Method, device, computer equipment and storage medium for generating operation and maintenance solution

Abstract

The present application relates to a method, device, computer equipment and storage medium for generating an operation and maintenance solution. The method includes: in response to the user's operation and maintenance problem, searching in the operation and maintenance knowledge base according to the semantic vector of the operation and maintenance problem, and outputting a preliminary operation and maintenance solution; updating the semantic vector according to the detailed problem information and the preliminary operation and maintenance solution input by the user based on the preliminary operation and maintenance solution, and searching in the operation and maintenance knowledge base according to the updated semantic vector, and outputting the updated operation and maintenance solution; iteratively executing the update process of the operation and maintenance solution until the preset conditions are met, and obtaining the operation and maintenance solution of the operation and maintenance problem. The use of this method can improve the accuracy of the operation and maintenance solution.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to a method, apparatus, computer equipment and storage medium for generating an operation and maintenance solution. Background Art

[0002] Operation and maintenance work is an important part of ensuring the normal operation of information systems. With the continuous expansion and complexity of information technology infrastructure, operation and maintenance personnel need to deal with a large number of complex system problems.

[0003] In the related art, in order to solve the problem quickly, an operation and maintenance knowledge base system is introduced, which provides structured and standardized operation and maintenance knowledge. When in use, the operation and maintenance knowledge base system searches the operation and maintenance knowledge base based on the keywords in the user input question, and generates a solution based on the retrieved documents or knowledge.

[0004] However, the solution generation method in the related art has a technical problem of low solution accuracy. Summary of the invention

[0005] Based on this, it is necessary to provide a method, device, computer equipment and storage medium for generating an operation and maintenance solution to address the above technical problems, which can improve the accuracy of the operation and maintenance solution.

[0006] In a first aspect, an embodiment of the present application provides a method for generating an operation and maintenance solution, including:

[0007] In response to the user's operation and maintenance problem, the operation and maintenance knowledge base is searched according to the semantic vector of the operation and maintenance problem, and a preliminary operation and maintenance solution is output;

[0008] The semantic vector is updated according to the refined problem information and the preliminary operation and maintenance solution input by the user based on the preliminary operation and maintenance solution, and according to the updated semantic vector, the operation and maintenance knowledge base is searched to output the updated operation and maintenance solution;

[0009] Iterate the update process of the operation and maintenance solution until the preset conditions are met and the operation and maintenance solution to the operation and maintenance problem is obtained.

[0010] In the operation and maintenance solution generation method provided by the embodiment of the present application, in response to the user's operation and maintenance problem, the operation and maintenance knowledge base is searched according to the semantic vector of the operation and maintenance problem, and a preliminary operation and maintenance solution is output, and then the semantic vector is updated according to the refined problem information and the preliminary operation and maintenance solution input by the user based on the preliminary operation and maintenance solution, and according to the updated semantic vector, the operation and maintenance knowledge base is searched, and the updated operation and maintenance solution is output, and the update process of the operation and maintenance solution is iteratively executed until the preset conditions are met to obtain the operation and maintenance solution of the operation and maintenance problem. In this method, after the user raises the operation and maintenance problem, the operation and maintenance knowledge base system first converts the operation and maintenance problem into a semantic vector, and searches the operation and maintenance knowledge base based on the semantic information of the operation and maintenance problem, so that the system can accurately understand the true meaning of the user's problem, match the semantic level, and output the answer most relevant to the operation and maintenance problem, thereby improving the accuracy of the operation and maintenance solution; at the same time, the user refines the problem based on the operation and maintenance solution of the previous round, conducts multiple rounds of dialogue with the operation and maintenance knowledge base system, iteratively updates the operation and maintenance solution, and the operation and maintenance knowledge base system outputs a more accurate and customized solution based on the answer and refined problem information of the previous round, further improving the accuracy of the operation and maintenance solution.

[0011] In one embodiment, before searching the operation and maintenance knowledge base according to the semantic vector of the operation and maintenance problem and outputting a preliminary operation and maintenance solution, the method further includes:

[0012] Perform text segmentation on operation and maintenance issues to obtain multiple short sentence problem information;

[0013] Encode each short sentence question information to obtain the question semantic vector of each short sentence question information;

[0014] The problem semantic vector of each short sentence problem information is determined as the semantic vector of the operation and maintenance problem.

[0015] In the operation and maintenance solution generation method provided in the embodiment of the present application, the operation and maintenance problem is segmented into text to obtain multiple short sentence problem information, and then each short sentence problem information is encoded to obtain the problem semantic vector of each short sentence problem information, and the problem semantic vector of each short sentence problem information is determined as the semantic vector of the operation and maintenance problem. In this method, when converting the operation and maintenance problem into a semantic vector, the operation and maintenance problem is first segmented into text, and the long text or complex operation and maintenance problem is effectively split to reasonably adjust the hierarchical structure of the data to ensure that each part can maintain semantic integrity, so that the subsequent retrieval in the operation and maintenance knowledge base can improve the retrieval efficiency and accuracy.

[0016] In one embodiment, the operation and maintenance knowledge base includes a plurality of operation and maintenance knowledge vector data; the operation and maintenance knowledge base is searched according to the semantic vector of the operation and maintenance problem, and a preliminary operation and maintenance solution is output, including:

[0017] According to each problem semantic vector and each operation and maintenance knowledge vector data, determining a similarity coefficient between each problem semantic vector and each operation and maintenance knowledge vector data;

[0018] According to the similarity coefficient between each problem semantic vector and each operation and maintenance knowledge vector data, the matching vector data of each problem semantic vector is determined;

[0019] Based on the matching vector data of each problem's semantic vector, output a preliminary operation and maintenance solution.

[0020] In the operation and maintenance solution generation method provided in the embodiment of the present application, the similarity coefficient between each question semantic vector and each operation and maintenance knowledge vector data is determined based on each question semantic vector and each operation and maintenance knowledge vector data, and then the matching vector data of each question semantic vector is determined based on the similarity coefficient between each question semantic vector and each operation and maintenance knowledge vector data, and then the preliminary operation and maintenance solution is output based on the matching vector data of each question semantic vector. In this method, by calculating the similarity coefficient between the question semantic vector of each short sentence question information and multiple operation and maintenance knowledge vector data in the operation and maintenance knowledge base, the operation and maintenance knowledge matching each question semantic vector can be quickly retrieved in the operation and maintenance knowledge base, and the preliminary operation and maintenance solution is quickly determined and output based on the retrieved operation and maintenance knowledge, thereby improving the generation efficiency of the preliminary operation and maintenance solution.

[0021] In one embodiment, a preliminary operation and maintenance solution is output based on the matching vector data of each problem semantic vector, including:

[0022] According to the matching vector data of each problem semantic vector, the operation and maintenance knowledge information corresponding to each matching vector data is obtained;

[0023] Process each operation and maintenance knowledge information to generate answers that match the semantic information of the operation and maintenance problem;

[0024] The answer that meets the semantic information of the operation and maintenance problem information is determined as a preliminary operation and maintenance solution, and the preliminary operation and maintenance solution is output.

[0025] In the operation and maintenance solution generation method provided in the embodiment of the present application, firstly, based on the matching vector data of each problem semantic vector, the operation and maintenance knowledge information corresponding to each matching vector data is obtained, and then each operation and maintenance knowledge information is processed to generate an answer that conforms to the semantic information of the operation and maintenance problem, and then the answer that conforms to the semantic information of the operation and maintenance problem information is determined as a preliminary operation and maintenance solution, and the preliminary operation and maintenance solution is output. In this method, firstly, based on the vector data that matches the semantic vector of each problem, the operation and maintenance knowledge information corresponding to each matching vector data is obtained from the operation and maintenance knowledge base, and each operation and maintenance knowledge information is integrated to process the fragmented operation and maintenance knowledge information into an answer that conforms to the semantic information of the current operation and maintenance problem, so that the generated preliminary operation and maintenance solution can be more consistent with the semantic information of the operation and maintenance problem, thereby improving the accuracy of the operation and maintenance solution.

[0026] In one embodiment, updating the semantic vector according to the refined problem information and the preliminary operation and maintenance solution input by the user based on the preliminary operation and maintenance solution includes:

[0027] Vectorizing the preliminary operation and maintenance solution to obtain solution vector data of the preliminary operation and maintenance solution; and problem encoding the refined problem information to obtain refined problem vector data of the refined problem information;

[0028] An updated semantic vector is determined based on the solution vector data and the refined problem vector data.

