that the above-mentioned embodiment of the invention can obtain direct answer data and reference data corresponding to operation and maintenance questions based on the trained large language model, and then predict the answer fine-tuning parameters based on user parameters and reference data, so as to correct the direct answer data based on the large language model to obtain more accurate corrected answer data, thereby being able to provide operation and maintenance question and answer services to operation and maintenance users more accurately and efficiently, improve the efficiency and convenience of operation and maintenance work, and reduce operation and maintenance costs and operation and maintenance errors.

[0115] As an optional embodiment, the large language model is trained by the following steps:

[0116] Collect standard operation and maintenance data and operation and maintenance question and answer template data, clean and split the data, extract valid question and answer pair data, perform part-of-speech analysis and entity naming task processing on the question and answer pair data, and store the processed question and answer pair data in the question and answer pair database;

[0117] Extract noun and verb entities from the question-answer pair database, vectorize them, and analyze related clustering data to obtain multiple mapping pairs. Use these mapping pairs to evaluate the understanding ability of the PLM pre-trained large language model. Build a knowledge dialogue set based on mapping pairs with negative evaluation results. Optionally, manually fine-tune parameters for the question-answer pair database, and construct a task dialogue set based on the question-answer pair data and the manually fine-tuned parameter annotations.

[0118] The training corpus and test corpus are constructed through the knowledge dialogue set and the task dialogue set. The training corpus is input into the large language model pre-trained by PLM for instruction fine-tuning, and the test corpus is input into the large language model to verify the accuracy. This process is repeated until the verification result of the large language model meets the preset requirements, and a trained large language model is obtained.

[0119] It can be seen that through the above optional embodiments, the training details of the large language model are limited, and the large language model can be fully trained based on the collected operation and maintenance data and manual answer fine-tuning parameter annotations to obtain a model that can accurately predict answer data and correct answer data, thereby helping to provide operation and maintenance question and answer services to operation and maintenance users more accurately and efficiently, improve the efficiency and convenience of operation and maintenance work, and reduce operation and maintenance costs and errors.

[0120] As an optional embodiment, the answer fine-tuning parameters or manual answer fine-tuning parameter annotations include at least one of an answer tone parameter, an answer information volume parameter, an answer citation material volume parameter, and an answer text style parameter.

[0121] It can be seen that through the above optional embodiments, the answer fine-tuning parameters or the content of the answer fine-tuning parameter annotations are limited, which can more accurately characterize the characteristics of the answers that users may be more acceptable to, so as to obtain more accurate corrected answer data in the future, and assist in providing operation and maintenance question and answer services to operation and maintenance users more accurately and efficiently, improve the efficiency and convenience of operation and maintenance work, and reduce operation and maintenance costs and operation and maintenance errors.

[0122] As an optional embodiment, the user parameters include user position, user department, user working experience, user gender and user education background.

[0123] It can be seen that through the above optional embodiments, the content of user parameters is limited, which can fully characterize the user's characteristics, so as to facilitate the subsequent accurate prediction of fine-tuning parameters, and assist in providing operation and maintenance question and answer services to operation and maintenance users more accurately and efficiently, thereby improving the efficiency and convenience of operation and maintenance work, and reducing operation and maintenance costs and operation and maintenance errors.

[0124] As an optional embodiment, the prediction module predicts the specific method of fine-tuning the answer parameters based on the user parameters and reference data using a neural network algorithm, including:

[0125] Based on user parameters, multiple similar user historical operation and maintenance question and answer records are screened from multiple candidate user historical operation and maintenance question and answer records;

[0126] All similar user historical operation and maintenance question and answer records and reference data are input into the trained fine-tuning parameter prediction neural network to obtain the output answer fine-tuning parameters.

[0127] It can be seen that through the above optional embodiments, multiple similar user historical operation and maintenance question and answer records can be screened out from multiple candidate user historical operation and maintenance question and answer records based on user parameters, so as to accurately predict fine-tuning parameters based on the trained fine-tuning parameter prediction neural network combined with similar user question and answer records and reference data, thereby assisting in providing operation and maintenance question and answer services to operation and maintenance users more accurately and efficiently, improving the efficiency and convenience of operation and maintenance work, and reducing operation and maintenance costs and operation and maintenance errors.

