Intelligent operation and maintenance work order question answering method and device based on large model, medium and equipment
Through the intelligent operation and maintenance work order question-and-answer method based on large models, the traditional problem of inefficient processing of IT operation and maintenance work orders is solved, and the in-depth and accurate understanding and quick response of operation and maintenance work orders is achieved, which improves operation and maintenance efficiency and users Experience, adapt to complex and non-standardized scenarios.
Patent Information
- Application Number
- CN202411156168.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-22
- Publication Date
- 2024-11-15
AI Technical Summary
Traditional manual processing of IT operation and maintenance work orders is inefficient and slow, and it is prone to processing bottlenecks, and it is difficult to adapt to complex scenarios and non-standardized expressions, resulting in increased operation and maintenance risks.
Using intelligent operation and maintenance work order question-and-answer method based on large models, the use of large-scale language models in the field of IT operation and maintenance to identify problem types, neural network algorithms identify user intentions, and named entity recognition technology to identify entities Information, combined with intelligent search algorithms, retrieve solutions from the operation and maintenance knowledge base and historical fault case database, generate reply content, and optimize the algorithm through continuous learning.
It has achieved a deep and accurate understanding of operation and maintenance work orders, reduced manual intervention, improved operation and maintenance efficiency and user experience, adapted to the rapidly changing environment, quickly gave appropriate reply content, and reduced operation and maintenance risks.
Smart Images

Figure CN118967107A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of operation and maintenance management technology, and in particular to a large model-based intelligent operation and maintenance work order question and answer method and device, medium, and equipment. Background Art
[0002] As enterprises become more information-based, the number of IT operation and maintenance work orders has increased dramatically, which has led to problems such as low efficiency and slow response in traditional manual processing methods. Manual processing of work orders not only consumes a lot of human resources, but also easily leads to processing bottlenecks during peak hours, affecting service quality. In addition, human factors are also prone to errors and omissions, increasing operation and maintenance risks.
[0003] Although there are some work order processing systems based on rules and keyword matching, they are limited in their ability to handle complex scenarios and non-standard expressions. These systems often rely on predefined rules and templates, have weak processing capabilities for emerging problems and unstructured data, and are difficult to adapt to rapidly changing environments. Summary of the invention
[0004] In response to at least one of the above technical problems, an embodiment of the present invention provides an intelligent operation and maintenance work order question and answer method, device, medium, and equipment based on a large model.
[0005] In a first aspect, the intelligent operation and maintenance work order question and answer method based on a large model provided by an embodiment of the present invention includes:
[0006] Obtain operation and maintenance work orders;
[0007] Preprocessing the operation and maintenance work order to obtain standardized and structured operation and maintenance work order information;
[0008] A large-scale language model in the IT operation and maintenance field is used to identify the problem type in the operation and maintenance work order information, a neural network algorithm is used to identify the user intent in the operation and maintenance work order information, and a named entity recognition technology is used to identify the entity information in the operation and maintenance work order information;
[0009] According to the problem type, the user intention and the entity information, an intelligent retrieval algorithm is used to search from the connected operation and maintenance knowledge base and the historical fault case library to obtain a solution;
[0010] According to the solution, the reply content corresponding to the operation and maintenance work order is determined, and the reply content is fed back to the user of the operation and maintenance work order.
[0011] In one embodiment, the preprocessing of the operation and maintenance work order to obtain standardized and structured operation and maintenance work order information includes:
[0012] Cleaning the text in the operation and maintenance work order and removing noise information in the text;
[0013] Sentence processing of long text in the cleaned and denoised text;
[0014] Perform part-of-speech tagging on each word in the cleaned and denoised text;
[0015] Convert non-standardized terms in the text to standard terms;
[0016] Perform security checks on the text and remove information that does not meet the security checks.
[0017] In one embodiment, the method further comprises:
[0018] Identifying the sentiment tendency of the operation and maintenance work order information;
[0019] Correspondingly, determining the reply content corresponding to the operation and maintenance work order according to the solution, and feeding back the reply content to the user of the operation and maintenance work order, includes:
[0020] Determine the reply content corresponding to the operation and maintenance work order according to the solution;
[0021] Determining a response tone corresponding to the response content according to the emotional tendency;
[0022] The reply tone is used to feed back the reply content to the user of the operation and maintenance work order in the form of voice.