[0029] In the operation and maintenance solution generation method provided in the embodiment of the present application, the preliminary operation and maintenance solution is vectorized to obtain solution vector data of the preliminary operation and maintenance solution; and the refined problem information is problem-encoded to obtain refined problem vector data of the refined problem information, and then the updated semantic vector is determined based on the solution vector data and the refined problem vector data. In this method, the preliminary operation and maintenance solution and the refined problem information are both converted into vector data by vectorizing the preliminary operation and maintenance solution and problem-encoding the refined problem information, and the updated semantic vector is obtained based on the two vector data, so that when the operation and maintenance knowledge base system answers the refined problem information, it searches in the operation and maintenance knowledge base in combination with the solution of the previous round, thereby improving the accuracy of answering the refined problem information, thereby improving the accuracy of updating the operation and maintenance solution.

[0030] In one embodiment, searching in an operation and maintenance knowledge base according to the updated semantic vector and outputting an updated operation and maintenance solution includes:

[0031] Determining matching vector data of the updated semantic vector according to the updated semantic vector and the operation and maintenance knowledge vector data in the operation and maintenance knowledge base;

[0032] Generate an answer that matches the semantic information of the refined question information based on the matching vector data of the updated semantic vector;

[0033] An answer that meets the semantic information of the refined question information is determined as an updated operation and maintenance solution, and the updated operation and maintenance solution is output.

[0034] In the operation and maintenance solution generation method provided by the embodiment of the present application, first, the matching vector data of the updated semantic vector is determined based on the updated semantic vector and the operation and maintenance knowledge vector data in the operation and maintenance knowledge base, and then the answer to the semantic information that conforms to the detailed problem information is generated based on the matching vector data of the updated semantic vector, and the answer to the semantic information that conforms to the detailed problem information is determined as the updated operation and maintenance solution, and the updated operation and maintenance solution is output. In this method, based on the updated semantic vector, the corresponding operation and maintenance knowledge vector data is matched in the operation and maintenance knowledge base, and the matching operation and maintenance knowledge is determined based on the matching operation and maintenance knowledge vector data, so that the answer to the semantic information that conforms to the detailed problem information is quickly generated based on the matching operation and maintenance knowledge, and the updated operation and maintenance solution is obtained, which improves the updating efficiency and accuracy of the operation and maintenance solution.

[0035] In one embodiment, the method further comprises:

[0036] Verify the operation and maintenance solution;

[0037] If the operation and maintenance solution is verified, the operation and maintenance solution and the problem information corresponding to the operation and maintenance solution are vectorized, and the vectorized operation and maintenance solution and problem information are stored in the operation and maintenance knowledge base to obtain an updated operation and maintenance knowledge base.

[0038] In the operation and maintenance solution generation method provided in the embodiment of the present application, the operation and maintenance solution is verified. If the operation and maintenance solution passes the verification, the operation and maintenance solution and the problem information corresponding to the operation and maintenance solution are vectorized, and the vectorized operation and maintenance solution and problem information are stored in the operation and maintenance knowledge base to obtain an updated operation and maintenance knowledge base. In this method, after the operation and maintenance knowledge base system outputs the final operation and maintenance solution, it verifies the operation and maintenance solution. If the verification passes, the operation and maintenance solution and the corresponding operation and maintenance information are stored in the operation and maintenance knowledge base to dynamically expand the operation and maintenance knowledge base, continuously integrate and update the latest operation and maintenance knowledge, maintain the system's adaptability to new problems and new technical solutions, and maintain efficient retrieval performance during the growth of the operation and maintenance knowledge base.

[0039] In a second aspect, the embodiment of the present application further provides an operation and maintenance solution generating device, including:

[0040] A preliminary solution generation module is used to respond to the user's operation and maintenance problems, search the operation and maintenance knowledge base according to the semantic vector of the operation and maintenance problems, and output a preliminary operation and maintenance solution;

[0041] A solution updating module is used to update the semantic vector according to the refined problem information and the preliminary operation and maintenance solution input by the user based on the preliminary operation and maintenance solution, and to search in the operation and maintenance knowledge base according to the updated semantic vector, and output the updated operation and maintenance solution;

[0042] The solution determination module is used to iteratively execute the update process of the operation and maintenance solution until the preset conditions are met and the operation and maintenance solution to the operation and maintenance problem is obtained.

[0043] In a third aspect, an embodiment of the present application further provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the steps of the method in any one of the embodiments of the first aspect above are implemented.

[0044] In a fourth aspect, the present application further provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method in any one of the embodiments of the first aspect above.

[0045] In a fifth aspect, the present application further provides a computer program product, comprising a computer program, which, when executed by a processor, implements the steps of the method in any one of the embodiments of the first aspect above.

[0046] The operation and maintenance solution generation method, apparatus, computer equipment and storage medium provided in the embodiments of the present application respond to the user's operation and maintenance problems, search in the operation and maintenance knowledge base according to the semantic vector of the operation and maintenance problem, output a preliminary operation and maintenance solution, and then update the semantic vector according to the detailed problem information and the preliminary operation and maintenance solution input by the user based on the preliminary operation and maintenance solution, and search in the operation and maintenance knowledge base according to the updated semantic vector, output an updated operation and maintenance solution, and iteratively execute the update process of the operation and maintenance solution until the preset conditions are met and the operation and maintenance solution to the operation and maintenance problem is obtained. In this method, after the user raises an operation and maintenance question, the operation and maintenance knowledge base system first converts the operation and maintenance question into a semantic vector, and searches the operation and maintenance knowledge base based on the semantic information of the operation and maintenance question, so that the system can accurately understand the true meaning of the user's question, perform semantic-level matching, and output the answer that is most relevant to the operation and maintenance question, thereby improving the accuracy of the operation and maintenance solution; at the same time, the user refines the question based on the operation and maintenance solution of the previous round, conducts multiple rounds of dialogue with the operation and maintenance knowledge base system, iteratively updates the operation and maintenance solution, and the operation and maintenance knowledge base system outputs more accurate and customized solutions based on the answers and refined problem information of the previous round, thereby further improving the accuracy of the operation and maintenance solution. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the related technologies, the drawings required for use in the embodiments or the related technical descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0048] Figure 1 An application environment diagram of a method for generating an operation and maintenance solution in an embodiment;

[0049] Figure 2 is an internal structure diagram of a computer device in one embodiment;

[0050] Figure 3 A flowchart of a method for generating an operation and maintenance solution in one embodiment;

[0051] Figure 4 A schematic diagram of a process for determining a semantic vector of an operation and maintenance problem in one embodiment;

[0052] Figure 5 A flowchart of outputting a preliminary operation and maintenance solution in one embodiment;

[0053] Figure 6 A flowchart of outputting a preliminary operation and maintenance solution in another embodiment;

[0054] Figure 7 A schematic diagram of a process of updating a semantic vector in one embodiment;

[0055] Figure 8 A flowchart of outputting an updated operation and maintenance solution in one embodiment;

[0056] Fig. 9 A schematic diagram of a process for updating an operation and maintenance knowledge base in one embodiment;

[0057] Fig.10 A flowchart of a method for generating an operation and maintenance solution in another embodiment;

[0058] Fig.11 It is a structural schematic diagram of an operation and maintenance solution generating device in one embodiment;

[0059] Fig.12 The figure is a timing relationship diagram of each module when processing a problem in an embodiment. DETAILED DESCRIPTION

[0060] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0061] The technical background of the embodiments of the present application is first described below.

[0062] Operation and maintenance work is an important part of ensuring the normal operation of information systems. With the continuous expansion and complexity of IT (Information Technology) infrastructure, operation and maintenance personnel need to deal with a large number of complex system problems, including hardware failures, software configuration errors, network security vulnerabilities, etc. Traditional operation and maintenance management methods usually rely on manual experience or querying operation and maintenance manuals to solve problems. However, with the accumulation of operation and maintenance knowledge, manually searching for relevant documents and troubleshooting solutions becomes increasingly inefficient, especially when the system complexity and the amount of knowledge base data increase dramatically.