[0128] As an optional embodiment, the specific method of the prediction module screening multiple similar user historical operation and maintenance question and answer records from multiple candidate user historical operation and maintenance question and answer records based on user parameters includes:

[0129] For each candidate user's historical operation and maintenance question and answer record, obtain the candidate user parameters and multiple question and answer record texts corresponding to the candidate user's historical operation and maintenance question and answer record;

[0130] Calculating a first similarity between the candidate user parameters and the user parameters;

[0131] Calculate the average of the similarities between all question and answer record texts and the current question and answer data to obtain a second similarity; optionally, the current answer data includes operation and maintenance question data and direct answer data;

[0132] Calculate the product of the first similarity and the second similarity to obtain the record parameters corresponding to the candidate user's historical operation and maintenance question and answer record;

[0133] Sort all candidate user historical operation and maintenance question and answer records from largest to smallest according to the record parameters to obtain a record sequence;

[0134] All candidate user historical operation and maintenance question and answer records whose first preset number in the record sequence and whose record parameters are greater than the parameter threshold are screened out to obtain multiple similar user historical operation and maintenance question and answer records.

[0135] It can be seen that through the above optional embodiments, user question and answer records can be screened by comparing the similarity between the user parameters and question and answer record text in the candidate user's historical operation and maintenance question and answer records and the current user parameters and question and answer data, so as to facilitate the subsequent accurate prediction and fine-tuning of parameters, and assist in providing operation and maintenance question and answer services to operation and maintenance users more accurately and efficiently, improve the efficiency and convenience of operation and maintenance work, and reduce operation and maintenance costs and operation and maintenance errors.

[0136] As an optional embodiment, the fine-tuning parameter prediction neural network is a CNN neural network, which is trained using a training data set including multiple training user operation and maintenance question and answer record data and corresponding reference material annotations and answer fine-tuning parameter annotations, and is trained based on the cross entropy loss function and gradient descent algorithm until convergence.

[0137] It can be seen that through the above optional embodiments, the network details and training details of the fine-tuning parameter prediction neural network are defined to accurately predict the answer fine-tuning parameters, assisting in providing operation and maintenance question and answer services to operation and maintenance users more accurately and efficiently, improving the efficiency and convenience of operation and maintenance work, and reducing operation and maintenance costs and operation and maintenance errors.

[0138] As an optional embodiment, the correction module inputs the answer fine-tuning parameters and the direct answer data into the large language model to obtain a specific method for correcting the answer data, including:

[0139] Based on the preset language organization rules, a fine-tuning command statement is determined according to the answer fine-tuning parameters;

[0140] The direct answer data and the fine-tuning command statement are combined into an input command statement;

[0141] An input command sentence is input to the large language model to obtain corrected answer data output by the large language model.

[0142] It can be seen that through the above-mentioned optional embodiments, the fine-tuning command statement can be determined according to the answer fine-tuning parameters through the preset language organization rules, so as to combine the direct answer data to form the input command statement, so that the large language model can directly understand and execute the fine-tuning command, obtain more accurate corrected answer data, and provide operation and maintenance question and answer services to operation and maintenance users more accurately and efficiently, improve the efficiency and convenience of operation and maintenance work, and reduce operation and maintenance costs and operation and maintenance errors.

[0143] Example 3

[0144] See also Figure 3 , Figure 3 This is another operation and maintenance question and answer data processing system based on a large language model disclosed in an embodiment of the present invention. Figure 3 The described operation and maintenance question-answering data processing system based on a large language model is applied to a data processing system / data processing device / data processing server (wherein the server includes a local processing server or a cloud processing server). Figure 3 As shown, the operation and maintenance question and answer data processing system based on the large language model may include:

[0145] A memory 301 storing executable program code;

[0146] a processor 302 coupled to the memory 301;

[0147] The processor 302 calls the executable program code stored in the memory 301 to execute the steps of the operation and maintenance question and answer data processing method based on the large language model described in the first embodiment.

[0148] Example 4

[0149] An embodiment of the present invention discloses a computer-readable storage medium storing a computer program for electronic data exchange, wherein the computer program enables a computer to execute the steps of the operation and maintenance question and answer data processing method based on a large language model described in Example 1.

[0150] Example 5

[0151] An embodiment of the present invention discloses a computer program product, which includes a non-transitory computer-readable storage medium storing a computer program, and the computer program is operable to enable a computer to execute the steps of the operation and maintenance question and answer data processing method based on a large language model described in Example 1.