[0023] In one embodiment, the intelligent retrieval algorithm is used to retrieve from the connected operation and maintenance knowledge base and historical fault case library according to the problem type, the user intention and the entity information to obtain a solution, including:
[0024] According to the problem type, the user intention and the entity information, an intelligent retrieval algorithm is used to search from the connected operation and maintenance knowledge base and the historical fault case library to obtain multiple solutions;
[0025] Using a relevance mechanism to score the relevance of the multiple solutions;
[0026] The solution with the highest relevance score is selected as the final solution.
[0027] In one embodiment, the method further comprises:
[0028] Obtaining contextual information of the operation and maintenance work order;
[0029] Correspondingly, determining the reply content corresponding to the operation and maintenance work order according to the solution includes: generating reply content that conforms to the context environment according to the solution and the context association information.
[0030] In one embodiment, the method further comprises:
[0031] If the intelligent retrieval algorithm fails to retrieve a solution from the connected operation and maintenance knowledge base and historical fault case library, a pre-trained large-scale language model is used to generate reply suggestions, and the reply suggestions are fed back to the user of the operation and maintenance work order.
[0032] In one embodiment, the method further comprises:
[0033] Obtaining user feedback information regarding the reply content;
[0034] Optimizing the intelligent retrieval algorithm according to the feedback information;
[0035] The knowledge learned from the feedback information is added to the operation and maintenance knowledge base.
[0036] According to the second aspect, the intelligent operation and maintenance work order question and answer device based on a large model provided by an embodiment of the present invention includes:
[0037] Work order acquisition module, used to obtain operation and maintenance work orders;
[0038] A preprocessing module, used to preprocess the operation and maintenance work order to obtain standardized and structured operation and maintenance work order information;
[0039] A work order understanding module is used to use a large-scale language model in the IT operation and maintenance field to identify the problem type in the operation and maintenance work order information, use a neural network algorithm to identify the user intent in the operation and maintenance work order information, and use named entity recognition technology to identify entity information in the operation and maintenance work order information;
[0040] A solution retrieval module is used to retrieve from the connected operation and maintenance knowledge base and historical fault case library using an intelligent retrieval algorithm according to the problem type, the user intention and the entity information to obtain a solution;
[0041] The reply determination module is used to determine the reply content corresponding to the operation and maintenance work order according to the solution, and feed back the reply content to the user of the operation and maintenance work order.
[0042] According to a third aspect, an embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed in a computer, enables the computer to execute the method provided in the first aspect.
[0043] According to a fourth aspect, an embodiment of the present invention provides a computing device, including a memory and a processor, wherein the memory stores executable code, and when the processor executes the executable code, the method provided in the first aspect is implemented.
[0044] The intelligent operation and maintenance work order question and answer method, device, medium, and equipment based on a large model provided in the embodiment of the present invention pre-process the operation and maintenance work order to obtain standardized and structured operation and maintenance work order information, and then use a large-scale language model in the IT operation and maintenance field to identify the problem type in the operation and maintenance work order information, use a neural network algorithm to identify the user intent in the operation and maintenance work order information, and use named entity recognition technology to identify the entity information in the operation and maintenance work order information, so as to achieve understanding of the operation and maintenance work order. Then, based on the above-mentioned information understood, and using an intelligent retrieval algorithm to search from the docked operation and maintenance knowledge base and historical fault case library, a solution is obtained, and then according to the solution, the reply content corresponding to the operation and maintenance work order is determined, and the reply content is fed back to the user of the operation and maintenance work order. It can be seen that the operation and maintenance work order is understood based on a large-scale language model, a neural network algorithm, and a named entity recognition technology in the IT operation and maintenance field. This process can achieve a deep and accurate understanding of the operation and maintenance work order. On the basis of deep and accurate understanding, the most appropriate solution is retrieved from the docked operation and maintenance knowledge base and the historical fault case library using an intelligent retrieval algorithm, and then the reply content is determined. It can be seen that the embodiments of the present invention can accurately analyze the information in the operation and maintenance work order, accurately match the solution, reduce manual intervention, improve operation and maintenance efficiency and user experience. For emerging problems and unstructured data, it can also quickly give appropriate responses to adapt to the rapidly changing environment. BRIEF DESCRIPTION OF THE DRAWINGS
[0045] Figure 1 It is a flowchart of an intelligent operation and maintenance work order question and answer method based on a large model in one embodiment of the present invention;
[0046] Figure 2 The present invention is a structural block diagram of an intelligent operation and maintenance work order question and answer device based on a large model in one embodiment of the present invention. DETAILED DESCRIPTION
[0047] In the first aspect, the embodiment of the present invention provides an intelligent operation and maintenance work order question and answer method based on a large model, see Figure 1 The method comprises the following steps S110 to S150:
[0048] S110, obtain the operation and maintenance work order;
[0049] Specifically, operation and maintenance work orders can be accessed through multiple channels, such as email, web forms, instant messaging tools, API interfaces, etc.