[0063] To solve the above problems, operation and maintenance knowledge base systems have been gradually introduced, providing structured and standardized operation and maintenance knowledge. Through intelligent knowledge retrieval in the operation and maintenance knowledge base, fault solutions can be quickly found. At present, in the field of intelligent retrieval of operation and maintenance knowledge base, there are several retrieval methods: (1) Keyword-based search engine. Keyword matching mechanism: This type of technology relies on the keywords entered by the user for retrieval. The system will return relevant documents or answers based on factors such as the frequency and position of the keywords. Advantages: These systems are relatively simple to build and use the traditional inverted index structure for keyword retrieval, which can provide faster retrieval speed on small and medium-sized data sets. Disadvantages: Keyword matching can only recognize precise words and cannot understand the subtle differences in semantics. When users ask questions using different words or expressions, the system may not be able to return accurate answers. In addition, as the data set increases, the retrieval performance of the system will significantly decrease. (2) Question and answer system based on template matching. Technical principle: The system uses predefined question and answer templates. When the user's question meets a certain template, the system directly returns the corresponding answer. Advantages: Template matching systems are very effective in dealing with some common and fixed questions and can provide instant answers. Disadvantages: The system is not flexible and can only handle a limited range of questions. Once the user's question format changes slightly (for example, using different vocabulary or grammatical structures), the system cannot match the correct answer. At the same time, this technology is difficult to handle complex and diverse operation and maintenance scenarios.

[0064] Based on the above two retrieval methods, the main problems existing in traditional technologies are summarized as follows: (1) Insufficient semantic understanding and low retrieval accuracy. Most existing retrieval systems, such as search engines based on keyword matching, can only match the input text at the character level, ignoring the semantic relationship between questions and answers. When the user's question does not completely match the keywords in the knowledge base or there are differences in expression, the system often cannot provide relevant answers, resulting in low retrieval accuracy. Reason: This is because the traditional system uses an inverted index method, which can only match the frequency and position of words, but cannot truly understand the semantics of natural language. This method is highly dependent on keywords, especially when dealing with long-tail problems or non-standardized expressions. (2) Slow retrieval speed when processing massive data. With the continuous expansion of the operation and maintenance knowledge base, the amount of data may become very large. When processing large amounts of data, the query response time of traditional keyword retrieval systems increases significantly, especially when the data needs to be frequently updated or expanded, the system performance bottleneck is particularly obvious. Reason: Although the traditional inverted index structure can provide faster query speeds on small and medium-sized data sets, the retrieval efficiency will drop significantly on large-scale data sets. Each time a search is performed, the system needs to scan the entire data set, which increases the query time. At the same time, the construction and update of the inverted index also consumes a lot of computing resources and cannot quickly adapt to changes in new data. (3) Lack of flexible handling of natural language questions. Existing systems usually require users to ask clear keywords or follow templates. If the user asks slightly differently, such as using synonyms, different word orders, or more complex language structures, the system may not be able to recognize or return accurate answers. Reason: This is because the existing template matching or keyword retrieval technology cannot handle the complex structure and semantic changes in natural language, and the system does not provide sufficient support for language flexibility and expression diversity. (4) Unable to quickly adapt to the dynamic changes of the knowledge base. The information in the operation and maintenance knowledge base needs to be updated frequently, especially in areas where technology is changing rapidly, and new operation and maintenance solutions, tools, and technologies are constantly emerging. It is difficult for traditional systems to integrate new information into the existing knowledge base in a timely manner, resulting in insufficient real-time performance of the knowledge base and affecting the user experience. Reason: The update mechanism of traditional systems often relies on manual input or periodic updates, and cannot achieve dynamic and real-time data updates. This makes it impossible for the system to provide updated solutions in a timely manner when facing the latest operation and maintenance problems. (5) Limited ability to handle long-tail and complex problems. In operation and maintenance work, in addition to common problems, there are also a large number of personalized or rare "long-tail problems", which usually have no fixed solutions or few data samples. Traditional systems cannot effectively handle these long-tail problems, making it difficult for operation and maintenance personnel to find suitable solutions. Traditional keyword matching systems rely on high-frequency problems in existing data. For long-tail problems, the system cannot provide effective retrieval support because data samples are scarce or non-existent.In addition, most existing question-and-answer systems provide simple answers to single questions and are unable to cope with multi-level, multi-step solutions to complex problems. (6) Insufficient human-computer collaboration and lack of interactivity. In complex operation and maintenance scenarios, a single question and answer cannot cover all situations, and operation and maintenance personnel need to confirm the background or details of the problem through multiple rounds of interaction. However, traditional systems are usually in a single question-and-answer mode and lack the ability to interact with users in multiple rounds and provide continuous guidance. Reason: Most existing retrieval systems return answers in one go and lack the design of multiple rounds of dialogue or guided question-and-answer, resulting in the system being unable to provide continuous support or advice when dealing with complex problems.

[0065] Based on this, the present application provides a method for generating an operation and maintenance solution. After the user raises an operation and maintenance problem, the operation and maintenance knowledge base system first converts the operation and maintenance problem into a semantic vector, and searches the operation and maintenance knowledge base based on the semantic information of the operation and maintenance problem, so that the system can accurately understand the true meaning of the user's problem, perform semantic-level matching, and output the answer that is most relevant to the operation and maintenance problem, thereby improving the accuracy of the operation and maintenance solution; at the same time, the user refines the problem based on the operation and maintenance solution of the previous round, conducts multiple rounds of dialogues with the operation and maintenance knowledge base system, iteratively updates the operation and maintenance solution, and the operation and maintenance knowledge base system outputs a more accurate and customized solution based on the answers and refined problem information of the previous round, thereby further improving the accuracy of the operation and maintenance solution.

[0066] The operation and maintenance solution generation method provided in the embodiment of the present application can be applied to Figure 1 In the application environment shown. Among them, the terminal 102 communicates with the server 104 through the network. The data storage system can store the data that the server 104 needs to process. The data storage system can be integrated on the server 104, or it can be placed on the cloud or other network servers. Optionally, a dialogue robot can be deployed in the server 104, and the user can enter the operation and maintenance problem in the terminal 102, and the server 104 will process the operation and maintenance problem and output the operation and maintenance solution. In addition, the user can also interact with the dialogue robot through the terminal in the form of voice. Among them, the terminal 102 can be but not limited to various personal computers, laptops, smart phones, tablets, microphones, speakers, Internet of Things devices and portable wearable devices. The Internet of Things devices can be smart speakers, smart TVs, smart air conditioners, smart car-mounted devices, etc. Portable wearable devices can be smart watches, smart bracelets, head-mounted devices, etc. The server 104 can be implemented with an independent server or a server cluster consisting of multiple servers.

[0067] The operation and maintenance solution generation method provided in the embodiment of the present application can also be applied to a computer device. The computer device can be an operation and maintenance knowledge base system, and its internal structure diagram can be as follows: Figure 2As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. The processor, memory and input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The database of the computer device is used to store data. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a method for generating an operation and maintenance solution is implemented. Those skilled in the art can understand that Figure 2 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0068] In an exemplary embodiment, Figure 3 As shown, a method for generating an operation and maintenance solution is provided, and the method is applied to an operation and maintenance knowledge base system as an example for description, including the following steps 201 to 203. Among them:

[0069] S201, in response to the user's operation and maintenance problem, searching in the operation and maintenance knowledge base according to the semantic vector of the operation and maintenance problem, and outputting a preliminary operation and maintenance solution.

[0070] Operation and maintenance questions refer to questions about operation and maintenance work asked by users in the operation and maintenance knowledge base system, including hardware failures, network failures, software configuration errors, network security vulnerabilities, software upgrade issues, performance issues, server management, etc. Among them, operation and maintenance questions can be entered by users in the interface provided by the operation and maintenance knowledge base system, and the input method can be manual input or voice input.

[0071] When the operation and maintenance knowledge base system receives an operation and maintenance question from a user, it needs to search the operation and maintenance knowledge base to find the answer to the operation and maintenance question from the operation and maintenance knowledge base. In the embodiment of the present application, when searching the operation and maintenance knowledge base, the semantic vector of the operation and maintenance question is used for searching, and the operation and maintenance question needs to be converted into a semantic vector to extract the semantic information of the operation and maintenance question.

[0072] In one embodiment, the semantic vector of the operation and maintenance problem can be obtained by inputting the operation and maintenance problem into a pre-trained language model, and the pre-trained language model processes the operation and maintenance problem to obtain the semantic vector of the operation and maintenance problem. In another embodiment, the semantic vector of the operation and maintenance problem can be obtained by processing the operation and maintenance problem using a bag-of-words model or a word embedding model to convert the operation and maintenance problem into a semantic vector.