[0152] The foregoing description of specific embodiments of the present disclosure is intended to illustrate a method for performing a multi-tasking process. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims may be performed in an order different from that described in the embodiments and still achieve the desired results. Furthermore, the processes depicted in the accompanying drawings do not necessarily require the specific order shown or the sequential order shown to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0153] The systems, devices, modules, or units described in the above embodiments may be implemented by computer chips or entities, or by products having certain functions. A typical implementation device is a computer. Specifically, the computer may be, for example, a personal computer, a laptop computer, a cellular phone, a camera phone, a smartphone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or a combination of any of these devices.

[0154] For the convenience of description, the above devices are described as being divided into various units according to their functions. Of course, when implementing this specification, the functions of each unit can be implemented in the same or multiple software and / or hardware.

[0155] Those skilled in the art will appreciate that the embodiments of this specification may be provided as methods, systems, or computer program products. Therefore, the embodiments of this specification may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Furthermore, the embodiments of this specification may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0156] This specification is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of this specification. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0157] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0158] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 The steps for the function specified in one or more boxes.

[0159] In a typical configuration, a computing device includes one or more processors (CPUs), input / output interfaces, network interfaces, and memory.

[0160] Memory may include non-permanent storage in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. Memory is an example of a computer-readable medium.

[0161] Computer-readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. The information can be computer-readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassettes, magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer-readable media does not include transitory media such as modulated data signals and carrier waves.

[0162] It should also be noted that the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, commodity, or apparatus that includes a series of elements includes not only those elements but also other elements not explicitly listed, or includes elements inherent to such process, method, commodity, or apparatus. In the absence of further limitations, an element defined by the phrase "comprises a ..." does not exclude the presence of other identical elements in the process, method, commodity, or apparatus that includes the element.

[0163] This specification may be described in the general context of computer-executable instructions, such as program modules, executed by a computer. Generally, program modules include routines, programs, objects, components, data structures, and the like that perform specific tasks or implement specific abstract data types. This specification may also be practiced in distributed computing environments where tasks are performed by remote processing devices connected through a communications network. In a distributed computing environment, program modules may be located in both local and remote computer storage media, including storage devices.

[0164] The various embodiments in this specification are described in a progressive manner. Similar parts between the various embodiments can be referred to in conjunction with each other. Each embodiment focuses on the differences between the other embodiments. In particular, the system embodiments are generally similar to the method embodiments, so the description is relatively simple. For relevant parts, refer to the description of the method embodiments.

[0165] Finally, it should be noted that the operation and maintenance question and answer data processing method and system based on a large language model disclosed in the embodiment of the present invention is only a preferred embodiment of the present invention, which is only used to illustrate the technical solution of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, ordinary technicians in this field should understand that it is still possible to modify the technical solutions recorded in the aforementioned embodiments, or to replace some of the technical features therein with equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A method for processing operation and maintenance question and answer data based on a large language model, characterized in that: The method comprises: Obtain the operation and maintenance problem data and user parameters of the target operation and maintenance user; Input the operation and maintenance question into the trained large language model to obtain direct answer data and reference data corresponding to the operation and maintenance question; According to the user parameters and the reference data, based on the neural network algorithm, the answer fine-tuning parameters are predicted, including: For each candidate user's historical operation and maintenance question and answer record, obtain the candidate user parameters and multiple question and answer record texts corresponding to the candidate user's historical operation and maintenance question and answer record; Calculating a first similarity between the candidate user parameter and the user parameter; Calculating an average of similarities between all the question and answer record texts and current question and answer data to obtain a second similarity; the current question and answer data includes the operation and maintenance question data and the direct answer data; Calculate the product of the first similarity and the second similarity to obtain the record parameters corresponding to the candidate user's historical operation and maintenance question and answer record; Sort all the candidate user's historical operation and maintenance question and answer records from largest to smallest according to the record parameters to obtain a record sequence; Filter out all the candidate user historical operation and maintenance question and answer records whose first preset number in the record sequence and whose record parameters are greater than the parameter threshold, to obtain a plurality of similar user historical operation and maintenance question and answer records; Inputting all the similar user historical operation and maintenance question and answer records and the reference data into the trained fine-tuning parameter prediction neural network to obtain the output answer fine-tuning parameters; The answer fine-tuning parameters and the direct answer data are input into the large language model to obtain modified answer data; the modified answer data is sent to the terminal of the target operation and maintenance user for display.