[0050] S120, preprocessing the operation and maintenance work order to obtain standardized and structured operation and maintenance work order information;
[0051] In one embodiment, the preprocessing of the operation and maintenance work order in S120 to obtain standardized and structured operation and maintenance work order information may include:
[0052] 1. Clean the text in the operation and maintenance work order and remove noise information in the text;
[0053] Among them, cleaning the text in the received operation and maintenance work orders and removing irrelevant characters, redundant spaces, formatting marks, etc. can improve the text quality and lay a good foundation for subsequent natural language processing.
[0054] Among them, noise information in the text, such as user personal information, sensitive data, etc., is removed to protect privacy and avoid interference with the model.
[0055] 2. Sentence processing of long text in the cleaned and denoised text;
[0056] That is, splitting long text into separate sentences or paragraphs so that the model can better understand the structure and meaning of the text.
[0057] 3. Perform part-of-speech tagging on each word in the cleaned and denoised text;
[0058] Among them, each word in the text is tagged with parts of speech to identify nouns, verbs, adjectives, etc., providing auxiliary information for semantic analysis.
[0059] 4. Convert non-standardized terms in the text into standard terms;
[0060] Specifically, non-standardized terms, abbreviations, slang, etc. are converted into standard terms that the system can understand to reduce ambiguity and improve the accuracy of the model. Abnormal patterns or outliers in the text can also be detected, such as unusual device behavior descriptions, abnormal service requests, etc., to quickly identify potential serious problems.
[0061] 5. Perform a security check on the text and remove information that does not meet the security check requirements.
[0062] That is, security checks are performed in the preprocessing stage to ensure that the work order content does not contain malicious code or information that violates security policies.
[0063] It can be seen that the above preprocessing can convert operation and maintenance work order information into standardized and structured data suitable for deep learning model processing, providing a solid foundation for the system's intelligent question and answer and automated execution, improving the system's processing efficiency, and improving the accuracy and reliability of answers.
[0064] S130, using a large-scale language model in the IT operation and maintenance field to identify the problem type in the operation and maintenance work order information, using a neural network algorithm to identify the user intent in the operation and maintenance work order information, and using named entity recognition technology to identify entity information in the operation and maintenance work order information;
[0065] S130 actually understands the operation and maintenance work order, deeply understands the content of the work order, and extracts valuable information from it.
[0066] Among them, the large-scale language model in the field of IT operation and maintenance is a large-scale language model that has been fine-tuned for a specific field and can achieve accurate parsing of operation and maintenance work orders. The large-scale language model is fine-tuned using data specific to the field of IT operation and maintenance to improve the model's understanding of operation and maintenance terminology, processes, and context. The large-scale language model in the field of IT operation and maintenance relies on the pre-processed work order information in the field of IT operation and maintenance to identify the types of questions raised in the work order, such as fault diagnosis, configuration consultation, performance optimization, etc., laying the foundation for providing accurate answers or performing operations.
[0067] Among them, named entity recognition technology, namely NER technology, can extract key information from the text, such as time, place, equipment name, service name, etc., so as to understand the content of the work order more accurately.
[0068] Among them, neural network algorithms, such as CNN and RNN algorithms, can identify user intentions and determine the purpose or needs of users submitting work orders, such as fault reports, service requests, information queries, etc.