[0073] The operation and maintenance knowledge base integrates a large amount of historical operation and maintenance issues, equipment failure records, operation and maintenance manuals, technical manuals, etc. The operation and maintenance knowledge base also includes vector data of this information, that is, the information is vectorized and the vectorized information is also stored in the operation and maintenance knowledge base.

[0074] When searching in the operation and maintenance knowledge base, a similarity comparison can be performed between the semantic vector of the operation and maintenance problem and the vector data in the operation and maintenance knowledge base, and the operation and maintenance knowledge corresponding to the vector data with the highest similarity to the semantic vector of the operation and maintenance problem is determined as a preliminary operation and maintenance solution, and the preliminary operation and maintenance solution is output.

[0075] S202, updating the semantic vector according to the detailed problem information and the preliminary operation and maintenance solution input by the user based on the preliminary operation and maintenance solution, and searching in the operation and maintenance knowledge base according to the updated semantic vector, and outputting the updated operation and maintenance solution.

[0076] In an embodiment of the present application, after receiving the preliminary operation and maintenance solution, the user will continue to ask questions based on the preliminary operation and maintenance solution and input detailed problem information to further clarify the background, details and requirements of the problem. The operation and maintenance knowledge base system will record the questions and answers answered last time. When receiving the detailed problem information input by the user based on the preliminary operation and maintenance solution, it will generate a new semantic vector in combination with the preliminary operation and maintenance solution, and search again in the operation and maintenance knowledge base based on the updated semantic vector to obtain the updated operation and maintenance solution, and output the updated operation and maintenance solution. Among them, when updating the semantic vector, the refined problem information and the preliminary operation and maintenance solution can be vectorized respectively to obtain the semantic vector of the refined problem information and the semantic vector of the preliminary operation and maintenance solution, and the two semantic vectors are added to obtain the updated semantic vector. The method of searching in the operation and maintenance knowledge base according to the updated semantic vector is the same as the retrieval method in the previous step, which will not be repeated here.

[0077] S203, iteratively executing the updating process of the operation and maintenance solution until the preset conditions are met and the operation and maintenance solution to the operation and maintenance problem is obtained.

[0078] After outputting the updated operation and maintenance solution, the user continues to have multiple rounds of conversations, continues to ask questions based on the updated operation and maintenance solution, and inputs updated detailed problem information. After receiving the updated detailed problem information, the operation and maintenance knowledge base system updates the semantic vector and updates the operation and maintenance solution based on the updated semantic vector until the user no longer continues the conversation after receiving the updated operation and maintenance solution. The iteration is then stopped to obtain the final operation and maintenance solution.

[0079] In the operation and maintenance solution generation method provided by the embodiment of the present application, in response to the user's operation and maintenance problem, the operation and maintenance knowledge base is searched according to the semantic vector of the operation and maintenance problem, and a preliminary operation and maintenance solution is output, and then the semantic vector is updated according to the refined problem information and the preliminary operation and maintenance solution input by the user based on the preliminary operation and maintenance solution, and according to the updated semantic vector, the operation and maintenance knowledge base is searched, and the updated operation and maintenance solution is output, and the update process of the operation and maintenance solution is iteratively executed until the preset conditions are met to obtain the operation and maintenance solution of the operation and maintenance problem. In this method, after the user raises the operation and maintenance problem, the operation and maintenance knowledge base system first converts the operation and maintenance problem into a semantic vector, and searches the operation and maintenance knowledge base based on the semantic information of the operation and maintenance problem, so that the system can accurately understand the true meaning of the user's problem, match the semantic level, and output the answer most relevant to the operation and maintenance problem, thereby improving the accuracy of the operation and maintenance solution; at the same time, the user refines the problem based on the operation and maintenance solution of the previous round, conducts multiple rounds of dialogue with the operation and maintenance knowledge base system, iteratively updates the operation and maintenance solution, and the operation and maintenance knowledge base system outputs a more accurate and customized solution based on the answer and refined problem information of the previous round, further improving the accuracy of the operation and maintenance solution.

[0080] Based on the above embodiment, an embodiment is provided to illustrate the process of determining the semantic vector of the operation and maintenance problem.

[0081] In an exemplary embodiment, Figure 4 As shown, before searching the operation and maintenance knowledge base according to the semantic vector of the operation and maintenance problem and outputting a preliminary operation and maintenance solution, the method further includes:

[0082] S301, performing text segmentation on the operation and maintenance problem to obtain multiple short sentence problem information.

[0083] In the embodiment of the present application, after receiving the user's operation and maintenance questions, the operation and maintenance knowledge base system performs text segmentation on the operation and maintenance questions to break down long questions or complex sentences into multiple short sentences, perform semantic optimization, and obtain multiple short sentence question information. Among them, the method of performing text segmentation on the operation and maintenance questions can be to use a recursive text segmenter to perform text segmentation on the operation and maintenance questions.

[0084] S302, performing question encoding on each short sentence question information to obtain a question semantic vector of each short sentence question information.

[0085] After segmenting the operation and maintenance problems, the operation and maintenance knowledge base system encodes the obtained multiple short sentence problem information to convert the operation and maintenance problems into semantic vectors.

[0086] In an embodiment of the present application, each short sentence question information can be input into a pre-trained large model, and the large model will analyze and process each short sentence question information to convert the natural language question into a semantic vector to obtain a question semantic vector for each short sentence question information.

[0087] S303: Determine the problem semantic vector of each short sentence problem information as the semantic vector of the operation and maintenance problem.

[0088] After encoding the short sentence question information, the question semantic vector of the short sentence question information is determined as the semantic vector of the operation and maintenance question, so as to be retrieved in the operation and maintenance knowledge base.

[0089] In the operation and maintenance solution generation method provided in the embodiment of the present application, the operation and maintenance problem is segmented into text to obtain multiple short sentence problem information, and then each short sentence problem information is encoded to obtain the problem semantic vector of each short sentence problem information, and the problem semantic vector of each short sentence problem information is determined as the semantic vector of the operation and maintenance problem. In this method, when converting the operation and maintenance problem into a semantic vector, the operation and maintenance problem is first segmented into text, and the long text or complex operation and maintenance problem is effectively split to reasonably adjust the hierarchical structure of the data to ensure that each part can maintain semantic integrity, so that the subsequent retrieval in the operation and maintenance knowledge base can improve the retrieval efficiency and accuracy.

[0090] Based on the above embodiment, an embodiment is provided to illustrate the process of outputting the preliminary operation and maintenance solution.

[0091] In an exemplary embodiment, Figure 5 As shown, the operation and maintenance knowledge base includes multiple operation and maintenance knowledge vector data; according to the semantic vector of the operation and maintenance problem, the operation and maintenance knowledge base is searched and a preliminary operation and maintenance solution is output, including:

[0092] S401, determining a similarity coefficient between each question semantic vector and each operation and maintenance knowledge vector data according to each question semantic vector and each operation and maintenance knowledge vector data.

[0093] In an embodiment of the present application, the operation and maintenance knowledge base includes multiple operation and maintenance knowledge vector data. When searching in the operation and maintenance knowledge base, the similarity between each question semantic vector and each operation and maintenance knowledge vector data is first calculated. For any question semantic vector, a distance-based similarity calculation method can be used to calculate the similarity between the question semantic vector and each operation and maintenance knowledge vector data. For example, Euclidean distance, Manhattan distance, Chebyshev distance, etc. can be used. In addition, the similarity calculation method can also use cosine similarity and Pearson correlation coefficient, etc.

[0094] S402: Determine matching vector data of each problem semantic vector according to a similarity coefficient between each problem semantic vector and each operation and maintenance knowledge vector data.

[0095] After calculating the similarity coefficient between each question semantic vector and each operation and maintenance knowledge vector data, for any question semantic vector, the matching vector data of the question semantic vector is determined according to the multiple similarity coefficients corresponding to the question semantic vector. For example, the operation and maintenance knowledge vector data corresponding to the maximum value among the multiple similarity coefficients corresponding to the question semantic vector is determined as the matching vector data corresponding to the question semantic vector.

[0096] S403: Output a preliminary operation and maintenance solution based on the matching vector data of each problem semantic vector.