2. The operation and maintenance question and answer data processing method based on a large language model according to claim 1 is characterized in that: The large language model is trained by the following steps: Collecting operation and maintenance standard data and operation and maintenance question and answer template data, cleaning and splitting the operation and maintenance standard data and operation and maintenance question and answer template data, extracting valid question and answer pair data, performing part-of-speech analysis and entity naming task processing on the question and answer pair data, and storing the processed question and answer pair data in a question and answer pair database; Noun entities and verb entities are extracted from the question-answer pair database, and related clustering data is analyzed after vectorization conversion of the noun entities and verb entities to obtain multiple sets of mapping pairs. The understanding ability of the PLM pre-trained large language model is evaluated through the multiple sets of mapping pairs, and a knowledge dialogue set is constructed based on the mapping pairs with negative evaluation results; manual answer fine-tuning parameter annotation is performed on the question-answer pair database, and a task dialogue set is constructed based on the question-answer pair data and the manual answer fine-tuning parameter annotation; The training corpus and test corpus are constructed through the knowledge dialogue set and the task dialogue set. The training corpus is input into the large language model pre-trained by PLM for instruction fine-tuning, and the test corpus is input into the large language model to verify the accuracy. This process is repeated until the verification result of the large language model meets the preset requirements to obtain a trained large language model.

3. The operation and maintenance question and answer data processing method based on a large language model according to claim 2 is characterized in that: The answer fine-tuning parameter or the manual answer fine-tuning parameter annotation includes at least one of an answer tone parameter, an answer information amount parameter, an answer citation material amount parameter, and an answer text style parameter.

4. The operation and maintenance question and answer data processing method based on a large language model according to claim 1 is characterized in that: The user parameters include user position, user department, user working experience, user gender and user education background.

5. The operation and maintenance question and answer data processing method based on a large language model according to claim 1 is characterized in that: The fine-tuning parameter prediction neural network is a CNN neural network, which is trained using a training data set including multiple training user operation and maintenance question and answer record data and corresponding reference material annotations and answer fine-tuning parameter annotations, and is trained based on the cross entropy loss function and gradient descent algorithm until convergence.

6. The operation and maintenance question and answer data processing method based on a large language model according to claim 1 is characterized in that: Inputting the answer fine-tuning parameter and the direct answer data into the large language model to obtain the modified answer data includes: Based on the preset language organization rules, determining the fine-tuning command sentence according to the answer fine-tuning parameters; Combining the direct answer data and the fine-tuning command sentence into an input command sentence; The input command sentence is input to the large language model to obtain corrected answer data output by the large language model.

7. An operation and maintenance question and answer data processing system based on a large language model, characterized in that: The system comprises: The acquisition module is used to obtain the operation and maintenance problem data and user parameters of the target operation and maintenance user; An input module, configured to input the operation and maintenance question into a trained large language model to obtain direct answer data and reference data corresponding to the operation and maintenance question; The prediction module is used to predict the answer fine-tuning parameters based on the user parameters and the reference data based on the neural network algorithm, including: For each candidate user's historical operation and maintenance question and answer record, obtain the candidate user parameters and multiple question and answer record texts corresponding to the candidate user's historical operation and maintenance question and answer record; Calculating a first similarity between the candidate user parameter and the user parameter; Calculating an average of similarities between all the question and answer record texts and current question and answer data to obtain a second similarity; the current question and answer data includes the operation and maintenance question data and the direct answer data; Calculate the product of the first similarity and the second similarity to obtain the record parameters corresponding to the candidate user's historical operation and maintenance question and answer record; Sort all the candidate user's historical operation and maintenance question and answer records from largest to smallest according to the record parameters to obtain a record sequence; Filter out all the candidate user historical operation and maintenance question and answer records whose first preset number in the record sequence and whose record parameters are greater than the parameter threshold, to obtain a plurality of similar user historical operation and maintenance question and answer records; Inputting all the similar user historical operation and maintenance question and answer records and the reference data into the trained fine-tuning parameter prediction neural network to obtain the output answer fine-tuning parameters; A correction module is used to input the answer fine-tuning parameters and the direct answer data into the large language model to obtain corrected answer data; the corrected answer data is used to be sent to the terminal of the target operation and maintenance user for display.

8. An operation and maintenance question and answer data processing system based on a large language model, characterized in that: The system comprises: a memory storing executable program code; a processor coupled to the memory; The processor calls the executable program code stored in the memory to execute the operation and maintenance question and answer data processing method based on a large language model as described in any one of claims 1 to 6.

Images