[0069] S140, searching from a connected operation and maintenance knowledge base and a historical fault case library using an intelligent retrieval algorithm according to the problem type, the user intention, and the entity information to obtain a solution;
[0070] Among them, the operation and maintenance knowledge base and the historical fault case base can be connected through a standardized data interface to ensure compatibility with the operation and maintenance knowledge base and the historical fault case base of different sources and formats, so as to facilitate information search. It is also necessary to ensure the real-time data update of the operation and maintenance knowledge base and the historical fault case base to ensure that the accessed information is the latest and most accurate.
[0071] Among them, the intelligent retrieval algorithm can adopt an intelligent retrieval algorithm based on the transformation and construction of LLAMA3, which can quickly retrieve solutions related to the problem type, the user intention and the entity information from a huge operation and maintenance knowledge base and a historical fault case library.
[0072] Among them, the intelligent retrieval algorithm provides services and interfaces such as semantic matching technology, case recommendation capabilities, knowledge graph construction, trend analysis and prediction, multi-dimensional labels, and user behavior analysis.
[0073] Among them, the operation and maintenance knowledge base and historical fault case library can provide powerful information support and decision-making assistance, improve the efficiency and quality of operation and maintenance work. At the same time, based on the accumulation and sharing of knowledge, it provides valuable intellectual resources for the operation and maintenance team.
[0074] In actual scenarios, a highly scalable framework can be created to achieve high scalability in the construction of a knowledge base, and the framework can be connected to the knowledge base quickly through a data interface to provide the ability to quickly connect.
[0075] In one embodiment, the step of searching from a connected operation and maintenance knowledge base and a historical fault case library using an intelligent retrieval algorithm according to the problem type, the user intent, and the entity information to obtain a solution in S140 may include:
[0076] S141, searching from a connected operation and maintenance knowledge base and a historical fault case library using an intelligent retrieval algorithm according to the problem type, the user intention, and the entity information, to obtain multiple solutions;
[0077] S142, using a relevance mechanism to perform relevance scoring on the multiple solutions;
[0078] S143. The solution with the highest relevance score is taken as the final solution.
[0079] It can be seen that the intelligent retrieval algorithm based on the transformation and construction of LLAMA3 can quickly search the knowledge base in multiple dimensions, such as severity, type, and scope of impact, and at the same time perform relevance scoring on different retrieved solutions, thereby selecting the solution with the highest relevance score as the final solution output.
[0080] S150. Determine the reply content corresponding to the operation and maintenance work order according to the solution, and feed back the reply content to the user of the operation and maintenance work order.
[0081] In one embodiment, the method provided by the embodiment of the present invention may further include:
[0082] Identifying the sentiment tendency of the operation and maintenance work order information;
[0083] Correspondingly, determining the reply content corresponding to the operation and maintenance work order according to the solution, and feeding back the reply content to the user of the operation and maintenance work order, includes:
[0084] Determine the reply content corresponding to the operation and maintenance work order according to the solution;
[0085] Determining a response tone corresponding to the response content according to the emotional tendency;
[0086] The reply tone is used to feed back the reply content to the user of the operation and maintenance work order in the form of voice.
[0087] Among them, emotional tendencies, for example, whether the user shows emotions such as urgency, dissatisfaction, etc., so that these factors can be considered when responding. It can be seen that the emotional color in the work order is evaluated, for example, whether the user shows emotions such as urgency, anxiety or dissatisfaction, so that the corresponding tone can be adopted when responding.
[0088] In one embodiment, the method provided by the embodiment of the present invention may further include:
[0089] Obtaining contextual information of the operation and maintenance work order;
[0090] Correspondingly, determining the reply content corresponding to the operation and maintenance work order according to the solution includes: generating reply content that conforms to the context environment according to the solution and the context association information.
[0091] It can be seen that by using context modeling technology to understand the contextual information in the work order text, we can improve our ability to understand complex problems, thereby generating responses that are in line with the context, satisfying the ability of multiple rounds of conversation, and facilitating the generation of coherent and consistent responses. It can be seen that integrating the solution with the contextual information of the work order can generate more accurate and personalized responses.