[0097] Based on the matching vector data of each problem semantic vector determined above, the answer to the operation and maintenance problem, i.e., the preliminary operation and maintenance solution, is determined, and the preliminary operation and maintenance solution is output. One possible implementation method is to determine the operation and maintenance knowledge corresponding to each matching vector based on the matching vector data of each problem semantic vector, determine the operation and maintenance knowledge as the preliminary operation and maintenance solution, and output the preliminary operation and maintenance solution.

[0098] In the operation and maintenance solution generation method provided in the embodiment of the present application, the similarity coefficient between each question semantic vector and each operation and maintenance knowledge vector data is determined based on each question semantic vector and each operation and maintenance knowledge vector data, and then the matching vector data of each question semantic vector is determined based on the similarity coefficient between each question semantic vector and each operation and maintenance knowledge vector data, and then the preliminary operation and maintenance solution is output based on the matching vector data of each question semantic vector. In this method, by calculating the similarity coefficient between the question semantic vector of each short sentence question information and multiple operation and maintenance knowledge vector data in the operation and maintenance knowledge base, the operation and maintenance knowledge matching each question semantic vector can be quickly retrieved in the operation and maintenance knowledge base, and the preliminary operation and maintenance solution is quickly determined and output based on the retrieved operation and maintenance knowledge, thereby improving the generation efficiency of the preliminary operation and maintenance solution.

[0099] Based on the above embodiment, another embodiment is provided to illustrate the process of outputting the preliminary operation and maintenance solution.

[0100] In an exemplary embodiment, Figure 6 As shown, according to the matching vector data of each problem semantic vector, a preliminary operation and maintenance solution is output, including:

[0101] S501, according to the matching vector data of each problem semantic vector, obtaining the operation and maintenance knowledge information corresponding to each matching vector data.

[0102] In the embodiment of the present application, after determining the matching vector data of each problem semantic vector, the operation and maintenance knowledge base system obtains the operation and maintenance knowledge information corresponding to each matching vector data in the operation and maintenance knowledge base, so as to output a preliminary operation and maintenance solution based on the operation and maintenance knowledge information. For example, a mapping relationship is established between each operation and maintenance knowledge information in the operation and maintenance knowledge base and the operation and maintenance knowledge vector data corresponding to each operation and maintenance knowledge information, and when obtaining the operation and maintenance knowledge information corresponding to each matching vector data, the operation and maintenance knowledge information is obtained based on the mapping relationship.

[0103] S502, processing each operation and maintenance knowledge information to generate an answer that meets the semantic information of the operation and maintenance problem.

[0104] After obtaining each piece of operation and maintenance knowledge information, each piece of operation and maintenance information is processed to obtain an answer that meets the semantic information of the operation and maintenance problem. In one achievable method, each piece of operation and maintenance knowledge information can be input into a pre-trained natural language processing large model, and the large model integrates and prunes the each piece of operation and maintenance knowledge information, so as to integrate the fragments of each piece of operation and maintenance knowledge information into a standardized answer that is complete in semantics and meets the current operation and maintenance problem. The integration process may include understanding the meaning of each piece of operation and maintenance knowledge information, finding the common themes and connection points of each piece of operation and maintenance knowledge information, determining the conjunctions or phrases between each piece of operation and maintenance knowledge information, adjusting the sentence structure of each piece of operation and maintenance knowledge information, merging each piece of operation and maintenance knowledge information, checking grammar and logic, deleting redundant parts, etc.

[0105] S503: Determine the answer that matches the semantic information of the operation and maintenance problem information as a preliminary operation and maintenance solution, and output the preliminary operation and maintenance solution.

[0106] After obtaining an answer that matches the semantic information of the operation and maintenance problem through processing, the answer that matches the semantic information of the operation and maintenance problem information is determined as a preliminary operation and maintenance solution, and the preliminary operation and maintenance solution is output.

[0107] In the operation and maintenance solution generation method provided in the embodiment of the present application, firstly, based on the matching vector data of each problem semantic vector, the operation and maintenance knowledge information corresponding to each matching vector data is obtained, and then each operation and maintenance knowledge information is processed to generate an answer that conforms to the semantic information of the operation and maintenance problem, and then the answer that conforms to the semantic information of the operation and maintenance problem information is determined as a preliminary operation and maintenance solution, and the preliminary operation and maintenance solution is output. In this method, firstly, based on the vector data that matches the semantic vector of each problem, the operation and maintenance knowledge information corresponding to each matching vector data is obtained from the operation and maintenance knowledge base, and each operation and maintenance knowledge information is integrated to process the fragmented operation and maintenance knowledge information into an answer that conforms to the semantic information of the current operation and maintenance problem, so that the generated preliminary operation and maintenance solution can be more consistent with the semantic information of the operation and maintenance problem, thereby improving the accuracy of the operation and maintenance solution.

[0108] Based on any of the above embodiments, an embodiment is provided to illustrate the process of updating the semantic vector.

[0109] In an exemplary embodiment, Figure 7 As shown, the semantic vector is updated according to the refined problem information and the preliminary operation and maintenance solution input by the user based on the preliminary operation and maintenance solution, including:

[0110] S601, vectorizing the preliminary operation and maintenance solution to obtain solution vector data of the preliminary operation and maintenance solution; and problem encoding the refined problem information to obtain refined problem vector data of the refined problem information.

[0111] In an embodiment of the present application, after the user inputs the refined problem information, the operation and maintenance knowledge base system searches the operation and maintenance knowledge base based on the operation and maintenance solutions output in the previous round and the currently input refined problem information to adjust and optimize the answers and ensure the accuracy of the solutions.

[0112] Based on this, before searching in the operation and maintenance knowledge base, it is necessary to process both the preliminary operation and maintenance solution and the refined problem information into vector information. The preliminary operation and maintenance solution is vectorized to convert the preliminary operation and maintenance solution into a vector to obtain the solution vector data of the preliminary operation and maintenance solution. Among them, the method of vectorizing the preliminary operation and maintenance solution can be to convert the preliminary operation and maintenance solution into vector data by sentence embedding. For example, the BERT model is trained in advance using historical operation and maintenance knowledge. After the training is completed, the preliminary operation and maintenance solution is input into the trained model for processing, and the model outputs the solution vector data of the preliminary operation and maintenance solution. In addition, in addition to the sentence embedding method, the vectorization processing can also be performed by word frequency-inverse document frequency, word embedding and other methods, which are not limited in the embodiments of the present application. The refined problem information is problem-encoded to convert the refined problem information into a semantic vector to obtain the refined problem vector data of the refined problem information. Among them, the method of problem encoding the refined problem information is the same as the aforementioned method of problem encoding the operation and maintenance problem, which will not be repeated here.

[0113] It should be noted that in order to improve retrieval efficiency and accuracy and ensure semantic integrity, when converting the preliminary operation and maintenance solution and refined problem information into vector data, the preliminary operation and maintenance solution and refined problem information can also be subjected to text segmentation, and then the information after text segmentation can be vectorized or problem encoded. In this way, it can be ensured that each part of the preliminary operation and maintenance solution and refined problem information can maintain semantic integrity, so as to improve efficiency and accuracy during retrieval.

[0114] S602: Determine an updated semantic vector according to the solution vector data and the refined problem vector data.

[0115] After obtaining the solution vector data and the refined problem vector data, the solution vector data and the refined problem vector data are concatenated to obtain an updated semantic vector.

[0116] In the operation and maintenance solution generation method provided in the embodiment of the present application, the preliminary operation and maintenance solution is vectorized to obtain solution vector data of the preliminary operation and maintenance solution; and the refined problem information is problem-encoded to obtain refined problem vector data of the refined problem information, and then the updated semantic vector is determined based on the solution vector data and the refined problem vector data. In this method, the preliminary operation and maintenance solution and the refined problem information are both converted into vector data by vectorizing the preliminary operation and maintenance solution and problem-encoding the refined problem information, and the updated semantic vector is obtained based on the two vector data, so that when the operation and maintenance knowledge base system answers the refined problem information, it searches in the operation and maintenance knowledge base in combination with the solution of the previous round, thereby improving the accuracy of answering the refined problem information, thereby improving the accuracy of updating the operation and maintenance solution.

[0117] Based on the above embodiment, an embodiment is provided to illustrate the process of outputting the updated operation and maintenance solution.