[0092] The above-mentioned evaluation of the emotional color in the work order and determination of context-related information are actually part of understanding the content of the work order. Only by fully and deeply understanding the content of the operation and maintenance work order can accurate information be provided to generate appropriate answers, making the operation and maintenance work more intelligent and automated, improving the efficiency and quality of the operation and maintenance work, and at the same time enhancing the user experience and improving user satisfaction and trust.
[0093] In one embodiment, the method provided by the embodiment of the present invention may further include:
[0094] If the intelligent retrieval algorithm fails to retrieve a solution from the connected operation and maintenance knowledge base and historical fault case library, a pre-trained large-scale language model is used to generate reply suggestions, and the reply suggestions are fed back to the user of the operation and maintenance work order.
[0095] It can be seen that if the intelligent retrieval algorithm cannot retrieve relevant solutions, the large-scale language model is used to generate reply suggestions. The large-scale language model is also a fine-tuned large model, reply suggestions, such as troubleshooting steps, suggested solutions, etc.
[0096] In one embodiment, the method provided by the embodiment of the present invention may further include:
[0097] Obtaining user feedback information regarding the reply content;
[0098] Optimizing the intelligent retrieval algorithm according to the feedback information;
[0099] The knowledge learned from the feedback information is added to the operation and maintenance knowledge base.
[0100] It can be seen that feedback and learning are an indispensable part of the intelligent operation and maintenance work order question and answer system. By collecting the opinions of operation and maintenance personnel on the reply content and the direct feedback of users, the intelligent retrieval algorithm is continuously optimized and learned.
[0101] It can be seen that a comprehensive feedback collection mechanism is established here, allowing operation and maintenance personnel and users to provide comments and suggestions on the reply content in a simple and intuitive way. Feedback information can come from multiple channels, such as user interface, API call, email notification, etc., to ensure that all feedback can be captured and analyzed. The collected feedback information is used to re-optimize the intelligent retrieval algorithm and adjust the parameters to improve the retrieval performance and accuracy. In addition, the knowledge learned from the feedback is updated to the operation and maintenance knowledge base to enrich the content of the operation and maintenance knowledge base and improve the accuracy of future replies.
[0102] In short, feedback and learning can ensure continuous learning and progress from practical applications, ultimately achieving self-optimization and continuous improvement, and providing users with more accurate, efficient and personalized services.
[0103] It can be seen that the embodiment of the present invention relies on large-scale pre-trained language model technology to completely innovate the traditional operation and maintenance workflow and achieve a leap in automation and intelligence in the processing of operation and maintenance work orders. The core of the embodiment of the present invention lies in deep understanding and efficient response, and the goal is to directly target the most critical part of the operation and maintenance work order - automatic analysis of the problem description and the immediate generation of accurate response strategies.
[0104] Specifically, the embodiment of the present invention firstly uses a highly optimized natural language processing algorithm to deeply analyze the input operation and maintenance work order. This process not only includes the accurate interpretation of the literal meaning, but also involves the intelligent identification of the implicit intention, the severity of the problem and the historical related events in the technical context. By utilizing the cross-domain knowledge coverage capability of the large model, it is possible to integrate and mine all-round information across technical documents, past cases, and even the latest technical forum discussions. Then, on the basis of accurate analysis, it further utilizes its built-in advanced semantic matching mechanism to quickly filter out the most appropriate answer from the huge operation and maintenance knowledge base and historical fault case library. This matching is not only based on keywords, but also goes deep into the semantic level to ensure that the most suitable solution path can be found even if the expression method is different. For those complex or rare problems, it uses its generative capabilities to automatically generate reply suggestions, which comprehensively consider the specific context of the problem, the effectiveness of historical solutions and the latest technical trends to ensure the accuracy and cutting-edge nature of the reply. In addition, the embodiment of the present invention also incorporates the mechanism of continuous learning and self-optimization. By continuously receiving feedback on the results of work order processing, self-adjusting parameters, and optimizing matching strategies, it can continuously evolve in actual applications and improve the accuracy and efficiency of problem solving. This closed-loop learning model ensures that the system becomes more intelligent over time, and operation and maintenance efficiency and user experience are significantly improved.