[0118] In an exemplary embodiment, Figure 8 As shown, according to the updated semantic vector, the operation and maintenance knowledge base is searched and the updated operation and maintenance solution is output, including:

[0119] S701, determining matching vector data of the updated semantic vector according to the updated semantic vector and the operation and maintenance knowledge vector data in the operation and maintenance knowledge base.

[0120] In the embodiment of the present application, after the semantic vector is updated, the updated semantic vector is similarly calculated with each operation and maintenance knowledge vector data in the operation and maintenance knowledge base to obtain the similarity coefficient between the updated semantic vector and each operation and maintenance knowledge vector data, and the operation and maintenance knowledge vector data corresponding to the maximum value of the similarity coefficient between the updated semantic vector and each operation and maintenance knowledge vector data is determined as the matching vector data of the updated semantic vector. Among them, the calculation method of the similarity coefficient is the same as the aforementioned method of determining the similarity coefficient between each problem semantic vector and each operation and maintenance knowledge vector data, which will not be repeated here.

[0121] S702, generating an answer that conforms to the semantic information of the refined question information according to the matching vector data of the updated semantic vector.

[0122] Exemplarily, according to the matching vector data of the updated semantic vector, the operation and maintenance knowledge information corresponding to the matching vector data is obtained, and based on the operation and maintenance knowledge information, an answer to the semantic information that conforms to the detailed question information is generated. If there are multiple matching vector data, there are also multiple operation and maintenance knowledge information, and it is also necessary to integrate and process the multiple operation and maintenance knowledge information to obtain an answer to the semantic information that conforms to the detailed question information. If there is only one matching vector data, the operation and maintenance knowledge information corresponding to the matching vector data can be determined as the answer to the semantic information that conforms to the detailed question information.

[0123] S703: Determine the answer that matches the semantic information of the refined question information as an updated operation and maintenance solution, and output the updated operation and maintenance solution.

[0124] After generating an answer that matches the semantic information of the refined question information, the answer that matches the semantic information of the refined question information is determined as an updated operation and maintenance solution, and the updated operation and maintenance solution is output.

[0125] In the operation and maintenance solution generation method provided by the embodiment of the present application, first, the matching vector data of the updated semantic vector is determined based on the updated semantic vector and the operation and maintenance knowledge vector data in the operation and maintenance knowledge base, and then the answer to the semantic information that conforms to the detailed problem information is generated based on the matching vector data of the updated semantic vector, and the answer to the semantic information that conforms to the detailed problem information is determined as the updated operation and maintenance solution, and the updated operation and maintenance solution is output. In this method, based on the updated semantic vector, the corresponding operation and maintenance knowledge vector data is matched in the operation and maintenance knowledge base, and the matching operation and maintenance knowledge is determined based on the matching operation and maintenance knowledge vector data, so that the answer to the semantic information that conforms to the detailed problem information is quickly generated based on the matching operation and maintenance knowledge, and the updated operation and maintenance solution is obtained, which improves the updating efficiency and accuracy of the operation and maintenance solution.

[0126] Based on the above embodiment, an embodiment is provided to illustrate the updating process of the operation and maintenance knowledge base.

[0127] In an exemplary embodiment, Fig. 9 As shown, the method also includes:

[0128] S801, verify the operation and maintenance solution.

[0129] In the embodiment of the present application, the operation and maintenance solution can be reviewed and verified by setting up a simulation environment without affecting the production environment, and conducting actual operation tests on the solution to observe whether the problem can be effectively solved. At the same time, the operation and maintenance solution can also be reviewed by experts. If the verification and review are passed, it means that the operation and maintenance solution is highly accurate and can be fully promoted for use.

[0130] S802: If the operation and maintenance solution is verified, the operation and maintenance solution and the problem information corresponding to the operation and maintenance solution are vectorized, and the vectorized operation and maintenance solution and problem information are stored in the operation and maintenance knowledge base to obtain an updated operation and maintenance knowledge base.

[0131] After the operation and maintenance solution is verified, the operation and maintenance solution and the problem information corresponding to the operation and maintenance solution are stored in the operation and maintenance knowledge base, and the operation and maintenance solution and the problem information corresponding to the operation and maintenance solution are vectorized, and the vectorized operation and maintenance solution and problem information are also stored in the operation and maintenance knowledge base to obtain an updated operation and maintenance knowledge base. In this way, the operation and maintenance knowledge base is dynamically expanded, and new operation and maintenance problems and operation and maintenance solutions are integrated in real time to ensure that new data can be retrieved immediately.

[0132] In the operation and maintenance solution generation method provided in the embodiment of the present application, the operation and maintenance solution is verified. If the operation and maintenance solution passes the verification, the operation and maintenance solution and the problem information corresponding to the operation and maintenance solution are vectorized, and the vectorized operation and maintenance solution and problem information are stored in the operation and maintenance knowledge base to obtain an updated operation and maintenance knowledge base. In this method, after the operation and maintenance knowledge base system outputs the final operation and maintenance solution, it verifies the operation and maintenance solution. If the verification passes, the operation and maintenance solution and the corresponding operation and maintenance information are stored in the operation and maintenance knowledge base to dynamically expand the operation and maintenance knowledge base, continuously integrate and update the latest operation and maintenance knowledge, maintain the system's adaptability to new problems and new technical solutions, and maintain efficient retrieval performance during the growth of the operation and maintenance knowledge base.

[0133] In addition, in an exemplary embodiment, the present application also provides a method for generating an operation and maintenance solution, such as Fig.10 As shown, the following steps may be included:

[0134] S901, in response to the user's operation and maintenance problem, perform text segmentation on the operation and maintenance problem to obtain a plurality of short sentence problem information.

[0135] S902, performing question encoding on each short sentence question information to obtain a question semantic vector of each short sentence question information.

[0136] S903: Determine the problem semantic vector of each short sentence problem information as the semantic vector of the operation and maintenance problem.

[0137] S904: Determine a similarity coefficient between each question semantic vector and each operation and maintenance knowledge vector data according to each question semantic vector and each operation and maintenance knowledge vector data.

[0138] S905: Determine matching vector data of each problem semantic vector according to the similarity coefficient between each problem semantic vector and each operation and maintenance knowledge vector data.

[0139] S906: Acquire operation and maintenance knowledge information corresponding to each matching vector data according to the matching vector data of each question semantic vector.

[0140] S907, processing each operation and maintenance knowledge information to generate an answer that meets the semantic information of the operation and maintenance problem.

[0141] S908: Determine the answer that matches the semantic information of the operation and maintenance problem information as a preliminary operation and maintenance solution, and output the preliminary operation and maintenance solution.

[0142] S909, vectorizing the preliminary operation and maintenance solution to obtain solution vector data of the preliminary operation and maintenance solution; and problem encoding the refined problem information to obtain refined problem vector data of the refined problem information.

[0143] S910, determining an updated semantic vector according to the solution vector data and the refined problem vector data, and searching in the operation and maintenance knowledge base according to the updated semantic vector, and outputting an updated operation and maintenance solution.

[0144] S911, iteratively execute the updating process of the operation and maintenance solution until the preset conditions are met and the operation and maintenance solution of the operation and maintenance problem is obtained.

[0145] The above-mentioned process of S901-S911 can refer to the description of the above-mentioned method embodiment, and its implementation principle and technical effect are similar, which will not be repeated here.

[0146] In addition, in one embodiment, the process from inputting a question to returning an answer in the operation and maintenance knowledge base system includes the following steps:

[0147] 1. User input questions: Users ask questions to the system using natural language.

[0148] 2. Recursive text segmentation: The system uses a recursive text segmenter to break long questions or complex sentences into multiple short sentences for semantic optimization.

[0149] 3. Question encoding: Through the pre-trained large model, the system converts natural language questions into semantic vectors to capture the semantic information behind the questions.

[0150] 4. Vectorized representation: The question is represented as a high-dimensional vector and its similarity is compared with the vector data in the knowledge base.

[0151] 5. Chroma vector index retrieval: Through vector retrieval technology, the system can quickly find historical questions and answers that are most similar to user questions in the huge operation and maintenance knowledge base.

[0152] 6. Generate question and answer results: Based on the matched results, the system generates preliminary answers.

[0153] 7. Multi-round dialogue: If the user's query is too complex, the system will conduct multiple rounds of dialogue with the user through the LangChain framework to further clarify the background and needs of the question.

[0154] 8. Return the final solution: The system combines the results of multiple rounds of dialogue to generate a standardized operation and maintenance solution.

[0155] This solution significantly improves the efficiency and accuracy of problem solving in operation and maintenance work by combining natural language processing, semantic matching and vectorized retrieval technology. The following are the main effects of this solution:

[0156] (1) Improved the accuracy of operation and maintenance problem retrieval

[0157] Traditional keyword-based search systems cannot accurately understand natural language questions raised by users, especially in the case of synonyms, word order changes and complex semantics, which are prone to mismatching. This solution introduces the semantic understanding ability of large models and combines vector search technology to accurately understand the true meaning of user questions and perform semantic matching.

[0158] Effect: No matter how the user expresses the operation and maintenance problem, the system can return the most relevant answer through semantic matching, avoiding retrieval failures caused by keyword mismatches in traditional systems.

[0159] (2) Significantly improve retrieval speed

[0160] Through Chroma vector retrieval technology, the system can perform efficient vectorized data retrieval in the huge operation and maintenance knowledge base, avoiding the problem of slow retrieval speed of traditional inverted index when facing large-scale data sets.

[0161] Effect: Even in a knowledge base with millions of operation and maintenance records, the system can still return accurate search results within seconds, greatly improving the work efficiency of operation and maintenance engineers.

[0162] (3) Enhanced dynamic scalability of the system

[0163] The system can continuously integrate and update the latest operation and maintenance knowledge base content to keep the system adaptable to new problems and new technical solutions. As new operation and maintenance solutions, troubleshooting cases and technical documents are added to the knowledge base, the system can index and retrieve them instantly.

[0164] Effect: The system can always maintain the latest operation and maintenance knowledge, help operation and maintenance engineers cope with the dynamically changing operation and maintenance environment, and provide real-time and effective solutions to new problems.

[0165] (4) Supporting multiple rounds of human-machine collaborative dialogue

[0166] By integrating the human-machine collaborative framework (LangChain), multiple rounds of dialogue can be conducted with users based on the initial answers to further clarify the background, details and requirements of the problem, and finally return a more accurate and customized solution. Especially for complex operation and maintenance problems, the system can work with operation and maintenance engineers to solve them.

[0167] Effect: The system can not only provide a single answer, but also help operation and maintenance personnel refine questions through multiple rounds of interaction in complex scenarios, thereby improving the accuracy and applicability of answers.

[0168] (5) Improved ability to handle long-tail problems

[0169] In operation and maintenance, in addition to common fault problems, there are also a large number of personalized and rare "long-tail problems". Traditional systems often find it difficult to effectively handle these problems, but this solution can effectively understand and retrieve long-tail problems through the deep learning capabilities of large models.

[0170] Effect: Even for questions with fewer samples in the knowledge base, the system can provide high-quality answers through the learning ability of the large model and semantic matching technology, helping operation and maintenance personnel to deal with complex or rare problems.

[0171] (6) Lowering the learning and usage threshold for operation and maintenance personnel

[0172] Traditional operation and maintenance systems usually require operation and maintenance personnel to use specific keywords or templated query methods, which have a high threshold for use. This solution supports natural language queries, and operation and maintenance personnel can ask questions in a more natural and flexible way, and the system can automatically understand and return the corresponding answers.

[0173] Effect: It reduces the time that operation and maintenance personnel spend on learning how to use the system, improves the usability of the system, reduces the technical requirements for operation and maintenance personnel, and improves overall work efficiency.

[0174] It should be understood that, although the various steps in the flowcharts involved in the above-mentioned embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-mentioned embodiments can include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.

[0175] Based on the same inventive concept, the embodiment of the present application also provides an operation and maintenance solution generation device for implementing the operation and maintenance solution generation method involved above. The implementation solution for solving the problem provided by the device is similar to the implementation solution recorded in the above method, so the specific limitations in one or more operation and maintenance solution generation device embodiments provided below can refer to the limitations of the operation and maintenance solution generation method above, and will not be repeated here.

[0176] In an exemplary embodiment, Fig.11 As shown, an operation and maintenance solution generation device 1 is provided, comprising: a preliminary solution generation module 10, a solution update module 20 and a solution determination module, wherein:

[0177] A preliminary solution generation module 10 is used to respond to the user's operation and maintenance problem, search the operation and maintenance knowledge base according to the semantic vector of the operation and maintenance problem, and output a preliminary operation and maintenance solution;

[0178] A solution updating module 20 is used to update the semantic vector according to the detailed problem information and the preliminary operation and maintenance solution input by the user based on the preliminary operation and maintenance solution, and to search in the operation and maintenance knowledge base according to the updated semantic vector, and output the updated operation and maintenance solution;

[0179] The solution determination module 30 is used to iteratively execute the updating process of the operation and maintenance solution until a preset condition is met and an operation and maintenance solution to the operation and maintenance problem is obtained.

[0180] In one embodiment, the operation and maintenance solution generating device 1 further includes:

[0181] The text segmentation module is used to segment the operation and maintenance problems into multiple short sentences.

[0182] A question encoding module is used to encode each short sentence question information to obtain a question semantic vector of each short sentence question information;

[0183] The vector determination module is used to determine the problem semantic vector of each short sentence problem information as the semantic vector of the operation and maintenance problem.

[0184] In one embodiment, the preliminary solution generation module 10 is further used for:

[0185] Based on the semantic vectors of each problem and the operation and maintenance knowledge vector data, determine the similarity coefficient between each problem semantic vector and each operation and maintenance knowledge vector data; based on the similarity coefficient between each problem semantic vector and each operation and maintenance knowledge vector data, determine the matching vector data of each problem semantic vector; based on the matching vector data of each problem semantic vector, output a preliminary operation and maintenance solution.

[0186] In one embodiment, the preliminary solution generation module 10 is further used for:

[0187] According to the matching vector data of the semantic vector of each question, the operation and maintenance knowledge information corresponding to each matching vector data is obtained; each operation and maintenance knowledge information is processed to generate an answer that conforms to the semantic information of the operation and maintenance question; the answer that conforms to the semantic information of the operation and maintenance question information is determined as a preliminary operation and maintenance solution, and the preliminary operation and maintenance solution is output.

[0188] In one embodiment, the scheme updating module 20 is further used for:

[0189] The preliminary operation and maintenance solution is vectorized to obtain solution vector data of the preliminary operation and maintenance solution; and the refined problem information is problem-encoded to obtain refined problem vector data of the refined problem information; and an updated semantic vector is determined based on the solution vector data and the refined problem vector data.

[0190] In one embodiment, the scheme updating module 20 is further used for:

[0191] According to the updated semantic vector and the operation and maintenance knowledge vector data in the operation and maintenance knowledge base, the matching vector data of the updated semantic vector is determined; according to the matching vector data of the updated semantic vector, an answer that conforms to the semantic information of the refined question information is generated; the answer that conforms to the semantic information of the refined question information is determined as an updated operation and maintenance solution, and the updated operation and maintenance solution is output.

[0192] In one embodiment, the operation and maintenance solution generating device 1 further includes:

[0193] Solution verification module, used to verify the operation and maintenance solution;

[0194] The knowledge base updating module is used to vectorize the operation and maintenance solution and the problem information corresponding to the operation and maintenance solution if the operation and maintenance solution is verified, and store the vectorized operation and maintenance solution and problem information in the operation and maintenance knowledge base to obtain an updated operation and maintenance knowledge base.

[0195] Each module in the above-mentioned operation and maintenance solution generation device can be implemented in whole or in part by software, hardware, or a combination thereof. Each of the above-mentioned modules can be embedded in or independent of a processor in a computer device in the form of hardware, or can be stored in a memory in a computer device in the form of software, so that the processor can call and execute the operations corresponding to each of the above modules.