[0105] It can be seen that the intelligent operation and maintenance work order question and answer solution based on a large model proposed in the embodiment of the present invention is not only a major technological innovation of the existing operation and maintenance model, but also an important step towards future intelligent operation and maintenance. It will greatly release the productivity of operation and maintenance personnel, accelerate the problem-solving process, and ensure that users enjoy unprecedented efficient and convenient service experience.
[0106] The embodiment of the present invention is based on the basic data of work orders and combines artificial intelligence technology to build an intelligent question-and-answer system, which involves automatically parsing the problems in the operation and maintenance work orders, accurately matching solutions or automatically generating replies, which can reduce manual intervention and improve operation and maintenance efficiency and user experience. The embodiment of the present invention can assist enterprises in building an operation and maintenance service system that is both agile and intelligent. It not only achieves significant advantages in labor cost control and fault response speed, but also fundamentally reshapes the user experience, allowing users to immediately obtain professional guidance like a dedicated consultant when encountering any technical obstacles. In the end,
[0107] In recent years, large-scale language models based on deep learning (for example, BERT and T5 of Transformer architecture) have made breakthrough progress in natural language understanding and generation tasks. These models can learn rich language representation and pattern recognition capabilities through pre-training on massive data, and can better understand and generate natural language text. Intelligent operation and maintenance work order processing solutions based on deep learning can significantly improve the efficiency, accuracy and automation level of IT operation and maintenance, and provide enterprises with more intelligent, efficient and secure operation and maintenance services. At the same time, it also reduces the workload of operation and maintenance personnel, allowing them to focus on more complex and strategic tasks.
[0108] The core of the embodiment of the present invention is to cleverly integrate big models, deep learning and natural language processing technology to achieve a comprehensive and multi-level analysis of operation and maintenance work order call problems, deeply explore the key information and potential needs in the work order description, accurately map them to the most appropriate solution library, or automatically generate feedback information that fits the situation in real time. This process not only greatly reduces the pressure of manual review and response, but also successfully shortens the problem-solving cycle. That is, by utilizing the powerful contextual understanding and pattern recognition capabilities of the big model, it can effectively respond to complex and changeable problem statements, ensuring that every suggestion or response provided is highly relevant and practical. In addition, a continuous optimization mechanism is also configured to learn from each interaction and continuously evolve itself, so that the accuracy and personalization of the service are increasingly improved.
[0109] In a second aspect, an embodiment of the present invention provides an intelligent operation and maintenance work order question and answer device based on a large model, see Figure 2 , the device 100 comprises:
[0110] The work order acquisition module 110 is used to acquire the operation and maintenance work order;
[0111] A preprocessing module 120, used to preprocess the operation and maintenance work order to obtain standardized and structured operation and maintenance work order information;
[0112] The work order understanding module 130 is used to use a large-scale language model in the IT operation and maintenance field to identify the problem type in the operation and maintenance work order information, use a neural network algorithm to identify the user intention in the operation and maintenance work order information, and use a named entity recognition technology to identify the entity information in the operation and maintenance work order information;
[0113] A solution retrieval module 140 is used to retrieve from the connected operation and maintenance knowledge base and historical fault case library using an intelligent retrieval algorithm according to the problem type, the user intention and the entity information to obtain a solution;
[0114] The reply determination module 150 is used to determine the reply content corresponding to the operation and maintenance work order according to the solution, and feed back the reply content to the user of the operation and maintenance work order.
[0115] In one embodiment, the preprocessing module is specifically used to: clean the text in the operation and maintenance work order and remove noise information in the text; perform sentence processing on the long text in the cleaned and denoised text; perform part-of-speech tagging on each word in the cleaned and denoised text; convert non-standardized terms in the text into standard terms; perform security checks on the text and eliminate information that does not meet the security checks.
[0116] In one embodiment, the device may further include:
[0117] An emotion recognition module, used to identify the emotion tendency of the operation and maintenance work order information;
[0118] Correspondingly, the reply determination module is specifically used to: determine the reply content corresponding to the operation and maintenance work order according to the solution; determine the reply tone corresponding to the reply content according to the emotional tendency; and use the reply tone to feedback the reply content to the user of the operation and maintenance work order in the form of voice.
[0119] In one embodiment, the solution retrieval module is specifically used to: use an intelligent retrieval algorithm to search from a connected operation and maintenance knowledge base and a historical fault case library according to the problem type, the user intent and the entity information to obtain multiple solutions; use a relevance mechanism to score the relevance of the multiple solutions; and use the solution with the highest relevance score as the final solution.
[0120] In one embodiment, the device may further include:
[0121] An associated acquisition module, used to acquire context associated information of the operation and maintenance work order;
[0122] Correspondingly, the reply determination module is specifically used to generate reply content that conforms to the context environment according to the solution and the context association information.
[0123] In one embodiment, the device may further include:
[0124] The reply generation module is used to generate reply suggestions using a pre-trained large-scale language model if the intelligent retrieval algorithm fails to retrieve a solution from the connected operation and maintenance knowledge base and historical fault case library, and feed back the reply suggestions to the user of the operation and maintenance work order.
[0125] In one embodiment, the device may further include:
[0126] The feedback learning module is used to obtain user feedback information on the reply content; optimize the intelligent retrieval algorithm according to the feedback information; and add the knowledge learned from the feedback information to the operation and maintenance knowledge base.
[0127] It is understandable that the explanation, specific implementation, beneficial effects, examples, etc. of the relevant contents in the device provided in the embodiment of the present invention can be found in the corresponding parts of the method provided in the first aspect, and will not be repeated here.
[0128] In a third aspect, an embodiment of the present invention provides a computer-readable medium, on which computer instructions are stored. When the computer instructions are executed by a processor, the processor executes the method provided in the first aspect.
[0129] Specifically, a system or device equipped with a storage medium can be provided, on which software program code that implements the functions of any of the above-mentioned embodiments is stored, and a computer (or CPU or MPU) of the system or device can be enabled to read and execute the program code stored in the storage medium.
[0130] In this case, the program code itself read from the storage medium can realize the function of any one of the above-mentioned embodiments, and thus the program code and the storage medium storing the program code constitute a part of the present invention.
[0131] The storage medium embodiments for providing the program code include a floppy disk, a hard disk, a magneto-optical disk, an optical disk (such as CD-ROM, CD-R, CD-RW, DVD-ROM, DVD-RAM, DVD-RW, DVD+RW), a magnetic tape, a non-volatile memory card, and a ROM. Alternatively, the program code can be downloaded from a server computer by a communication network.
[0132] In addition, it should be clear that the functions of any of the above embodiments can be implemented not only by executing the program code read by the computer, but also by enabling an operating system operating on the computer to complete part or all of the actual operations based on instructions from the program code.
[0133] In addition, it can be understood that the program code read from the storage medium is written to a memory provided in an expansion board inserted into the computer or to a memory provided in an expansion module connected to the computer, and then based on the instructions of the program code, a CPU installed on the expansion board or expansion module is enabled to perform part or all of the actual operations, thereby realizing the functions of any of the above-mentioned embodiments.
[0134] It is understandable that the explanation, specific implementation methods, beneficial effects, examples, etc. of the relevant contents in the computer-readable medium provided in the embodiment of the present invention can be found in the corresponding parts of the method provided in the first aspect, and will not be repeated here.
[0135] In a fourth aspect, an embodiment of the present specification provides a computing device, including a memory and a processor, wherein the memory stores executable code, and when the processor executes the executable code, it implements the method in any embodiment of the specification.
[0136] It is understandable that the explanation, specific implementation, beneficial effects, examples, etc. of the relevant contents in the computing device provided in the embodiment of the present invention can be found in the corresponding parts of the method provided in the first aspect, and will not be repeated here.
[0137] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.
[0138] Those skilled in the art should be aware that in one or more of the above examples, the functions described in the present invention can be implemented by hardware, software, widgets, or any combination thereof. When implemented by software, these functions can be stored in a computer-readable medium or transmitted as one or more instructions or codes on a computer-readable medium.
[0139] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made on the basis of the technical solution of the present invention should be included in the scope of protection of the present invention.