[0196] In an exemplary embodiment, the embodiment of the present application provides an intelligent retrieval system for operation and maintenance knowledge base based on a large model, which aims to help operation and maintenance engineers quickly obtain accurate operation and maintenance solutions through semantic matching, vector retrieval and natural language processing technology. The overall architecture of the system is divided into several core modules, including a data integration module, a natural language processing module, a semantic understanding and vectorization module, a Chroma vector retrieval module, a question and answer generation module, and a human-computer collaboration module. Among them, the data integration module converts the original operation and maintenance documents, manuals, problem records and other data into a structured knowledge base through recursive text segmentation technology. The natural language processing module parses the natural language questions input by the user and understands the semantics of the user's questions through a large model. The semantic understanding and vectorization module converts natural language into vector representation based on the Encoder-Decoder architecture and performs vector retrieval. The Chroma vector retrieval module quickly finds the most relevant answers in a large data set through efficient vector retrieval technology. The question and answer generation module generates standardized answers that meet the current questions based on the retrieved historical questions and answers. The human-computer collaboration module interacts with the user in multiple rounds through the LangChain framework to further improve the generation of answers. In addition, the timing relationship diagram of each module when processing the problem is as follows Fig.12 As shown. Natural language processing module: receives natural language questions input by users and performs semantic understanding. Chroma vector retrieval module: vectorizes questions and performs similarity retrieval in the vector index database. Question and answer generation module: generates preliminary answers based on the matched historical answers. LangChain multi-round dialogue module: conducts multiple rounds of interactions with users, adjusts and optimizes answers, and ensures the accuracy of solutions.

[0197] The above modules can be implemented in whole or in part by software, hardware or a combination thereof. The above modules can be embedded in or independent of a processor in a computer device in the form of hardware, or stored in a memory in a computer device in the form of software, so that the processor can call and execute operations corresponding to the above modules.

[0198] In an exemplary embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the computer program, the following steps are implemented:

[0199] In response to the user's operation and maintenance problem, the operation and maintenance knowledge base is searched according to the semantic vector of the operation and maintenance problem, and a preliminary operation and maintenance solution is output;

[0200] The semantic vector is updated according to the refined problem information and the preliminary operation and maintenance solution input by the user based on the preliminary operation and maintenance solution, and according to the updated semantic vector, the operation and maintenance knowledge base is searched to output the updated operation and maintenance solution;

[0201] Iterate the update process of the operation and maintenance solution until the preset conditions are met and the operation and maintenance solution to the operation and maintenance problem is obtained.

[0202] The implementation principles and technical effects of each step implemented by the processor in the embodiment of the present application are similar to the principles of the above-mentioned operation and maintenance solution generation method, and will not be repeated here.

[0203] In one embodiment, a computer readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, the following steps are implemented:

[0204] In response to the user's operation and maintenance problem, the operation and maintenance knowledge base is searched according to the semantic vector of the operation and maintenance problem, and a preliminary operation and maintenance solution is output;

[0205] The semantic vector is updated according to the refined problem information and the preliminary operation and maintenance solution input by the user based on the preliminary operation and maintenance solution, and according to the updated semantic vector, the operation and maintenance knowledge base is searched to output the updated operation and maintenance solution;

[0206] Iterate the update process of the operation and maintenance solution until the preset conditions are met and the operation and maintenance solution to the operation and maintenance problem is obtained.

[0207] The implementation principles and technical effects of the various steps implemented when the computer program in the embodiment of the present application is executed by the processor are similar to the principles of the above-mentioned operation and maintenance solution generation method, and will not be repeated here.

[0208] In one embodiment, a computer program product is provided, comprising a computer program, which, when executed by a processor, implements the following steps:

[0209] In response to the user's operation and maintenance problem, the operation and maintenance knowledge base is searched according to the semantic vector of the operation and maintenance problem, and a preliminary operation and maintenance solution is output;

[0210] The semantic vector is updated according to the refined problem information and the preliminary operation and maintenance solution input by the user based on the preliminary operation and maintenance solution, and according to the updated semantic vector, the operation and maintenance knowledge base is searched to output the updated operation and maintenance solution;

[0211] Iterate the update process of the operation and maintenance solution until the preset conditions are met and the operation and maintenance solution to the operation and maintenance problem is obtained.

[0212] The implementation principles and technical effects of the various steps implemented when the computer program in the embodiment of the present application is executed by the processor are similar to the principles of the above-mentioned operation and maintenance solution generation method, and will not be repeated here.

[0213] It should be noted that the personnel information (including but not limited to personnel equipment information, personnel personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the personnel or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.

[0214] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but are not limited to this.

[0215] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0216] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. A method for generating an operation and maintenance solution, characterized in that: The method comprises: In response to the user's operation and maintenance problem, search the operation and maintenance knowledge base according to the semantic vector of the operation and maintenance problem, and output a preliminary operation and maintenance solution; updating the semantic vector according to the refined problem information input by the user based on the preliminary operation and maintenance solution and the preliminary operation and maintenance solution, and searching in the operation and maintenance knowledge base according to the updated semantic vector, and outputting an updated operation and maintenance solution; The updating process of the operation and maintenance solution is iteratively executed until the preset conditions are met and the operation and maintenance solution of the operation and maintenance problem is obtained.

2. The method according to claim 1, characterized in that Before searching the operation and maintenance knowledge base according to the semantic vector of the operation and maintenance problem and outputting a preliminary operation and maintenance solution, the method further includes: Perform text segmentation on the operation and maintenance problem to obtain multiple short sentence problem information; Performing question encoding on each of the short sentence question information to obtain a question semantic vector for each of the short sentence question information; The problem semantic vector of each of the short sentence problem information is determined as the semantic vector of the operation and maintenance problem.

3. The method according to claim 2, characterized in that The operation and maintenance knowledge base includes a plurality of operation and maintenance knowledge vector data; the operation and maintenance knowledge base is searched according to the semantic vector of the operation and maintenance problem, and a preliminary operation and maintenance solution is output, including: Determine, according to each of the problem semantic vectors and each of the operation and maintenance knowledge vector data, a similarity coefficient between each of the problem semantic vectors and each of the operation and maintenance knowledge vector data; Determining matching vector data of each of the problem semantic vectors according to a similarity coefficient between each of the problem semantic vectors and each of the operation and maintenance knowledge vector data; The preliminary operation and maintenance solution is output based on the matching vector data of each of the problem semantic vectors.

4. The method according to claim 3, characterized in that Outputting the preliminary operation and maintenance solution according to the matching vector data of each of the problem semantic vectors includes: According to the matching vector data of each of the problem semantic vectors, obtaining the operation and maintenance knowledge information corresponding to each of the matching vector data; Processing each of the operation and maintenance knowledge information to generate an answer that conforms to the semantic information of the operation and maintenance problem; The answer to the semantic information that matches the operation and maintenance problem information is determined as the preliminary operation and maintenance solution, and the preliminary operation and maintenance solution is output.

5. The method according to any one of claims 1 to 4, characterized in that: The updating of the semantic vector according to the refined problem information input by the user based on the preliminary operation and maintenance solution and the preliminary operation and maintenance solution includes: Performing vectorization processing on the preliminary operation and maintenance solution to obtain solution vector data of the preliminary operation and maintenance solution; and performing problem encoding on the refined problem information to obtain refined problem vector data of the refined problem information; An updated semantic vector is determined according to the solution vector data and the refined problem vector data.

6. The method according to any one of claims 1 to 4, characterized in that: The step of searching the operation and maintenance knowledge base according to the updated semantic vector and outputting the updated operation and maintenance solution includes: Determining matching vector data of the updated semantic vector according to the updated semantic vector and the operation and maintenance knowledge vector data in the operation and maintenance knowledge base; Generating an answer that conforms to the semantic information of the refined question information according to the matching vector data of the updated semantic vector; The answer to the semantic information that meets the refined question information is determined as the updated operation and maintenance solution, and the updated operation and maintenance solution is output.

7. The method according to any one of claims 1 to 4, characterized in that: The method further comprises: Verify the operation and maintenance solution; If the operation and maintenance solution is verified, the operation and maintenance solution and the problem information corresponding to the operation and maintenance solution are vectorized, and the vectorized operation and maintenance solution and problem information are stored in the operation and maintenance knowledge base to obtain an updated operation and maintenance knowledge base.

8. A device for generating an operation and maintenance solution, characterized in that: The device comprises: A preliminary solution generation module is used to respond to the user's operation and maintenance problem, search the operation and maintenance knowledge base according to the semantic vector of the operation and maintenance problem, and output a preliminary operation and maintenance solution; A solution updating module, configured to update a semantic vector according to the refined problem information input by the user based on the preliminary operation and maintenance solution and the preliminary operation and maintenance solution, and to search in an operation and maintenance knowledge base according to the updated semantic vector, and output an updated operation and maintenance solution; The solution determination module is used to iteratively execute the updating process of the operation and maintenance solution until the preset conditions are met and the operation and maintenance solution of the operation and maintenance problem is obtained.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images