Claims
1. An intelligent operation and maintenance work order question and answer method based on a large model, characterized in that: include: Obtain operation and maintenance work orders; Preprocessing the operation and maintenance work order to obtain standardized and structured operation and maintenance work order information; A large-scale language model in the IT operation and maintenance field is used to identify the problem type in the operation and maintenance work order information, a neural network algorithm is used to identify the user intent in the operation and maintenance work order information, and a named entity recognition technology is used to identify the entity information in the operation and maintenance work order information; According to the problem type, the user intention and the entity information, an intelligent retrieval algorithm is used to search from the connected operation and maintenance knowledge base and the historical fault case library to obtain a solution; According to the solution, the reply content corresponding to the operation and maintenance work order is determined, and the reply content is fed back to the user of the operation and maintenance work order.
2. The method according to claim 1, characterized in that The preprocessing of the operation and maintenance work order to obtain standardized and structured operation and maintenance work order information includes: Cleaning the text in the operation and maintenance work order and removing noise information in the text; Sentence processing of long text in the cleaned and denoised text; Perform part-of-speech tagging on each word in the cleaned and denoised text; Convert non-standardized terms in the text to standard terms; Perform security checks on the text and remove information that does not meet the security checks.
3. The method according to claim 1, characterized in that Also includes: Identifying the sentiment tendency of the operation and maintenance work order information; Correspondingly, determining the reply content corresponding to the operation and maintenance work order according to the solution, and feeding back the reply content to the user of the operation and maintenance work order, includes: Determine the reply content corresponding to the operation and maintenance work order according to the solution; Determining a response tone corresponding to the response content according to the emotional tendency; The reply tone is used to feed back the reply content to the user of the operation and maintenance work order in the form of voice.
4. The method according to claim 1, characterized in that The method of searching from a connected operation and maintenance knowledge base and a historical fault case library using an intelligent retrieval algorithm according to the problem type, the user intention and the entity information to obtain a solution includes: According to the problem type, the user intention and the entity information, an intelligent retrieval algorithm is used to search from the connected operation and maintenance knowledge base and the historical fault case library to obtain multiple solutions; Using a relevance mechanism to score the relevance of the multiple solutions; The solution with the highest relevance score is selected as the final solution.
5. The method according to claim 1, characterized in that Also includes: Obtaining contextual information of the operation and maintenance work order; Correspondingly, determining the reply content corresponding to the operation and maintenance work order according to the solution includes: generating reply content that conforms to the context environment according to the solution and the context association information.
6. The method according to claim 1, characterized in that Also includes: If the intelligent retrieval algorithm fails to retrieve a solution from the connected operation and maintenance knowledge base and historical fault case library, a pre-trained large-scale language model is used to generate reply suggestions, and the reply suggestions are fed back to the user of the operation and maintenance work order.
7. The method according to claim 1, characterized in that Also includes: Obtaining user feedback information regarding the reply content; Optimizing the intelligent retrieval algorithm according to the feedback information; The knowledge learned from the feedback information is added to the operation and maintenance knowledge base.
8. An intelligent operation and maintenance work order question and answer device based on a large model, characterized in that: include: Work order acquisition module, used to obtain operation and maintenance work orders; A preprocessing module, used to preprocess the operation and maintenance work order to obtain standardized and structured operation and maintenance work order information; A work order understanding module is used to use a large-scale language model in the IT operation and maintenance field to identify the problem type in the operation and maintenance work order information, use a neural network algorithm to identify the user intent in the operation and maintenance work order information, and use named entity recognition technology to identify entity information in the operation and maintenance work order information; A solution retrieval module is used to retrieve from the connected operation and maintenance knowledge base and historical fault case library using an intelligent retrieval algorithm according to the problem type, the user intention and the entity information to obtain a solution; The reply determination module is used to determine the reply content corresponding to the operation and maintenance work order according to the solution, and feed back the reply content to the user of the operation and maintenance work order.
9. A computer-readable storage medium, characterized in that: A computer program is stored thereon, and when the computer program is executed in a computer, the computer is caused to execute the method according to any one of claims 1 to 7.
10. A computing device, characterized in that: The method comprises a memory and a processor, wherein the memory stores executable codes, and when the processor executes the executable codes, the method according to any one of claims 1 to 7 is implemented.
Citation Information
Cited By
Network security operation and maintenance method and device
CN119728195A
Application maintenance method and device based on artificial intelligence, equipment and medium
CN119781800A
Information feedback processing method based on artificial intelligence, electronic equipment and storage medium
CN120561645A
